key: cord-022252-9yiuuye3
authors: Mims, Cedric A.; Dimmock, Nigel J.; Nash, Anthony; Stephen, John
title: Mechanisms of Cell and Tissue Damage
date: 2013-11-17
journal: Mims' Pathogenesis of Infectious Disease
DOI: 10.1016/b978-0-12-498262-8.50015-1
sha: 
doc_id: 22252
cord_uid: 9yiuuye3

nan

The impact on the host of microbial damage depends very much on the tissue involved. Damage to muscle in the shoulder or stomach wall, for instance, may not be serious, but in the heart the very existence of the host depends on a strong muscle contraction continuing to occur every second or so, and here the effect of minor functional changes may be catastrophic. The central nervous system is particularly vulnerable to slight damage. The passage of nerve impulses requires normal function in the neuronal cell membrane, and viruses especially have important effects on cell membranes. Also a degree of cellular or tissue oedema that is tolerable in most tissues may have serious consequences if it occurs in the brain, enclosed in that more or less rigid box, the skull. Therefore, encephalitis and meningitis tend to cause more severe illness than might be expected from the histological changes themselves. Oedema is a serious matter also in the lung. Oedema fluid or inflammatory cell exudates appear first in the space between the alveolar capillary and the alveolar wall, decreasing the efficiency of gaseous exchanges. Respiratory function is more drastically impaired when fluid or cells accumulate in the alveolar air space.* The effect of tissue damage is much less in the case of organs such as the liver, pancreas or kidney, which have considerable functional reserves. More than two-thirds of the liver must be removed before there are signs of liver dysfunction. Cell damage has profound effects if it is the endothelial cells of small blood vessels that are involved. The resulting circulatory changes may lead to anoxia or necrosis in the tissues supplied by these vessels. Here too, the site of vascular lesions may be critical, effects on organs such as the brain or heart having a greater impact on the host, as discussed above. Rickettsiae characteristically grow in vascular endothelium, and this is an important mechanism of disease production. By a combination of direct and immunopathological factors there is endothelial swelling, thrombosis, infarcts, haemorrhage and tissue anoxia. This is especially notable in the skin, and forms the basis for the striking rashes in typhus and the spotted fevers. These skin rashes, although important for the physician are less important for the patient than similar lesions in the central nervous system or heart. It is damage to cerebral vessels that accounts for the cerebral disturbances in typhus; involvement of pulmonary vessels causes pneumonitis, and involvement of myocardial vessels causes myocardial oedema. In Q fever, rickettsiae sometimes localise in the endocardium, and this causes serious complications.

Sometimes an infectious agent damages an organ, and loss of function in this organ leads to a series of secondary disease features. The signs of liver dysfunction are an accepted result of infections of the liver, just as paralysis or coma is an accepted result of infection of the central nervous system. Diabetes may turn out to be caused by infection of the islets of Langerhans in the pancreas. Coxsackie and other virus infections of the islets of Langerhans can certainly cause diabetes in experimental animals, and coxsackieviruses have been associated with juvenile diabetes in man.

There are many diseases of unknown aetiology for which an infectious origin has been suggested. Sometimes it is fairly well established that an infectious agent can at least be one of the causes of the disease, but in most instances it is no more than a hypothesis, with little or no good evidence. For conditions as common and as serious as multiple sclerosis, cancer and rheumatoid arthritis it would be of immense importance if a microorganism were incriminated, since this would give the opportunity to prevent the disease by vaccination. Accordingly, there is a temptation to accept or publicise new reports even though the evidence is weak or the observations poorly controlled. As if to warn us about this and remind us of possibilities from environmental toxins, Parkinson's disease, a chronic neurological condition in which there is a loss of neurons in a sharply defined region of the brain (substantia nigra), can be caused by exposure to the chemical MPTP. One example which raises the possibility that subtle CNS disturbances may be caused by viruses is experimental infection with Borna disease virus. This virus was used to infect tree shrews (Tupaia glis) which are primitive primates. There is little overt disease, but afterwards the male is no longer able to enact the ritual courtship behaviour which (as students well know), is an essential preliminary to mating in all primates, and the frustrated male usually ends up bitten by the female. Thus it can be said that infection with Borna disease virus renders the male psychologically sterile. Presumably the virus in some way alters the functioning of neurons concerned in this particular pathway. All other behavioural and physiological aspects appear normal. Borna disease virus is not known to occur in man, but speculation about an analogous human situation is fuelled by the finding of Borna disease virus-specific antibodies in patients with psychiatric/ behavioural disorders. In an entirely different clinical context, infection of a particular strain of rats with Borna disease virus causes immense obesity, the underlying physiological basis of which is not understood. Since the aetiology of such diseases raises interesting problems in pathogenesis, the present state of affairs is summarised in Table 8 .1, which includes some of the human diseases whose infectious origin is probable, possible, conceivable, or inconceivable.

Causal connections between infection and disease states are particularly difficult to establish when the disease appears a long time after infection. It was not too difficult to prove and accept that the encephalitis that occasionally occurs during or immediately after measles was due to measles virus. But it was hard to accept that a very rare type of encephalitis (subacute sclerosing panencephalitis or SSPE), occurring up to 10 years after apparently complete recovery from measles, was also due to measles virus and this was only established after careful studies and the eventual difficult isolation of a mutant form of measles virus from brain cells. 'Slow' infections, in which the first signs of disease appear a long time after infection are now an accepted part of our outlook. The disease kuru occurred in New Guinea and was transmitted from person to person by cannibalism. The incubation period of the disease in man appears to be 12-15 years, and was caused by a nonisolatable infectious agent that grew in the brain. This was established when the same disease appeared in monkeys several years after the injection of material from the brain of Kuru patients. A similar agent called scrapie (see Ch. 7) infects sheep, mice and other animals and also has an incubation period representing a large portion of the life span of the host. In both Kuru and SSPE the agent was eventually shown to be present in the brains of patients. So far this has only been demonstrated indirectly as, despite strenuous efforts, the causative agent has yet to be isolated. If in a slow infection the microorganism that initiated the pathological process is no longer present by the time the disease becomes manifest, then the problem of establishing a causal relationship will be much greater. This may possibly turn out to be true for diseases like multiple sclerosis and rheumatoid arthritis. Liver cancer in humans and leukaemia in mice, cats, humans and cattle can be caused by slow type virus infections. Cancer or leukaemia appears as a late and occasional sequel to infection. The virus, its antigens or fragments of its nucleic acid are detectable in malignant cells.

One important factor that often controls the speed of an infectious process and the type of host response, is the rate of multiplication of a microorganism.* Different infectious agents show doubling times * Every infection is a RACE between the spread and multiplication of the microbe and the generation of an antimicrobial response by the host. A day or two's delay in this response may let the microbe reach the critical levels of growth that give tissue damage and disease. varying from 20 min to 2 weeks, and some of these are listed in Table 8 .2.

Often the rate of multiplication in the infected host, in the presence of antimicrobial and other limiting factors and when many bacteria are obliged to multipy inside phagocytic cells, is much less than the optimal rate in artificial culture. Clearly a microorganism with a doubling time of a day or two will tend to cause a more slowly evolving infection and disease than one that doubles in an hour or less. It is uncommon for an infectious agent to cause exactly the same disease in all those infected. Its nature and severity will depend on infecting dose and route, and on the host's age, sex, nutritional status, genetic background, and so on. Many infections are asymptomatic in more than 90% of individuals, clinically characterised disease occurring in only an occasional unfortunate host, as 'the tip of the iceberg'.* Asymptomatically infected individuals are important because they are not identified, move normally in the community, and play an important part in transmission.

This chapter deals with demonstrable cell and tissue damage or dysfunction in infectious diseases. But one of the earliest indications of illness is malaise, 'not feeling very well'. This is distinct from fever or a specific complaint such as a sore throat and although it is difficult to define and impossible to measure, we all know the feeling. It can precede the onset of more specific signs and symptoms, or accompany them. Sometimes it is the only indication that an infection is taking place. Almost nothing is known of the basis for this feeling. Toxins', of course, have been invoked and the earliest response to pyrogens (see pp. 298-300) before body temperature has actually risen, may play a part. Interferons may have something to do with it because pure preparations of human a or ß Interferons cause malaise and often headaches and muscle aches after injection into normal individuals. If interferon is eventually recognised as an important antimicrobial force we may have to regard these side effects as unfortunate but acceptable. Soluble mediators of immune and inflammatory responses, such as interleukin-1 (see Glossary) or other cytokines may also play a part. In some infectious diseases weakness and debility are prominent during convalescence. This can be especially notable following influenza and hepatitis, but its basis is as mysterious as in the case of malaise.

The infections that matter are those causing pathological changes and disease. Before giving an account of the mechanisms by which these changes are produced, it is important to remember that many infectious agents cause little or no damage in the host. Indeed, it is of some advantage to the microorganism to cause minimal host damage, as discussed in Ch. 1. Virus infections as often as not fall into this category. Thus, although infection with rabies or measles viruses nearly always causes disease, there are many enterovirus, reovirus and myxovirus infections that are regularly asymptomatic. Even viruses that are named for their association with disease (poliomyelitis, influenza, Japanese B encephalitis) often give an antibody response as the only sign of infection in the host. Tissue damage is too slight to cause detectable illness. There is also a tendency for persistent viruses to cause no more than minor or delayed cellular damage during their persistence in the body, even if the same virus has a more cytopathic effect during an acute infection, e.g. adenoviruses, herpes simplex (see Ch. 10). A few viruses are remarkable because they cause no pathological changes at all in the cell, even during a productive infection in which infectious virus particles are produced. For instance, mouse cells infected with LCM or leukaemia virus show no pathological changes. A mouse congenially infected with LCM virus shows a high degree of immune tolerance, and all tissues in the body are infected. Throughout the life of the animal, virus and viral antigens are produced in the cerebellum, liver, retina etc. without discernible effect on cell function. But sometimes there are important functional changes in infected cells which lead to a pathological result. For example, the virus infects growth hormoneproducing cells in the anterior pituitary. Although the cells appear perfectly healthy, the output of growth hormone is reduced, and as a result of this, suckling mice fail to gain weight normally and are runted.

When bacteria invade tissues, they almost inevitably cause some damage, and this is also true for fungi and protozoa. The extent of direct damage, however, is sometimes slight. This is true for Treponema pallidum, perhaps because the lipopolysaccharide-protein components that might have induced inflammatory responses, are not exposed on the surface of the bacteria. It produces no toxins, does not cause fever, and attaches to cells in vitro without harmful effects. Leprosy and tubercle bacilli eventually damage and kill the macrophages in which they replicate, but pathological changes are to a large extent caused by indirect mechanisms (see below). In patients with untreated lepromatous leprosy, the bacteria in the skin invade blood vessels, and large numbers of bacteria, many of them free, may be found in the blood. In spite of the continued presence of up to 10 5 bacteria m l -1 of blood there are no signs or symptoms of septicaemia or toxaemia. Mycobacterium leprae can be regarded as a very successful parasite that induces very little host response in these patients, even when the bloodstream is invaded. The resident bacteria inhabiting the skin and intestines of man and animals do not invade tissues and are normally harmless; indeed, as discussed in Ch. 1, they may benefit the host. Bacteria such as meningococci and pneumococci, whose names imply pathogenicity, spend most of their time as harmless inhabitants of the normal human nasopharynx: only occasionally do they have the opportunity to invade tissues and give rise to meningitis or pneumonia.

Cell and tissue damage are sometimes due to the direct local action of the microorganism. However, it is not at all clear how viruses cause the death of cells. Many virus infections result in a shutdown of RNA synthesis (transcription), protein synthesis (translation) and DNA synthesis in the host cell, but usually these are too slow to account for the death of the cell. After all, cells like neurons never synthesise DNA, and the half life of most proteins and even RNAs is at least several hours. A possible alternative mechanism is the alteration of the differential permeability of the plasma membrane. This is important as the cell has a high internal K + concentration and low N a + concentration, while the reverse is true of body fluids. Viruses do alter membrane permeability, but the unresolved question is whether or not this is responsible for the death of the cell or is merely an after-effect.

It now appears that at least some virus infections (HIV, adenovirus, influenza virus, Sindbis virus) cause the cells to commit suicide by a mechanism called 'programmed cell death' or 'apoptosis'. This is the natural process by which the body controls cell numbers and rids itself of superfluous or redundant cells during development. A familiar example is a tadpole 'losing' its tail. Cells do not disintegrate but round up, and are then removed by phagocytes. A possible example of apoptosis is seen in the common cold when caused by the rhino virus group of picorna viruses. Rhinoviruses infect nasal epithelial cells, and at an early stage the cells round up, fall off the mucosal surface and are carried away, often with their cilia still beating, in a stream of fluid induced by the infection.* This leaves areas of raw mucosa, with the exposed underlying tissues inflamed, oedematous, and susceptible to infection by the normally harmless resident bacteria.

The gross pathology of cells infected with viruses appears generally nonspecific, very like that induced in cells by the toxins of diphtheria bacilli and streptococci, or by physical and chemical agents. The most common and potentially reversible change, the oedema seen as 'cloudy swelling' by routine histology, is associated with membrane permeability changes. Changes in the endoplasmic reticulum, mitochondria and polyribosomes are seen by electron microscopy at this stage. Later the nuclear chromatin moves to the edge of the nucleus ('margination' of chromatin) and becomes condensed (pycnosis), but the cell has already died by this time. There are two more characteristic types of morphological change produced by certain viruses, and these were recognised by histologists more than 50 years ago. The first are inclusion bodies, parts of the cell with altered staining behaviour which develop during infection. They often represent either cell organelles or virus factories in which viral proteins and/or nucleic acids are being synthesised and assembled. Herpes group viruses form intranuclear inclusions, rabies and poxviruses intracytoplasmic inclusions, and measles virus both intranuclear and intracytoplasmic inclusions. The second characteristic morphological change caused by viruses is the formation of multinucleate giant cells. This occurs, for instance when HIV 'fusion' proteins (gpl20-gp41) present in nascent virus particles budding from an infected cell attach to CD4 receptors in the plasma membranes of neighbouring cells; membranes then fuse and multinucleate cells are formed. It also happens in measles and certain herpes virus infections.

Before leaving the subject of direct damage by viruses, one supreme example will be given. Here the direct damage is of such a magnitude that the susceptible host dies a mere six hours after infection. If Rift Valley Fever virus, an arthropod-borne virus infecting cattle, sheep and man in Africa, is injected in very large doses intravenously into mice, the injected virus passes straight through the Kupffer cells and endothelial cells lining liver sinusoids (see Ch. 5) and infects nearly all hepatic cells. Hepatic cells show nuclear inclusions within an hour, and necrosis by four hours. As the single cycle of growth in hepatic cells is completed, massive liver necrosis takes place, and mice die only six hours after initial infection. The host defences in the form of local lymph nodes, local tissue phagocytes etc. are completely overcome by the intravenous route of injection, and by the inability of Kupffer cells to prevent infection of hepatic cells. Direct damage by the replicating virus destroys hepatic cells long before immune or interferon responses have an opportunity to control the infection. This is the summit of virulence. The experimental situation is artificial but it illustrates direct and lethal damage to host tissues after all host defence mechanisms have been overwhelmed.

Most viruses, rickettsiae and chlamydia damage the cells in which they replicate, and it is possible that some of this damage is due to the action of toxic microbial products. This action, however, is confined to the infected cell, and toxic microbial products are not liberated to damage other cells. Mycoplasma (see Table A .2) can grow in special cell-free media, but in the infected individual they generally multiply while attached to the surface of host cells. As studied in culture and on the respiratory epithelium they 'burrow' down between cells, inhibit the beat of cilia and cause cell necrosis and detachment. The mechanism is not clear. If a complete lawn of mycoplasma covers the surface of the host cell, some effect on the health of the cell is to be expected, but it is possible that toxic materials are produced or are present on the surface of the mycoplasma.

Dental caries provides an interesting example of direct pathological action. Colonisation of the tooth surface by Streptococcus mutans leads to plaque formation, and the bacteria held in the plaque utilise dietary sugar and produce acid (see p. 35). Locally produced acid decalcifies the tooth to give caries. Caries, arguably the commonest infectious disease of Western man, might logically be controlled by removing plaque, withholding dietary sugar, or vaccinating against Streptococcus mutans. However, fluoride in the water supply or in toothpaste has been the method of choice, and has been very successful. It acts by making teeth more resistant to acid. Bacteria generally damage the cells in which they replicate, and these are mostly phagocytic cells (see Ch. 4). Listeria, Brucella and Mycobacteria are specialists at intracellular growth, and the infected phagocyte is slowly destroyed as increasing numbers of bacteria are produced in it. Bacteria such as staphylococci and streptococci grow primarily in extracellular fluids, but they are ingested by phagocytic cells, and virulent strains of bacteria in particular have the ability to destroy the phagocyte in which they find themselves, even growing in the phagocytes, as described in Ch. 4. Many bacteria cause extensive tissue damage by the liberation of toxins into extracellular fluids. Various toxins have been identified and characterised. Most act locally, but a few cause pathological changes after spreading systemically through the body.

This is a huge and growing part of our subject and we need to define the term toxin, a task which is more difficult than one might think. An attempt was made by Bonventre who in 1970 defined toxins as a 'special class' of poisons which differ from, for example, cyanide or mercury by virtue of their microbial origin, protein structure, high molecular weight, and antigenicity. This view is too embracing, because it includes proteins of doubtful significance in disease, and also too restrictive, because it excludes nonprotein toxic complexes such as endotoxin. Another suggestion is that toxin must include all naturally occurring substances (of plant, animal, bacterial or whatever origin) which when introduced into a foreign host are adverse to the well-being or life of the victim. This, too, is unsatisfactory because some substances-potent toxins within the scope of this definition-are being used in some contexts as therapeutic agents! Perhaps it is pointless to strive for an all-embracing definition, although the obvious differences between bacterial and fungal toxins warrant the continued use of the appropriate prefix. For example, bacterial toxins are usually of high molecular weight and hence antigenic, whereas fungal toxins tend to be low molecular weight and not antigenic.

The problem of definition is compounded because there are substances (aggressins) which help to establish an infective focus as well as those whose action is uniquely or largely responsible for the disease syndrome. Also there are substances known to be produced by bacteria in vitro, whose properties on a priori grounds make them potential determinants of disease, but which have not been shown to play a role in vivo.

We will concentrate on those toxins known to be, or likely to be, involved in some aspect of disease causation. It is not difficult to demonstrate the production of'toxins' by bacteria in culture, as judged by some bioassays. Primary consideration will be given to those substances which are produced under ecologically significant conditions (i.e. in the natural host or relevant animal model) and cause (also in biologically relevant systems) damage to cells or tissues thereby contributing to disease.

Exotoxins are produced and then either secreted by, or released upon lysis from, both Gram-positive and Gram-negative bacteria. They are proteins, some of which are enzymes. When liberated locally they can cause local cell and tissue damage. Those that damage phagocytic cells and are therefore particularly useful to the microorganism have been described in Ch. 4. Those that promote the spread of bacteria in tissues have been referred to in Ch. 5. A description of some of the more interesting exotoxins follows.

Proteases and hyaluronidases, which help the spread of bacteria through tissues have already been mentioned in Ch. 5. Here we consider toxins which act on extracellular substances and are responsible for many of the main features of the diseases caused by the infecting organism. Pseudomonas aerwgmosa-elastase, and one of at least six proteases of Legionella pneumophila, both induce fibrinopurulent exudation in the rat lung (a model for P. aeragmosa-induced pneumonia in human cystic fibrosis) and the guinea-pig lung (a model for legionnaires' disease) respectively. These characteristics almost certainly arise from the release of oligopeptides from extracellular matrix components of the host which are chemotactic for leucocytes and fibroblasts. The L. pneumophila protease is the same major secretory protein (the 38 000 M r zinc metalloprotease) already considered in Ch. 4 in relation to survival within macrophages.

Staphylococcal exfoliatin is important in staphylococcal 'scalded skin syndrome' (SSSS), a disease of newborn babies. The disease is characterised by a region of erythema which usually begins around the mouth and, in 1-2 days, extends over the whole body. During this period, small yellowish exudative lesions often appear. The most striking feature of the disease, however, is that the epidermis, although apparently healthy, can be displaced and wrinkled like the skin of a ripe peach by the slightest pressure. Soon large areas of the epidermis become lifted by a layer of serous fluid and peel at the slightest touch. Large areas of the body rapidly become denuded in this way and the symptoms resemble those of massive scalding. While the toxin causes cleavage of desmosomes (specialised cell membrane thickenings through which cells are attached to each other) in the stratum granulosum, it may also act intracellularly. Despite numerous attempts to characterise the biological activity of exfoliatin, the genetically predicted serine protease and/or lipase activity has never been demonstrated.

Some enterotoxigenic E. coli elaborate families of low molecular weight heat stable (ST) peptides as well as heat labile (LT; cholera-like) toxin. STs bind to a receptor which then activates a tightly coupled membranebound guanylate cyclase in gut cells, resulting in the transmission of a signal to the inside of the cell, thereby elevating cGMP, or some other second message. As described later in the section on diarrhoea this gives rise to efflux of ions, and hence water, from enterocytes.

We know that these toxins are able to traverse membranes since their targets are intracellular; we still know very little about the details of how they achieve this. These proteins have common features: one component (usually described as the B fragment or subunit) binds to and interacts with a cell receptor and promotes the uptake of an active component (A fragment or subunit) in which resides the biological activity which confers toxicity. There are several ways in which one can classify these toxins.

(1) Biologically-some kill cells (cytotoxic toxins) whereas others (sometimes called cytotonic toxins) 'deregulate' cells.

(2) Biochemically-some belong to a well defined biochemical group recognised by their ability to cleave NAD + into nicotinamide and ADP-ribose moieties. They are designated ADP-ribosyl (ADPR) transferases since they transfer ADPR to different target proteins; this group straddles groups 1 and 3. (3) Genetically-the third possibility reflects the genetic origin of these toxins: (i) production from a single gene of a single peptide which undergoes post-translational modification into A and B fragments which are covalently linked; (ii) production from separate genes of A and B subunits which noncovalently associate into stable complexes; (iii) production from separate genes of different proteins which do not associate into stable complexes but which must act in concert to express toxicity. These are known as binary toxins; again, group 3 straddles group 1.

Examples of each of these groups will be given illustrating how they work, their importance in disease, and surprisingly (!) how some of the most toxic substances known to man are being used or may be used for therapeutic purposes.

Diphtheria toxin (DT). Corynebacterium diphtheriae organisms multiply on the epithelial surfaces of the body (nose, throat, skin) but do not penetrate deeply into underlying tissues. The infection on the body surface causes necrosis of mucosal cells with an inflammatory exudate and the formation of a thick 'membrane' (hence the name C. diphtheriae: Gr., diphthera = membrane) and if the infection spreads into the larynx there may be respiratory obstruction. The toxin probably assists colonisation of the throat or skin by killing epithelial cells and polymorphs. Active immunisation with diphtheria toxoid has made diphtheria a clinical rarity in developed countries. DT is disseminated from the infection site and has important actions, especially on the heart and nervous system.

DT is encoded by lysogenic phage ß but its expression is controlled by the host bacterium under conditions of iron stress. It is synthesised and processed as shown in Fig. 8 .1.

There is intense interest in seeking to understand the mechanisms of uptake of the active moieties of diphtheria and other toxins whose target is intracellular. This is driven by the desire to understand fundamental mechanisms in cell biology and to develop selective Cytotoxic therapies' in clinical medicine. The mechanism of internalisation of the active A Under the acidic conditions of the endosome, some DT-A is released by reduction, DT-B undergoes conformational change and interacts with the endosomal membrane resulting in the opening of cation channels. The latter are not absolutely necessary, but their formation is associated with 10 times greater uptake of DT-A. In contrast, most other toxins which enter the cytosol appear to do so by more complicated routes which initially involving endocytosis. Shiga toxin (and ricin, a toxic plant lectin believed to have been located in the tip of an umbrella and used to poison a Bulgarian spy!) are not translocated across the endosomal membrane but are transported to the trans-Golgi network and all the way back to the endoplasmic reticulum. Somewhere in the exocytotic pathway conditions exist for the translocation of Shiga toxin (ShT) into the cytoplasm. ShT acts by modifying 28S ribosomal rRNA.

For cholera toxin, it used to be thought that A l ( Fig. 8 .5) moved down through the central hole of the ring-shaped complex of CT-B which was anchored to the membrane by five ganglioside GM1 binding sites. Structural work on this family of proteins shows that this is not physically possible. The finding that the C-terminal sequence lysine-aspartateglutamate-leucine (the KDEL* motif) in the A subunit of cholera toxin and related sequences in E. coli heat labile toxin and Pseudomonas exotoxin A raises the possibility that transport to the endoplasmic reticulum may also be necessary for these A subunits to reach the cytoplasm.

A new and exciting development is beginning. For decades, attempts have been made to make immunotoxins by coupling native toxins to antibodies specific to some surface antigen on tumour cells, with little practical success, as yet. However, scientists have now genetically engineered DT by substituting that portion of the DT structural gene encoding the native toxin receptor-binding domain with modified cDNA encoding one of a variety of cytokines and growth factors including IL-2, IL-4, IL-6, EGF and α-MSH. The resultant fusion toxin bears a new 'cellular address', but retains all of the other biological properties of the native DT molecule as well as a three-dimensional structure nearly identical with native DT. The IL-2 receptor-binding domain fusion toxin has undergone Phase I/II clinical trials for the treatment of IL-2 receptor positive haematological malignancies (and other IL-2 related conditions) and has been shown to be safe and well tolerated. Such constructs are potent against a murine model of IL-2 receptor positive lymphoma, and the hope is that a durable remission of disease will be achieved in humans. (An era of new 'magic bullets'?) P. aeruginosa exotoxin A. P. aeruginosa, already alluded to above, is common in soil and water and can occasionally be isolated from the faeces of normal, healthy individuals. It is virtually harmless for healthy adults, but its ability to multiply in almost any moist environment and its resistance to many antibiotics have made the bacterium a major cause of * The KDEL motif is normally found in proteins which having been processed in the Golgi are trapped by the endoplasmic reticulum which recognises the KDEL motif. This prevents the protein being lost to the cell via exocytotic trafficking and results in its translocation into the cytoplasm. The schema shows a round of peptide elongation and illustrates the key role played by two enzymes, EF1 and EF2. DTA and PEA each ADP-ribosylates diphthamide (a modified histidine) in EF2-GTP which can no longer translocate the newly elongated peptide from the A site to the P site. ShT A fragment is a specific N-glycosidase which cleaves an adenine residue from near the 3'-prime end of the 28S ribosomal RNA. This depurination results in failure of EF1-dependent binding of aminoacyl-tRNA to site A and hence inhibits protein synthesis. Poliovirus achieves selective inhibition of host protein synthesis at an earlier stage than is depicted here. Host mRNA is first modified (capped) then bound to the small ribosomal subunit; poliovirus mRNA is not capped. The function of a cap-binding protein, which recognises and binds host mRNA to the ribosome, is inhibited by a poliovirus virion protein thereby allowing differential translation of virus messenger RNA. EF-Ια: nucleotide-binding protein. hospital-acquired infection, particularly among patients with impaired host defence mechanisms such as those with chronic illness, genetic immunodeficiencies, those under treatment with immunosuppressive drugs, or patients suffering from extensive burns. P. aeruginosa causes localised infection in the urinary tract, respiratory tract, burns and wound infections. In severely debilitated patients these localised infections may develop into general septicaemia, with mortality in such cases approaching 100%. The mechanisms of pathogenicity of P. aeruginosa are complicated and unclear because (unlike C. diphtheriae) the organism elaborates several potentially toxic extracellular products, including a phospholipase, several proteases, lipase, haemolysin, enterotoxin, lipopolysaccharide endotoxin, elastase, exoenzyme S (PES), and exotoxin A (PEA); evidence definitely implicates the last three. Elastase has already been mentioned above (p. 207). PES is without question a virulence determinant, but at present its biological mode of action is not wholly understood. PEA is in many ways like DT. It inhibits protein synthesis by cleavage of NAD + and subsequent ADP-ribosylation of EF2 at the same diphthamide residue, is synthesised predominantly during the decline phase of cell growth, and its production appears to be dependent upon the concentration of iron in the medium. Its M r is 66 580, and it can be activated in cell-free systems by proteolysis, or by reduction with thiols in the presence of urea and other chaotropic agents. All the enzymic activity of the toxin resides in a proteolytically derived fragment (M r approximately 26000). However, dissimilarities exist between the two toxins. Firstly, the active A fragment resides in the COOH-terminal portion of the molecule, whereas the diphtheria A fragment is at the NH 2 -terminal end. Secondly, the B fragment of PEA recognises a different receptor to that of DT, and there are no sequence homologies or serological crossreactivity between the two toxins.

Shiga toxin. This toxin is a subunit protein with an AB 5 type structure (1 A subunit; B, an oligomer of 5 subunits). Certain biotypes of E. coli make Vero toxins (VTs) or Shiga-like toxins (SLTs) which are identical (SLT1) or near identical (SLT2) with Shiga toxin. The role of Shiga toxin in bacillary dysentery is discussed below. SLT-producing strains of E. coli are responsible for haemorrhagic colitis and haemolytic urea syndrome. Very recent crystallographic studies have shown a remarkable similarity in organisational structure of the B subunits of Shiga toxin and those of the cholera toxin family discussed below, despite the near complete absence of sequence homology. The B subunits attach to a ganglioside structure Gb3 or Gb4 (cholera toxin attaches to GM1) on susceptible cell surfaces. ShT-A fragment is a specific N-glycosidase which cleaves an adenine residue from near the 3'-prime end of 28S ribosomal RNA. This depurination results in failure of EFl-dependent binding of aminoacyl-tRNA to site A and hence inhibits protein synthesis ( Fig. 8.3 ). This is based on studies on cultured cells, and how such cytotoxic activity gives rise to the enterotoxic activity-watery diarrhoea seen in cases of dysentery preceding the bloody Mode of action of actin-ADP-ribosylating toxins. C. botulinum C2 toxin component 211 binds to the cell membrane followed by C2I. The latter is internalised and upsets the equilibrium between polymerisation and depolymerisation of actin. ADP-ribosylation of actin inhibits its polymerisation and turns G-actin into a capping protein which binds to the fast growing (concave) ends of actin filaments. Capping of the concave ends increases the critical concentration for actin polymerisation. Since the slow-growing (pointed) ends of actin filaments are free, depolymerisation of actin occurs at these ends. Released actin is substrate for the toxin and will be withdrawn from the treadmilling pool of actin by ADPribosylation, i.e. trapped. Both reactions will finally induce the breakdown of the microfilament network. (Taken with permission of authors (K. Aktorieseia/.) and publisher (Springer-Verlag GmbH & Co. KG, Heidelberg, Germany) from Botulinum C2 toxin. Clostridium botulinum C2 toxin is not a neurotoxin but belongs to a family of cytotoxins {Clostridium perfringens iota toxin, Clostridium spiroforme toxin, Clostridium difficile ADP ribosyltransferase) whose common features are that they are binary toxins and that they target cytoskeletal actin. Of these C. botulinum C2 is the best studied at present. Component C2I is the equivalent of the A subunit and C2II, the B subunit. They are produced in parallel although different strains produce different ratios. More C2 is produced during sporulation than in the vegetative phase of growth. C2II is activated by proteolysis to produce a 74 000 M r protein which binds to cell surfaces and thereby exposes a binding site for C2I (M r ca. 45 000) either on the cell surface or on attached C2II. Once inside, C2I ADP-ribosylates monomeric actin and not polymerised F actin. The substrate specificity is high. The toxin will modify a specific arginine residue present in certain isoforms of actin. Neither subunit is toxic on its own, but together the LD 50 dose for a mouse is ca. 5-50 ng. (Other examples of such 'binary' activity include anthrax toxin and staphylococcal leucocidin). The two components may be injected into different sites or by different routes but toxic effects always occur at the site of injection of C2II.

This mode of action, summarised in Fig. 8 .4, is believed to underlie the biological manifestations of toxicity which, in rats, include hypotension, haemorrhaging in and fluid accumulation around the lungs as a result of vascular permeability changes. The toxin is currently being used as a tool in cell biology to study cytoskeletal systems and dissect mechanisms of cell signal transduction.

Cholera and related toxins. There is a growing number of toxins in this category. Easily the most studied is cholera toxin (CT), a major research area for three decades, which was originally stimulated by the desire to make a more effective vaccine against cholera. Ironically, effective immunity is best mediated by antibacterial antibodies which prevent colonisation of the gut with V. cholerae rather than antitoxin-neutralising antibodies. Cholera toxin is an AB 5 subunit toxin ( Fig. 8.5 ). Enterotoxigenic E. coli (ETEC) strains produce two heat labile toxins: LTI and LTII. LTII exists as two minor variants, LTIIa and LTIIb, the latter requiring trypsin treatment for full activation in the Y-1 adrenal cell assay. The B subunits of CT, LTIIa and LTIIb recognise and are bound to cell surface gangliosides GM1, GDlb and GDI a respectively and mediate the translocation of A l subunits into the cytoplasm; LTI binds to either GM1 or a glycoprotein. The same reaction occurs as was described above for DT: NAD + is cleaved into nicotinamide and ADP-ribose (ADPR) moieties, with two important differences:

(1) ADPR transferase is greatly enhanced by a group of proteins (both membrane bound and soluble) of ca. 20 000 M r designated ADPribosylation factors (ARFs; of as yet unknown physiological significance); (2) ADPR is transferred to the a s subunit of the G s regulator of adenylate cyclase as illustrated in Fig. 8 .6.

This results in the elevation of cAMP levels in the cell with the consequences described below in the section on diarrhoea.

Bordetella pertussis toxin (pertussigen) and adenylate cyclasehaemolysin (AC-Hly). Whooping cough (pertussis) is a severe respiratory tract infection characterised by prolonged paroxysmal coughing, attacks of which continue long after infection has cleared. The disease is capable of striking all ages but is particularly prevalent and severe in young children, where hospitalisation is required in about 10% of cases.

The causative agent of pertussis, Bordetella pertussis, is transmitted aerially from the respiratory tract of an infected individual to that of a susceptible host. The organism attaches (probably via its filamentous haemagglutinin, pili, and a 69 000 M r outer membrane protein) to, and colonises the mucosal surface between the cilia, and multiplies there during the incubation period of the disease which is commonly around seven days. The infection then manifests as a slight fever and catarrh which is often indistinguishable from a common cold. However, one to two weeks later bouts of uncontrollable coughing begin. It is this paroxysmal coughing, along with the notorious 'whoop' as the child attempts to draw breath, which characterises the disease. The paroxysmal coughing stage often lasts for several weeks and no treatment fully effective in controlling the symptoms. The only proven means of controlling whooping cough is vaccination but, in the UK at least, sporadic reports of vaccine-induced brain damage in infants has diminished public acceptance of the vaccine. It should be noted that permanent encephalopathy (brain damage) is a recognised though rare consequence of whooping cough infection. Much current work is being devoted to producing immunogenic, completely nontoxic preparations of pertussis toxin by genetic manipulation of t h e gene encoding t h e S I s u b u n i t ( Fig. 8 .5). In clinical trials in Italy, such engineered vaccines h a v e been shown to be both safe a n d effective a s j u d g e d by antibody titres to pertussis toxin.

Several toxic substances h a v e been isolated from Bordetella pertussis including a h e a t labile toxin, tracheal cytotoxin, endotoxin, adenylate cyclase-haemolysin (AC-Hly), a n d pertussis toxin. T h e latter h a s m a n y biological activities-histamine sensitisation, leucocytosis promotion, islet (a) The production of cyclic AMP by adenyl cyclase. Cyclic AMP is an important second messenger invoked in the intracellular amplification of many cellular responses to external signals including hormones. The nature of the physiological response reflects the biochemical differentiation of the cell responding to the stimulus. For example, in gut cells the response would be altered ion transport and hence fluid secretion; in muscle cells it would be glycogen breakdown in response to the call for more energy. The production of cAMP is controlled both positively and negatively at two different levels. The central cycle represents a normal membranebound hormone receptor and the heterotrimeric (aß7) G protein regulator complex which is activated upon binding of hormone to the receptor. There are two receptors, each with regulatory G proteins: one system responds to stimulatory and the other responds to inhibitory stimuli; only one system is shown. In gut cells these receptors would be on the basolateral (nonluminal) side of enterocytes enabling responses to stimuli from the circulation. (b) The second level of control involves endogenous GTPase properties of both stimulatory (a s ) and the inhibitory (a A ) subunits of the G protein regulator which may be outlined as follows. Stimulation: agonist stimulation of its receptor results in the dissociation of the a s ß7 G protein from the receptor and of the subunits from each other, and the binding of GTP to a s . a s -GTP will productively interact with the catalytic unit of adenyl cyclase (not shown) and stimulate the production of cAMP from ATP. Endogenous GTPase activity in a s results in a delayed conversion of a s -GTP to inactive a s -GDP which can no longer stimulate the cyclase but allows the reassociation of the trimeric G protein and loss of GDP. Inhibition: antagonist stimulation of its receptor results in an analogous situation for inhibiting cyclase activity via c^: inhibitory cti-GTP inactivates the cyclase; cti-GDP does not inactivate the cyclase.

(c) The level of cAMP may be affected by physiological stimuli (which stimulate removal and modification of the G-protein complex) or by perturbation of the normal regulatory cycle as illustrated, by cholera toxin (CT) or pertussis toxin (PTx).

(d) Cholera toxin acts first by interacting with its receptor resulting in the internalisation of the active A l subunit. A l ADP-ribosylates a s -GTP which promotes continued dissociation of the heterotrimer; also the endogenous GTPase is no longer functional hence the stimulation of the cyclase continues. LT toxins act in a similar manner.

(e) Pertussigen acts first by interacting with its receptor resulting in the internalisation of the active SI subunit. SI ADP-ribosylates the ai-GDPß7-heterotrimer which can no longer associate with the receptor (or lose GDP) to undergo another cycle of GTP activation; activated cyclase can no longer be turned off.

(Adapted, with kind permission of author (Gierschik, P.) and publisher (Springer-Verlag GmbH & Co. KG, Heidelberg, Germany) from Figure 4 in 'ADP-ribosylation of signal-transducing guanine nucleotide-binding proteins by pertussis toxin', Current Topics in Microbiology andlmmunology (1992) 175,78, edited by Klaus Aktories. activation and adjuvanticity-and is believed to be the most important but by no means the only virulence determinant of B. pertussis: it is called pertussigen and is the basis of the vaccine against whooping cough. Pertussigen is a complex subunit toxin whose biochemical mode of action is identical to that of CT except that the target is the aj subunit of the Gj regulator of adenylate cyclase (see Fig. 8 .6). It is not yet clear how such biochemical activity relates to the clinical syndrome.

AC-HLy has two functional domains. One confers adenylate cyclase activity and another haemolytic activity. Its haemolytic activity arises from its pore-forming properties in cell membranes. Pore formation is probably responsible for the translocation of the adenylate cyclase portion of the complex. AC-HLY is also known as cyclolysin and belongs to the group of toxins whose common feature is the presence of an array of a nine amino acid repeats; they have been designated RTX (repeats in toxin) toxins. The prototype is the haemolysin of E. coli which is important in extraintestinal infections caused by this organism. Other toxins of this group include the haemolysin from Proteus vulgaris, and the leukotoxin from Pasteurella haemolytica.

Anthrax toxin. Anthrax is a disease of animals, particularly sheep and cattle, and to a lesser extent man, caused by infection with Bacillus anthracis. Infection takes place following the ingestion of spores, the inhalation of spores, or by the entry of spores through abraded skin. The spores germinate and then the bacteria form a toxin that increases vascular permeability and gives rise to local oedema and haemorrhage. Infection of the skin in man leads to the formation of a lesion (malignant pustule; a black eschar, hence B. anthracis; Gr. anthrakos = coal) consisting of a necrotic centre surrounded by vesicles, blood-stained fluid and a zone of oedema and induration. In severe infections there is septicaemia with toxic signs, loss of fluid into tissues, with widespread oedema and eventually death. Anthrax toxin, the discovery of which by Smith and Keppie in the early 1950s was a major milestone in our subject, consists of three components, none of which are toxic by themselves: factor I (oedema factor or EF), factor II (protective antigen or PA) and factor III (lethal factor or LF). The toxin is largely responsible for local lesions, kills phagocytes, enters the circulation and accounts for the toxicity, oedema and death caused by B. anthracis. Virulent strains also have a capsular polypeptide composed of D-glutamic acid, which inhibits opsonisation and phagocytosis (see Ch. 4). Anthrax in man occurs mainly in those whose work brings them into contact with infected animals. It is not a common disease in the UK, and the usual source of infection is imported bones, hides, skins, bristles, wool and hair, or imported fertilisers made from the blood and bones of infected animals. Most of what has just been described was discovered more than three decades ago but more recently new insights have been obtained as to the biochemical basis of the mode of action of anthrax toxin.

EF was shown to be an inactive form of adenylate cyclase which is activated by calmodulin; PA and LF were not active. However combinations of PA and EF (but not PA and LF) caused an elevation of cAMP. Thus anthrax toxin (at least the PA/EF combination) is a binary toxin. The response observed is more rapid and greater than that induced by cholera toxin. In contrast to cholera toxin, the anthrax complex is the cyclase, acts without a lag phase and its effects are instantly reversible upon washing toxin-treated cells. This suggests that the enzyme is either rapidly degraded after internalisation, or that it is not completely internalised but WsfhiW II cAMP ATP Fig. 8 .7 Mode of action of anthrax toxin. Factor II (PA) interacts with the cell membrane. After proteolytic cleavage sites are exposed which bind factor I (EF) and facilitate internalisation of factor I (EF). This internalisation takes place via an endocytic step, with factor I being released into the cytosol. Factor I must be rapidly inactivated since washing toxin-treated cells results in a rapid loss of adenylate cyclase activity. Factor I interacts with calmodulin (CAL) to become an active adenylate cyclase enzyme. Interaction of factor I with factor II and subsequent internalisation is blocked by prior binding of factor III (LF) to factor II. associated with the cell membrane in a manner allowing it to be readily removed. These facts together with the observations that LF would block the activity of EF are summarised in Fig. 8 .7. This explains how LF blocks the effect of EF. They either compete for the same sites on PA, in which case the explanation is self-evident, or they bind to separate sites on PA which are situated such that occupancy of one occludes access to the other.

One can extend these observations made in vitro to the in vivo situation. The model provides a basis for understanding the protective function of PA. Antibodies to PA could either block the initial attachment to the cell membrane of target organs in vivo, or prevent the binding of EF to PA, or both. The fact that cAMP is known to be a potent secretagogue could explain how the oedematous reaction occurs when PA and EF are injected into test animals, i.e. EF is, after activation, an oedema factor. Moreover, in biological tests, addition of LF (factor III) depresses oedema production as would be predicted by this model. The observation that PA was probably fixed first may now be placed on a firmer basis since binding of PA is a prerequisite for the expression of EF (and by implication) LF activities. The case for the view that, in experimental anthrax in guinea-pigs, animals died (as do humans) of secondary shock, is also placed on a firmer basis with perhaps one important reservation. Smith and coworkers claimed that factor III (LF) decreased oedema but increased lethality; the former but not the latter observation is explicable by the model which is therefore too simplistic to explain the effects of the holotoxin. It is likely that LF exerts its activity together with PA on other cells in vivo. It has in fact been shown that LF in combination with PA is cytolytic for macrophages and macrophage-like cell lines-the probable basis of the earlier observation that anthrax toxin kills phagocytes and perhaps justification for talking about a plurality of anthrax binary toxins.

Clostridium tetani spores germinate in an infected wound and produce their toxin. Spores are ubiquitous in faeces and soil, require the reduced oxygen tension for germination provided locally in the wound by foreign bodies (splinters, fragments of earth or clothing) or by tissue necrosis as seen in most wounds, the uterus after septic abortion, or the umbilical stump of the newborn.* The site of infection may be a contaminated splinter just as well as an automobile or battle injury. All strains of Clostridium tetani produce the same toxin. It also reaches the central nervous system by travelling up other peripheral nerves following bloodborne dissemination of the toxin through the body. The motor nerves in the brain stem are short and therefore the cranial nerves are among the first to be affected, causing spasms of eye muscles and jaw (lockjaw). There is also an increase in tonus of muscles round the site of infection, followed by tonic spasms. In generalised tetanus there is interference with respiratory movements, and without skilled treatment the mortality rate is about 50%. Botulism* caused by Clostridium botulinum, a widespread saprophyte present in soil and vegetable materials. C. botulinum contaminates food, particularly inadequately preserved meat or vegetables, and produces a powerful neurotoxin. The toxin is destroyed at 80°C after 30 min-of great importance to the canning industry-and there are at least seven antigenically distinct serotypes (A-G) produced by different strains of bacteria but which have a pharmacologically similar mode of action. It is absorbed from the intestine and acts on the peripheral nervous system, interfering with the release of acetylcholine at cholinergic synapses of neuromuscular junctions. Somewhere between 12 and 36 hours after ingestion there are clinical signs suggesting an acute neurological disorder, with vertigo, cranial nerve palsies and finally death a few days later with respiratory failure. 1 * Botulus (Latin) = sausage. In 1793 a large sausage was eaten by 13 people in Wildbad in Germany; all became ill, and six died. The disease was subsequently referred to as botulism. t In recent years a less typical form of botulism has been described in small infants. The spores, present in honey applied to rubber teats, appear to colonise the gut, so that the toxin is produced in vivo after ingestion.

Tetanus and botulinum toxins are two of the most toxic substances known to man: 1 g of botulinum toxin will kill 10 10 mice. A great deal is now known about the genetics of both these toxins. Tetanus toxin and botulinum toxin type G are encoded by plasmids, botulinum toxins types C and D are encoded by phages that infect clostridia, and botulinum type A is encoded in the bacterial genome; the genes have been sequenced. Both toxins are synthesised as single peptides and are released from bacteria upon lysis. They are proteolytically activated by endogenous proteases to yield dichain derivatives consisting of -S -S -linked L (M r 50 000) and H (M r 100 000) chains. The recently adopted nomenclature reflects our current understanding of their peptide structures ( Fig. 8.8 ).

Very recently, huge strides have been made in our knowledge of the mode of action of these toxins ( Fig. 8.9 ). There is very little overall sequence homology between BoNTs and TeTx; this could account for their immunological characteristics and the animal types normally affected. However, there is one structural feature common to all these neurotoxins located in the middle of the light chain and that is a zincbinding motif. In fact these neurotoxins are zinc endopeptidases. Their targets are protein constituents of the synaptic fusion machinery responsible for the exocytotic release of neurotransmitters. The intracellular target cleaved by BoNTs B, D, and F and TeTx is synaptobrevin (VAMP, vesicle-associated membrane protein) present in small synaptic vesicles. BoNTs A and E cleave SNAP-25 (synaptosomal associated protein of 25 000M r ) a highly conserved protein known to be involved in exocytosis. BoNT C acts on syntaxin, a synaptic membrane protein also involved in exocytosis.

However, some puzzles remain to be resolved. Tetanus and botulinum toxins enter neuronal tissue preferentially at motoneuronal endplates, but the nature of the 'physiologically relevant ectoacceptors' is still not known. It remains unclear why botulinum toxin acts directly at the site of uptake and not, as observed with tetanus toxin, in the central nervous system, although a considerable amount of botulinum toxin (like tetanus toxin) is retrogradely transported. Botulinum toxin blocks the release of acetyl choline at neuromuscular junctions to cause flaccid paralysis. Likewise, one might ask why tetanus toxin fails to act at the motoneuronal junction at concentrations which would completely block release of neurotransmitter from GABAergic synapses. To reach inhibitory interneurons-its principal site of action-tetanus toxin must leave α-motoneurons after the primary uptake step, traverse the synaptic cleft to interneurons, leave those again in order to become finally internalised again from presynaptic membranes-a route identical to that of several neurotropic viruses. It acts by blocking the release of inhibitory transmitters (glycine or GAB A) resulting in a failure to relax the affected muscle-pathophysiological 'tetanus'. Only in rare cases does it act peripherally like botulinum toxin.

The most toxic substances known to man are now being used as therapeutic agents to treat focal dystonias such as neck twists and eye squints, or eyelid closure and, more recently, some childhood palsies. The preparations are made from BoNT A and consist of toxin-haemagglutinin complex-the form in which the toxin is usually produced by the organism. This complex is less toxic than purified neurotoxin when administered parenterally but relatively more toxic when given orally; the haemagglutinin apparently protects the neurotoxin from proteolytic degradation in the gut. The effects of BoNT in relieving muscle spasms are not permanent but last for several months. Treatment has to be repeated. To date the successes significantly outweigh the failures and no long-term adverse effects have been reported. Antibody to the toxin has not so far been detected in the sera of the majority of patients.

Some toxins destroy membranes by virtue of their proteolytic activities, and some by their ability to degrade lipid components, while others are pore-forming or detergent-like in their mode of action.

In addition to their action on protein components of lung connective tissue referred to above, Pseudomonas aeragmosa-elastase and the zinc metalloprotease of Legionella pneumophila are believed to destroy cell membranes by their proteolytic activity. This is the probable reason for the haemorrhage associated with lung infections caused by these pathogens, i.e. effects on type I alveolar epithelial and endothelial cells.

Clostridium perfringens α-toxin. A large number of bacterial enzymes are phospholipases some of which, but by no means all, are important toxins. The best example is the α-toxin of Clostridium perfringens, the organism most commonly associated with gas gangrene. It is strictly anaerobic and occurs as a normal inhabitant in the large intestines of man and animals; its spores are ubiquitous in soil, dust and air. C. perfringens does not multiply in healthy tissues, but grows rapidly when it reaches devitalised and therefore anaerobic tissues. This could be after contamination of a natural wound with soil or dust, particularly on battlefields or in automobile accidents, or after contamination of a surgical operation site with clostridia from the patient's own bowels or skin. After abortions, particularly in the old days before antibiotics, intestinal clostridia often gained access to necrotic or devitalised tissues in the uterus and set up life-threatening infections. Invasion of the blood was common and soon resulted in death, the clostridia localising and growing in internal organs such as the liver after death. C. perfringens has various enzymes that enable it to break down connective tissue materials, including collagen and hyaluronidase, thereby facilitating spread of the infection along tissue planes. Most of the enzymes are toxic (a) Reflex arc (top). Mechanism for inhibiting the antagonists to a muscle contracting in response to stretch. Muscles are reciprocally innervated with sensory and motor neurons, although for clarity this is shown only for the protagonist muscle. On stretch, the stretch receptors generate an impulse which is transmitted along the afferent sensory (S) neuron of the protagonist (P) muscle. This SP neuron enters the spinal cord by the dorsal root and synapses with the motor neuron supplying the protagonist muscle (MP) and with an interneuron (I) which in turn synapses with the motor neuron supplying the antagonist muscle (MA); the efferent motor neurons leave the spinal cord by the ventral root. At the SP/MP synapse an excitatory transmitter is released which induces an impulse in MP which leads to contraction of protagonist muscle. However, excitation of I causes release of an inhibitory transmitter at the I/MA synapse which leads to relaxation of the antagonist muscle. Note that the basic reflex arc has been shown for simplicity but TeTx acts mainly on voluntary muscles.

(b) A simplified version of the biochemical events occurring in synapses (lower left). Excitatory and inhibitory synapses, neurotransmitter release and action. Gly, glycine; R, receptors of neuro transmitters; X, hitherto uncharacterised (candidates include glutamate, dopamine, ATP, substance P, and somatostatin).

to host cells and tissues, but α-toxin is easily the most important one. It is dermonecrotic, haemolytic (a feature seen mainly in tissues close to the focus of infection but sometimes responsible for large-scale intravascular haemolysis in infected patients), causes turbidity in lipoprotein-rich solutions and is lethal. While it is still true that these activities are all due to one molecular species, they are not different expressions of the one enzymic activity.

Historically, C. perfringens α-toxin was the first bacterial toxin to be characterised as an enzyme: it is a phospholipase C (PLC) which removes the head group, phosphoryl choline, from phosphatidyl choline and from sphingomyelin. It is of undoubted importance in gas gangrene. Toxoid prepared by formalin-treated toxin will protect sheep against infection caused by C. perfringens. However, one might ask why all enzymes with such biochemical specificity are not equally toxic or important in determining virulence; there are several reasons which can be put forward. By the combined use of monoclonal antibodies and molecular genetic analyses we now know that there are at least two functional domains in this C. perfringens α-toxin. If one compares C. perfringens α-toxin with the phosphatidyl choline-preferring, nontoxic phospholipase of Bacillus cereus, one finds that two-thirds of the N-terminal sequence shows homology with the entire sequence of B. cereus PLC. This portion of C. perfringens α-toxin retains its PLC activity but not its haemolytic and lethal activities. The C-terminal part is not haemolytic, not enzymatically active and not cytotoxic for mouse lymphocytes, but is necessary for conferring toxicity on the N-terminal part of the protein. In fact, the C-terminus is a potent immunogen that will solidly protect mice -and hopefully man-against experimental infection with C. perfringens. Surprisingly, the C-terminal domain of the nontoxic C. bifermentans enzyme shows sequence similarity with that of its C. perfringens α-toxin counterpart. The nontoxicity of this enzyme is ascribed to its comparatively much lower turnover rate, i.e. it is a much less efficient enzyme.

Recently, attention has been drawn to the distinct possibility that while haemolysis may be initially induced by the PLC activity of C. perfringens α-toxin, it is possible that actual disruption of the membrane is due to the activation by the toxin of other phospholipases (phosphatidylinosotol-4,5-biphosphate (PIP2) specific phospholipase and phospholipase D) present in the erythrocyte membrane. Some now believe (c) Sites of neurotoxin action (lower right). The predominant site of action of TeTx is the intermotor neuron synapse; the exocytotic machine is interfered with by the endopeptidase action of TeTx on VAMP. BoNT acts at the neuromuscular junction inhibiting the release of acetyl choline (Ach) by its proteolytic action on VAMP (types B, D and F), or SNAP (types A and E), or syntaxin (type C).

(Amplified from Figs 18 and 19, 'Bacterial Toxins', 2nd Edition, by J. Stephen and R. A. Pietrowski (1986), pp. 60 and 62, Van Nostrand Reinhold (UK).) that the basis of toxicity is not simply a cytolytic one, but rather a consequence of its ability, in sublytic doses, to release inosotol triphosphate (IP3) and activate the arachidonic acid cascade.

There are other pathogenic clostridia that cause gas gangrene and produce similar toxins. Infected tissues show inflammation, oedema and necrosis, not necessarily with the formation of gas, and the illness can be mild or very severe according to the extent of bacterial spread and the nature and quantity of toxins that are formed and absorbed. Since the bacteria grow and produce their toxins only in devitalised tissues, the most important form of treatment is to remove such tissues. Clostridia are strictly anaerobic, and exposure of the patient to hyperbaric oxygen (pure oxygen at 2-3 atmospheres in a pressure chamber) has been found useful in addition to chemotherapy.

Staphylococcal ß-toxin. Staphylococcal ß-haemolysin is known to be produced in vivo. In Ch. 4 studies with isogenic mutants were described which indicate that it is important in killing neutrophils. It probably has the narrowest substrate specificity among the phospholipases, and is a hotr-cold haemolysin: lysis of erythrocytes occurs only on cooling after incubation at 37°C. The phenomenon, though of doubtful significance in vivo, has attracted attention and generated speculation about its mechanism. Perhaps the most likely explanation is that, when cooled below their phase-transition temperature, the remaining phospholipids undergo quasi-crystalline formation, thereby generating intramembranous stresses incompatible with structural integrity.

These proteins (made by some 19 species, not all of which are pathogens) have been called oxygen-labile haemolysins because they are reversibly inactivated on standing in air, and their most studied property is their ability to lyse red cells. SH-activation of crude preparations is necessary for the expression of haemolytic activity but they are also active towards a variety of cell types. Thiol-activated cytolysins share the following properties: they are cross-neutralised by hyperimmune sera, lyse a wide range of species of erythrocyte, exhibit similar pH and temperature optima for cytolytic activity, are lethal and cardiotoxic, and lose activity on incubation with erythrocyte ghosts. They are inactivated irreversibly by small amounts of cholesterol. Interaction with cholesterol is the key primary event in their interaction with susceptible membranes which leads to the impairment of the latter; cholesterol plays no further part in the subsequent damage process. A schematic outline of how they damage cell membranes is shown in Fig. 8.10 .

Several facts have recently reopened considerable interest in this group. First, the recognition of what these substances do to host defence cells when presented in sublethal doses. For example, pneumolysin from S. pneumoniae inhibits the respiratory burst in neutrophils, inhibits antibody synthesis in B cells and activates the classical complement pathway in the absence of antibodies. Since complement is an important defence mechanism against the pneumococcus in vivo this could lead to depletion of complement levels and abrogate protection. In fact, immunisation of mice against pneumolysin affords some protection against challenge infection. Second, in addition to its role in promoting invasive bacteraemia, pneumolysin has also been implicated in causing sensorineural deafness associated with meningitis caused by the pneumococcus; this perception is based on very recent work with a guinea-pig Other pore-forming toxins T h e R T X toxins h a v e a l r e a d y b e e n referred to above in connection w i t h B. pertussis A C -H l y . B u t t h e r e a r e o t h e r pore-forming toxins.

Staphylococcal δ-toxin. This toxin acts in a manner similar to that of the SH-activated cytolysins, with an important difference: there is no initial specific binding. δ-Toxin initially forms small pores and then islands of membrane or large micelles; this gives rise to its perceived detergent-like properties. There is a family of closely related δ-toxins which inhibit the growth of gonococci. It is not often that one can ascribe a positive function to a toxin which is beneficial for the organism producing it. In this case 8-toxin(s) could have important ecological significance in the mixed culture situation that is characteristic of the real microbial world. Of great interest is the synergy that δ-toxin displays. Sublytic amounts of δ-toxin causes release of cell constituents without lysis. But, only 0.01 haemolytic units of staphylococcal ß-toxin will cause lysis of cells in the presence of 0.004 lytic units of δ-toxin. This synergistic interaction could be the way in which staphylococcal toxins, which rarely exert their lethal effects in the majority of infections, exercise important cytolytic effects. Of less obvious significance is the fact that δ-toxin is a poor antigen; for a long time its antigenicity was controversial. If δ-toxin were to prove of crucial importance as a cytolytic potentiator, then this could also partly explain why natural acquired immunity to staphylococcal infection is either non-existent or sufficiently low as to be easily overcome.

Staphylococcal a-and 7-toxins. Staphylococci produce a range of toxins several of which we have already met. The α-toxin is considered as the main cytolysin produced by S. aureus. Like Streptolysin-0 and staphylococcal δ-toxin, it is secreted as a water-soluble protein and undergoes self-induced oligomerisation on cell membranes to form pores. It has recently been suggested that transmembrane channels formed by the hexameric form allow the penetration of the monomeric form which interferes with certain steps in translation. In systemic staphylococcal infections death is most probably due to the potent α-toxin but in localised pyogenic infections-such as mastitis in cattle, goats, rabbits and mice-its role is most likely one of killing phagocytes or conferring survivability on intracellular bacteria. 7-Toxin is a binary toxin with a haemolytic action on rabbit erythrocytes. There is evidence that α, β and 7 haemolysins and leucocidin are produced in vivo during natural and experimental infections. This is based on the presence of antibodies to these haemolysins in subjects undergoing staphylococcal infection and the extraction of toxin from peritoneal abscesses in mice. However, the precise roles that these or any other staphylococcal toxins play in localised infections is not clear.

Atrophie rhinitis is a disease of pigs in which the principal clinical sign is atrophy of the nasal turbinate bones and shortening of the snout.

Scientists have isolated and purified a toxin believed to be important in the causation of this disease. This is a remarkable toxin on at least two counts. It shows an incredible predilection for the target tissue in that it causes turbinate atrophy in gnotobiotic pigs, even when injected intraperitoneally; it also causes necrosis of spleen and thymus. Perhaps even more surprising is the fact that it is a potent mitogen, hardly the kind of activity that one would associate with atrophy! One can only assume that it upsets the dynamic balance between the laying down and resorption of mesenchymal bone tissue by osteoblasts and osteoclasts respectively.

The remainder of this section deals with other important (or potentially important) toxins which have not already been referred to in earlier chapters or in the section above.

Several staphylococcal toxins have already been referred to above. Staphylococci are ubiquitous and responsible for a large group of diseases which range from the relatively harmless skin pimple through abscesses, impetigo, food poisoning, osteomyelitis, mastitis, primary pneumonia, staphylococcal scalded skin syndrome (SSSS) and toxic shock syndrome (TSS) to fatal septicaemias. In the case of SSSS a toxin is known and is described on p. 208.

Toxic shock syndrome toxin (TSST-1). TSS is seen characteristically in menstruating women whose tampons harbour multiplying staphylococci. It is due to a toxin called toxic shock syndrome toxin 1 (TSST-1; one of the so-called staphylococcal enterotoxins, see next paragraph). Toxic shock syndrome is characterised by sudden onset of fever, vomiting, diarrhoea, an erythematous rash followed by peeling of the skin, hypotensive shock, impairment of renal and hepatic functions and occasionally death. The main symptoms of the disease have been reproduced in rabbits by implanting chamber-enclosed TSS-strains in the rabbit uterus or peritoneum or by injection of TSST-1 into rabbits. Complex changes are observed including: haemorrhage in kidney and liver; congestion and haematomas in the lungs; leakage of blood into the thymus; and fluid in the pericardial sac and in the gut lumen. These effects in rabbits are very similar to those seen in humans and would certainly explain the shock and diarrhoeal syndrome so characteristic of the disease. All of this indicates that the primary effect of TSST-1 is on the integrity of capillary walls. The lethal effect of TSST-1 is enhanced considerably by endotoxin (see below). Such synergy is also known for streptococcal erythrogenic toxin and LPS, but the mechanism(s) involved in this complex synergy is not clear.

Although relatively uncommon in the UK, staphylococcal food poisoning accounts for over 40% of all cases of food poisoning in the United States. The disease is characterised by vomiting and diarrhoea commencing 1-6 h after consumption of contaminated food, especially dairy produce. Symptoms usually last no longer than 24 h. Like botulism, staphylococcal food poisoning is commonly caused by the ingestion of food containing preformed toxins, known collectively as the staphylococcal enterotoxins. They are serotypically heterogeneous single-chain globular proteins, of molecular weight between 28 and 35 kDa, secreted by certain strains of S. aureus. One of these, serotype F is identical to TSST-1 and staphylococcal pyrogenic exotoxin C. These toxins are not classical enterotoxins acting on intestinal cells but 'neurotoxins' activating receptors on the abdominal viscera, the stimulus reaching the vomiting centre via the vagus nerve.

Several of the staphylococcal toxins referred to above (TSS-1 and enterotoxins), and also the erythrogenic toxins A and C of Streptococcus pyogenes (see below) have an additional important action. They are among the most powerful T cell mitogens known, acting at picomolar concentrations. They bind to MHC class II molecules on antigenpresenting cells and interact with the Vß chain of the T cell receptor (see Chs 6 and 7). This causes proliferation and cytokine release by the entire subset of T cells bearing that particular Vß chain-2-20% of all T cells are affected, whereas only 0.001-0.01% of T cells respond to a given regular antigen. This represents an important interference with a coordinated immune response, and the widespread polyclonal activation and cytokine release can be regarded as a microbial strategy, a 'diversion' of host immune defences. In addition, if the superantigen reacts with developing T cells (with the correct Vß chain) in the thymus, these cells are deleted. It seems probable, therefore, that this is a more important biological function of these toxins than the one responsible for the characteristic disease, which may be no more than an 'accidental' phenomenon. It turns out that similar molecules are formed by mycoplasma and by certain retroviruses (e.g. the Mis antigen of mouse mammary tumour virus).

Scarlet fever is one of the many conditions caused by streptococcal infection and may accompany a streptococcal sore throat or occasionally a streptococcal wound infection. There is a generalised erythematous rash together with fever and a sore throat. The rash is due to an erythrogenic toxin produced by certain strains of Streptococcus pyogenes. Erythrogenic toxin is a low molecular weight protein produced in a complex form with hyaluronic acid, which acts as a carrier. The protein consists of two parts. One is heat labile, carries the determinants of immunological specificity which give rise to three serotypes, A, B and C, and is responsible for primary toxicity manifestations, including pyrogenicity, low lethality, cytotoxic effects on cultured spleen macrophages, and suppression of the reticuloendothelial and immune systems. The second part is heat stable, antigenically common to types A, B and C, and is responsible for secondary toxicity; that is, hypersensitivity effects, including skin hyperreactivity (the basis of the rash), myocardial necrosis, enhancement of pyrogenicity and lethality, and enhanced host response to other injurious agents. Thus, it is not possible to define the biological activity of 'erythrogenic' toxins without considering the immunological state of the host. An individual may suffer no reddening of the skin upon intradermal injection of erythrogenic toxin (negative Dick test) because a high neutralising titre of antibody blocks the primary toxiphore or because of a lack of hypersensitivity to the common antigen.

Streptolysins O and S are dealt with in Ch. 4.

There is a plethora of toxins produced by numerous clostridial pathogens important in both human and veterinary medicine. The clostridial genus comprises a large number of toxigenic species, some of which are known to produce several toxins, extracellular enzymes, and other factors which are as yet recognised only by a letter of the Greek alphabet. All the α-toxins are dermonecrotic and the others have haemolytic, enzymatic properties. Toxin production in vitro is used as the basis of typing clostridia, and the picture is complex. No attempt will be made to describe every disease in man or animals associated with clostridia; only those are selected which best serve to illustrate the involvement of some recognisable toxins. In the case of sheep diseases-lamb dysentery, struck, enterotoxaemia, black disease, braxy, black quarter-some of the best evidence for implicating relevant toxins comes from field studies using multivalent vaccines (based on toxoids of these toxins).

Clostridium perfringens type C causes pig bei in man, essentially due to the production of ß-toxin. This is a rare disease in developed societies but a public health hazard in Papua New Guinea. Four factors are responsible: the ubiquity of C. perfringens type C in the soil and faeces of man and pigs; the relatively low immunogenicity of ß-toxoids in young children, the group most at risk; the high carbohydrate, low protein nature of the staple diet; and the sporadic consumption of large quantities of pork on occasions of celebration. The latter dietary change promotes a proliferation of clostridia in the intestine which may lead to intestinal gangrene and death. ß-Toxin damages the mucosa, reduces mobility of villi and causes more bacteria to become attached to the villi. More toxin is absorbed and the mucosa and underlying the intestinal wall become necrotic, leading to death in many cases. The influence of diet is additionally important in t h a t low protein diets cause decreased secretion of pancreatic proteolytic enzymes and sweet potato contains a trypsin inhibitor. These conditions promote the survival of ß-toxin which is highly sensitive to proteolytic inactivation. Immunisation with ß-toxoid preparations has dramatically lowered the incidence of this fatal disease in children.

Gas gangrene in man may be caused by several bacterial species separately or in concert. These include Clostridium perfringens type A, C. novyi types A and B, and C. septicum; C. perfringens and its a-toxin have already been discussed. Far less is known about C. novyi and C. septicum and their toxins in gas gangrene in man. Much more is known about the role these organisms play in diseases of animals, and multivalent vaccines confer a very high degree of immunity (particularly to sheep) against several clinically identifiable but separate diseases. A few of these diseases are described below.

Clostridium perfringens type B causes lamb dysentery. This is an acute, fatal disease of young lambs occurring during the first week of life and caused by absorption of toxin(s) generated by C. perfringens type B in the small intestine.

Clostridium perfringens type C causes struck in sheep, a disease occurring in the Romney marshes of Kent, but rare in other areas in the world. The pathological changes observed differ markedly from other enterotoxaemias and include enteritis.

Clostridium perfringens type D enterotoxaemia in sheep is another acute fatal disease. The most constant lesion is subendocardial haemorrhage around the mitral valve. C. perfringens type D e-toxin or its protoxin are recoverable from intestinal contents.

Clostridium novyi type B causes black disease of sheep or infectious necrotic hepatitis. This is an acute infectious disease of sheep (occasionally cattle) caused by the absorption of the α-toxin elaborated by the organism in necrotic foci in the liver, and is nearly always associated with invasion of the liver by immature liver flukes. How C novyi gets to the liver in the first place is not known but it is readily demonstrable in livers of normal sheep in areas where the disease is prevalent. Experimental reproduction in guinea-pigs of a similar disease is possible by the combined action of C. novyi spores and liver fluke infestation.

Clostridium novyi type D causes a rapidly fatal disease in cattle (redwater disease) similar to, and regarded by some as an atypical manifestation of, black disease. The characteristic lesions include jaundice, various haemorrhagic manifestations, and anaemic infarcts in the liver; active liver fluke infestation may or may not be present. In culture this organism produces ß-toxin, which explains the haemoglobinuria, but no a-toxin.

Clostridium septicum causes braxy in sheep. The role of C. septicum in this acute, fatal disease is assumed because of its association with the characteristic haemorrhagic inflammatory lesion in the abomasum. The disease has not been reproduced experimentally with C. septicum but can be prevented by immunisation with sterile toxoids derived from this organism.

Clostridium chauvoei causes black quarter in sheep and cattle. This is a gas gangrene type infection of muscles and associated connective tissues in cattle and sheep; C. chauvoei is also the causative agent of parturient gas gangrene in sheep. The initial stimulus which activates the infection in cattle is not known, since the disease is hardly ever associated with any overt wounding. Washed spores alone do not cause disease when injected, but do in conjunction with a tissue-necrotising agent. In sheep, wounding caused by parturition, castration, tailing, shearing, vaccination, as well as accidental damage will create a focus within which C. chauvoei can multiply.

There are numerous examples now known of synergistic reactions between toxins of the same or different species. Bacillus cereus makes a phosphatidyl choline-preferring phospholipase and a sphingomyelinase which are separately nontoxic; in concert they are haemolytic and termed cereolysin A-B. The examples of staphylococcal a-and δ-toxins and of staphylococcal TSST-1 and endotoxin, have already been alluded to above. Streptococcal erythrogenic toxins increase sensitivity to other streptococcal factors (for example, Streptolysin-O) and also to Gramnegative endotoxin. The susceptibility of rabbits to endotoxin is increased 100 000 times, and adult cynomolgus monkeys die within 24 h, when injection of low levels of steptococcal toxin is followed 3 h later by an otherwise sublethal dose of endotoxin. This may be because streptococcal toxin creates a state of hypersusceptibility to a wide variety of stressful agents. Perhaps, in view of the high incidence of exposure of man and his domestic animals to streptococci, we should actively investigate possible toxin-mediated synergy in mixed infections involving streptococci.

Another entirely different type of example is that of the increase in toxicities of staphylococcal a-and 7-toxins, diphtheria toxin and endotoxin for neonatal ferrets preinfected with influenza virus. Increases were 14-, 3-, 219-and 84-fold respectively. No increase in viral replication was observed. Neonates died suddenly without clinical symptoms as in human babies dying from sudden infant death syndrome (SIDS). Pathological examination showed inflammation of the upper respiratory tract, lung oedema and collapse, and early bronchopneumonia in animals receiving the dual challenge but not those receiving either toxin or virus on their own. Thus, some bacterial toxins in conjunction with influenza virus could be one of the several causes of SIDS.

Many fungi contain substances that are harmful when taken by mouth, and there are two diseases that result from the ingestion of food containing preformed fungal toxins. As with C. botulinum, the disease is caused without the need for infection. Aspergillus flavus infects ground nuts (monkey nuts) and produces a very powerful toxin (aflatoxin). Contaminated (badly stored) ground nuts used to prepare animal feeds caused the death of thousands of turkeys and pigs in the UK in 1960 and the survivors of intoxication nearly all developed liver cancer. Human disease has not yet been associated with this toxin. Clavicepspurpurae is a rust fungus affecting rye, and it produces toxins (ergotamine especially) that give rise to ergot poisoning when contaminated grain is eaten. Mushrooms and toadstools have long been recognised as sources of poisons and hallucinogens.

Unlike the toxins already discussed in this chapter, there is a group of toxins which are distinct structural components and are not released into the surrounding medium in any quantity except upon death and lysis of the bacteria. Several protein toxins (e.g. Clostridium difficile toxin A, tetanus toxin) are released during the decline phase of batch culture-probably on autolysis-and these are classified as exotoxins.

Here we deal with toxins which are known to comprise well-recognised structural entities which on a priori grounds must have key functions in the organism: they are found in the outer membranes of Gram-negative organisms. There are two chemically distinct types of toxin considered: lipopolysaccharide (endotoxin; LPS) and protein. The bulk of this section is taken up with endotoxin.

Many pathogenic organisms, however, are pathogenic by virtue of possessing various types of surface structure important in conferring virulence. These include, for example, adhesins which are important in colonising body surfaces or a variety of surface molecules (which may or may not be inside capsules) which render them resistant to phagocytosis. But the majority of adhesins and anti-phagocytic determinants are themselves nontoxic.

The Gram-negative bacterial cell wall is subject to considerable variations in both the composition of LPS and in the number and nature of the proteins found in the outer cell membrane. Apart from the examples given in Ch. 7 in relation to the gonococcus, such phenotypic variation in LPS has rarely been examined in the context of pathogenicity. However, the examination of cell-bound proteins of Yersinia pestis from organisms grown in vivo led to the discovery of a toxin lethal for mice and guinea-pigs.

Plague is one of the most deadly diseases of man and has, over several thousands of years, claimed millions of lives. In the fourteenth century, 'the black death' wiped out a quarter of the population of Europe before spreading through the Middle East and Asia. Fortunately, however, the last 60 years or so have seen a drastic decrease in outbreaks of plague, though the threat of another epidemic is still with us.

The causative organism of plague, Yersinia pestis, is primarily a parasite of rodents in which it is endemic in many areas of the world. Only when man comes into close proximity with infected rodents do outbreaks of human plague occur. The disease is spread from rat to rat and from rat to man by fleas. Rodents in the terminal stages of infection with Y. pestis suffer massive bacteraemia and so when a flea sucks the blood of an infected animal it swallows large numbers of bacteria. When the rodent eventually dies, the flea leaves the corpse and awaits a new host. In the meantime, however, Y. pestis multiplies rapidly in the alimentary tract of the insect, often completely blocking the proventriculus. The flea becomes voraciously hungry and feeds on any suitable host, which sooner or later is man. However, because the flea's alimentary canal is blocked by bacteria it cannot feed efficiently and usually succeeds in imbibing blood only to regurgitate it, now contaminated with Y. pestis, back into the wound. During the next 2-8 days, organisms become localised and multiply in the regional lymph nodes, causing considerable swelling. These swellings, which are often in the axillary and inguinal regions because the most common sites for flea-bites are the arms or legs, are referred to as primary buboes (Greek boubon = groin); hence the name bubonic plague. Secondary buboes may develop, followed by necrosis of the lymphoid tissue and vascular damage giving rise to haemorrhaging in many organs and tissues. These pathological changes are accompanied by prostration, high fever and delirium, followed in the terminal stages of the disease by shock and death. It is also possible for the disease to be transmitted from human to human by the aerosol route: exhaled organisms from heavily infected patients are capable of infecting others giving rise to pneumonic plague.

The principal features of human plague can be reproduced in guineapigs and mice. Monkeys show shock-like signs only during the terminal period of 6-10 h, when they become quiet, progressively weak, prostrate and hypothermic; for the previous 2-4 days infected animals are lively and vigorous. In the terminal stages blood pressure drops rapidly but there is no evidence of oligaemia (low blood volume) caused by haemorrhage, or of oedema, suggesting that vascular collapse must be associated with a vasodilatory factor(s), resulting in pooling of blood. In this respect monkeys differ from humans and guinea-pigs.

The symptoms of plague-high fever and vascular damage -are characteristic of intoxication with endotoxin. However, it is extremely unlikely that endotoxin alone is the main toxin involved in plague. It is much more likely to act in conjunction with one or more other potentially toxic fractions from Y. pestis.

Plague murine toxin is a protein which, although highly lethal for mice and rats, is relatively nontoxic for guinea-pigs, rabbits, dogs and monkeys. A completely separate guinea-pig toxin complex exists comprising at least two cell wall/membrane protein components, one of which will kill mice, although both are needed to kill guinea-pigs. However, the nature of the toxin or toxins of Y. pestis and their role in the human disease syndrome are still far from clear. The classical approach to such a problem is to prepare a specific toxoid and to determine whether injection of this confers immunity to the disease. Needless to say, the severity of human plague renders such experiments impossible. Until such questions can be answered, we are left to argue whether, in the context of plague, man resembles a mouse, a monkey or a guinea-pig.

Endoxotins are part of the outer membrane of Gram-negative bacteria. It has been known for many years that the cells (alive or dead) or cell extracts of a wide variety of Gram-negative bacteria are toxic to man and animals. The literature on this subject is vast, sometimes confusing and often controversial; here we can give no more than a brief outline. Some of the diseases in which endotoxin may play an important role include typhoid fever, tularaemia, plague and brucellosis, and a variety of hospital-acquired infections caused by opportunistic Gram-negative pathogens which include Escherichia coli, Proteus, Pseudomonas aeruginosa, Enterobacter, Serratia and Klebsiella. In addition, endotoxin has been intensively studied as a possible causative agent of shock arising from post-operative sepsis or other forms of traumatic injury in which the normal flora of the gut is often the source of endotoxin.

The toxins we have considered so far have been protein (or at least part protein) in nature but, in contrast, endotoxin is a complex lipopolysaccharide. It is also much more heat stable than protein toxins and much less easily toxoided. In addition to lethality, endotoxin displays a bewildering array of biological effects.

The complex nature of the multi-layered Gram-negative bacterial envelope is shown in Fig. 8 .12 (see also Fig. 4.3) . The outer membrane is composed of a bimolecular leaflet arrangement as are other membranes but has a different composition from the cytoplasmic membrane. The lipopolysaccharide (LPS) is unique in nature, only found in Gramnegative bacteria, and is, or contains within it, what we designate endotoxin. Immunoelectron microscopy indicates that LPS exists in the outer leaflet of the membrane and extends outward up to 300 nm; it is on, rather than in, the cell. Thus it is evident that the term endotoxin is a misnomer which derives from the era when toxins were considered to be either exotoxins, which were synthesised and secreted by the viable organism, or endotoxins, which were intracellular and released only upon lysis. Moreover, extraction with EDTA shows that approximately 50% of LPS is held noncovalently linked in the membrane. Extraction with a variety of different solvents yields material which is highly heterogenous and of apparent molecular weight 1-20 x 10 6 . However, treatment with pyridine or addition of detergents reduces the polydispersity. The endotoxic glycolipid from the rough mutant of Salmonella minnesota, R 595, has an M r of 5900 for the basic unit, from which complex aggregate structures are derived.

Lipopolysaccharide consists of three regions: polysaccharide side chains, core polysaccharide, and lipid A which consists of a di-glucosamine backbone to which long chain fatty acids are linked (Fig. 8.12 ). The relationship of this type of molecule to the outer membrane is also shown in Fig. 8 .12. The long chain fatty acids interdigitate between the phospholipids in the outer leaflet and may also be linked (or interact) with lipoproteins, which in turn may or may not be covalently anchored to the rigid peptidoglycan (PG). The polysaccharide side chains project outwards. This structure is not invariant. For example, many organisms when first isolated give rise to colonies with a smooth appearance on agar but on subculture produce colonies with a rough appearance. In general, 'smooth' strains of pathogenic species are more virulent than rough strains. This S->R conversion is accompanied by a loss of region I side chains, which contain the deoxy and dideoxy sugars found in these LPS complexes. In addition to these somewhat drastic changes involving loss of side chains, it is possible to induce major compositional changes by manipulating the growth rate of these organisms in a chemostat. Thus the LPS of Salmonella enteriditis, when grown with a mean generation time of 20 min is nearly totally deficient in tyvelose (a dideoxy sugar), possesses 85% of the galactose and 150% of the glucose contents of LPS obtained when the generation time is 50 min. These genotypic S organisms exhibit an R-phenotype in terms of their vastly reduced O-agglutinability (see below); such observations are potentially very important in the context of the in vivo phenotype and pathogenicity, since it is well known that the growth rate of Salmonella typhimurium in mice is 10-20 times lower than in vitro. Examples have already been given in Ch. 7 of changes in LPS structures in vivo in relation to Neisseria gonorrheae and Neisseria meningitidis.

The extent to which lipid A is common between different genera is uncertain, but it is not likely to vary tremendously. The core polysaccharide structure is the same or very similar within groups of the Enterobacteriaciae: thus polysaccharides from salmonellae are similar to each other, but differ from those of E. coli strains. However, within a group such as the salmonellae, there is a wide variation in the composition and detailed structures of the side chains, a fact which is exploited in the Kauffman-White scheme for classifying salmonellae, giving rise to several thousand serotypes. The side chains carry the O-somatic antigen specificities of which there are far more than can readily be accounted for on the basis of the known number of sugars involved in the basic repeating units. In the side chains are found a range of depxy and dideoxy sugars. The general principles governing the relationship between the various chemotypes and serotypes are now well understood; the multiplicity of antibody specificities evoked may be explained in terms of antibodies which can recognise different aspects of one three-dimensional structure.

Lipid A is the primary toxiphore, but the polysaccharide plays an important part in conferring solubility upon, and optimising the size of micellar aggregates of LPS, hence affecting biological activity. However, the immune status of the test animal may affect toxicity: as normal animals produce antibodies to the antigenic determinants on the surface of normal gut organisms (including O-somatic antigens), some of the biological effects of endotoxin may be mediated by hypersensitivity mechanisms.

The most powerful evidence that lipid A is the primary toxiphore comes from studies on smooth (S) and rough (R) mutants whose biosynthetic capabilities are blocked at various points. This established that neither the O-side chains, nor the core polysaccharide are necessary for endotoxicity. Pure lipid A can be made toxic by complexing it to a hydrophilic carrier like bovine serum albumin. One can make synthetic lipid A preparations which are toxic and assume space-filling configurations, which make it easy to see how they could fit into and be part of a bimolecular leaflet arrangement in the outer membrane.

The range of biological properties of endotoxin is quite bewildering and the mode(s) of action very complicated. Included among those effects which might play a role in Gram-negative bacterial infections, are abortion, pyrogenicity, tolerance (not immune tolerance), the Schwartzmann phenomenon, hypotension and shock, and lethality, but the precise part played by LPS in these phenomena in Gram-negative infections is far from clear. LPS causes the release of vasoactive substances, activates the alternative pathway of the complement cascade, and also activates factor XII (Hageman factor), the first step of the coagulation cascade, which sometimes results in disseminated intravascular coagulation (p. 258). Many, perhaps nearly all, the actions of LPS are due to the stimulation of cytokine release from macrophages and other cells. There is an effect on the circulation, leading ultimately to vascular collapse. The vascular regions most affected differ from species to species; in man and sheep the main changes are found in the lungs. LPS has powerful immunological actions, which is surely no accident; as well as activating the complement system, it induces IL-1 production and is a potent B cell mitogen. Man is one of the most sensitive of all species to the pyrogenic action of endotoxin. A dose of 2 n g per kg of body weight injected intravenously into man causes the release of an endogenous pyrogen (interleukin-1, see Glossary) and tumour necrosis factor (TNF) from macrophages, which act on the hypothalamus to give an elevation of body temperature within an hour. It is possible that the pyrogenic action of LPS helps to generate fever in Gram-negative bacterial infections, but LPS is not the only bacterial factor capable of inducing a febrile response.

In spite of all these toxic actions, there have been suggestions that some of the responses to LPS (by macrophages, polymorphs) could be advantageous to the host, possibly assisting in the recognition and destruction of bacteria. Could it be that host responses to LPS are, like the complement or the clotting systems, useful in moderation but harmful in excess? There are reports that when animals with less vigorous responses to LPS are infected they suffer fewer symptoms, but permit greater growth of bacteria.

Very large numbers of Gram-negative bacteria are normally present in the intestines (see Ch. 2), their continued death and exit in the faeces being balanced by multiplication in the lumen. There is a continuous, inevitable low-grade absorption of endotoxin from the intestine.* Absorbed (endogenous) endotoxin enters the portal circulation and is taken * In addition, various antigens are absorbed in small quantities from the intestine, and in normal individuals antibodies are formed against various food proteins and to some extent against resident intestinal bacteria (see Ch. 2). KupfFer cells remove any antigen-antibody complexes formed locally in the intestine and prevent them from entering the systemic circulation. up and degraded by reticuloendothelial cells, mainly Kupffer cells in the liver. Continuous exposure to endotoxin probably has profound effects on the immune system and on the histology of the intestinal mucosa, stimulating development of the immune system in the immature individual, but there are no obvious pathogenic consequences. Normal people have low levels of antibody to endotoxin as a result of this continuous exposure. The sick individual may be much more susceptible to endogenous endotoxin, perhaps because of defects in removal by Kupffer cells.

After trauma or after genito-urinary instrumentation endotoxin is detectable in peripheral blood by the Limulus test,* but this leads to no particular signs or symptoms. When large amounts of endotoxin enter the blood there are profound effects on blood vessels with peripheral vascular pooling, a drastic fall in blood pressure, collapse and sometimes death. Thus, if enough endotoxin enters the blood during massive Gramnegative bacterial sepsis, the vasomotor action of endotoxin becomes important and shock intervenes.! In experimental animals endotoxin also causes vasodilation and haemorrhage into the intestinal mucosa, and sometimes haemorrhage into the placenta with abortion, but these actions do not appear to be important in all Gram-negative bacterial infections.

To summarise, endotoxin, although studied so carefully and for so long, has not yet been shown to play a definitive role as a toxin in the pathogenesis of any infectious disease. However, there is a growing body of opinion that seeks to implicate endotoxin in the inflammatory response induced in meningitis caused by Gram-negative bacteria. But, in spite of its effects on various host defence systems including polymorphs, lymphocytes, macrophages, complement, and on endothelial cells and platelets, its overall role in infection is still not clear. It can, however, cause shock when Gram-negative bacteria invade the blood. It is for this reason that considerable effort in recent years has gone into the development of antilipid A antibodies for use as therapeutic agents to combat shock in such situations; the success rate is only partial and the expense enormous. For that reason several groups are seeking to exploit the wealth of chemical and biophysical information available on LPS in attempts to develop synthetic derivatives which would neutralise the biological activity of lipid A. We await the outcome of such research. However, the characteristics of the O-antigen polysaccharide are sometimes important in determining virulence (pp. 241-2). * A sensitive test for endotoxin based on the ability of endotoxin to induce gelation of a lysate obtained from the blood cells of the horseshoe crab, Limulus polyphemus.

Considerable space has been given to toxins because they are being intensively investigated as possible virulence determinants. The account illustrates the complexity of host-microbe interactions when analysed at the molecular level. Most toxins are liberated from the microbial cell and can be studied with greater facility than many of the more elusive determinants of pathogenicity. But remember that microbes that replicate inside host cells are less likely to form powerful toxins because they cannot afford to damage at too early a stage the cell in which they are multiplying. Thus, toxins are not prominent products in intracellular infections due to mycobacteria, Brucella, Rickettsiae, or Chlamydia, and viruses do not form toxins. Although a single molecule of a toxin like diphtheria toxin is enough to kill a cell, other toxins may do no more than impair cell function when present in sublethal concentrations. This can lead, for instance, to defective function in immune or phagocytic cells. Low concentrations of the streptococcal streptolysins* will inhibit leucocyte chemotaxis.

The ability to form toxins, whether encoded by plasmids or the microbial genome, is subject to selective forces. If toxin production puts a microorganism at a serious disadvantage it will tend to disappear. If it is advantageous it will be maintained, and will spread through the microbial population, just as the genetic changes that confer resistance to antimicrobial drugs are selected for when these drugs are widely used. It is therefore not unreasonable to ask how many of the well-known toxins are actually useful to the microbe as well as being important in causing disease in the host (Table 8. 3). However, microbes that multiply extracellularly must produce a variety of enzymes and other molecules involved in nutrition, adherence to substrate, and so on. In the case of free-living microbes these substances, as well as substances that damage or interfere with competing organisms, are of major importance. Probably they cannot all be discarded when the parasitic mode of life is adopted. Many will have a toxic action. Yet, for the infecting microbe, these substances remain as unfortunate necessities, of no particular advantage and perhaps a disadvantage, in the parasitic way of life.

In infectious diseases there is nearly always a certain amount of direct microbial damage to host tissues, as discussed above. Host cells are destroyed or blood vessels injured as a direct result of the action of * The streptolysins are responsible for ß-haemolysis. Most haemolysins will kill phagocytes. ) . Inflammatory materials are liberated from necrotic cells, whatever the cause of the necrosis. Also many bacteria themselves liberate inflammatory products and certain viruses cause living infected cells to release inflammatory mediators. Therefore it is not always clear how much of the inflammation is directly microbial rather than host in origin.* But inevitably the host (see Ch.

3) generates inflammatory and other tissue responses, and these responses sometimes account for the greater part of the tissue changes. Pathological changes can then be regarded as occurring indirectly as a result of these responses to the infection. Inflammation causes redness, swelling, pain and sometimes loss of function of the affected part (see Ch. 6) and is generally a major cause of the signs and symptoms of disease. Indirect damage attributable to the host immune response is discussed separately below. In most diseases direct and indirect types of damage both make a contribution to pathological changes, but in a given disease one or the other may be the most important.

In a staphylococcal abscess the bacteria produce inflammatory materials, but they also kill infiltrating polymorphs whose lysosomal enzymes are thereby liberated and induce further inflammation. This type of indirect nonimmunological damage is sometimes important in streptococcal infections. Virulent streptococci produce various toxins that damage phagocytes, and also bear on their surfaces substances that impede phagocytosis (see Ch. 4). Nevertheless, with the help of antibody, all streptococci are eventually phagocytosed and killed and the infection terminated. Unlike the staphylococci, however, killed group A streptococci pose a digestive problem for phagocytic cells. The peptidoglycan component of the streptococcal cell wall is very resistant to digestion by lysosomal enzymes. When streptococci are injected into the skin of a rabbit, for instance, streptococcal peptidoglycans persist in macrophages for as long as 146 days. Hence macrophages laden with indigestible streptococcal cell walls tend to accumulate in sites of infection. Lysosomal enzymes, including collagenase, leak from these macrophages, causing local destruction of collagen fibres and the connective tissue matrix. Macrophages secrete many other substances some of which may contribute to cell and tissue damage (see also p. 84). Many macrophages eventually die or form giant cells, sometimes giving rise to granulomatous lesions (see p. 292). In this way persistent streptococcal materials sometimes cause chronic inflammatory lesions in the infected host. An additional immunopathological contribution to the lesions is to be expected if the host is sensitised to peptidoglycan components. Other pathogenic microorganisms that are digested with difficulty by phagocytes include Listeria, Shigella, Candida albicans and, of course, Mycobacteria, but the importance of this in the pathogenesis of disease is not generally clear.

The expression of the immune response necessarily involves a certain amount of inflammation, cell infiltration, lymph node swelling, even tissue destruction, as described in Ch. 6. Such changes caused by the immune response are classed as immunopathological. Sometimes they are very severe, leading to serious disease or death, but at other times they play a minimal part in the pathogenesis of disease. With the possible exception of certain vertically transmitted virus infections and the transmissible 'prion' dementias (see Ch. 10), there are signs of an immune response in all infections. Therefore it is to be expected that there will nearly always be some contribution of the immune response to pathological changes.* Often the immunological contribution is small, but sometimes it forms a major part of the disease. For instance, in tuberculosis the pathological picture is dominated by the operation of a strong and persistent CMI response to the invading bacillus. In the classical tubercle a central zone of bacilli with large mononuclear and giant cells, often with some necrosis, is surrounded by fibroblasts and lymphocytes. Mononuclear infiltrations, giant cells and granulomatous lesions (see p. 292) are characteristic pathological features of tuberculosis. There are no recognised toxins formed by tubercle bacilli, and there seems to be no single antigen or other component that accounts for virulence. Bacterial glycolipids (e.g. 'cord factor'), resistance to H 2 0 2 (see pp. 80-81) and ability to utilise host Fe (see p. 352) have been correlated with pathogenicity, and inhibition of phagosome-lysosome fusion in macrophages (see pp. 94-95) by release of unidentified bacterial components would also contribute to pathogenicity. However, none of these factors is by itself absolutely necessary for virulence, which in such a complex, ancient parasite is likely to be multifactorial. As a gene library for M. tuberculosis is slowly built up there will be opportunities for clearer definition of virulence determinants. When macrophages are killed by intracellular mycobacteria the lysosomal enzymes and other materials released from the degenerating cell contribute to chronic inflammation as in the case of the streptococcal lesions referred to above. * A number of different microbial antigens are produced during most infections (see Ch. 6) and the possible immunological reactions are therefore numerous. For instance, at least 18 types of circulating malarial antigen are found in heavily infected individuals.

The mere enlargement of lymphoid organs during infectious diseases is a morphological change that can often be regarded as pathological. The lymph node swelling seen in glandular fever, for instance, is an immunopathological feature of the disease, and the same can be said of the striking enlargement of the spleen caused by chronic malaria and other infections in the condition known as tropical splenomegaly.

As often as not the relative importance of direct microbial damage as opposed to immune and nonimmune inflammatory reactions have not yet been determined, but the picture is clearer in most of the examples given below.

In one important human disease, pathological changes are certainly immunopathological in nature, but not enough is known about it to classify the type of reaction (see Table 8 .4). This disease is rheumatic fever, which follows group A streptococcal infections of the throat. It is the commonest form of heart disease in many developing countries. Antibodies formed against a streptococcal cell wall or membrane component also react with the patient's heart muscle or valves, and myocarditis develops a few weeks later. Many strains of streptococci have antigens that cross-react with the heart, and repeated infections with different streptococci cause recurrent attacks of rheumatic fever. There is genetic predisposition to the disease, based either on a particular antigen present in the heart of the patient or on a particular type of antibody response. Chorea, a disease of the central nervous system, is a rare complication of streptococcal infection and antistreptococcal antibodies have been shown to react with neurons in the caudate and subthalamic nuclei of the brain. A number of microorganisms have antigens similar to host tissue components (p. 173) so that in the course of responding immunologically to such infections the host is vulnerable to autoimmune damage (see ankylosing spondylitis, p. 200). The antibodies to host components such as DNA, IgG, myofibrils, erythrocytes etc. that are seen in trypanosomiasis, Mycoplasma pneumoniae, and EB virus infections appear to result from polyclonal activation of B cells (see p. 183). It is not clear how important these autoimmune responses are in pathogenesis, but they reflect fundamental disturbances in immunoregulation.

Four types of immunopathology can be distinguished according to the classification of allergic reactions by Coombs and Gell, and microbial immunopathology will be described under these headings (see Table 8 .4).

These depend on the reactions of antigens with reaginic (IgE) antibodies attached to mast cells, resulting in the release of histamine, leukotrienes (see p. 68) and heparin from mast cells, and the activation of serotonin and plasma kinins. If the antigen-antibody interaction takes place on a large enough scale in the tissue, the histamine that is released can give rise to anaphylactic shock, the exact features depending on the sensitivity and particular reaction of the species of animal to histamine. Guinea-pigs suffer from bronchospasm and asphyxia, and in man there are similar symptoms, sometimes with a fall in blood pressure and shock. This type of immunopathology, although accounting for anaphylactic reactions to horse serum or to penicillin, is not important in infectious diseases. When the antigen-IgE antibody interaction takes place at the body surface there are local inflammatory events, giving rise to urticaria in the skin, and hayfever or asthma in the respiratory tract. This local type of anaphylaxis may play a part in the pathogenesis of virus infections of the upper respiratory tract (e.g. common cold, respiratory syncytial virus infections of infants), or in skin rashes in infectious diseases.

Type 1 reactions are common in helminth infections perhaps because IgE antibodies have an important role in protection against these parasites. A dramatic Type 1 reaction can follow rupture of a hydatid cyst of Echinococcus granulosus (the dog tapeworm). Slow leakage of worm antigens means that mast cells are sensitised with specific IgE antibody, and the sudden release of antigen can cause life-threatening anaphylaxis. When the larvae of Ascaris lumbricoides pass through the lung on their journey from blood to intestine, they can give rise to IgEmediated respiratory symptoms, with infiltration of eosinophils.

Reactions of this type occur when antibody combines with antigen on the surface of a tissue cell, activates the complement sequence or triggers cytotoxicity by K cells (NK cells or phagocytes with Fc receptors). K (killer) cell cytolysis is referred to as antibody-dependent cellular cytotoxicity (ADCC). The antibody-coated cell is destroyed. As discussed in Ch. 6, the same reaction on the surface of a microorganism (e.g. enveloped virus) constitutes an important part of antimicrobial defences, often leading to the destruction of the microorganism. Cells infected with viruses and bearing viral antigens on their surface are destroyed in a similar way.

Clearly the antibody-mediated destruction of infected cells means tissue damage, and it perhaps accounts for some of the liver necrosis in hepatitis B, for instance, and probably in yellow fever. Infected cells can also be destroyed by sensitised lymphocytes or NK cells independently of antibody (see below).

In certain infections antibodies are formed against host erythrocytes and these cells are particularly sensitive to lysis. The haemolysis in malaria is caused by antibodies to parasite-derived antigens that have attached to red cells, rather than by autoantibodies to red cells themselves. In pneumonia due to Mycoplasma pneumoniae (atypical pneumonia), antibodies (cold agglutinins) are formed against normal human group O erythrocytes. Haemolytic anaemia is occasionally seen, and there is reticulocytosis (see Glossary) in 64% of patients. The lesions in the lungs are perhaps based on cell-mediated immunopathological reactions.

The combination of antibody with antigen is an important event, initiating inflammatory phenomena that are inevitably involved in the expression of the immune response. In the infected host, these inflammatory phenomena are most of the time of great antimicrobial value (see Ch. 6). But there are nevertheless immunopathological features of the infection, and immune complex reactions sometimes do a great deal of damage in the infected individual. The mechanisms by which antigenantibody reactions cause inflammation and tissue damage are outlined in Fig. 8.13 . IgA immune complexes are less harmful. Antigens absorbed from the intestine can combine locally with IgA antibody and the complex then enters the blood, to be filtered out in the liver and excreted harmlessly in bile (see p. 147).

When the antigen-antibody reaction takes place in extravascular tissues, there is inflammation and oedema with infiltration of polymorphs. If soluble antigen is injected intradermally into an individual with large amounts of circulating IgG antibody, the antigen-antibody reaction takes place in the walls of skin blood vessels, and causes an inflammatory response. The extravasating poly morphs degenerate and their lysosomal enzymes cause extensive vascular damage. This is the classical Arthus response. Antigen-antibody reactions in tissues are not usually as serious as this, and milder inflammatory sequelae are more common as in the case of allergic alveolitis (see below), or the red zone seen round the borders of a smallpox vaccination site after seven or eight days. In the latter example circulating antibodies pass through vessel walls, meet vaccinia virus antigen in the dermal tissues and an inflammatory response is generated. A similar mild response can be induced experimentally to cause a reaction known as cutaneous anaphylaxis (see Glossary), the test antigen being injected into the skin and reacting with blood-borne antibody. The resulting inflammation is detected by the visible local leakage of plasma proteins from blood vessels, circulating plasma proteins having been coloured by the intravenous injection of Evans' blue.

When the antigen-antibody reaction takes place in the blood to give circulating immune complexes, the sequelae depend to a large extent on size and on the relative proportions of antigen and antibody. If there is a large excess of antibody, each antigen molecule is covered with antibody and is removed rapidly by reticuloendothelial cells, which have receptors for the Fc portion of the antibody molecule (see Ch. 4). When equal amounts of antigen and antibody combine, lattice structures are produced, and these form large aggregates whose size ensures that they are also rapidly removed by reticuloendothelial cells. If, however, complexes are formed in antigen excess, the poorly coated antigen molecules are not removed by reticuloendothelial cells. They continue to circulate in the blood and have the opportunity to localise in small blood vessels elsewhere in the body. The mechanism is not clear, but complexes are deposited in the glomeruli of the kidneys, the choroid plexuses, joints and ciliary body of the eye. Factors may include local high blood pressure and turbulent flow (glomeruli), or filtering function of vessels involved (choroid plexus, ciliary body). In the glomeruli the complexes pass through the endothelial windows (Fig. 8 .14) and come to lie beneath the basement membrane. The smallest-sized complexes pass through the basement membrane and seem to enter the urine. This is probably the normal mechanism of disposal of such complexes from the body. Immune complexes are formed in many, perhaps most, acute infectious diseases. Microbial antigens commonly circulate in the blood in viral, bacterial, fungal, protozoal, rickettsial etc. infections. When the immune response has been generated and the first trickle of specific antibody enters the blood, immune complexes are formed in antigen excess. This is generally a transitional stage soon giving rise to antibody excess, as more and more antibody enters the blood and the infection is terminated. Sometimes the localisation of immune complexes and complement in kidney glomeruli is associated with a local inflammatory response.* There is an infiltration of polymorphs, swelling of the glomerular basement membrane, loss of albumin, even red blood cells, in the urine and the patient has acute glomerulonephritis. This is seen following streptococcal infections, mainly in children (see below). As complexes cease to be formed the changes are reversed, and complete recovery is the rule. Repeated attacks or persistent deposition of complexes leads to irreversible damage, often with proliferation of epithelial cells following the seepage of fibrin into the urinary space.

Under certain circumstances complexes continue to be formed in the blood and deposited subendothelially for long periods. This happens in certain persistent microbial infections in which microbial antigens are continuously released into the blood but antibody responses are only minimal or of poor quality (see below). Complexes are deposited in glomeruli over the course of weeks, months or even years. The normal mechanisms for removal are inadequate. The deposits, particularly larger complexes containing high molecular weight antigens or antibodies (IgM) are held up at the basement membrane and accumulate in the subendothelial space together with the complement components. As deposition continues, they gradually move through to the mesangial space ( Fig. 8.14) where they form larger aggregates. Mesangial cells, * See also footnote p. 255; cells in kidney glomeruli, in joint synovium and in choroid plexuses bear Fc or C3b receptors. This would favour localisation in these tissues.

of Cell and Tissue Damage one of whose functions is to deal with such materials, enlarge, multiply and extend into the subepithelial space. If these changes are gradual there are no inflammatory changes, but the structure of the basement membrane alters, allowing proteins to leak through into the urine. Later the filtering function of the glomerulus becomes progressively impaired. In the first place the glomerular capillary is narrowed by the mesangial cell intrusion. Also, the filtering area is itself blocked by the mesangial cell intrusion, by the accumulation of complexes (Fig. 8.14) , and by alterations in the structure of the basement membrane. The foot processes of epithelial cells tend to fuse and further interfere with filtration. The pathological processes continue, some glomeruli ceasing to produce urine, and the individual has chronic glomerulonephritis. Circulating immune complex deposition in joints leads to joint swelling and inflammation but in choroid plexuses there are no apparent pathological sequelae. Circulating immune complexes are also deposited in the walls of small blood vessels in the skin and elsewhere, where they may induce inflammatory changes.* The prodromal rashes seen in exanthematous virus infections and in hepatitis B are probably caused in this way. If the vascular changes are more marked they give rise to the condition called erythema nodosum, in which there are tender red nodules in the skin, with deposits of antigen, antibody and complement in vessel walls. Erythema nodosum is seen following streptococcal infections and during the treatment of patients with leprosy. When small arteries are severely affected, for instance in some patients with hepatitis B, this gives rise to periarteritis nodosa.

Immune complex glomerulonephritis occurs as an indirect immunopathological sequel to a variety of infections. First there are certain virus infections of animals. The antibodies formed in virus infections generally neutralise any free virus particles, thus terminating the infection (see Ch. 6), but the infection must persist if antigen is to continue to be released into the blood and immune complexes formed over long periods. Non-neutralising antibodies help promote virus persistence because they combine specifically with virus particles, fail to render them noninfectious, and at the same time block the action of any good neutralising antibodies that may be present. Immune complexes in antigen excess are formed in the blood when the persistent virus or its antigens circulates in the plasma and reacts with antibody which is present in relatively small amounts. Virus infections with these characteristics are * It is not clear how inflammation is caused. Complement activation would presumably take place while complexes were circulating in the blood. Perhaps the complexes bind more antibody after they have localised, or alternatively it is possible that free antigen circulates in the first place, localises, and later binds antibody to generate mediators of inflammation. included in Table 8 .5. In each instance complexes are deposited in kidney glomeruli and sometimes in other blood vessels as described above. In some there are few if any pathological changes (LDV and leukaemia viruses in mice) probably because there is a slow rate of immune complex deposition, whereas in others glomerulonephritis (LCM virus in mice, ADV in mink) or vasculitis (ADV in mink) is severe.

A persistent virus infection that induces a feeble immune response forms an ideal background for the development of immune complex glomerulonephritis, but there are no known viral examples in man. There are one or two other microorganisms that occasionally cause this type of glomerulonephritis and it is seen, for instance, in chronic quartan malaria and sometimes in infective endocarditis. In both these examples microbial antigens circulate in the blood for long periods. But immune complex deposition does not necessarily lead to the development of glomerulonephritis, and immune complexes are detectable in the glomeruli of most normal mice and monkeys. Even in persistent virus infections the rate of deposition may be too slow to cause pathological changes as with LDV and leukaemia virus infections of mice (see Table  8 .5). During the acute stage of hepatitis B in man, when antibodies are first formed against excess circulating viral antigen (hepatitis B surface antigen), immune complexes are formed and deposited in glomeruli. But the deposition is short-lived and there is no glomerulonephritis. Persistent carriers of the antigen do not generally develop glomerulonephritis, because their antibody is usually directed against the 'core' antigen (nephrotic syndrome in secondary syphilis) Unknown causative agents Man + + + + of chronic glomerulonephritis of the virus particle, rather than against the large amounts of circulating hepatitis B surface antigen.

Kidney failure in man is commonly due to chronic glomerulonephritis, and this is known to be mostly of the immune complex type, but the antigens, if they are microbial, have not yet been identified.

Immune complex glomerulonephritis occurs in man as an important complication of streptococcal infection, but this is usually acute in nature with inflammation of glomeruli, as referred to above. Antibodies formed against an unknown component of the streptococcus react with circulating streptococcal antigen, perhaps also with a circulating host antigen, and immune complexes are deposited in glomeruli. Streptococcal antibodies cross-reacting with the glomerular basement membrane may contribute to the picture. Deposition of complexes continues after the infection is terminated, and glomerulonephritis develops a week or two later. The streptococcal infection may be of the throat or skin, and Streptococcus pyogenes types 12 and 49 are frequently involved.

When certain antigens are inhaled by sensitised individuals and the antigen reaches the terminal divisions of the lung, there is a local antigen-antibody reaction with formation of immune complexes. The resulting inflammation and cell infiltration causes wheezing and respiratory distress, and the condition is called allergic alveolitis. Persistent inhalation of the specific antigen leads to chronic pathological changes with fibrosis and respiratory disease. Exposure to the antigen must be by inhalation; when the same antigen is injected intradermally, there is an Arthus type reaction (see p. 252).

There are a number of microorganisms that cause allergic alveolitis. Most of these are fungi. A disease called farmer's lung occurs in farm workers repeatedly exposed to mouldy hay containing the actinomycete Micromonospora faeni. Cows suffer from the same condition. A fungus contaminating the bark of the maple tree causes a similar disease (maple bark stripper's disease) in workers in the USA employed in the extraction of maple syrup. The mild respiratory symptoms occasionally reported after respiratory exposure of sensitised individuals to tuberculosis doubtless have the same immunopathological basis.

In addition to their local effects, antigen-antibody complexes generate systemic reactions. For instance, the fever that occurs at the end of the incubation period of many virus infections is probably attributable to a large-scale interaction of antibodies with viral antigen, although extensive CMI reactions can also cause fever. The febrile response is mediated by endogenous pyrogen (interleukin-1) and tumour necrosis factor (TNF) liberated from polymorphs and macrophages, as described on p. 298. Probably the characteristic subjective sensations of illness and some of the 'toxic' features of virus diseases are also caused by immune reactions and liberation of cytokines.

Systemic immune complex reactions taking place during infectious diseases very occasionally give rise to a serious condition known as disseminated intravascular coagulation. This is seen sometimes in severe generalised infections such as Gram-negative septicaemia, meningococcal septicaemia, plague, yellow fever and other haemorrhagic arthropod-borne virus diseases. Immune complex reactions activate the enzymes of the coagulation cascade ( Fig. 8.13 ), leading to histamine release and increased vascular permeability. Fibrin is formed and is deposited in blood vessels in the kidneys, lungs, adrenals and pituitary. This causes multiple thromboses with infarcts, and there are also scattered haemorrhages because of the depletion of platelets, prothrombin, fibrinogen etc. Systemic immune complex reactions were once thought to form the basis for dengue haemorrhagic fever. This disease is seen in parts of the world where dengue is endemic, individuals immune to one type of dengue becoming infected with a related strain of virus. They are not protected against the second virus, although it shows immunological cross-reactions with the first one. Indeed the dengue-specific antibodies enhance infection of susceptible mononuclear cells, so that larger amounts of viral antigen are produced (see p. 160). It was thought that after virus replication, viral antigens in the blood reacted massively with antibody to cause an often lethal disease with haemorrhages, shock and vascular collapse. However, it has proved difficult to demonstrate this pathophysiological sequence, and the role of circulating immune complexes and platelet depletion remains unclear. Perhaps in this and in some of the other viral haemorrhagic fevers the virus multiplies in capillary endothelial cells. Disease seems due to cytokines liberated from infected mononuclear cells.

Immune complex immunopathology is probable in various other infectious diseases. For instance, the occurrence of fever, polyarthritis, skin rashes and kidney damage (proteinuria) in meningococcal meningitis and gonococcal septicaemia indicates immune complex deposition. Circulating immune complexes are present in these conditions. Certain African arthropod-borne viruses with exotic names (Chikungunya, O'nyong-nyong) cause illnesses characterised by fever, arthralgia and itchy rashes, and this too sounds as if it is immune complex in origin. Immune complexes perhaps play a part in the oedema and vasculitis of trypanosomiasis and in the rashes of secondary syphilis.

Sensitive immunological techniques are available for the detection of circulating complexes and for the identification of the antigens and antibodies in deposited complexes. The full application of these techniques will perhaps solve the problem of the aetiology of chronic glomerulonephritis in man.

The mere expression of a CMI response involves inflammation, lymphocyte infiltration, macrophage accumulation and macrophage activation as described in Ch. 6, and can therefore by itself cause pathological changes. The CMI response to infection dominates the pathological picture in tuberculosis, with mononuclear infiltration, degeneration of parasitised macrophages, and the formation of giant cells as central features. These features of the tissue response result in the formation of granulomas (see Glossary) which reflect chronic infection and accompanying inflammation. There is a ding-dong battle as the host attempts to contain and control infection with a microorganism that is hard to eliminate. The granulomas represent chronic CMI responses to antigens released locally. Various other chronic microbial and parasitic diseases have granulomas as characteristic pathological features. These include chlamydial (lymphogranuloma inguinale), bacterial (syphilis, leprosy, actinomycosis), and fungal infections (coccidiomycosis). Antigens that are disposed of with difficulty in the body are more likely to be important inducers of granulomas. Thus, although mannan is the dominant antigen of Candida albicans, glucan is more resistant to breakdown in macrophages and is responsible for chronic inflammatory responses.

The lymphocytes and macrophages that accumulate in CMI responses also cause pathological changes by destroying host cells. Cells infected with viruses and bearing viral antigens on their surface are targets for CMI responses as described in Chs 6 and 9. Infected cells, even if they are perfectly healthy, are destroyed by the direct action of sensitised T lymphocytes, which are demonstrable in many viral infections. In spite of the fact that the in vitro test system so clearly displays the immunopathological potential of cytotoxic T cells, this is not easy to evaluate in the infected host. It may contribute to the tissue damage seen, for instance, in hepatitis B infection and in many herpes and pox virus infections. Antigens from Trypanosoma cruzi are known to be adsorbed to uninfected host cells, raising the possibility of autoimmune damage in Chagas' disease, caused by this parasite.* It is also becoming * Chagas' disease, common in Brazil, affects 12 million people, and is transmitted by blood-sucking bugs. After spreading through the body during the acute infection, the parasitaemia falls to a low level and there is no clinical disease. Years later a poorly understood chronic disease appears, involving heart and intestinal tract, which contain only small numbers of the parasite but show a loss of autonomic ganglion cells. An autoimmune mechanism is possible (see p. 173), because a monoclonal antibody to T. cruzi has been obtained that cross-reacts with mammalian neurons. clear that cells infected with certain protozoa (e.g. Theileria parva in bovine lymphocytes) have parasite antigens on their surface and are susceptible to this type of destruction. Little is known about intracellular bacteria.

The most clearly worked out example of type 4 (CMI) immunopathology is seen in LCM virus infection of adult mice. When virus is injected intracerebrally into adult mice it grows in the meninges, ependyma and choroid plexus epithelium, but the infected cells do not show the slightest sign of damage or dysfunction. After 7-10 days, however, the mouse develops severe meningitis with submeningeal and subependymal oedema, and dies. The illness can be completely prevented by adequate immunosuppression, and the lesions are attributable to the mouse's own vigorous CD8 + T cell response to infected cells. These cells present processed LCM viral peptides on their surface in conjunction with MHC I proteins, and sensitised CD8 + T cells, after entering the cerebrospinal fluid and encountering the infected cells, generate the inflammatory response and interference with normal neural function that cause the disease. The same cells destroy infected tissue cells in vitro, but tissue destruction is not a feature of the neurological disease. In this disease the CD8 + T cells probably act by liberating inflammatory cytokines. It may be noted that the brain is uniquely vulnerable to inflammation and oedema, as pointed out earlier in this chapter. The infected mouse shows the same type of lesions in scattered foci of infection in the liver and elsewhere, but they are not a cause of sickness or death. LCM infection of mice is a classical example of immunopathology in which death itself is entirely due to the cell-mediated immune response of the infected individual. This response, although apparently irrelevant and harmful, is nevertheless an 'attempt' to do the right thing. It has been shown that immune T cells effectively inhibit LCM viral growth in infected organs. However, a response that in most extraneural sites would be useful and appropriate turns out to be self-destructive when it takes place in the central nervous system.

Another type of T cell-mediated immune pathology is illustrated by influenza virus infection of the mouse. When inoculated intranasally, the virus infects the lungs and causes a fatal pneumonia in which the airspaces fill up with fluid and cells. The reaction is massive and the lungs almost double in weight. Effectively the animal drowns. The cause is an influx of virus-specific CD8 + T cells. Normally when an appropriate number of T cells had entered the lungs, the T cells would issue a feedback response to prevent such over-accumulation, but it is thought that influenza virus infects the T cells and inhibits this control process, so that the lungs are eventually overwhelmed. The action of the virus is subtle as it does not multiply in or kill the infected T cells, and it is presumed that it undergoes limited gene expression.

One human virus infection in which a strong CMI contribution to pathology seems probable is measles. Children with thymic aplasia show a general failure to develop T lymphocytes and cell-mediated immunity, but have normal antibody responses to most antigens. They suffer a fatal disease if they are infected with measles virus. Instead of the limited extent of virus growth and disease seen in the respiratory tract in normal children, there is inexorable multiplication of virus in the lung, in spite of antibody formation, giving rise to giant cell pneumonia. This indicates that the CMI response is essential for the control of virus growth. In addition there is a total absence of the typical measles rash, and this further indicates that the CMI response is also essential for the production of the skin lesions. There is evidence that cell-mediated immune responses also make a contribution to the rashes in poxvirus infections.

Sometimes in infectious diseases there are prominent pathological changes which are not attributable to the direct action of microbes or their toxins, nor to inflammation or immunopathology. The stress changes mediated by adrenal cortical hormones come into this category. Stress is a general term used to describe various noxious influences, and includes cold, heat, starvation, injury, psychological stress and infection. An infectious disease is an important stress, and corticosteroids are secreted in large amounts in severe infections (see also Ch. 11). They generally tend to inhibit the development of pathological changes, but also have pronounced effects on lymphoid tissues, causing thymic involution and lymphocyte destruction. These can be regarded as pathological changes caused by stress. It was the very small size of the thymus gland as seen in children dying with various diseases, especially infectious diseases, that for many years contributed to the neglect of this important organ, and delayed appreciation of its vital role in the development of the immune system.

Appreciation of the effects of stress on infectious diseases and the immune response in particular has led to the establishment of the science of neuroimmunology. Properly controlled experiments are difficult to mount but there is a persuasive and growing body of evidence which shows that the nervous system affects the functioning of the immune system. The pathways of this communication are still poorly understood. Work on Mycobacterium bovis grew out of observations from the turn of the century that stress appears to increase the death rate in children with TB. In one type of experiment mice were stressed by being kept in a restraining device where movement was virtually impossible. This resulted in the reduction of expression of MHC class II antigens on macrophages, which correlated with increased susceptibility to infection. Similarly stressing mice infected with influenza virus caused several immunosuppressive events including reduction of inflammatory cells in the lung, and decreased production of interleukin-2. Suppression of antibody responses is found in people suffering a type of stress familiar to students-examinations! The best responses to hepatitis B vaccine in students immunised on the third day of their examinations were found in those who reported the least stress. Finally, in a double blind trial at the Common Cold Research Unit in England with five different respiratory viruses, it was ascertained in human volunteers that stress gave a small but statistically significant increased likelihood of an individual developing clinical disease.

Pathological changes are sometimes caused in an even more indirect way as in the following example. Yellow fever is a virus infection transmitted by mosquitoes and in its severest form is characterised by devastating liver lesions. There is massive mid-zonal liver necrosis following the extensive growth of virus in liver cells, resulting in the jaundice that gives the disease its name. Destruction of the liver also leads to a decrease in the rate of formation of the blood coagulation factor, prothrombin, and infected human beings or monkeys show prolonged coagulation and bleeding times. Haemorrhagic phenomena are therefore characteristic of severe yellow fever, including haemorrhage into the stomach and intestine. In the stomach the appearance of blood is altered by acid, and the vomiting of altered blood gave yellow fever another of its names, 'black vomit disease'. Haemorrhagic phenomena in infectious diseases can be due to direct microbial damage to blood vessels, as in certain rickettsial infections (see p. 127) or in the virus infection responsible for haemorrhagic disease of deer. They may also be due to immunological damage to vessels as in the Arthus response or immune complex vasculitis, to any type of severe inflammation, and to the indirect mechanism illustrated above. Finally there are a few infectious diseases in which platelets are depleted, sometimes as a result of their combination with immune complexes plus complement, giving thrombocytopenia and a haemorrhagic tendency (see also disseminated intravascular coagulation, p. 258). Thrombocytopenic purpura is occasionally seen in congenital rubella and in certain other severe generalised infections.

Infection during pregnancy can lead to foetal damage or death not just because the foetus is infected (pp. 302-3), but also because of infection and damage to the placenta. This is another type of indirect pathological action. Placental damage may contribute to foetal death during rubella and cytomegalovirus infections in pregnant women.

Certain viruses undoubtedly cause tumours (leukaemia viruses, human papillomaviruses, several herpes viruses in animals) and this is to be regarded as a late pathological consequence of infection. As was discussed in Ch. 7 the tumour virus genome can be integrated into the host cell genome whether a tumour is produced or not, so that the virus becomes a part of the genetic constitution of the host. Sometimes the host cell is transformed by the virus and converted into a tumour cell, the virus either introducing a transforming gene into the cell, activating expression of a pre-existing cellular gene, or inactivating the cell's own fail-safe tumour suppressor gene. The transforming genes of DNA tumour viruses generally code for T antigens which are necessary for transformation, and the transforming genes of RNA tumour viruses are known as one genes.* Transformation has been extensively studied in vitro, and the features of the transformed cell described (changed surface and social activity, freedom from the usual growth restraints).

Simultaneous infection with two different microorganisms would be expected to occur at times, merely by chance, especially in children. On the other hand, a given infection generates antimicrobial responses such as as interferon production and macrophage activation which would make a second infection less likely. Dual infections are commonest when local defences have been damaged by the first invader. The pathological results are made much more severe because there is a second infectious agent present. This can be considered as another mechanism of pathogenicity. Classical instances involve the respiratory tract. The destruction of ciliated epithelium in the lung by viruses such as influenza or measles allows normally nonpathogenic resident bacteria of the nose and throat, such as the pneumococcus or Haemophilus influenzae, to invade the lung and cause secondary pneumonia. If these bacteria enter the lung under normal circumstances, they are destroyed by alveolar macrophages or removed by the mucociliary escalator. In at least one instance the initial virus infection appears to act by interfering with the function of alveolar macrophages. Mice infected with parainfluenza 1 (Sendai) virus show greatly increased susceptibility to infection with Haemophilus influenzae, and this is largely due to the fact that alveolar macrophages infected with virus show a poor ability to phagocytose and kill the bacteria. Specialised respiratory pathogens such as influenza, measles, parainfluenza or rhinoviruses damage the naso-* One genes (oncogenes) are also present in host cells, where they play a role in normal growth and differentiation, often coding for recognised growth factors (e.g. human platelet-derived growth factor). They can be activated and the cell transformed when tumour viruses with the necessary 'promoters' are brought into the cell. The one genes of the RNA tumour viruses themselves originate from cellular oncogenes which were taken up into the genome of infecting viruses during their evolutionary history.

pharyngeal mucosa and can lead in the same way to secondary bacterial infection, with nasal catarrh, sinusitis, otitis media or mastoiditis. The normal microbial flora of the mouth, nasopharynx or intestine are always ready to cause trouble if host resistance is lowered, but under normal circumstances they hinder rather than help other infecting microorganisms (see Ch. 2). One interesting example of exacerbation of infection occurs in mice dually infected with influenza virus and microorganisms such as Streptococcus aureus or Serratia marcescens. Under these conditions animals suffer a more severe viral infection. This results from the need to proteolytically cleave the viral haemagglutinin protein which is done by a cellular enzyme. If the appropriate protease is in short supply or lacking completely, virions are formed but they are not infectious. Under these circumstances the haemagglutinin can be cleaved extracellularly by microbial proteases with resulting increased amounts of infectious virus and disease.

As a final example of dual infections, microorganisms that cause immunosuppression can activate certain pre-existing chronic infections. In measles, for instance, there is a temporary general depression of CMI; tuberculin-positive individuals become tuberculin negative, and in patients with tuberculosis the disease is exacerbated. In AIDS (see p. 175) immunosuppression by HIV activates a variety of pre-existing persistent infections.

Diarrhoea deserves a separate section, since it is one of the commonest types of illness in developing countries and a major cause of death in childhood. Particularly in infants, who have a very high turnover of water relative to their size, the loss of fluid and salt soon leads to life-threatening illness. It is estimated that on a global scale diarrhoea is responsible for 3-6 million deaths per year in children under five years old. In villages in West Africa and Guatemala the average 2-3-year-old child has diarrhoea for about two months in each year.* Diarrhoea also interacts with malnutrition and can cause stunted growth, defective immune responses and susceptibility to other infections (pp. 343-5). Diarrhoea is also a common affliction of travellers from developed countries, * Diarrhoea on a massive scale is not always confined to developing countries. There was a major outbreak of Cryptosporidium infection in Milwaukee, USA, in 1993 with more than 400 000 cases; 285 of these were diagnosed in the laboratory and they suffered watery diarrhoea (mean 12 stools a day) for a mean of nine days. The small (4-5 μιη) oocysts, probably from cattle, had entered Lake Michigan, and then reached the community water supply because of inadequate filtration and coagulation treatment. and business deals, athletic successes and holiday pleasures can be forfeited on the toilet seats of foreign lands. The most reliable prophylaxis is to 'cook it, peel it, or forget it'. Most attacks of diarrhoea are self-limiting. Fluid and electrolyte replacement is a simple, highly effective, life-saving treatment that can be used without determining the cause of the diarrhoea. Oral rehydration therapy (ORF) means giving a suitable amount of salt and sugar in clean water and this is something that can be done by the mother.

Diarrhoea means the passage of liquid faeces,* or faeces that take the shape of the receptacle rather than have their own shape. This could arise because of increased rate of propulsion by intestinal muscles, giving less time for reabsorption of water in the large bowel, or because there was an increase in the amount of fluid held or produced in the intestine. In many types of infectious diarrhoea the exact mechanism is not known. Diarrhoea, on the one hand, can be regarded as a microbial device for promoting the shedding and spreading of the infection in the community, or on the other hand as a host device to hasten expulsion of the infectious agent. Diarrhoea is a superb mechanism for the dissemination of infected faeces (see p. 52) and there is no doubt that strains of microbes are selected for their diarrhoea-producing powers. The advantages to the host of prompt expulsion of the infectious agent was illustrated when volunteers infected with Shigella flexneri were given Lomotil, a drug that inhibits peristalsis. They were more likely to develop fever and had more difficulty in eliminating the pathogen.

In recent years significant strides have been made in our understanding of the pathophysiology of diarrhoeal disease. First, a quick resume of the normal structure and function of gut before attempting to understand the processes whereby it may be perturbed. The main function of the gut is the active inward transport of ions and nutrient solutes which is followed by the passive movement of water ( Fig. 8.15 ). The driving force is the Na + /K + ATPase situated in the basolateral membrane of enterocytes on the villus (Fig. 8.16 ) which maintains a low intracellular [Na" 1 "] thus creating the electrochemical gradient favourable for Na + entry and, a high regional [Na + ] in the intercellular spaces; Cl~ follows Na + . A similar situation exists in crypt cells: Na + /K + ATPase drives secretion. The key difference is the location of the carrier systems responsible for the facilitated entry of the actively transported species. In villus cells the carriers are present in the brush border, whereas in crypt cells they are located in the basal membrane: this is responsible for the vectorial aspects of ion/fluid traffic in villus/crypt assemblies. However, it is clear that several factors in addition to enterocytes are involved in * Liquid faeces are not abnormal in all species. The domestic cow experiences life-long diarrhoea, but presumably does not suffer from it. (a) Two methods of N a + cotransport are shown involving a glucose-linked symport and two coupled antiports; the latter results in the cotransport of Cl~. The coupled antiports are functionally linked via H + and HC0 3~, the relative concentrations of which are a reflection of metabolic activity. These processes occur within the same cells but are shown separately for clarity. The driving force for N a + uptake is the low Na + concentration maintained by the Na + /K + pump (ATPase) which creates the electrochemical gradient which promotes the inward movement of Na + ; Cl~ follows N a + by diffusion. Water is drawn osmotically across the epithelium paracellularly (i.e. across tight junctions) and/or transcellularly, the former pathway accounting for approximately 80% of fluid movement.

(b) Secretion is the result of the coupled entry of N a + and Cl~ across the basolateral membrane. N a + is recycled by the Na + /K + pump and Cl~ exits by diffusing down an electrochemical gradient and across the undifferentiated crypt cell apical membrane; N a + follows Cl~ and water follows passively. If by whatever means, the level of NaCl were to increase in such cells (as for example in rotavirus infection of neonatal mice, see below) then one could perceive how this could give rise to a Cl~ driven hypersecretion of water.

Note, (i) The driving force results from the same mechanism that powers absorption i.e. the N a + / K + pump located in the basolateral membrane; it is the location of the 'port' 'diffusion' systems that determines the vectorial aspects of ion movement, (ii) The tight junctions are less tight in the crypts than villi. (iii) The apical membrane of the crypt cell is undifferentiated and only acquires micro villi during ascent into villous regions, φ : Na + /K + pump; O: sym-, anti-port or diffusion channel. r e g u l a t i n g fluid t r a n s p o r t in t h e gut; t h e s e include t h e enteric n e r v o u s s y s t e m a n d t h e a n a t o m y of t h e microcirculation. T h e l a t t e r plays a profoundly i m p o r t a n t role in t h e u p t a k e of fluid. This is i l l u s t r a t e d in Fig. 8 .16, which shows t h e existence of zones of g r a d e d osmotic Fig. 8 .16 Schematic representation of a villus. Note the central arterial vessel (AV) which arborizes at the tip into a capillary bed drained by a subepithelial venous return (VR). Movement of sodium into VR creates a concentration gradient between VR and AV causing absorption of water from AV and surrounding tissue. This results in a progressive increase in the osmolarity of incoming blood moving into the tip region through to VR. Tip osmolarity is about three times higher than normal. This counter current system has been demonstrated in man and can be inferred in mice from the morphology of red blood cells which changes during ascent of the same vessel from base to tip regions of villi. The shaded areas indicate a vertical increase in osmolarity. Left crypt: represents normal physiological secretion. Right crypt: represents hyper secretion.

ENS, enteric nervous system, depicted schematically and not anatomically.

potential. At the tips of villi in adult human gut, osmolalities range from 700 to 800mOsmkg _ 1 H 2 0 , which would generate huge osmotic forces. Thus, current perceptions are that enterocytes are responsible for generating this gradient and the blood supply acts as a countercurrent multiplier which amplifies the gradient in a manner analogous to the loops of Henle in the kidney. The hypertonic zone has been demonstrated directly in whole villi of infant mice in terms of the changing morphology of erythrocytes: in the lower regions of villi they show characteristic discoid morphology, whereas in the upper region they are crenated, indicating a hyperosmotic environment. The hypertonicity is dissipated if the blood flow is too slow and washed out if too fast. It is the villus unit rather than enterocytes by themselves which is responsible for fluid uptake. Another consequence of the microcirculatory anatomy is that villus tip regions are relatively hypoxic. In addition, neonatal brush borders contain disaccharidases (principally lactase) which break down nonabsorbable disaccharides (e.g. lactose) into constituent absorbable monosaccharides. Villus tips and crypts are regarded as the anatomical sites of physiological absorption and secretion respectively. Fluid transport is a bidirectional process in the healthy animal with net absorption in health and net secretion in disease. The balance between absorption and secretion is poised at different points throughout the intestinal tract reflecting differences in both structure and function. Proximal small intestine is relatively leaky; in contrast the colon is a powerfully absorptive organ.

Finally, crypts are the principal sites of cell regeneration, replacing cells which migrate up the epithelial escalator. The epithelium is renewed in approximately 3-5 days. At villus tips senescent cells are shed. Diarrhoeal disease can result from interference with almost any one, or combination of these systems. Diarrhoea-producing microbes are listed in Table 8 .6. Some examples are considered in detail below. 

Rotaviruses are known to invade intestinal epithelial cells and cause diarrhoea in man, foals, dogs, pigs, mice etc. Extensive multiplication takes place and very large amounts of virus (10 11 particles g _ 1 ) are shed in faeces. The conventional wisdom is that tips of villi especially are affected, leading to reduced absorption of fluid from the lumen. In addition destruction of enterocytes leads to a loss in lactase resulting in an accumulation of lactose in the gut causing an osmotic flux of fluid into the intestine. A major study of rota virus-induced diarrhoea in neonatal mice provides a different model of this important disease of children. Oral infection of baby mice with a murine strain of rotavirus resulted in virus replication in enterocytes of the small intestine. Before there had been detectable virus replication, villi became ischaemic, enterocytes severely damaged and villi shortened. In gut enterocytes this virus was demonstrably not cytopathic; the cell damage and villus responses were almost identical to that seen in experimentally induced ischaemia in physiological experiments. At the peak of virus replication, which was coincident with maximum shortening of villi, the gut was still absorptive, glucose transport intact and sufficient lactase activity remained to deal with the normal lactose load delivered to the stomach. After this the gut became secretory and diarrhoea clinically evident. Villi were rapidly regenerated, but hyperaemia and diarrhoea persisted until the regeneration of the hypertonic zones as judged by the morphology of the red blood cells. Peak diarrhoea coincided with the resynthesis of new villi (as judged by thymidine kinase levels, increased mitotic activity and morphometric analyses), and increased levels of intracellular NaCl in the zone of cells where cell division was most active. Transiently high levels of NaCl have also been observed in dividing cells in culture. Since ischaemic and hyperaemic villi were occasionally seen in control villi it may well be the case that the pathological changes reflect an exaggerated synchronised response of basic circulatory control mechanisms, which are part of the normal homeostatic mechanisms of villus physiology. Thus, the pathophysiology of rotavirus-induced diarrhoea in neonatal mice may be summarised as follows: the reduction of red blood cells flowing through villi in the early stages of infection instigates hypoxia and hence atrophy of villi. The degree of atrophy in this infection is not associated with lack of absorption. The ensuing villus repair processes induce hypersecretion. The increase in blood flow throughout the remaining course of the infection reduces the hypertonicity of villi which impairs water absorption and thus prolongs diarrhoea. The preceding description of the self-limiting diarrhoea induced by rotavirus in neonatal mice is probably applicable to many diarrhoeas. However, the observed pathology may be different according to age, species, or the inducing pathogen. For example, in rotavirus-infected lambs, villus atrophy and crypt hypertrophy occur (the latter indicative of crypt cell division) but as in mice, infected lambs are not lactose intolerant.

In rotavirus-infected piglets, crypt hypertrophy occurs but villus atrophy is severe, the animals are lactose intolerant and mortality high; a similar situation exists for the coronavirus, transmissible gastroenteritis (TGE) virus. The latter has often been used as the model for infantile diarrhoea but the question is whether human infants are more like piglets or lambs. Clinical studies have shown that recovery from mild, acute gastroenteritis of rota virus origin occurs within two weeks irrespective of the carbohydrate ingested. Clearly, the severity of disease and the clinical outcome will depend on the extent of Vertical' villus/ crypt involvement and the regions of intestine infected. When villus erosion is severe, then lactose may cause an 'osmotic' purge or be fermented by intestinal bacteria to short chain fatty acids which stimulate secretion in the colon. Astroviruses, Norwalk virus, certain coronaviruses, certain adenoviruses and probably toroviruses all cause gastroenteritic disease by infecting enterocytes. However, parvoviruses cause severe intestinal disease in dogs by virtue of their predilection for the mitotically active crypt cells; this causes the near complete erosion of villi similar to that seen after exposure to sublethal doses of irradiation.

The classic paradigm for bacterial watery diarrhoea is cholera caused by Vibrio cholerae in the small intestine. V. cholerae attaches to enterocytes of the proximal small bowel and is capable of producing at least three toxins: classical cholera toxin (CT); a toxin which disrupts the zonula occludens tight junction (designated ZOT); and another less well defined, auxilliary cholera enterotoxin (designated ACE). As a result of toxic action, water and electrolytes are lost through the intact epithelial cells into the small intestine. As the multiplying bacteria increase in numbers and more and more epithelial cells are affected, the absorptive capacity of the colon is overwhelmed and there is profuse watery diarrhoea, as much as l l h -1 in severe cases.* The massive loss of isotonic fluid with excess of sodium bicarbonate and potassium leads to hypovolaemic shock, acidosis and haemoconcentration. Anuria develops, and the collapsed, lethargic patient may die in 12-24 h. Lives are saved by replacing the lost water and salts. The effect of toxin on an intestinal epithelial cell is long lasting, but the patient recovers as affected cells are shed and replaced in the normal fashion. The infection is particularly severe in children who easily develop low levels of plasma potassium. However, on a global scale this greatly feared disease, cholera, is only responsible for less than 1% of the total deaths due to diarrhoea.

Administration of as little as 5 μg of purified CT alone to a healthy volunteer will reproduce the hugely dramatic, potentially lethal, purge of near isotonic fluid emphasising the central importance of CT in the causation of disease. The effect of ZOT has been demonstrated by electron microscopy in human biopsies, but the relative importance of ZOT and ACE in human disease has not been quantified. How does CT work? As already outlined (pp. 216-19) the A subunit of this toxin is an ADPribosyl transferase whose target is the a s subunit of the regulatory G protein which governs the expression of adenylate cyclase and hence the production of cAMP. This in turn results in the perturbation of the transport systems leading to a net efflux of ions and hence water. Things are slightly more complicated, because elevation of cAMP occurs in villus tip cells (hence affecting absorption) but not in crypt cells. Probably a signal is generated in intoxicated tip enterocytes which is transmitted to the crypt region by the ENS causing release of neurotransmitters (e.g. serotonin, acetyl choline, and vasoactive peptide) which act on enterochromaffin (secretory) cells and on intestinal smooth muscle thereby augmenting crypt secretion and increasing peristalsis. Studies have been made of sequential histological changes in human biopsies, taken from a series of patients from the onset of symptoms to recovery. It was clear that fluid secretion was not the result of massive desquamation of the epithelium. Histological changes included 'engorgement of the capillaries' (i.e. interference with the blood supply), 'dilation of lacteals, vacuolation of enterocytes, exhaustion atrophy of enterocytes' (remember this is a noninvasive pathogen), 'accelerated shedding of cells, and increased mitotic activity'. Most of these changes are similar to those described above for experimental murine rotavirus at post-peak diarrhoea and it seems that the range of histological reactions that can occur in the small intestine are limited so that qualitatively the picture is similar in different diarrhoeas.

Despite the undoubted importance of CT in the causation of the disease, and the potent antigenicity of CT, it is now recognised that protective immunity is very largely antibacterial. It is stopping effective colonisation which is important rather than neutralisation of the toxin. This has been partially achieved by using killed whole cell vaccines. Several attempts have been made in the laboratory to genetically manipulate virulent strains (in practice this means deleting or inactivating the known toxin genes) such that the attenuated strain will colonise the gut and stimulate local immune responses and thereby prevent colonisation of the gut by virulent strains. To date, attenuated strains have been developed which fulfil these criteria but induce a mild transient diarrhoea which has prevented their adoption into vaccination programmes. However the recent emergence of V. cholerae strain 0139 (a new, third type) in India and Bangladesh is a reminder that today's solution may not be adequate for tomorrow's problems. Current vaccines are ineffective against this new strain.

The picture withi£. coli is complicated since there are several biotypes of this pathogen which include: ETEC (enterotoxigenic E. coli, the principal cause of travellers' diarrhoea), EPEC (enteropathogenic E. coli), EIEC (enteroinvasive E. coli), and EHEC (enterohaemorrhagic E. coli) variants (see Ch. 2). The situation with ETEC is very similar to V. cholerae as far as heat labile toxin (LT, which is very closely related to CT) is concerned. It is additionally complicated by the fact that some strains of ETEC also make heat stable toxins (STs), which are nonantigenic low molecular weight peptides which activate membrane-bound guanylate cyclase which causes the production of cGMP which is the functional equivalent of cAMP. The 'classical' EPEC lesion is the formation of a characteristic pedestal-like lesion in the brush border of enterocytes with only limited invasion of the mucosa. Just how it causes diarrhoea is not wholly clear. EIEC is almost indistinguishable from Shigella, and causes similar disease. EHEC strains, which belong mainly (but not exclusively) to serogroup 0157:H7, cause a spectrum of intestinal disease ranging from asymptomatic carriage through mild diarrhoea to severe haemorrhagic colitis. They also cause extraintestinal infections such as haemolytic uraemic syndrome (HUS). The initial lesion seen in the gut is similar to that caused by EPEC but EHEC do not remain localised there: they multiply in the lamina propria and glandular crypts. While there is no conclusive proof of the involvement of SLT in the pathogenesis of EHEC diarrhoea there is a correlation between SLT production and the severity of diarrhoeal disease or HUS caused by this organism. Salmonella spp. cause acute gastroenteritis and systemic typhoid disease in man, as well as other important infections in domestic animals. Virulence in Salmonella, especially in their natural hosts (see Table 8 .7, footnote) isxomplex and mediated by numerous genes, some of which are present in otherwise nonpathogenic organisms such as E. coli. Presumably all these bacteria share nutritional and environmental problems, whether inside or outside a host, that call for such genes. Those restricted to Salmonella are more likely to be critical in vivo. The clinical features of Salmonella-induced diarrhoea (gastroenteritis) in humans and systemic typhoid infections are quite different. For example, gastroenteritis may follow 8-36 h after ingestion of contaminated food, whereas typhoid follows an incubation period of 10-20 days. Diarrhoea, (which is usually watery, but may be severe, and sometimes bloody) is the predominating feature of gastroenteritis, whereas in adults, constipation is seen in the early clinical stages of typhoid; diarrhoea may occur much later. Although fever may occur in gastroenteritis, in typhoid this may be so severe as to cause delirium. Current perceptions are that gastroenteritis results from the initial interactions of Salmonella (any one of many serotypes but frequently typhimurium or enteriditis, and/or their products) with the gut mucosa, whereas typhoid fever is produced by S. typhi organisms which translocate the mucosa, survive within macrophages, multiply and release endotoxin which triggers the highly complex endotoxin cascade. Gastroenteritis is usually self-limiting, whereas in untreated typhoid mortality can be as high as 10%. The pathophysiology of fluid secretion caused by Salmonellae is highly complex and as yet incompletely understood. The only two biological parameters which have been shown to correlate with induction of fluid secretion in a rabbit model (the best available small laboratory animal for human gastroenteritis) is the ability to invade the gut mucosa, and the induction of a leucocyte (mainly polymorphonuclear cell) influx into the mucosa and lumen of the gut. Fluid secretion does not correlate per se with the ability to make an enterotoxin: avirulent as well as virulent strains make a cholera-like toxin though neither appear to release the toxin readily. The latter is therefore either not primarily involved, or is involved only after release from bacteria by means other than normal protein secretion mechanisms. These questions are as yet unresolved and the reader is referred to the reference list for a fuller discussion of these complex matters.

Currently, the major diarrhoeagenic pathogen of the small intestine in the developed world is Campylobacter jejuni. The latter are bacteria present in wild birds, chickens and in the faeces of up to 10% of healthy cows in the UK. Milk becomes contaminated and those who drink raw (unpasteurised) infected milk develop diarrhoea, sometimes with dysentery (blood and pus in the stools) and fever. The incubation period is 2-7 days and diarrhoea occurs after bacterial replication in the upper small intestine (jejunum). However, we are only at the beginning of understanding the detailed mechanisms and determinants responsible for disease causation by this pathogen.

Giardia lamblia is a noninvasive protozoan pathogen of the human small bowel which causes a spectrum of infection ranging from asymptomatic carriage through acute watery diarrhoea to chronic diarrhoea and malabsorption. Giardia has a simple life cycle, existing in two forms: the multiplying trophozoite which infects mammalian hosts to cause disease and the environmentally resistant cyst. After ingestion, excystation is triggered by the pH in the stomach, and the trophozoite colonises the small bowel; the putative adherence factors have been described in Ch. 2. The precise mechanisms involved in subsequent stages responsible for diarrhoea, malabsorption and cystation are less well understood.

Entamoeba histolytica causes lysis of target cells apparently by direct contact with the cell membrane. This pathogen produces under in vitro conditions a spectacular array of potential (but as yet unproven) virulence determinants including: proteases that round up cells, poreforming proteins, collagenases and oligosaccharidases and neurotransmitter-like compounds; the latter can induce intestinal fluid secretion. Some of these factors have been implicated as the determinants responsible for liver abscess formation.

Not all diarrhoeal disease is the result of small bowel dysfunction. Two important pathogens of the colon will be considered-Clostridium difficile and Shigella dysenteriae. C. difficile causes a spectrum of disease Varied Hepatitis a There are more than 1000 serotypes of Salmonella, distinct from Salmonella typhi and Salmonella paratyphi. They are primarily parasites of animals, ranging from pythons to elephants, and their importance for man is their great tendency to colonise domestic animals. Pigs and poultry are commonly affected, and human disease follows the consumption of contaminated meat or eggs. b Other campylobacters cause sepsis, abortion and enteritis in animals. ranging from asymptomatic carriage through mild antibiotic-associated diarrhoea (AAD) to fatal pseudomembranous colitis (PMC). Normal flora play a major part-by means which are still not completely understood-in suppressing outgrowth of resident C. difficile spores. Disruption of the normal flora, by antibiotic treatment for example, results in vegetative growth and production of several toxins of which C. difficile toxin A seems to be the most important in disease. This toxin is responsible for the secretory and inflammatory response in the intestine, and epithelial disruption. It probably acts by disrupting tight junctions between cells, and by causing chemotactic and other responses in polymorphs. Diarrhoea could be due to the destruction of the 'absorptive epithelium' and failure to cope with the normal fluid load delivered by the small intestine, or an active type of secretion caused by replacement of the damaged epithelium, or both.

The second example is that of dysentery caused by Shigellae spp., in particular S. dysenteriae and S. flexneri. We have already dealt with the capacity of Shigellae to invade (Ch. 2). Invasiveness is a vital part of the virulence armoury of Shigellae spp.; non-invasive mutants are a virulent. The pathogen invades the gut via M cells in Peyer's patches and thence adjacent microvillus-bearing colonocytes. What is the role played by Shiga toxin? Mutants of S. dysenteriae devoid of the ability to make Shiga toxin still retained the ability to cause lethal fulminant dysentery (low volume fluid production, pus cells, mucus) in macaque monkeys, and they invaded, multiplied within and rapidly killed Hela cells. The major difference between the mutants and the toxin-positive wild type was that the latter caused a more severe disease with more haemorrhage, giving rise to blood in the stools, and, in some instances greater destruction of mucosal tissues. Evidently the production of Shiga toxin is not a sine qua non for causing bacillary dysentery: it exacerbates the disease.

Although much research has been focused on toxins, their mode of action, and their role in disease, it is useful to compare different types of intestinal infection and to refer to the concept of food poisoning. Types of intestinal infection are set out in Table 8 .7. Food poisoning is a loosely used term, and usually refers to illnesses caused by preformed toxins in food, or sometimes to illnesses that come on within a day or so after eating contaminated food. Food may be contaminated with plant poisons, fungal poisons (e.g. poisoning due to Amanita phalloides), fish poisons,* heavy metals, as well as with bacterial toxins or bacteria.

Sourcebook of Bacterial Toxins

The virology and immunology of lymphocytic choriomeningitis virus infection

* Ingestion of scombroid fish (mackerel etc.) containing large amounts of histamine or similar substances leads to headache, flushing, nausea and vomiting within an hour

Response of man to infection with Vibrio cholerae. 1. Clinical, serologic and bacteriologic responses to a known inoculum

Epitopes of streptococcal M proteins shared with cardiac myosin

Bacterial Infections of Respiratory and Gastrointestinal Mucosae

Pathogenesis and immunology of Treponema pallidum

T-lymphocyte stimulation by microbial antigens

Lessons from diarrhoeal diseases, demography to molecular pharmacology

Clostridium botulinum toxins: a general review of involvement in disease, structure, mode of action and preparation for clinical use

Disseminated intravascular coagulation: a review

Campylobacterpyloridis and gastritis: association with intercellular spaces and adaptation to an environment of mucus as important factors in colonization of the gastric epithelium

Typhoid fever: pathogenesis and Immunologie control

Comparison of the alpha-toxin genes of Clostridium perfringens type A and C strains: evidence for extragenic regulation of transcription

Pathogenic mechanism ofNeisseriagonorrhoeae: observations on damage to human fallopian tubes in organ cultures by gonococci of colony Type I or Type 4

Rift Valley Fever virus in mice VI: Histological changes in the liver in relation to virus multiplication

Viral aetiology of diseases of obscure origin

Campylobacter jejuni. Current status and future trends

Cytolytic pore-forming proteins and peptides: is there a common structural motif?

Treatment of spasmodic torticollis with local injections of botulinum toxin

Pathogenesis of diseases caused by Entamoeba histolytica: studies of adherence, secreted toxins and contactdependent cytolysis

Pituitary dwarfism in mice persistently infected with lymphocytic choriomeningitis virus

Tetanus and botulinum-B neurotoxins block neurotransmitter release by proteolytic cleavage of synaptobrevin

Tetanus toxin is a zinc protein and its inhibition of neurotransmitter release and protease activity depend on zinc

Experimental Salmonella typhimurium induced gastroenteritis

Bacterial phospholipases C

The development of respiratory syncytial virus-specific IgE and the release of histamine in naso-pharyngeal secretions after infection

Immune complexes in human diseases