key: cord-0039041-ksz8clc1
authors: Harris, J.R.
title: The ultrastructure of multinucleate giant cells
date: 2002-11-11
journal: Micron
DOI: 10.1016/0968-4328(93)90070-h
sha: 9305b53bc46cb7fbf8d7cb0abffac31b5f97b2df
doc_id: 39041
cord_uid: ksz8clc1

A survey of the available ultrastructural data on physiologically and pathologically occurring and virally-induced multinucleate giant cells (MNGCs) is presented. Emphasis is initially placed upon the bone osteoclast, the skeletal muscle myotube and the placental syncytiotrophoblast. The widespread occurence of MNGCs in a range of pathological situations is discussed, with emphasis upon the broad involvement of the macrophage in inflammatory responses. Many viruses produce cell fusion in vivo and in vitro when cell cultures are infected. Several examples are given. A clear distinction is drawn between viral fusion from “without” and viral fusion from “within” the cell. The cytopathic effect (CPE) of the animal and human retroviruses is discussed in considerable detail. The in vivo and in vitro formation of lymphocytic and macrophage MNGCs by HIV-1 is given extensive coverage. The possible significance of the presence of brain MNGCs of macrophage/microglial origin as a cellular feature of AIDS dementia is discussed. A new hypothesis is advanced relating to the possible role of endogenous C-type retrovirus in the physiological fusion of the invasive placental cytotrophoblasts to create the syncytiotrophoblast. The evolutionary and developmental significance of such an event in relation to the evolution of the placental mammals is discussed. The possible importance of MNGC formation in the depletion of the CD4(+) population of T-lymphocytes in vivo in the clinical progression of the AIDS-related complex and AIDS is related to the potent fusogenic effect of HIV-1 when cell cultures are infected.

Interest in viral adhesion to cells has been in the forefront of virological studies for many years, stemming from the widespread haemagglutinating activity of viruses (Howe and Lee, 1972) and the use of Sendai virus for the study of membrane fusion, using both mammalian and avian erythrocytes and nucleated cells (Hosaka, 1988; Ringertz and Savage, 1976) . Extensive biochemical studies have been performed using chemical cell fusion agents, such as polyethylene glycol (PEG) (Kersting et al., 1989; Lucy and Ahkong, 1989; Wang et al., 1982) and lignin derivatives (Sorimachi et al., 1990) , and also by the electrofusion procedure (Gallagher et al., 1990; Zheng and Chang, 1991) , but these approaches will not be dealt with within this review. Although many studies have been directed towards the production of viable binucleate heterokaryons (Harris, 1970) it is clear that multinucleate giant cells (MNGCs) will also be formed, but that these are likely to be lost during subsequent cell culture since they do not remain viable for long periods of time.

In normal tissues there are very few examples of naturally occurring syncytia or MNGCs. This is undoubtedly due to the extremely specialized histological and physiological conditions that in general would have to be created to maintain their viability. Cardiac muscle is not a true syncytium, although it is considered to be a functional syncytium, with its highly ionically permeable gap junctions providing electrical communication between the cells and intercalated discs providing tissue strength. Multinucleate skeletal muscle myotubes, on the other hand, are true syncytia derived from the developmental fusion of myoblasts, leaving the nuclei spaced longitudinally and somewhat flattened, just beneath the myolemma. The cyst-like insect spermatocyte syncytium (Philips, 1970) may not be a true syncytium, whereas it is accepted that in mammalian spermatogenesis true syncytial clones are present during the development of the spermatagonia and during meiotic division yielding the spermatocytes and spermatids (Dym and Fawcett, 1971) .

Similarly, the maturation of a functional group of monocyte/macrophages designated to become osteoclasts is believed to occur due to controlled fusion events (Pierce et al., 1991; Marks and Popoff, 1988; Udagawa et al., 1990 ). An interesting situation exists in insect eggs, which has been most clearly documented for Drosophila. The zygote nucleus undergoes several mitotic divisions within the centre of the egg, without separate cells being formed. This is followed by the progressive migration of nuclei within the ooplasm to the surface of the egg, where mitosis continues in the absence of cytokinesis. The cellular structure so formed is termed the syncytial blastoderm and it ultimately becomes the cellular blastoderm, due to infolding of the oocyte membrane between the monolayer of surface nuclei. The formation and lifespan of the placental syncytiotrophoblast, by fusion of the embryonic precursor cells of the developing mammalian embryo, termed the primitive syncytial trophoblastic cells and later the placental cytotrophoblasts or Langhans' cells, presents an extremely specialized situation (Jones and Fox, 1991) . The uniquely prolific trophoblastic cells surround the blastocyst cavity and form a specialized region, the syncytial plate, the invasive precursor to the placenta. In all mammals the syncytiotrophoblast clearly has a limited lifespan, in all species correlating functionally with the length of pregnancy. Remarkably, at varying stages of the synctiotrophoblast lifespan it will be shown to bear several ultrastructural features (Section II.B) which parallel those of the large syncytia of HIVinduced lymphoid cells (Section IV.G and H) forming several days post infection in suspension cultures.

MNGCs occur in viL, o in a number of pathological disorders (Chambers, 1978) , particularly in tumours and diseases of viral origin. Significantly, the mononuclear macrophage appears to be involved in many instances of inflammatory lung infection, and is believed to account for the microglial MNGCs found within the brains of AIDS patients expressing neurological complications (AIDS encephalopathy/dementia) (Sharer et al., 1985; Takeya et al., 1991) . Whether or not syncytial formation in vivo is of importance in relation to the slow depletion of the CD4 + (T4) T-lymphocyte population during the progression of ARC/AIDS is not at the moment completely clear, but this remains a distinct possibility (Haseltine, 1990 (Haseltine, , 1991 . MNGCs have been detected in the lymph nodes (Beneviste et al., 1988) , lung, spleen and brain of AIDS patients and primates infected with SIV, but not in buffy coat lymphocytes. It is reasonable to expect that any bi-or tri-nucleate cells present in the circulation would be selectively removed by passage through the spleen and thereby go undetected in samples taken from the peripheral circulation (Harris et al., 1989; Haseltine, 1990) . Undoubtedly, because of the very low numbers of activated viral-producing cells during the earlier stages of AIDS/ARC, there will be a vast excess (some 1:500 to 1:50,000) of non-infected or unactivated latently infected circulatory and tissue CD4 + T-lymphocytes available for fusion.

It is widely accepted that the multinucleate osteoclast ( Fig. 1) , responsible for bone resorption, is derived from extra-skeletal granulocyte-macrophage mononuclear precursors, by a process of cell fusion rather than cell division without cytokinesis (Marks et al., 1983; Pierce et al., 1991) . Most of the evidence for this proposal comes from light microscopical data (Udagawa et al., 1990) , which although convincing when applied to in vitro culture systems is less so when derived from in vivo samples purporting to show bone osteoclast or odentoclast formation. That the osteoclast is derived from a blood-borne precursor monocytic cell has been conclusively shown by the production of osteoclast-like cells from monocyte cultures from chicken hatchlings (Ostoby et al., 1980) . In this study a combination of light microscopy, scanning and transmission electron microscopy was used. Nevertheless, the available transmission electron microscopical evidence for the macrophage-macrophage fusion event is extremely limited (Sasaki et al., 1989a, b; Sutton and Weiss, 1966) and not entirely convincing. Figure 2 , taken from Wise et al. (1985) , shows highly suggestive evidence for the fusion of macrophage/monocytes with a bone osteoclast. The direct contact between the monocyte plasma membrane Fig. 1 . An electron micrograph showing an osteoclast. Four nuclei (numbered 1 to 4) are shown in this section. This multinucleate cell is attached to both the bone (B) and calcified cartilage (C) by clear zones (CZ) and a ruffled border (R). The bone and cartilage surfaces below the ruffled border are frayed (asterisks) compared to smooth surfaces beneath the clear zones. Active osteoclasts have a vacuolated cytoplasm next to the ruffled border. These vacuoles (V) are large and indeed visible by light microscopy, and their presence is a reliable indicator of bone resportion. Large venous sinuses (S) are often seen next to osteoclasts. An osteocyte (5) is seen in the lower part of the figure. The scale bar indicates 10/~m. From Marks and Popoff (1988) , with permission. Monocyte number 4 appears to be in the process of fusing with one of the osteoclasts (i.e. between the arrowheads). The scale bar indicates 5/~m. Modified from Wise et al. (1985) , with permission. and the plasma membrane of the multinucleate osteoclast is indicated, but the differential staining density of the cytoplasm of the two cells suggests that cytoplasmic mixing has not yet occurred. Cultured monocyte/macrophages have been found to form MNGCs, when stimulated by interleukin-4 (McInnes and Renick, 1988) or insulin-like growth factor-I (Scheven and Hamilton, 1991) . Similarly, Kassem et al. (1991) have produced osteoclasts in longterm bone marrow cultures in the presence of 1,25-dihydroxyvitamin D 3. It is to be hoped that aliquots taken for electron microscopy from such osteoclast-forming cultures may clarify the early stages of macrophage fusion in the near future.

It has been shown by James et al. (1991) using monoclonal antibodies, that true osteoclasts can be distinguished from apparently similar MNGCs derived from a variety of different human tissues. Indeed cytochemical differences between osteoclasts and other MNGCs were also detected by , strongly supporting derivation from the macrophage lineage yet indicating separate development and functional specialization of the osteoclast. In vivo, the osteoclast possesses unique structural features, Fig. 3 . (a) Part of an osteoclast (O) in the coronal portion of a dental crypt displaying an extremely pronounced ruffled border (R) adjacent to the bone (B). The identity of the cell at the LHS of the figure is uncertain, although its numerous mitochondria suggest that it may also be an osteoclast, and that areas of contact may be forming between the two cells, seen more clearly in (b) (arrowheads). Scale bars indicate 10/~m (a) and 2.5/1m (b). Modified from Marks and Cahill (1986) , with permission.

intimately related to its functional role in bone resorption (Figs 1 to 3) . Nuclei are distorted, with marginated heterochromatin. Facing the bone surface or calcified cartilage is a ruffled border of flexuous filopodia or microvilli, behind which is a clear cytoplasmic zone, free of organelles. Vacuolation of the osteoclast cytoplasm is a common feature, usually interpreted as being indicative of active bone resorption. The membrane surface of the osteoclast not facing the bone is smoother, with a sparse coating of short filopodia/ microvilli. Detail of the undulating ruffled border of an osteoclast, actively resorbing bone during tooth eruption, is shown in Fig. 3 . The creation of an acidic calcium-solubilizing environment by the osteoclast is due to the metabolic activity of the ruffled border H +-ATPase (for further details see Pierce et al., 1991) . A three-dimensional TEM study of the mouse osteoclast has been performed by Doman and Wakita (1991) , who were able to distinguish ultrastructurally between migrating and bone resorbing osteoclasts. Emphasis was placed in this investigation upon the morphological classification of the clear zone between the central nucleated region of the MNGCs and the ruffled border. Knowledge of osteoclast structure and function is of vital importance for the further understanding of the hormonal control of mineral resorption in bone and teeth, as well as of bone diseases such as osteoporosis.

The ultrastructural and cytochemical changes that occur throughout the 280-day lifespan of the human placenta have been described in detail by Jones and Fox (1991) . Thus, only selected ultrastructural features will be presented here, with emphasis upon the MNGC theme and comparison with the other MNGCs included within this review. The continuing growth of the syncytiotrophoblast (Fig. 4) by fusion with underlying cytotrophoblasts is the accepted mechanism for its formation, throughout approximately the first six months of pregnancy. Contrary to the osteoclast and myotube, the fusion event is much easier to demonstrate electron microscopically in placental tissue than is the case for the bone osteoclast and myoblast, and convincing evidence for cellular fusion has been available for at least 25 years (Carter, 1964; Pierce and Midgley, 1963; Pierce et al., 1964) . Nevertheless, it must be remembered that a vast amount of data of a histological nature, on the mammalian embryo, the normal and cancerous placenta exists at the light microscopical level (Elston, 1987; Larsen et al., 1991; Kurmen, 1991; Shanklin, 1990) . Figure 5 shows part of a first trimester human placenta. The highly vacuolated syncytial layer, containing distorted nuclei with a peripheral patchy heterochromatin distribution, lies adjacent to the cellular cytotrophoblast (Langhans' cell) layer, with a cell in mitosis. Often a marked difference between the low electron density of the chromatin in the metabolically active cytotrophoblast nuclei and that of the electron dense inactive syncytiotrophoblast nuclei is clearly apparent. This comparative feature can be somewhat variable in the first trimester placenta, as shown in Fig. 6 , where the difference between the cytotrophoblast and the syncytiotrophoblast nuclei is less marked. The second trimester placental syncytiotrophoblast is characterized by possessing an extremely pronounced "scalloping" of the microvillar surface (Fig. 7) , together with increased underlying cytoplasmic vacuolation, indicative of the cellular activity associated with the provision of the metabolic and nutritional requirements of the growing foetus. That cellular fusion is continuing to occur at a significant level throughout the second trimester is indicated by detection of cells of intermediate category, showing phenotypes very similar to the syncytiotrophoblast, but with desmosomal junctions still present (Fig. 8a) . Thus, the process of cell fusion is unquestionably indicated by these residual desmosomal junctions, around which the plasma membrane disintegrates ( Fig. 8a and b) giving complete cytoplasmic continuity, i.e. true fusion as opposed to merely cellular adherence. The physiological benefit to be gained from having a fused cell/syncytial layer immediately facing the maternal circulation is that a more complete and continually dynamic barrier to penetration by maternal cells is achieved than could be the case with a cellular layer maintained by desmosomes. Although there is little ultrastructural evidence for the presence of tight junctions (Jones and Fox, 1991), they may well play an important role in the species which possess a cellular trophoblast layer.

Towards full term, the syncytiotrophoblast nuclei progressively cluster into the so-termed syncytial "knot", with exclusion of the intervening cytoplasm and they become increasingly shrunken and electron dense ( Fig. 9a ). At this stage the amount of syncytiotrophoblast cytoplasm which is devoid of nuclei can become extensive (cf. Section VI.E and G). Syncytial nuclear "knots" in the post-mature placenta show varying levels of pyknosis (Fig. 9b) . Nuclear breakdown may even be detected within the syncytial "knots" (Fig. 10a) , an occurrence which is even more apparent under conditions of Rhesus incompatability (Fig. 10b ). (Figures 4 to l0 have been kindly made available by Dr Carolyn J. P. Jones.) The cytotrophoblast and syncytiotrophoblast appear to be susceptible to infection by HIV, although the involvement of the CD4 antigen is not decided. Thus, the placenta may provide the direct route of foetal infection by HIV from the mother (Jauniaux et al., 1988) .

Some considered speculations as to the possible endogenous retroviral induction of cytotrophoblast fusion and its significance for syncytiotrophoblast formation will be advanced within the framework of a new hypothesis for the evolution of placental mammals (Harris, 1991 ) (see Section V.A, below). 

During embryological development skeletal muscle is formed by the end-to-end fusion of the myoblasts.

This fusion is believed to involve cellular metalloendoproteases (Lucy and Ahkong, 1989; Pearson and Epstein, 1982) . Migration of the centrally located myoblast nuclei to the surface of the myotube occurs, together with nuclear flattening, resulting generally in the nuclei being well spaced and located immediately beneath the myolemma (Fig. 11 ). Some periodic invagination of the nuclear envelope of skeletal muscle nuclei is apparent, which sometimes appears to correlate with the surrounding actinomyosin Z-line period- icity. Occasionally two or more nuclei may be closely apposed (Fig. 12 ), but this is not generally the case. In mature skeletal muscle, "satellite" myoblast-like cells lie alongside the myotubes, beneath the basal lamina and may continue to be incorporated into the myotube bundle; undoubtedly they represent a source of primitive cells for muscle regeneration (Moraczewski et al., 1988) . Ultrastructurally, myoblast fusion has not been conclusively demonstrated, but convincing evidence has recently come from the in vivo studies of Robertson et al. (1990) who performed investigations on the regeneration of experimentally damaged skeletal muscle. The breakdown of the regenerative myogenic cell plasma membrane within the cytoplasm of the 9, Fig. 9 . Typical human placental syncytiotrophoblast nuclear "'knots" at lull term (a) and in a post-mature placenta (b). In (a) an apparent syncytial bridge between two villi is shown. In (b) the nuclei are at a more advanced stage of pyknosis than those in (a). The scale bars indicate 5 l~m. Prints kindly provided by Dr C. J. P. Jones.

fused cells was clearly shown (Fig. 13) , together with the laying down of organized myofibrils. The absence of glycocalyx and basal lamina appeared to be necessary to enable plasma membran~membrane contact to occur at the early stage of the fusion process. The somewhat specialized situation in the developing and regenerating lymph-heart muscles in avian and amphibian embryos (Runyantsev and Krylova, 1990) serves to indicate the developmental fusion of satelite myoblasts.

In skeletal muscle atrophies, degeneration of the myofibrils occurs without depletion of the number of nuclei per myotube (Tom6 and Fardeau, 1986) . Under these conditions (e.g. atrophied muscle from an AIDS patient) marked clustering of the nulcei does, however, occur (Figs 14 and 15 ), with pronounced myofibrillar Fig. 10 . The pyknotic degeneration of nuclei in a placental syncytial "knot", with almost complete exclusion of cytoplasm from between the nuclei (a) and in the case of Rhesus incompatibility when more marked nuclear breakdown occurs (b). The scale bars indicate 2.5 ~m. Prints kindly provided by Dr C. J. P. Jones.

disorganization. As with the full term syncytiotrophoblast and cultured MNGCs, there may be almost complete exclusion of cytoplasm from between the clustered nuclei (Fig. 15 ). An account of muscular changes in the AIDS patient has recently been presented by Buskila and Gladman (1990) , but does not place significant emphasis upon ultrastructural aspects of muscle wasting.

In pathological tissues there are numerous examples of MNGCs, ranging from the binucleate Reed-Sternberg cell in Hodgkin's disease, the Langhans' tuberculous lesion and the foamy lipid-containing bone marrow binucleate and multinucleate macrophage in Gaucher's disease, through to the multinucleate giant epithelial (Dail and Hammar, 1988) . Many tumours express MNGCs and they have also been detected in granulomatous inflammatory conditions such as rheumatoid arthritis (Bardosi et al., shown in tuberculoid granuloma, osteogenic sarcoma (osteosarcoma) (Ghadially and Mehta, 1970) , sarcoid granuloma (Ghadially, 1982; Kirkpatrick et al., 1988) , sarcoidosis (Carr, 1980) , turnout of the urinary bladder (Amir and Rosenmann, 1990) , giant cell tumour of the pancreas (Goldberg et al., 1991 (Goldberg et al., ), 1989 , giant cell arteritis (temporal arteritis) (Wawryk fibrous histiocytoma (Smith et al., 1990) and a maliget al., 1991) and the cardiac Aschoff cell in acute nant giant cell tumour of the uterus (Magni et al., rheumatic fever. In giant cell myocarditis (Komada 1991) . Foreign-body subcutaneous granulomas conet al., 1991) , which is also associated with various taining MNGCs have been studied by conventional types of inflammatory cells, the MNGCs can closely thin sectioning (Cain et al., 1981 ; Papadimitriou and resemble degenerate cardiac muscle cells. Giant cell Archer, 1974; Papadimitriou and Rigby, 1979 ; van carcinomas, characterized by the presence of MNGCs, der Rhee et al., 1979) and also using the quick occur within epithelial tumours (Black and Epstein, freeing deep etching method (Baba et al., 1991 (Baba et al., ). 1974 ) and at several visceral sites, such as breast, Emphasis has been placed in these studies upon the thyroid, pancreas, lung, ovary and kidneys. The cytoskeletal distribution and organization within the pleomorphic giant cell carcinoma of the lung (Wang MNGCs, which in general appear to retain feaet al., 1976) has been shown to contain unique ultra-tures characteristic of the macrophage. In breast structural features, which may represent extreme carcinoma there have been numerous reported cases anaplasia. Giant cell carcinomas containing histiocyte of the presence of MNGCs with an osteoclast-like or osteoclast-like giant cells often bear a strong nature (Oemus and Timmel, 1990; Trojani et al., structural resemblance to the giant cell osteoclas-1989) . Figure 16 shows an example of a MNGC tomas of bone, as is the case with MNGCs in within a breast carcinoma. Such MNGCs can also many primary visceral giant cell tumours. Thus, occur in areas remote from the tumour and in nonit is clear that the monocyte/macrophage appears turnout-bearing breasts, so it is reasonable to proto have a pronounced tendency to be involved pose that growth factor(s) released by the tumour Fig. 12 . Part of a mouse muscle myotube where two nuclei are located in close apposition to one another, immediately beneath the myolemma (a). At higher magnification (b) the undulating nuclear envelope is clearly presented, together with the thin rim of heterochromatin adjacent to the inner nuclear membrane/nuclear lamina. The scale bars indicate 5/~m (a) and 2 #m (b).

mass have had a secondary stimulatory action on the fusion of monocyte/macrophages. The role of cytokines in the fusion of macrophages (M6st et al., 1990) indicates that interferon-gamma is essential and that cellular expression of the adhesion molecule LFA-1 (CD18) is also involved. Athanasou et al. (1989) consider that these breast carcinoma MNGCs are distinct from both osteoclasts and other types of inflammatory foreign body giant cells. Tumours of the bone, which involve macrophage polykaryons in bone resorption have been defined (Kanchisa et al., 1991; Mie et al., 1991) , and also the giant cell tumour of the tendon sheath (Athanasou et al., 1991) . Other bone tumours which display MNGCs, such as some osteogenic sarcomas and tumours of the tendon sheath may, however, also be of osteoblastic and fibrous histiocytic origin, since they do not always display clear osteoclastic characteristics (Marks and Chambers, 1991; Stark et al., 1983) . Somewhat like the macrophage, the cytotrophoblast also appears to be involved in the formation of syncytiotrophoblastic-like giant cell tumours (cf. gonadal and extragenital germ cell tumours such as the teratocarinomas and choriocarcinomas) (Zimmerli and Hedinger, 1991) which can also form metastatic carcinomas of the gastrointestinal tract, where there is focal trophoblastic differentiation (Saigo et al., 1981) . Indeed, the placental site trophoblastic tumour can be highly aggressive, with extensive infiltration of the myometrium and metastasis to the lungs (Orrell and Sanders, 1991) . Although there are reported cases of multinucleated cells being present in certain haematological conditions (Kurabayashi et al., 1989) , it is not always clear whether a multi-lobulated nucleus within a single enlarged cell is being discussed rather than a truly MNGC. An interesting in vivo investigation of methylcholanthrene-induced primary tumours (Fortuna et al., 1989) has suggested that cell fusion is of significance in relation to primary tumour formation, but that the fused cells were progressively lost.

In the case of virally-induced MNGCs it is important to emphasize that in all instances a major contribution has been made by light microscopy of human and animal biopsy samples and of virally infected cell culture lines (Malherbe and Strickland-Cholmley, 1980) . In most cases, electron microscopy is unfortunately lacking, but it is clear that the cytopathic action of fusogenic viruses when applied to cell cultures often conforms very well with the in vivo actions of the virus. The term Cytopathic Effect (CPE) of a particular virus originally had a rather broad meaning, but has come to refer predominantly to the fusogenic action of a virus in producing MNGCs in vivo and in vitro; it will be used in this more restricted context throughout this review. Thus, the CPE should be distinguished from the overall cytolytic or cytocidal action of a virus, such as the cellular destruction of the anterior horn neurones by poliovirus or production of lysis of cultured cells. It is also important to clearly distinguish between virally-induced cell fusion from "without" and from "within" the cell.

All fusogenic viruses need to be able to bind to a cell surface receptor, usually on the specific cell type to which the virus has an affinity. Then follows fusion of the virus with the plasma membrane or endocytosis, internalization and acidic endosomal fusion of the viral membrane, infection of the cell, viral replication and subsequent release. The fusogenic action of a membraneenveloped virus may occur directly when one or more viruses fuse with the plasma membrane of two cells, thereby providing a cytoplasmic bridge between the cells (fusion from "without"). Alternatively, newly synthesized viral protein may need to be expressed on the cell surface to enable fusion to take place, in this case by attachment to receptors on non-infected cells or latently infected cells not yet expressing viral proteins in their plasma membrane (fusion from "within"). The former situation has been elegantly described for the Sendai virus-induced fusion of chicken erythrocytes with HeLa cells in the classical studies of Henry Harris (1970, Fig. 15 . A cluster of nulcei within a myotube showing marked myofibrillar (Mf) disorganization, in a muscle biopsy sample taken from the atrophied skeletal muscle of an AIDS patient. The nuclei show no indication of flattening and have clustered as a "knot" with almost total exclusion of cytomembranes, mitochondria and myofibrils. The scale bars indicate 5 pm (a) and 1 l~m (b).

1974) and also for chicken erythrocytes and hamster melanoma cells (Zakai et al., 1974) . Both viable and UV irradiated Sendai virus have this fusogenic capacity,

indicating that viral replication is not required. This approach provided the foundation for extensive studies on heterokaryon formation and the whole field of monoclonal antibody production, together with the extensive field of viral-induced membrane-membrane fusion from without (White and Blobel, 1989) . The second situation appears to be of particular importance in the case of herpes-and retroviral-induced cell fusion, where viral integration, replication and expression of viral proteins at the cell surface (Desportes et al., 1989) is required to induce MNGC formation (Kalgemaraman et al., 1990) . In this situation viral inactivation by UV or gamma irradiation (Chaterjee and Hunter, 1980; Kitchen et al., 1989) prevents the fusion event, as assessed by the in vitro cellular syncytium-formation assay (e.g. Nagy et al., 1983) . It should be emphasized that this cellular assay for retroviral fusogenicity (Hoshino et al., 1983; Klement et al., 1969) has also become an extremely valuable tool for the investigation of anti-HIV agents (including antibodies, viral surface antigens, peptides homologous to the attachment epitope(s) of viral surface glycoproteins, cloned and synthetic analogues to the cellular CD4 + receptor, and drugs) (Baba et al., 1990; Bremermann and Anderson, 1990; Gruters et al., 1987; Lifson et al., 1991; Nara et al., 1987; Owens et al., 1990; Pal et al., 1991; Tochikura et al., 1988) . Of great significance, within the present context, has been the demonstration that the incorporation of the HIV envelope glycoprotein gene into the cellular genome results in the fusion of these genetically modified cells with CD4 + unmodified cells (Ashorn et al., 1990 ; see also Aoki et al., 1991 and Kost et al., 1991) . Biochemical aspects of virally-induced cell fusion from without have been thoroughly reviewed by Gallagher et al. (1980) ; Kielian and Jungerwirth (1990) et al. (1989) and White and Blobel (1990) . Emphasis has generally been placed upon the specific viral membrane fusion proteins and the peptide epitopes responsible for attachment to cellular surface receptors, together with the environmental pH required for the viral-cell and cell-cell fusion process. This biochemical progression has been followed more recently throughout investigations on the retroviral surface glycoproteins and cellular CD4 receptor.

The family Herpesviridae contains a significant number of animal and human viruses of veterinary and clinical importance. Three subfamilies have been defined: (1) the Aiphaherpesvirinae, which contain the genera Simplex viruses (e.g. herpes simplex virus HSV and varicella zoster virus VZV of man), (2) the Betaherpesvirinae, which contain the genus Cytomegaloviruses (e.g. cytomegalovirus CMV of man) and (3) the Gammaherpesvirinae, with the genus Lymphocryptoviruses (e.g. Epstein-Barr virus, EBV, of man). All herpes viruses are enveloped DNA viruses and are responsible for numerous conditions in man. It is significant that both CMV and EBV are extremely common viruses, with a high proportion of the population processing circulatory antibody, and that they both infect lymphoid cells. In the immunocompromised transplant patient, CMV has been shown to cause cellular fusion and may actively promote pneumonia. HSV and VZV, on the other hand, infect primarily epithelial cells and have been found to produce enlarged vacuolated (balloon) cells, which fuse into syncytial masses within the HSV or VZV vesicular lesion. The genome of the herpesvirinae is integrated into the cellular genome, a feature sometimes expressed by 

Members of the family Paramyxovirinae are renound for their ability to produce cell fusion in vivo and in vitro. Three genera have been defined: (1) the Paramyxoviruses (e.g. human parainfluenza types 1 to 4, mumps virus, Sendai virus, Newcastle disease virus and Simian virus 5 SV5), (2) the Morbilliviruses (e.g. measles, canine distemper and rinderpest) and finally (3) the Pneumoviruses (e.g. respiratory syncytial virus RSV and pneumonia virus of man).

All paramyxoviruses are enveloped RNA viruses, but no integration of the viral genome into the cellular genome occurs, since these viruses lack reverse transcriptase. In vivo, measles virus is known to be responsible for the Warthin Finkeldy "mulberry giant cells" (Warthin, 1931) and is also the causative virus in subacute sclerosing panencephalitis (SSPE) (Brown and Thormar, 1976) , involving skin, mucosal and respiratory surfaces, urinary, intestinal tract and lymphoid inflammations, all with detectable MNGCs. Respiratory syncytial virus is responsible for "giant cell pneumonia" (Dail and Hammar, 1988) with very large MNGCs detectable in the respiratory tract, these being particularly frequent in young children. A recent report of syncytial giant cell hepatitis (Phillips et al., 1991) has emphasized the severe clinical consequences of paramyxovirus-induced hepatitis in man, and the need for more careful viral classification in severe sporadic hepatitis. Parainfluenza virus Type 3 has been found to occur as an opportunistic infection in the immunodeficient AIDS patient, and was thought to potentiate the pronounced presence of MNGCs detected at autopsy. In the case of Sendai virus (Haemagglutinating virus of Japan) and all other members of the paramyxovirinae, it is clear that a direct cellular fusogenic activity is expressed by the viral membrane glycoprotein, which does not require biosynthesis and expression of new viral proteins at the surface of the infected cell for fusion to occur (Hosaka, 1988) . An early, yet elegant, electron microscopical study of SV5 (Compans et al., 1964) demonstrated pronounced syncytium formation when a baby hamster kidney cell line (BHK21-F) was infected, yet no such CPE was detected with monkey kidney cells. This variation in CPE was interpreted as being indicative of differing viral release mechanisms. Within the BHK21-F syncytia, nuclei became markedly arranged in rows, rather than clusters, and cytoplasmic vacuolation was not pronounced (cf. Section VI.E and F). A detailed, but primarily light microscopical study of the organization of cytoplasmic fibers within BHK-21 cells following fusion with SV5 and also PEG has been presented by Wang et al. (1982) . In general the two forms of syncytium induction gave rise to similar MNGC features although the central alignment of the nuclei was more pronounced in the virally-induced syncytia.

The large family of animal and human Retroviridae (synonomous with the Retraviridae, Oncoviridae, Leucoviridae and Oncornavirinae) contains three subfamilies ( Fig. 18) : (1) (1991) and Mergia and Luciw (1991) .

The cellular pathology of the different retrovirus subfamilies and their genera presents a rather variable picture, with some overlap of characteristics. The overall retroviral classification is also based upon molecular biological and morphological criteria, such as the conformation of the mature virion core, the clarity with which the surface glycoprotein is revealed and the cellular location of the forming virion within infected cells, as described diagrammatically in Fig. 18 (Bouillant and Becker, 1984; Frank, 1987a lacking the glycoprotein membrane (also termed the Oncovirus A, is released and remains intracisternally, as shown in Fig. 19 ), whereas the complete Oncovirus B is produced by budding/exocytosis of the Oncovirus A through the plasma membrane (Fig. 18a, b) . Having enveloped the immature viral core with a glycoprotein membrane, maturation of the core follows after exocytosis (Ohtsuki et al., 1987) . Cultures infected with Oncovirus B do not generally have a tendency to exhibit cell fusion and MNGC formation (Sarkar, 1989) although Okoi et al. (1990) found this to occur with murine mammary tumour virus. The genus Oncovirus C, on the other hand, has a strong tendency to induce malignancies in vivo, and in cell cultures infected cells fuse with non-infected cells (Bouillant et al., 1975 (Bouillant et al., , 1980 Frank, 1987b; Hoshino et al., 1983; Klement et al., 1969; Monozaki et al., 1990; Nagy et al., 1983; Ogura et al., 1977; Timar et al., 1987) . Figure 20 shows a cultured mouse myeloma cell line in which a MNGC has been induced by the presence of murine leukaemia virus. A group of free virions can be seen adjacent to the MNGC, but there is little evidence of viral budding. Part of a MNGC induced by bovine leukaemia virus (BLV) is shown in Fig. 21 . The ribonucleoprotein core of type C oncoviruses usually forms only at the plasma membrane of infected cells, with subsequent budding and maturation (Fig. 18c ). An interesting report by Aziz et al. (1989) indicated that a defective mouse leukaemia virus had the ability to cause a severe immunodeficiency disease, indicating strong parallel with the lentiviruses. Viruses of the genus oncovirus D do not induce tumours, but syncytia are detected in vitro (Bohannon et al., 1991) and in vivo together with a persistent non-lytic infection. The spherical immature core of type D oncoviruses, which possesses a marked subunit structure, initially forms in the cytoplasm and buds through the plasma membrane ( Fig. 18d ) (Chatterjee and Hunter, 1980; Gelderblom, 1987) . The Lentivirinae are all characterized by their slow in vivo pathogenesis, following integration via reverse transcriptase (RT) into the cellular genome as proviral DNA. It appears that very often some additional lymphocytic activation/stimulation (cytokine/immunological/viral) is required to induce lentiviral replication with subsequent budding and release from infected lymphoid cells, but it is now believed that the macrophage may indeed be the prime host cell of the lentivirinae (Narayan and Clements, 1989). Considerable evidence is accumulating to indicate that in vivo both animal and human lentiviruses have the property of inducing cell fusion of both macrophages and lymphocytes. In vitro, cells infected with lentiviruses all have the property of inducing cell fusion and syncytiogenesis with non-infected cells carrying the appropriate receptor for the viral surface glycoprotein (Daniel et al., 1984 (Daniel et al., , 1985 . Indeed, the early studies on HIV (then termed HTLV-III/LAV) were to some extent slowed by this extreme cytopathic effect, which is more pronounced for HIV than for (Fig. 22) . The cell fusion process is indicated (arrowheads), despite the fact that very little virus is budding from the surface of the MNGC. Large lysosomes containing electron dense granules are particularly abundant and considerable disorganization of the cytomembranes and cytoskeleton is apparent. Under these conditions marked clustering of the nuclei occurs (Fig. 23) , with almost total exclusion of mitochondria and cytomembranes. Lentiviral release occurs via a very pronounced localized arc-like thickening of the plasma membrane where the core precursor is forming, with subsequent budding of the electron dense, spherical, immature virion and maturation of the core (Fig. 18e) after release from the cell (Frank, 1987c) . The exact stages of the lentiviral core maturation are, as yet, not precisely defined in ultrastructural terms (Goto et al., 1990; Katsumoto et al., 1987 Katsumoto et al., , 1988 Katsumoto et al., , 1990 , although much progress has recently been achieved (Hoglund et al., 1992) . The spumaviruses do not induce malignancies, but induce a persistent infection, despite circulatory antibody. A lyric in vivo cytopathic effect is observed, initially expressed by the presence of highly vacuolated "foamy" MNGCs, which subsequently become necrotic and die. Similar features are often observed in infected cell cultures. Bovine syncytial virus (BSV) readily induces MNGC formation in culture, as shown in Fig. 24 . Characteristic vacuolation of the cytoplasm is apparent, with formation/budding of incomplete virions occurring internally. Indeed the immature spumavirus is believed to form in the cytoplasm; exocytosis then imparts a viral membrane which contains particularly pronounced surface glycoproteins (Fig. 18f) (Gelderblom and Frank, 1987) .

Although numerous enveloped RNA viruses not mentioned above, including the orthomyxoviruses, togaviruses and rhabdoviruses all produce acid pHdependent cell fusion from without (White and Blobel, 1989) , the available EM data is sparce. An example of a recent LM and molecular study comes from Gallagher et al. (1991) who investigated the broad pH dependency of cell fusion induced by corona mouse hepatitis virus type 4 (MHV4). Variant viruses containing genetically altered fusion-active surface glycoprotein, produced at a specific region of the polypeptide, were found to require acidic pH.

The pox viruses have been known for many years to induce cell fusion (e.g. Ichikashi and Dales, 1971) . Whilst viral envelope proteins are again involved in the fusion event, the precise role of the pox virus surface tubules and their p58 protein (Stern and Dales, 1976), which may represent an excessive production of membrane protein, remains unclear.

Other viral groups, such as the alphaviruses (Koblet, 1990) , are also known to produce cell fusion, but the lack of published electron microscopical data precludes further discussion. Fig. 25 . A possibly artefactual "'multinucleate zone", detected within a lymph node biopsy sample from an AIDS pateint. The surface of the syncytium is not detectable, suggesting that plasma membrane breakdown may have occurred due to sample autolysis prior to fixation. The lymphoid nuclei show no signs of activation within a germinal centre, a feature usually indicative of HIV infection, and the mitochondria appear to be extremely swollen. The scale bar indicates 5/~m.

The two most marked and possibly significant clinical features which have emerged from the study of biopsy and autopsy samples taken from patients with ARC or AIDS are (1) the marked diminution in the population of circulatory CD4 ÷ (helper) T-lymphocytes as the disease progresses from the latent infection to the more active phase of HIV replication and release (Lucey et al., 1991; de Wolf et al., 1988) and (2) the presence of MNGCs in a number of tissues, most particularly the brain, lymph nodes, spleen and lungs. Whilst the demonstration of these MNGCs, of macrophage/monocyte (microglial) and lymphocyte origin, has been based primarily upon their detection by light microscopy (Budka et al., 1987; Byrnes et al., 1983; Dickson et al., 1990; Joshi et al., 1984; Sharer et al., 1985; Wiley et al., 1986) , an increasing amount of data is becoming available utilizing TEM to assess the presence of MNGCs together with viral particles (Felice et al,, 1987; Gendelman et al., 1989; Lindboe and Froland, 1988; Naito et al., 1989; Orenstein and Jannotta, 1988) . One problem with the EM of biopsy and autopsy samples often relates to the quality of the tissue and its fixation, which can result in the apparent presence of MNGCs simply by the loss of plasma membrane due to autolysis, but with retention of nuclear integrity. Degeneration or swelling of other cellular organelles, such as mitochondria, will usually be an indication of this cellular damage (Fig. 25) . In the situation shown, it was not possible to define the surface of a MNGC, thus, indicating apparent multinucleation due to tissue autolysis. Indeed. the EM screening of lymph node and spleen biopsy samples from AIDS patients for activated viral-producing lymphocytes, i.e. with viral budding and the presence of immature and mature virions, has proved to be an extremely difficult task, unless activated germinal centres have been previously defined by light microscopy (le Tourneau et al., 1986) . Even though the number of infected lymphocytes is somewhat higher than in the periferal circulation it still remains a very low ratio. However, the detection of virus-producing monocytes from brain biopsies has proved to be somewhat less difficult (Fig. 26) . The natural extension of the above in vivo observations was the production of primary cell cultures from lymphocytes and macrophages taken from HIV-positive people and AIDS/ARC patients (Bigi et al., 1990; Gendelman et al., 1989; Ho et al., 1986; Klatzmann et al., 1984; Watkins et al., 1990) together with the establishment of the HIV producer cell lines (Popovic et al., 1984) . A significant observation has been that HIV-infected quiescent lymphocytes generally require additional cellular stimulation with a mitogen such as PHA in order for viral production to be detected. Similarly, cultured primary macrophages on stimulation (Gendelmann et al., 1988; Ho et al., 1986; Pautrat et al., 1990) will readily produce and release virus, within single cell and MCGCs (Fig. 27) . The receptivity/permissiveness of HTLV-I or HTLV-II transformed (transactivated) lymphoid cells to HIV infection followed by rapid viral replication and release, combined with the formation of MNGCs as the culture time progresses (often together with a population of mitotic single uninfected and infected cells) has yielded valuable information on the nature of viral glycoprotein Gpl20, the cellular CD4 antigen and CD4~GpI20 interaction (Clapham et al., 1987; Harada et al., 1985; Hussey et al., 1988; Ho et al., 1991; Lifson et al., 1986a Lifson et al., , 1986b Montefiori and Mitchell, 1987; Nara et al., 1987; Sodroski et al., 1986; Yoffe et al., 1987) . There is here a clear parallel between the C-type retrovirally-transformed cultured lymphoid cells and the large transformed single lymphocytes seen in lymph node biopsy samples from AIDS patients (Byrnes et al., 1983) , which often lie alongside MNGCs. It is an apparent physiological contradiction that mitotic and immunostimulation may actually make T-iymphocytes more receptive to HIV infection! The similarity of the HIV-syncytium forming assay and the MNGC/syncytium formation by cells infected with other retroviruses (human and animal) should be noted. The fusion of infected cells with non-infected cells, provides the technical basis for many investigations directed towards immunological or pharmaceutical prevention of the cytopathic action of HIV, as a possible target for future clinical treatment (Bremermann and Anderson, 1990; Fund et al., 1987; Gruters et al., 1987; Hanada et al., 1991; Lifson et al., 1991; Matsui et al., 1990; Momota et al., 1991; Owens et al., 1990; Pal et al., 1991; Srinivas et al., 1991; Suzuki et al., 1989; Tochikura et al., 1988; Wells et al., 1991) .

In many of the above mentioned in vitro studies, emphasis has been placed upon the assessment of MNGC formation by light microscopy. The Montagnier group (Klatzmann et al., 1984; Montagnier et al., 1984) were the first to include electron micrographs of a HIV-(then LAV) induced syncytia in lymphoid cell 

(see below, Section IV.E and IV.F). A distinction should, however, be firmly drawn between the mechanism of HIV release by single lymphoid cells and macrophages, the former liberate virus only at the cell surface, whereas the single and multinucleate macrophage tends to liberate virus primarily into cytoplasmic vacuoles (Gendelman et al., , 1989 Jannotta, 1988; Pautret et al., 1990) (Figs 26 and 27 ), but Watanabe et al. (1991) observed viral budding at the macrophage surface.

Electron microscopical studies on the HIV-1 infection of suspension cultures of the permissive human C8166 HTLV-I transformed lymphoblastoid cell line have been performed by Harris et al. (1989) . From approximately two days post infection of the cell culture, viral-producing cells can be detected. This HIV-1 release is seen to occur only at the cell surface (Fig. 28) ; compare with the intravacuolar viral release in the macrophage (Figs 26  and 27) . The ultrastructural detail of viral budding and maturation, will not be dealt with further and the reader is referred to the recent articles by Christie and Almeida (1988) , Clavel and Orenstein (1990) , Gelderblom (1991 , Goto et al. (1990 ), Hoglund et al. (1992 and Katsumoto et al. (1987, 1988, 1990) . It should be noted that although viral release apparently occurs evenly over the surface of infected cells in some instances, in many cases discrete patches containing large numbers of viruses at varying stages of release can be encountered (Fig. 29) , with large regions of plasma membrane devoid of virus. Released virions have a pronounced tendency to remain close to the plasma membrane following budding (Fig. 30) , where HIV protease-dependent core maturation then proceeds.

The major difficulty facing those who wish to make a precise assessment of the early events of cell fusion by transmission electron microscopy, centres around the problem of being able to distinguish between close contact of adjacent cells and true adherence with commencement of breakdown/fusion of the two plasma membranes, thereby creating cytoplasmic continuity between the cells. The problem is somewhat worse for monolayer cultures than for suspension cultures, since the latter often possess a surface microvillar coat and do not usually have any spontaneous tendency to aggregate. Dispersion and slow centrifugation of fixed cells in low melting temperature agarose is desirable for EM processing, rather than centrifugal pelleting, since the latter can induce artificial close contact between cells. Even very large fragile syncytia are well preserved and dispersed by this agarose procedure. Thus, with the extensively microvillated C8166 lymphoblastoid cells, processed in agarose, it can be safely assumed that when two cell surfaces are in close proximity, with one or both cells exhibiting viral production, these cells may well be in the process of undergoing fusion. The presence of large numbers of single cells alongside the fusing cells provides further justification for the validity of this interpretation. The detection of lesions in the two plasma membranes then provides direct evidence for this fusion process, which does not require the presence of viral particles at the fusion site. Rather, it does require the presence of newly synthesized viral glycoprotein Gpl20 (Deportes et al., 1989) to attach to available CD4 + receptors. Figures 31  and 32 show varying stages of the cell fusion process Fig. 31 . The early stages of cell fusion in a HIV-l-infected C8166 suspension culture (day 5). Cells initially make contact with their cell processes (arrowheads) (a and b), usually without any clear bridging by surface virions, as expected for fusion from "within". Although it is sometimes clear (b) that one cell is infected (I) and releasing virus and the other cell is not releasing virus (NI), in many cases this distinction is by no means simple. Scale bars indicate 5 #m. and in Fig. 33 the breakdown of the two plasma membranes at the zone of contact is indicated at higher magnification. By five days post infection, C8166 cultures are found to contain a complete range of multinucleate cells, (i.e. binucleate to MNGC containing many tens of nuclei) together with large numbers of actively dividing mononuclear cells. This diversity of cellular material can best be demonstrated by light microscopy of Toluidine blue-stained semi-thin plastic sections (Fig. 34) , which does indeed provide a low magnification overview of much of the data presented in Section IV.

It is a somewhat arbitrary decision as to when a multinucleate cell becomes a multinucleate giant cell. A  Fig. 32 . A slightly later stage of cell fusion between HIV-l-infected C8166 cells. The region of plasma membrane contact (a and b) has extended (arrowheads) as a prelude to complete fusion. The scale bars indicate 2.5/~m. reasonable proposition might be when the fused cell contains upwards to 10 nuclei, but this fails to emphasize the extremely large MNGCs that can be formed under appropriate culture conditions, and which maintain "viability" for several days. Examples of individual multinucleate cells containing, two, three, four, five and six nuclei etc., have been readily obtained (Figs 35 and 36) . Undoubtedly the continued fusion of single cells and small multinucleate cells is the mechanism by which the larger MNGCs are formed (Fig. 37) . It is likely, however, that there is an active phase of cell fusion, followed by progressive metabolic changes within the MNGCs which reduces the capacity for further fusion and ultimately leads to loss of viability. This statement is based upon the observation that the electron-and toluidine blue-dense MNGCs containing pyknotic nuclei do not apparently continue to fuse with single cells or small multinucleate cells (see below, Section IV.H).

One of the most characteristic features of HIV-1 induced MNGCs is their pronounced vacuolation. This can be detected within multinucleate cells containing upwards of approximately 10 nuclei (Figs 38 and 39 ) through to those containing upwards of 100 nuclei, which then may contain very large numbers of small Fig. 33. Higher magnifications (a and b) showing zones of membrane fusion between adjacent C8166 cells (small arrowheads). Although breakdown of the plasma membrane is not rapid, as fusion progresses it becomes impossible to detect the former plasma membranes (between large arrowheads). Scale bars indicate 2 pm (a) and 0.5 glm (b).

vacuoles or enormous vacuoles which exhibit a "ballooning" effect ( Fig. 34 and see below, Section IV.E). This vacuolation should be contrasted with the known CPE of the Spuma/Foamy viruses, which produce marked vacuolation within single cells and MNGCs. The relatively small MNGCs (Figs 38 and 39 ) continue to release HIV-1 in very large quantities from the syncytial surface, at discrete patches. The microvillation of the surface diminishes to some extent with MNGC growth, indeed part of the surface may become almost smooth, whilst other regions retain microvilli.

As the number of nuclei incorporated into the multinucleate C8166 lymphoblastoid cells increases, certain clearly definable cytoplasmic changes occur, in addition to vacuolation and surface microvillation. The nuclei have a pronounced tendency to cluster, centrally in the smaller MNGCs and then in groups located some distance from the surface, in the larger MNGCs (Fig. 40) . Mitochondria also cluster more centrally, in a zone immediately outside the nuclear cluster. This produces a marked depletion of mitochondria in a broad zone of cytoplasm extending to the MNGC surface (Fig. 41) . Single HIV-infected cells appear to contain a normal complement of rough endoplasmic reticulum (rER), but organized cytomembranes appear to be lost in the MNGCs, except in close vicinity to the nuclei, where rER membranes can be seen (Fig. 41b) . Indeed, the mitochondrion-depleted cytoplasmic zone appears to be rich in free ribosomes with disorganized microfilaments and microtubules (Fig. 42) . The appearance of small and large vacuoles in the MNGCs, initially in close apposition to the clustered nuclei, creates the foamy balloon-like structures, indicated by the light micrographs shown in Fig. 34 . By electron microscopy it is readily apparent whether or not these vacuoles contain HIV. Clearly, the process of cell fusion must give rise to a gross excess of internalized plasma membrane. There is, therefore, the possibility that this plasma membrane could be re-utilized to create the vacuolar membranes within the MNGCs, possibly along with Golgi and/or smooth endoplasmic reticulum membrane. Immunolabelling data is required to answer these important questions.

That single HIV-I infected C8166 cells and small MNGCs release virus by budding from the surface has already been emphasized, in contradistinction to the single macrophage which releases primarily into internal vacuoles. We have found that medium sized MNGCs will very often be found to continue liberating virus at the surface (Harris et al., 1989) , whilst also budding virus into the internal vacuoles (Fig. 43) . With increasing size and vacuolation, by far the larger quantity of virus continues to be liberated internally rather than at the surface, as shown by Dowsett et al. (1978) . These workers failed to make any comment on the progressive change of HIV release from MNGC surface to vacuoles, with respect to MNGC size. Sometimes the large vacuoles can be almost completely filled with mature virions (Fig. 44) and on other occasions vacuoles are almost empty (Fig. 43) . Highly disorganized regions containing a network of small vacuoles can contain extremely large numbers of immature or mature virions (Fig. 45) . The massive level of intrasyncytial viral release indicated in Figs 44 and 45 can often give rise to the overlapping of budding sites at the vacuolar membranes. This results in the release of a significant number of larger multi-cored viruses with deviant shapes. The cores in these deviant virions are often elongated tubes with a constant diameter, rather than the more usual conical shape, but cf. Klimkait et al. (1990) , Christie and Almeida (1988) and Clavel and Orenstein (1990) .

In the HIV-1 infected C8166 cell and the MNGCs that are actively liberating virus, at their surface or into vacuoles, it can be assumed that large amounts of viral i Fig. 36 . A tetranucleate C8166 cell, showing the first signs of the "rounding-up" phenomenon that occurs to the MNGC as more nuclei are incorporated (cf. Fig. 34 ). The scale bar indicates 2.5/tin.

proteins and envelope glycoprotein are being synthesized by the cell. This follows from the initial insertion of the retroviral progene within this transactivated/permissive cell line followed by the production of viral mRNAs. This production of viral mRNA is under the control of the two HIV activators of viral production, the tat and rev proteins. Under appropriate cell culture conditions, this can result in production of very large amounts of viral mRNAs, in turn responsible for the massive viral production described above. It has been claimed by Koga et al. (1990) , from immunolabelling data, that accumulations of Gpl60 occur at the nuclear pore complexes of CD4 + cells transfected with the Gpl60 gene. This is a rather difficult observation to reconcile with viral budding and also with the known mRNP nuclear-cytoplasmic translocation process that is believed to occur at the nuclear pore complexes. Our own data (Harris, 1990; Harris et al., 1990) have shown that in the single cells and MNGCs actively producing large quantities of virus, a morphological change is detectable at the nuclear envelope. The nuclear pore complexes increase in prominance, this being expressed by a thickening, i.e. extension into the cytoplasm and an increase in electron density (Fig. 46) , compared to cells not liberating HIV-1. It is suggested that this is an ultrastructural indication of the increased viral mRNA production in the nucleus and mRNP translocation at the nuclear pore complexes. Clearly, further detailed ultrastructural and immunolabelling studies are required to obtain supportive data for this proposal, which is nevertheless, reasonably in accord with current ideas on the nucleocytoplasmic translocation of mRNP complexes (Mehlin et al., 1991) .

Following the phase of active viral production and release within the MNGCs, progressive nuclear changes occur, in parallel with those in the cytoplasm, both being an indication of the gradual loss of viability. The initially relatively smooth surfaced nuclei within the MNGCs become progressively shrunken and misshapen, this being shown firstly by nuclear envelope invaginations (Fig. 47a) and an increase in electron density with spongy reticulation of the nucleoli (Fig. 47b) . Although all the nuclei may not have entered the MNGC at exactly the same time, there is remarkable uniformity in nuclear morphology within any one MNGC, suggesting that a short-lived active/dynamic phase of cell fusion terminates relatively quickly (hours rather than days). Shrinkage of the nuclei creates profiles very similar to those often seen within tumour or degenerate tissues (Fig. 48) . At this later stage an increase in both nuclear and cytoplasmic electron density is noticeable. It should be emphasized that these metabolically inactive nuclei no longer exhibit the pronounced nuclear pore complexes mentioned above. 

The long-term viability of the HIV-1 induced MNGCs in suspension culture is clearly limited to a few days, but the continued presence of single cells can give rise to a steady production of smaller MNGCs, as indicated by the range of cellular material present (Fig. 34) • The marked increase in cytoplasmic electron density that occurs as the older, generally larger, MNGCs appears to be due to a progressive ageing phenomenon rather than rapid cytolysis. Electron microscopical specimens reveal increasing nuclear clustering and pyknosis (Figs 49 and  50) , which parallels in a remarkable manner that occurring in the syncytiotrophoblast at the later stages of pregnancy (Figs 9 and 10) . It should, of course, be mentioned that nutritionally, the relatively thin yet extensively spread syncytiotrophoblast is in a much superior situation, since it is in direct contact with the maternal blood supply. Also, other metabolic and hormonally controlled causes for syncytiotrophoblast degeneration which limit its survival, will be functionally linked to normal foetal development.

The ultimate degeneration of the MNGCs in suspension culture occurs when the sites of internal vacuolation disperse as a massive "'loamy mass" (Fig. 51) . The other, non-vacuolated, electron-dense cytoplasmic regions exhibit marked surface folding and abnormal villation. This can give rise to a multiplicity of surface microvilli (Fig. 51) , which undoubtedly indicate the rapidly approaching syncytial "death".

The somewhat limited data presented above serves to indicate that the HIV-1 infected C8166 lymphoblastid suspension culture system expresses a complex dynamic situation within which cellular events relating to: (1) viral integration, replication and release, (2) cell fusion, (3) MNGC formation and (4) MNGC ageing and death can be investigated at the ultrastructural level. The survey presented represents only a small fraction of the available data, some of which has already been published elsewhere (Harris et al., 1989 : Harris, 1990 ).

Embryologists working at the end of the last century, such as K611iker, recognized that the blastocyst was the tissue or organ from which the unique development of all mammals proceeds• Today we can be a little more precise, within the context of the extensive knowledge of embryological histology, placental ultrastructure and pathology, by saying that it is reasonable to define the trophoblast, the cellular layer of the blastocytst cavity enclosing the truly formative embryonic cells, as representing the most distinctive mammalian characteristic at this early stage of development. To quote Davies and Hesseldahl (1971) "There is no homologue to the trophoblast in submammalian forms. The trophoblastic shell must be regarded as a peculiar mammalian structure developed solely in response to the demands of viviparity and prolonged gestation in the uterus". The important question that must be posed is, why did the rapidly growing trophoblast develop, presumably at some critical stage in animal evolution just prior to the divergence of the mammalia (Aitken et al., 1979; Pierce and Midgley, 1963; Pierce et al., 1964) ? All early stage trophoblasts are multicellular, but in the majority of mammals a discrete trophoblastic pole or syncytial plate develops, which becomes the forming placenta and site of uterine attachment. That the trophoblast has a remarkable growth rate and a "tumour-like" invasive nature in the region of the syncytial plate is believed to account for the continual expansion of the placenta by fusogenic incorporation of the underlying trophoblast cells, which are later defined as the Langhans' cells or cytotrophoblasts. The fact that certain mammals (ungulates and some rodents) do not have a fused-cell syncytiotrophoblast apparently does not preclude the function of the invasive trophoblast or its role in the creation of the placenta and the foetal/maternal barrier (Wynn, 1971 ), but clearly more up-to-date histological and ultrastructural studies are required in this comparative sphere. In the large majority of mammals, trophoblastic cell fusion does occur and we must ask why it happens and what physiological benefits the animal gains from the presence of the syncytiotrophoblast rather than a multicellular trophoblast.

From the early 1970s through to the present day there have been repeated electron microscopical observations claiming the presence of retroviral particles and reverse transcriptase in animal and human placental tissue (Feldman, 1979; Feldman et al., 1983; Johns and Renager, 1990; Imamura et al., 1976; Nelson et al., 1978; Panem, 1979; Smith and More, 1988; Ueno et al., 1983) . In most cases these are C-type retrovirus particles, which bud from the plasma membrane, although occasionally A-type intracisternal particles have also been detected (Enders, 1971) . Most of these data were produced within the context of the intensive search throughout the 1970s for human tumour-forming viruses and has been greatly overshadowed by the more recent and highly productive work on the animal leukaemia viruses and human leukaemia/lymphoma viruses HTLV-I and HTLV-II, and indeed that on the animal and human lentiviruses. The observation must, nevertheless, be considered to be correct and highly indicative of the presence of one or more endogenous retroviruses, expressed selectively within placental tissue. More specifically, retrovirus-like particles have been found to be budding from the syncytiotrophoblast, particularly at the border with the underlying cytotrophoblasts, the histological zone at which cellular fusion is postulated to occur. Isolation of infectious virions from palcental tissue has proved to be difficult (see comments by Weiss, 1982) , but Stromberg and Beneviste (1983) claimed to be able to isolate an endogenous retrovirus from rhesus monkey trophoblast. Supplementary to this morphological evidence for the presence of endogenous retrovirus in placental cytotrophoblasts and syncytiotrophoblast, much data of a biochemical, immunological and molecular biological nature has accumulated which supports the overall implications of the ultrastructural data. Reverse transcriptase has been detected together with a number of viral antigens, located particularly within the syncytiotrophoblast, using immunofluorescence microscopy (Suni et al., 1981 (Suni et al., , 1988a . It is eminently reasonable to propose that if the integrated progene of such an endogenous virus is indeed replicating, the production of intracellular trophoblastic viral proteins could occur at the early stages of embryological development. Rather than remaining latent, as is also an inherent property of many retroviruses, there could be biosynthesis of the complete range of viral gene products or selected gene products. These may include the viral envelope glycoprotein, which will be incorporated into the plasma membrane of the infected cell, prior to viral budding. This could, in turn, have the capacity, as do other C-type oncoviruses and lentiviruses discussed above (Section III.D), to induce cell fusion by interaction with an appropriate receptor exposed on the surface of the plasma membrane of neighbouring cells (see also Davies and Chilton, 1978) . It has been suggested by Montagnier et al. 0984 ) that the extreme ability of certain members of the retroviridae to produce cytopathic cell fusion may indicate a general effect of retroviral glycoproteins at the plasma membrane of infected cells, which might deeply affect their specialized biological functions. Thus, a clear parallel can be drawn between the in vitro cell culture situation and in vivo situation, where MNGCs are formed following retroviral infections, and that are here postulated to occur in the trophoblast and placenta as the growing syncytiotrophoblast incorporates more cytotrophoblasts. Indeed the morphological features of the syncytiotrophoblast throughout its growth phase and degeneration are remarkably similar to those shown by cultured MNGCs (i.e. surface microvillation, cytoplasmic vacuolation, nuclear clustering and pyknosis). Furthermore, early attempts at growing the blastocyst/trophoblast in organ culture (Blandau, 1971; Lopta et al., 1982; Mohr and Trounson, 1982) and more recently the cell culture of cytotrophoblast-derived cell lines, indicates that they possess an inherent aggregative and fusogenic potential, and in many cases spontaneously form multinucleate syncytiotrophoblasts. Inexplicably, Friedman and Skehan (1979) appeared to retain the even then outdated concept that nuclear division in cytotrophoblast cultures occurred by mitosis without cytokinesis and was responsible for the formation of MNGCs in cultures treated with methotrexate, an error carried even further in parts of the more recent developmental biology literature (Gilbert, 1988) .

It is clear that in evolutionary and biological terms something extremely significant occurred at the time when the placental mammals diverged from all other animal species, including the marsupial mammals. A hypothesis has been advanced (Harris, 1991) which proposes that at some stage in animal evolution, prior to the divergence of the placental mammals, one or more embryos became infected at a very early intrauterine stage with a "primitive" retrovirus. The immediate effect of this infection may have been (1) the integration of the viral progene into the formitive embryo cells, including the germline cells, (2) the induction of a tumour-like proliferative growth of certain embryonic cells, thereby creating a primitive rapidly growing trophoblast, possessing a blastocyst cavity within which the truly embryo-forming cells are situated and protected. Whether or not a generalized trophoblastic myometrial invasiveness was present at this stage can only be speculated upon, but the creation of the early syncytial plate, in direct contact with the myometrium, by a cell fusion process is proposed. In the light of the known fusogenic potential of certain retroviral progene glycoprotein products, it is reasonable to speculate that this specific trophoblastic fusion could have occurred at this early developmental stage, a feature subsequently retained by all placental mammals. It is worth bearing in mind that in the early mammalian embryo of some 100 plus cells, approximately 90% contribute to the rapidly growing trophoblast, whereas only about 5% represent the formative (embryo) and 5% the primary endoderm (Davies and Hesseldahl, 1971 ). In the main, the parallel drawn between the tumourlike nature of the invasive trophoblast and indeed the placenta, is widely accepted by pathologists (Aitken et al., 1979; Shanklin, 1990) . When the placenta is discarded at full term of pregnancy, it may well do so primarily because of the degenerative metabolic and hormonal changes that occur within the syncytiotrophoblast resulting in a loss of its viability (cf. cultured MNGCs). Gynaecologists define various invasive pathological conditions of the myometrium, generally termed placental site tumours, nodular pseudotumour or trophoblastoma, where deep infiltration of the myometrium occurs. Clusters of MNGCs of trophoblastic origin can be detected, together with inflammatory cells (Shanklin, 1990) . The presence of C-type retroviruses in a variety of germ cell ovarian and testicular tumours cells (Anderson et al., 1991~ Bronson et al., 1984 : Beilby, 1985 : Boller et al., 1983; L6wer et al., 1984, 1987) , which in many cases also have a considerable tendency towards cell fusion, provides peripheral support for the above proposal. Nevertheless, the control of trophoblast metalloproteinases and invasiveness, and possibly syncytiotrophoblast formation, has been suggested by Graham and Lala (1991) to be influenced by transforming growth factors fl and ill-In terms of biological symbiosis, an endogenous virus or proto-oncogene (Adamson, 1987) needs to be perpetuated from generation to generation. This would indeed need to be the case for the germline integrated The nucleus is only slightly mis-shapen, the nucleolus is relatively normal and the chromatin has not yet become condensed. Note the absence of pronounced nuclear pore complexes (cf. Fig. 46 ). In (b) a nucleus at a slightly further stage of degeneration is shown. The undulating surface contour of the nucleus indicates some shrinkage and the nucleolus is exhibiting marked reticulation (arrowheads). Heterochromatin blocks are beginning to appear at the nuclear surface (arrows). The scale bars indicate 2.0 ~Lm. provirus DNA of the postulated retrovirus, despite the fact that the extra-embryonic placental tissue which most markedly expresses the presence of the retroviral genome and its products and is significantly influenced by their presence (i.e. trophoblast growth, invasiveness and fusogenic potential) is (fortunately) discarded at the end of each pregnancy. Fertilization may indeed provide the stimulation for the developmental expression of the postulated endogenous virus/viruses within each early embryo and impart untold benefit by enabling one or more embryos to progress through the first stages of growth within the protective and nutritive confines of the uterus. Ultimately the animal then sheds the histologically organized, yet undoubtedly tumour-like, vascular and endocrinal placental growth. Thus, a situation of true symbiosis is proposed, having benefits for both the perpetuation of the retroviral genome and the survival potential of the embryos of all palcental mammals.

The relatively slow progression of the ARC and AIDS, indicated clinically by the inversion of the Tlymphocyte CD4/CD8 (T4/T8) helper/suppressor ratio, is difficult to account for in terms of an immunological cytolytic response. Recent concepts may indicate an autoimmune (Weimer et al., 1991) or immunosuppressive (Habeshaw et al., 1990 ) response directed against CD4 + cell production and function may be involved rather than an immunolytic reaction. A hypothesis for T-cell dysfunction and depletion within the AIDS patient has recently been advanced by Ameisen and Capron (1991) , within the framework of "programmed cell death" or apoptosis (Laurent-Crawford et al., 1991) . On the other hand, it is certainly not beyond the bounds of possibility that the slow incorporation of activated viral-producing CD4 + lymphocytes, available from the pool of latently infected cells (McElrath et al., 1991; Zack et al., 1991) , into binucleate or trinucleate cells following fusion with uninfected lymphocytes could account for the CD4 + (helper) T-lymphocyte depletion (Fauci, 1988; Haseltine, 1990 Haseltine, , 1991 Harris et al., 1989) . This would be very difficult to detect histologically, for the reasons previously advanced (Section I), except when larger MNGCs are formed. That the level of CD4 + T-lymphocyte activation is an important factor is implicit and is clearly likely to be increased in the immunocompromised practising homosexual and drug addict. In addition, secondary lymphocytic viral infection with CMV, EBV (Bigi et al., 1990) or human Herpes 6 virus (Lusso et al., 1991) , may potentiate HIV replication and release. That CD4 + T-cell stimulation and phosphokinase C activation is indeed important in the fusogenic response to cells expressing the HIV-1 envelope glycoprotein has been emphasized by the work of Mohaghaghpour et al. (1991) , and Fig. 49 . Sections of electron-dense degenerate MNGCs containing shrunken pyknotic nuclei with large '~spongy'" reticulated nucleoli. The cytoplasm has undergone extensive vacuolation and there is grossly abnormal surface villation. Compare closely with the nuclear "knots" present in full term placental syncytiotrophoblast (Figs 9 and 10) . The scale bars indicate 5 pm. Fig. 50 . A single nucleus within a degenerate MNGC, exhibiting marked surface undulation, a large "'spongy" reticulated nucleolus and expanded nuclear envelope intracisternal space. The surrounding cytoplasm is grossly abnormal, indicating the approach of syncytial death. The scale bar indicates 1.0 ~m. Fig. 51 . The massive "foamy" degeneration of a MNGC into a large network of HIV-1 containing vacuoles (cf . Figs 34 and 45) .

The scale bar indicates 10 iLm. syncytium formation has been shown to be associated with an increase in cellular oleic acid (Apostolov et al., 1989) indicating an increase in membrane fluidity. Mounting evidence suggests that virulent HIV isolates from immunodeficient individuals expressing AIDS possess a greater syncytium-forming CPE than those from earlier stages of the ARC and AIDS (Cheng-Mayer et al., 1988; Fenyo et al., 1988 , Fiore et al., 1990 Tersmette et al., 1988 Tersmette et al., , 1989 . The rapid evolution of HIV in vivo appears to continually generate subpopulations of virions exhibiting an increasing cytopathic potential throughout the progression of AIDS (Habeshaw et al., 1990) . A note of caution has, however, come from Pantaleo et al. (1991) who consider that there may be some dissociation of HIV syncytium-forming capacity and HIV spreading, and that the cellular assay for inhibition of syncytium formation may not necessarily always be an indicator of the suppression of infection. Furthermore, the envelope glycoprotein of an infectious noncytopathic strain of HIV-2 has been found to be unable to induce syncytium formation . Contrary to this, Watanabe et al. (1991) have detected a chimpanzee-passaged HIV-1 isolated which is cytopathic to chimpanzee CD4 ÷ cells in vitro and in vivo, but does not cause development of disease. A differential discrimination between HIV-1 infection and syncytium formation has also been advanced by Lifson et al. (1991) , based upon the CD4 (81-92) amino acid sequence and its interaction with GP120. Nevertheless, it is very significant that the V3 loop of the HIV-1 envelope glycoprotein (GP120) has been found to be intimately involved with both syncytium formation and viral replicative capacity (de Jong et al., 1992) . Also, mutation of the transmembrane glycoprotein GP41 of HIV-1 has been claimed to interfere with cell fusion and infectivity (Freed et al., 1992) .

Despite the above emphasis upon the possible role of cell fusion in CD4 + T-lymphocyte depletion, it must be remembered that the macrophage lineage is also a major target for HIV (Gendelman et al., 1989; Gendrault et al., 1991; Pautret et al., 1990; Wiley et al., 1986) and that an increasing body of evidence shows the presence of neuronal multinucleate macrophage/microglial cells in AIDS dementia. A similar situation exists in the brain of Rhesus monkeys infected with SIV (Lackner et al., 1991) . At the moment there is no suggestion that such macrophage derived MNGCs in the brain of AIDS patients have a direct role in the creation of encephalopathy (Dickson et al., 1991; Takeya et al., 1991; Vazeux, 1990) . Rather, it is thought that the release of certain inflammatory compounds or neurotoxins from HIV-infected macrophages (Guilian et al., 1990; Pulliam et al., 1991) may be responsible for the neurological complications of AIDS.

The diverse content of this review serves to highlight the widespread significance of MNGCs in normal tissues and in both viral and non-viral pathology. An attempt has been made to survey the principal areas of interest and overall comparisons have been made, where possible. Although the involvement of one or more early retroviruses in the evolution of the placental mammals might be considered to be a highly speculative suggestion, the hypothesis advanced here and elsewhere (Harris, 1991) , serves to emphasize this important evolutionary question, that has for too long been neglected. Modern cellular and molecular biology have between them the available tools to answer some of the questions posed and provided answers to the possible involvement of integrated retroviral proDNA in the processes of evolution, development and oncogenesis.

It is hoped that greater appreciation of the existence and importance of MNGCs will follow from the publication of this review. Also, that further interest will be shown in the application of transmission electron microscopy to the study of the cytopathic effects of retroviruses. In combination with immunolabelling, this approach has an even greater potential, largely unexplored in this field of study. The possible involvement of C-type retroviruses in cell fusion and the formation of the Reed-Sternberg cell in Hodgkin's disease has been recently advanced by Sinkovics (1991), a phenomenon that appears to parallel the syncytium-forming capacity of many the retroviruses discussed throughout this review. It should, perhaps be stressed finally, that no suggestion is advanced above for the involvement of retroviruses in myoblast fusion during myotube formation or in macrophage fusion to yield the osteoclast, although for the latter in certain pathological situations (i.e. HIV-induced encephalopathy and mammary cancer) some role for such viruses in macrophage fusion may be implied.

Expression of proto-oncogenes in the placenta

Origin and formation of the placenta

Cell dysfunction and depletion in AIDS. The programmed cell death hypothesis

Osteoclast-like giant cell tumor of the urinary bladder

Endogenous origin of defective retrovirus-like particles from a recombinant chinese hamster ovary cell line

Syncytium formation of human and non-human cells by recombinant vaccinia viruses carrying the HIV env gene and human CD4 gene

Syncytia formation in HIV-1 infected cells is associated with an increase in cellular oleic acid

Human immunodeficiency virus envelope glycoprotein/CD4-mediated fusion of nonprimate cells with human cells

The origin and nature of stromal osteoclast-like multinucleated giant cells in breast carcinoma: implications for tumour osteolysis and macrophate biology

Bone resorption by macrophage polykaryons of giant cell tumors of tendon sheath

Severe immunodeficiency disease induced by a defective leukaemia virus

Sulphated polysaccharides as potent inhibitors of HIVinduced syncytium formation: A new strategy towards AIDS chemotherapy

Three-dimensional study of the cytoskeleton in macrophages and multinucleate giant cells by quick-freezing and deep-etching method

Are glycoconjugates and their endogenous receptors involved in the fusion of mononuclear macrophages resulting in multinucleate giant cells?

Germ-cell tumours of the ovary

Inocculation of baboons and macaques with SIV/Mne, a primate lentivirus closely related to HIV-2

High expression of multinucleate giant cells in cultures of peripheral cells HIV patients

Formation of multinucleate giant cells in organized epitheliod cell granulomas

Culture of guinea pig blastocyst

Isolation of a type D retrovirus from B-cell lymphomas of a patient with AIDS

Structural organization of unique retrovirus-like particles budding from human teratocarcinoma cell lines

Ultrastructural studies of a visna-like syncytia producing virus from cattle with lymphocytosis

Ultrastructural comparison of oncovirinae (Type C), spumavirinae, and lentivirinae: three subfamilies of retroviridae found in farm animals

Production of tumors in hamsters by cells of a pig uterine tube line

Type C virus production by a continuous line of pig oviduct cells (PFT)

Replication of the bovine immunodeficiency-like virus in diploid and aneuploid cells: permanent, latent and virus-productive infections in vitro

The H1V cytopathic effect--potential target for therapy

Production of virions with retrovirus morphology by human embryonal carcinoma cells in vitro

Non-productive subacute sclerosing panencephalitis (SSPE) virus of human and ferret: an ultrastructural study

Brain pathology induced by infection with the human immunodeficiency virus (HIV)

Musculoskeletal manifestations of infection with human immunodeficiency virus

Value of lymph node biopsy in unexplained lymphadenopathy in homosexual men

Centrioles, microtubules and microfilaments in activated mononuclear and multinucleate macrophages from rat peritoneum: Electron-microscopic and immunofluorescence microscopic studies

Sarcoid macrophage giant cells. Ultrastructure and lysosome content

Morphological evidence of syncytial formation from the cytotrophoblast cells

Multinucleate Giant Cells

Fusion of normal primate cells: a common biological property of the D-type retroviruses

Biological features of HIV-I that correlate with virulence in the host

Further studies of HIV morphology by negative staining

Human immunodeficiency virus infection of monocytic and T-lymphocytic cells: Receptor modulation and differentiation induced by phorbol ester

A mutant of human immunodeficiency virus with reduced RNA packaging and abnormal particle morphology

An electron microscopic study of moderate and virulent virus cell interactions of the parainfluenza virus SV5

Virus-like particles in embryos and female reproductive tract

A new type D retrovirus isolated from macaques with an immunodeficiency syndrome

Isolation of T-cell tropic HTLV-III-Iike retrovirus from macaques

Comparative embryology of mammalian blastocysts

Expression of HIV antigens at the surface of infected T4 cells: lmmunoelectron microscopic evidence of an immunogenic phase prior to the viral release

Biology of Disease: microglia in human disease, with an emphasis on acquired immune deficiency syndrome

The three-dimensional structure of the clear zone of a cultured osteoclast

Syncytia--a major site for the production of human immunodeficiency virus

Further observations on the numbers of spermatogonia, spermatocytes and spermatids connected by intracellular bridges in the mammalian testis

Gestational trophoblastic disease

The fine structure of the blastocyst

The human immunodeficiency virus: infectivity and mechanisms of pathogenesis

Virus particles in the basal plate of rhesus monkey and baboon placenta

C-type virus particles in placentas of rhesus monkeys after maternal treatment with recombinant leukocyte A interferon

Human immunodeficiency virus (HIV): An ultrastructural study

Distinctive replicative and cytopathic characteristics of human immunodeficiency virus isolates

H1V-1. Variability and progression to AIDS

Spumaviruses: A group of complex retroviruses

Cell fusion in tumor development and progression: Occurrence of cell fusion in primary methylcholanthrene-induced tumorigenesis

Retroviridae

Oncovirinae: type C retroviruses

A mutation on the HIV type 1 transmembrane glycoprotein gp41 dominantly interferes with fusion and infectivity

Morphological differentiation of human choriocarcinoma cells induced by methotrexate

Monoclonal antibodies that neutralize HIV-1 virions and inhibit syncytium formation

Ultrastructure of hybrids derived from electric pulse fusion of human HeLa cells and murine 3T3 4E cells

Alteration of the pH dependence of coronavirus-induced cell fusion: effect of mutations in the spike glycoprotein

Molecular and biological parameters of membrane fusion

Oncovirinae: type D oncovirus

Assembly and morphology of HIV--potential effect of structure on viral function

Spumavirinae

Morphogenesis and morphology of HIV

Efficient isolation and propagation of human immunodeficiency virus on recombinant colony-stimulating factor 1-treated monocytes

The macrophage in the persistence and pathogenesis of HIV infection

Interaction of cultured human Kupffer cells with HIV-infected CEM cells: an electron microscopic study

Ultrastructural Pathology of the Cell and Matrix

Ultrastructure of osteogenic sarcoma

Osteoclastlike giant cell tumor of the pancreas immunophenotypic similarity to giant cell tumor of bone. Human Path. 22

The budding of defective HIV-type 1 particles from cell clones persistently infected with HIV-I

Mechanism of control of trophoblast invasion in situ

Interference with HIV-induced syncytium formation and viral infectivity by inhibitors of trimming glucosidase

Secretion of neurotoxins by mononuclear phagocytes infected with HIV-I

AIDS pathogenesis. HIV envelope and its interaction with cell proteins

Inhibition of infection with HIV type-1 by sulfated gangliosides

Infection of HTLV-III/LAV in HTLV-I-carrying cells MT-2 and MT-4 and application in a plaque assay

Cell Fusion

Ultrastructural changes in nuclei within HIV-l-induced cultured cell syncytia

The evolution of placental mammals

Viral release from HIV-l-induced syncytia of CD4 + C8166 Cells

Ultrastructural changes in HIV-induced syncytia of cultured cells

Molecular biology of HIV-1. ln: AIDS and the New Viruses, Dalgleish

Molecular biology of the human immunodeficiency virus type 1

Infection of monocyte/macrophages by human T lymphotropic virus type IIl

Conformational epitope on Gpl20 important in CD4 binding and HIV-1 neutralization identifed by a human monoclonal antibody

Spatial visualization of the maturing HIV-1 core and its linkage to the envelope

Sendai virion structure and its interaction with cellular membranes

Detection of lymphocytes producing human retrovirus associated with adult T-cell leukemia by syncytia induction assay

Virus~rythrocyte interactions

A soluble CD4 protein selectively inhibits H1V replication and syncytium formation

Viral Cytopathology

The origin of osteoclasts: evidence clinical implications and investigative challenges of an extraskeletal source

Ultrastructure of alveolar bone during tooth eruption in the dog

The giant cells recruited by subcutaneous implants of mineralized bone particles and slices in rabbits are not osteoclasts

Bone cell biology: the regulation of development, structure, and function in the skeleton

Effects of succinylated concanavalin A on infectivity and syncytial formation of human immunodeficiency virus

Latent HIV-1 infection in enriched populations of blood monocytes and T-cells from seropositive patients

Interluekin 4 induces cultured monocyte/macrophages to form giant multinucleated cells

Transport of the Balbiani ring granules through nuclear pores in Chironomus tetans

Replication and regulation of primate foamy viruses

Osteoclast origin of giant cells in giant cell tumors of bone. Ultrastructural and cytochemical study of six cases

Early activation events render T-cells susceptible to HIV-l-induced syncytia formation--role of protein kinase C

Comparative ultrastructure of hatched human, mouse and bovine blastocysts

Inhibition of HIV type-l-induced syncytium formation and cytopathicity by complestatin

Selective syncytium formation by murine leukemia virus in rat 3Y1 fibroblasts transformed by adenovirus type-12 or its E1A gene

A new human Tlymphotropic retrovirus: characterization and possible role in lymphadenopathy and acquired immune deficiency syndromes

Persistent coinfection of T lymphocytes with HTLV-II and HIV and the role of syncytium formation in HIV-induced cytopathic effect

Phorbol ester binding in isolated muscle satellite cell compared to fetal myogenic cells from rat

Cytokineinduced generation of multinucleated giant cells in vitro requires interferon and expression of LFA-1

The env gene of an infectious noncytopathic HIV-2 is deficient in syncytium formation

Human T-cell Leukemic virus type 1: Induction of syncytia and inhibition by patients sera

Ultrastructural behaviour of human immunodeficiency virus (HIV) in multinucleated giant cells in the brain of a japanese hemophiliac presenting

Simple, rapid, quantitative syncytium forming microassay for the detection of HIV neutralizing antibody

Biology and pathogenesis of lentiviruses

Normal human placentas contain RNA-directed DNA polymerase activity like that in viruses

Chondroosteoid breast tumor with multinucleate giant cells

Cross neutralization of ovine and bovine C-type leukemia virus-induced syncytia formation

Ultrastructural and immunological studies of virus particles observed in a T-cell line derived from a mouse spontaneous lymphoma

Human immunodeficiency virus and papovavirus infections in acquired immunodeficiency syndrome: an ultrastructural study of three cases

Cytoplasmic assembly and accumulation of human immunodeficiency viruses types 1 and 2 in recombinant human colony-stimulating factor-l-treated human monocytes: an ultrastructural study

A particularly aggressive placental site trophoblastic tumour

The osteoblast and osteoclast cytodifferentiation

Oligopeptide inhibitors of HIVinduced syncytium formation

Effect of Evans Blue and Typan Blue on syncytia formation and infectivity of human immunodeficiency virus type I and type II in vitro

C-type virus expression in the placenta

Dissociation between syncytia formation and HIV spreading--suppression of syncytia formation does not necessarily reflect inhibition of HIV infection

The morphology of murine foreign body multinucleate giant cells

The detection of a contractile apparatus in murine multinucleate giant cells

HIV type-1 infection of U937 cells promotes cell differentiation and a new pathway of viral assembly

Muscle development

Insect sperm: their structure and morphogenesis

Syncytial giant-cell hepatitis

Osteoclasts: structure and function

The origin and function of human syncytiotrophoblast giant cells

An ultrastructural study of differentiation and maturation of trophoblast in the monkey

Stimulation of macrophage growth and multinucleate cell formation in rat bone marrow cultures by insulin-like growth factor-l Biochc~

Human Pa/h. 16, 760. Sinkovics. J. G., 199 I. Hodgkin's disease revisited: Reed--Sternberg cells as natural hybridomas

Expression of C-type viral particles at implantation in the marmoset monkey

Deep fibrous histiocytoma with giant cells and bone metaplasia

Role of the HYLV-III!LAV envelope in syncytium formation and cytopathicity

Multinucleation of macrophages with water-solubilized lignin derivatives in t>irro

Virus-induced cell fusion

Inhibition of virus-induced cell fusion by apolipoprotein A-l and its amphipathic peptide analogs

Human osteogenic sarcoma: Fine structure of theosteoblastic type. l'l~ras~ruct

Protein-mediated membrane fusion

Function of cytoplasmic fibers in syncytia

Giant cell carcinoma of lung, A light and electron microscopical study

Occurrence of numerous giant cells in the tonsils and pharyngeal mucosa in the prodromal stage of measles

Analysis of adhesion molecules in the immunopathogenesis of giant cell arteritis

A chimpanzeepassaged human immunodeficiency virus isolate is cytopathic for chimpanzee cells but does not induce disease

Specific tropism of HIV-1 for microglial cells in primary human brain cultures

Ultrastructural studies on medi-visna virus

Electron microscopical studies on equine infectious anemia virus (EIAV)

Autoantibodies against CD4 cells are associated with CD4 helped in HIV-infected patients

Human T-cell retroviruses

Inhibition of human immunodeficiency virus type l-induced cell fusion by recombinant interferons

Cell-to-cell fusion

Cellular localization of human in immunodeficiency virus infection within the brains of acquired immune deficiency syndrome patients

Ultrastructural features of the dental follicle associated with formation of the tooth eruption pathway in the dog

Immunological implications of comparative placental ultrastructure

Fusion as a mediator of cytolysis in mixtures of uninfected CD4 + lymphocytes and cells infected by human immunodeficiency virus

Characterization of cell fusion in XC cells induced by Sancur murinus mammary rumor virus

HIV-I entry into quiescent primary lymphocytes: molecular analysis reveals a labile, latent viral structure

Fusion of erythrocytes and other cells with retention of erythrocyte cytoplasm. Nuclear activation in chicken erythrocyte-melanoma heterokaryons

Reorganization of cell structures during cell fusion

A quantitative cytochemical investigation of osteoclasts and multinucleate giant cells

Hyperplasia and hypertrophy of Leydig cells associated with testicular germ cell tumors containing syncytiotrophoblastic giant cells

Acknowledgements--I would like to acknowledge the assistance and encouragement received from Professor Werner W. Franke and his colleagues in the Institute of Cell and Tumor Biology at the German Cancer Research Center, Heidelberg, throughout the preparation of this review. HIV-1 infected lymphoblastoid cell cultures were maintained by Mr A. Kitchen (N. E. Thames Regional Transfusion Centre, Brentwood, Essex) and processed for electron microscopy by Mr G. Tovey (London School of Hygiene and Tropical Medicine). I am extremely grateful to the many scientists worldwide who have freely made available electron micrographs for inclusion in this review, and for their direct assistance in some cases, thereby enabling this broad survey on MNGCs to be possible.

 Ichikawa, Y. and Dales, S., 1971 . Biogenesis of poxviruses: interrelationship between hemagglutinin production and polykariocytosis. Virology 46, 533-543. Imamura, M., Phillips, P. E. and Mellors, R. C., 1976 Jauniaux, E., Nessmann, C., Imbert, C., Meuris, S., Puissant, F. and Hustin, J., 1988 . Morphological aspects of the placenta in HIV pregnancies.Placenta 9, 633-642.Johns, T. C. and Renagar, R. H., 1990 . Ultrastructural morphology and relaxin immunolocalization in giant trophoblast cells of the golden hamster placenta. Am. J. Anat. 189, 167-178. Jones, C. J. P. and Fox, H., 1991