key: cord-0807815-13dey1ao
authors: Pagano, Nathan; Laurent-Rolle, Maudry; Hsu, Jack Chun-Chieh; Vogels, Chantal BF; Grubaugh, Nathan D; Manuelidis, Laura
title: Long SARS-CoV-2 nucleocapsid sequences in blood monocytes collected soon after hospital admission
date: 2021-02-19
journal: bioRxiv
DOI: 10.1101/2020.12.16.423113
sha: 9b2d050e65a3d7d22e93d357ba3934e1d220d9c7
doc_id: 807815
cord_uid: 13dey1ao

Many viruses infect circulating mononuclear cells thereby facilitating infection of diverse organs. Blood monocytes (PBMC) are being intensively studied as immunologic and pathologic responders to the new SARS-CoV-2 virus (CoV19) but direct evidence showing CoV19 in monocytes is lacking. Circulating myeloid cells that take up residence in various organs can harbor viral genomes for many years in lymphatic tissues and brain, and act as a source for re-infection and/or post-viral organ pathology. Because nucleocapsid (NC) proteins protect the viral genome we tested PBMC from acutely ill patients for the diagnostic 72bp NC RNA plus adjacent longer (301bp) transcripts. In 2/11 patient PBMC, but no uninfected controls, long NCs were positive as early as 2-6 days after hospital admission as validated by sequencing. Pathogenic viral fragments, or the infectious virus, are probably disseminated by rare myeloid migratory cells that incorporate CoV19 by several pathways. Predictably, these cells carried CoV19 to heart and brain educing the late post-viral pathologies now evident.

Some acute lytic viral infections release free viral particles into the bloodstream where they are easily assayed with modern molecular techniques. However, a large variety of viruses travel within white blood cells such as retroviruses, e.g., HIV (1, 2), and flaviviruses, e.g., Dengue (3) . Some viruses carried in white blood cells are associated with chronic disease. Poliovirus, a "neurotropic" enterovirus originally thought to spread directly through nerves, instead transits from gut to white blood cells to spleen, and only later infects brain (4) . There are at least 13 different classes of DNA and RNA viruses that infect peripheral blood mononuclear cells (PBMC) that can be a reservoir for persistence (5) in addition to the infectious Creutzfeldt-Jakob Disease (CJD) agent (6) . Like HTLV1, a retrovirus that can be sequestered in brain myeloid microglia, the CJD agent in white blood cells progressively increases after primary infection of gut dendritic (myeloid lineage) cells to later show up in highly infectious myeloid microglia of brain (7, 8) .

Experimentally, the CJD agent requires cell-to-cell contact for infection (9) , probably via viral synapses, the same mechanism or conduit used for T to myeloid cell transmission of HIV (1).

Moreover, chronic HIV infection of brain microglia is linked to neurocognitive disorders and dementia, as is the very different ~20-25nm CJD infectious particle (10, 11) . These observations emphasize the general principle that infected migrating myeloid cells can take up residence in and perpetuate infection and chronic pathology in the brain and other organs. The presence of nucleocapsid CoV19 RNA, especially in migrating myeloid cells, might explain some of the ensuing brain and heart pathologies that were predictable and that have become increasingly evident during this pandemic Coronavirus 2019 Disease .

We were aware that PBMC could contain extremely low CoV19 RNA from infection limited to only rare myeloid cell types. Those cell types can be less abundant than the 1-2% dendritic cells in a PBMC population that is typically dominated by lymphocytes (70-90%). Many different myeloid cell-types are increasingly appreciated and classified, and many of them show wide phenotypic flexibility (12) . Moreover, total cellular RNA is very low in PBMC, ~1/40th of that produced by epithelial and cultured tumor cells such as HeLa cells and NIH/3T3 cells.

Nevertheless, there were three additional reasons to pursue this study. First, coronaviruses are complex, and can elicit immune responses that damage brain. For example, the mouse hepatitis coronavirus (MHV) induces a post-viral autoimmune demyelinating disease that is a model for Multiple Sclerosis, and other coronavirus strains, such as CoV19, might elicit a different type of post-viral brain pathology. Indeed, prior to the evolution of CoV19, related respiratory 4 coronaviruses that cause human colds have been neuroinvasive, with viral RNA demonstrated in brain parenchyma as well as in myeloid microglia in culture (13) .

We proposed to study CoV19 in PBMC in April 2020 because little was known about potential immediate viral or late immunologic CoV19 effects on the brain. We suspected that a subset of blood monocytes, not just neural olfactory spread, might be a conduit for CoV19 into brain with subsequent development of neuropsychiatric symptoms. A wide variety of neurologic complications and neuropathology have now been published, e.g., thromboemboli, infarction, radiologic changes consistent with an autoimmune encephalitis (14) and even the presence of CoV19 in neurons (15) . Additionally, routine autopsies of lethal sudden deaths during uncharacterized winter "viral" infections can show classical viral acute lymphocytic and myeloid infiltrates in a normal heart and the recent collapse of healthy young athletes appears to be associated with CoV19 vascular dissemination with microthrombi in the heart and/or brain.

Finally, LM had the opportunity to join a zoom meeting at Yale with doctors from Wuhan who shared their data and experience with us early in 2020. One slide showed a striking interstitial pneumonia dominated by many cells consistent with a myeloid lineage. This further encouraged us to find if PBMC infection could be the source of multiorgan spread. We here identify >300nt of CoV19 nucleocapsid RNA in PBMC. These cells can be a conduit for dissemination of the virus, or viral elements, to other organs. Myeloid cells can use several known pathways for CoV19 uptake facilitating its protection and pathological persistence.

PBMC from 7 male and 7 female patients ranging from 24-82 years old and 4 random uninfected control volunteers were evaluated. The 14 patient PBMC were collected 2-6 days after hospital admission with positive CoV19 swab test. To avoid RNA extraction losses in low RNA PBMC, exhaustive DNA removal was not pursued. PBMC recovery can vary, especially in patients with other underlying illnesses. Thus, all PBMC RNA extracts (from control and CoV19 positive patients) were tested for GAPDH by RT/qPCR to assess amounts of RNA in each; 11 patient samples were adequate for further nucleocapsid (NC) assays, and showed a mean of 69% (+35%SD) of the "normal" controls. The other 3 patients yielded only 1/10 th the control RNAs and were insufficient for analyses. CoV19 nucleocapsid RNA was initially evaluated with the standard clinical CDC:N1 primer pair/probe that amplifies a 72bp sequence beginning 14nt downstream from the NC start codon (in red, Table I This indicated this 72nt NC RNA was present in a low percentage of PBMC. One other patient sample tested in parallel (14*) was ambiguous in this test with a Ct of 37.19 in one of three replicates. All other patient samples were negative for the 72nt RNA. We repeated these 72bp RT/qPCRs in duplicate, and only patient 18* was comparably positive with Cts of 34.2 and 34.7.

The 4 control normal samples (designated 1n-4n), and the 11 patient samples (all designated by an *), were subsequently tested for longer adjacent stretches of the NC that covered an additional 301bp as detailed in Table I (see Methods) . A single forward primer (F3) was used to generate both a shorter 251bp sequence (with primer R5) as well as a 301bp sequence (with primer R3). This allowed sequencing from two different PCR reactions for validation.

Melting profiles of RT/qPCR products with these primer pairs, using serial dilutions of positive controls along with H2O negative controls, revealed significantly different melting temperatures (Tm) for each primer pair product. Fig. 1 shows the clearly different Tm profiles of the 72bp versus the 251 and 301bp amplifications. The CDC:N1 primers melted at 81.45 o C (+0.07 SD) and had a broad profile. In contrast, the 251bp F3/R5 product gave a significantly higher and sharper Tm of 83.7 o C (+ 0.06 SD). As expected, the longest 301bp product had the highest Tm (84.19 o C+0.13 SD) with a very sharp and unambiguous curve. This allowed clear identification of real positives from background primer-dimers and other nonspecific artefacts by their Tms. In the CDC:N1 and F3/R3 graphs of standards, the water negative control shows no true melt and only a single profile was obtained. However, in less purified samples Tm peak artefacts were sometimes seen and these typically occurred in PBMC RNA extracts that contained DNA. These nonspecific and variable peaks sometimes precluded simple valid digital Cts. We therefore used the following criteria to validate CoV19-specific peaks: 1) a Tm or peak profile of the PBMC RT/qPCR that matched the positive controls, 2) a minimum of both a positive Tm peak and positive gel band of correct length verified in >2 separate aliquots of each patient RNA sample, and 3) sequencing of patient gel bands that matched those from bands of parallel controls including the nucleocapsid genome (G), the e3 reference virus, as well as CoV19 infected Vero cell RNA (V+). Uninfected Vero cell RNA (V-) was also used as a negative total cell RNA control. 18* gel band shown is also positive, and from a different aliquot that was run with it's parallel H2O (w) control (lanes [13] [14] . The 14* band at 251bp is weak but visible, and shows a smear of background that is consistent with non-DNA. Higher loads of this sample on another gel showed a very strong 251bp band (data not shown). Moreover, the RT/qPCR peak profile of the 14* amplified product shown above also revealed an overwhelmingly dominant peak with a Tm matching the other positive controls (see However, this artefactual peak's Tm and profile was clearly different in each of these two aliquots. Fig. 2A (lane 9) , it showed the same dominant Tm peak as the nucleocapsid genome (G), and its Tm peak was clearly much stronger than the nonspecific background. The small artefactual peaks of patient 23* were clearly artefactual by Tm, and were easily thresholded.

Accurate thresholding were more problematic for the longer 301bp product NC primers, (Fig. 3B) . In this profile, H2O displays an broad nonspecific peak starting at ~80 o C and this is 7 clearly due to a primer dimer of ~50bp. This dimer also is seen in patient samples 14* (purple) and 18* (red) beneath the largest H2O peak. The positive 14* sample's 301bp fluorescent peak lies under the other positive peaks, including the NC control (in green), and has an indistinguishable Tm of 84.19+0.13SD the same as other positive controls in Fig. 1 . Another apparently identical nonspecific gel band of ~350bp is seen in both the 14* and 18* samples (Fig. 2, lanes 11 and 14) yet each of these peaks had clearly different Tms (Fig. 3B, arrows) . This confirmed that these 14* and 18* peaks differed in sequence, making cross-contamination of the two samples unlikely. This 

The fact that no one else has reported direct reproducible detection of CoV19 RNA sequences in PBMC indicates spread via these cells has not been widely considered even though 8 such cells provide the most fundamental mechanism for disseminating many viruses to distant organs. The use of NC primers extending from the 72bp NC RNA revealed strong and unambiguous evidence that long CoV19 NC sequences were present in 2/11 patient PBMC. The two positive patients (14* and 18*) were validated by sequences that were completely homologous three different positive internal controls, as well as CoV19 NC sequences in the database. Further studies to determine which PBMC cell type concentrates CoV19 sequences is also critical. While we were limited to a small sample of patient PBMC, with only ~6e4 cell RNA for 2-3 repeat replicate studies each long NC, it is apparent that only a few select cells (<1/200) carry these sequences, we presume these are a subset of myeloid cells. Unfortunately, we could not obtain more of our positive patient samples to assess combined in-situ CoV19 antigen/nucleic acid hybridization tests with cell-type specific markers. Nor could we determine if any of the positive cells carry fully infectious CoV19, and not just the nucleocapsid RNA, as by culture or susceptible animal inoculation. Nevertheless, the idea that myeloid lineage circulating cells can be the conduit for dissemination and progressive or chronic disease is linked to their ability to penetrate and take up residence in tissues through the vascular endothelium. Their internalized viral transcripts may also be better protected in brain, and at the same time, as antigen presenting cells, incite 9 inflammatory neurodegenerative changes. Studies of other SARS and MERS coronaviruses have addressed these ACE2 and ACE2-independent monocyte and macrophage pathways (15) . These coronaviruses also induce a number of ultrastructural membrane modifications in cultured cells that may organize and protect viral assemblies (16) . The mechanism of CoV19 uptake is likely to be multifactorial in animals with many interrelated cell types, and the huge genetic complexity of coronaviruses should accommodate more than one entry mode. For myeloid cells, uptake of viruses via endocytic membranes with internal sequestration is commonplace, possibly nonspecifically, as a response to recognition of the large CoV19 particle structure, or by myeloid to infected epithelial or endothelial cell attachment. Finally, a selected subset of human migratory myeloid cells in blood have demonstrated Angiotensin converting enzyme 2 (ACE2) that the CoV19 spike protein uses as a receptor for cellular entry (17) . Therapeutic approaches that target such cells may prevent early organ dissemination of Cov19, in part or whole, and its associated pathological sequelae. A total of 18 YNHH peripheral blood monocytes (PBMC) samples were studied. The PBMC were collected and prepared by YNHH nurses, technicians and the Yale IMPACT Research Team from fresh blood and PBMC were separated using standard Ficoll-Paque™ methods, then then frozen at -80 o C in DMSO-human serum; they had a nominal estimate of 5e6 cells/ml/sample. We selected PBMC samples taken as close to admission as possible to avoid unknown effects of any hospital treatments or progressive disease on viral replication or spread.

All patient PBMC samples here were taken within the first week after admission and age range, gender and predisposing risk factors were broadly represented. Patients who subsequently were admitted to the ICU or intubated were excluded.

Individual PBMC in DMSO-human serum were thawed briefly in a 37 o C water bath in the neuropathology biohazard hood BSL2+ lab, mixed gently by pipetting using 1ml pipettor tips, and transferred to 8ml MEM at 22 o C with only 4-6 samples processed per day. The diluted samples were centrifuged at 700g x 10min at 22 o C. The NEB Total Miniprep kit was then used as suggested with the following modifications. First, lysis of cells was carefully monitored to minimize breaking large chromatin strands. After discarding each diluted serum-DMSO supernatant, the pellet was tapped to suspend it in residual MEM, 700µl lysis buffer was added and cells allowed to lyse for ~30" followed by gentle up/down pipetting as needed to disperse lysates using a 1ml tip. Second, only the first "g" column that removes large genomic DNA was used and further DNAse treatment and column purification was avoided because total yields of PBMC RNA could be <1µg in 5e6 cells. The g column supernatant then bound to the RNA binding column and eluted in ~90µl H2O; the top ~65µl away from any residual silica fines or molecular aggregates that were collected and this supernatant recentrifuged. Half (30µl) was used for GAPDH quantification and anti-viral response studies (ML-R and JC-CH) as well as for a standard analysis in triplicate of the CDC 72nt NC primer-reporter assay by the Grubaugh lab (18) .

The other half was divided into subaliquots and frozen for further nucleocapsid (NC) tests. PBMC RNA extracts represented maximally 2.5e4 cells/µl assuming a 100% recovery.

A CoV19 nucleocapsid (NC) RNA, a full length CoV19 viral control at e3 from UTMB, and mock infected and CoV19 infected total cell RNA were used as controls. The CoV19 RNA standard for the NC segment was generated as described (18) and CoV19 infected (V+) RNAs were used as paired total cellular RNA controls.

RT/qPCR: All 18 control and patient PBMC samples were initially tested in triplicate RT/qPCR in the Grubaugh lab (18) using the CDC primer/reporter pairs as described and 11 compared to dilutions of the NC RNA and viral e3 RNA as standards. These standards were also used for assays done in duplicate on both Stratasys and MyGo mini qPCR machines. In addition, control mock-infected (V-) and infected (V+) Vero cell RNAs were used as paired standards with longer NC primer pairs. In these, as well as GAPDH RT/qPCR, we used the Luna Universal One-

Step with polishing and melt as for GAPDH above, again using the NEB LUNA One-step RT/PCR kit.

The longer 301bp reverse R3 primer with a lower Tm was the same except the 45 cycle phase used 57 o C x 10" for annealing followed by a 59 o C x 30" extension. These reactions typically were done in a total of 14µl with 2-3µl PBMC extract to conserve sample for repeat confirmations.

Purification and sequencing of the NC 251bp and 301bp bands: Each RT/qPCR (0.5 or 1 µl sample) was resolved on mini 2.5% agarose TAE gels, the DNA stained with SYBR gold and the fluorescence imaged digitally in color using a blue emission transilluminator box with an orange barrier filter. Band purification was done as described (20) . Briefly, the 251bp and 301bp RT/qPCR bands were cut from 2.5% preparative gels (~5µl sample load), crushed in individual mini weigh boats under parafilm and transferred to 1.5ml polypropylene tubes. Then 200-300µl H2O added prior to freeze-thawing and centrifugation at 13,000g for 10'. Eluted supernatants were collected with a 10µl tip to avoid small gel fragments. H2O elution/freezing was repeated two more times and SYBR gold stained DNA was followed. More than 70% of the starting band fluorescence appeared in the supernatants while the final residual agarose fluorescence was weak.

An aliquot of each eluate was taken for PCR amplification and sequencing using the above NC primers, followed by a second round of band purification. SYBR gold did not interfere with sequencing; some eluates concentrated by lyophilization were washed with 70% EtOH before suspending in 50 µl H2O. These had no fluorescence and yielded the identical sequences.

FIG. 1: Melt temperature (Tm) of RT/qPCR nucleocapsid (NC) products using different NC primer pairs. The short CDC:N1 product of 72bp has the lowest Tm while the longest F3/R3 pair (301bp) has the highest Tm. Each primer pair Tm product is significantly different (see text). Some nonspecific bands are sample specific, e.g., patient 12* (~150bp band in A) Each duplicate in B is also from a different run with a different aliquot of that extract. A dot is at 300bp in the 100bp ladder marker lanes. SYBR Gold fluorescent DNA is inverted for clarity.

Tm of representative peaks from above gel where A) shows the 251bp product. A few smaller peaks with different Tms than the 251bp band could be thresholded. Note the consistency of all NC positive band peaks. The low fluorescent dimer at a Tm at ~81 o C is seen in 2 samples (longer black arrow) and a longer faint product, visible on the gel (>300bp) in 23* is also apparent (short orange arrow) at a Tm of ~89 o C. B) demonstrates background and non-specific primer dimer peaks, including in the H2O control (aqua) that interfered with quantitative Cts. The higher Tm peaks in 14* and 18* on gels looked identical, but had big Tm differences indicating they were sample specific artefact bands. Table I : NC amplified sequences in color downstream from the MET start codon and shows the standard 72 primers and sequence specific FAM probe in red. The 251 and 301bp products sequences are colored with respective primer pairs underlined in different colors. R3 is in green and R5 in orange. The 251bp product is in aqua and the 301bp longer product ends at R5. These

Illuminating the Role of Vpr in HIV Infection of Myeloid Cells

The HIV Reservoir in Monocytes and Macrophages

Interplay between Keratinocytes and Myeloid Cells Drives Dengue Virus Spread in Human Skin

Poliomyelitis virus in blood of orally infected monkeys and chimpanzees

Monocytes and Macrophages as Viral Targets and Reservoirs

Viremia in experimental Creutzfeldt-Jakob disease

Neuroinvasion by a Creutzfeldt-Jakob disease agent in the absence of B cells and follicular dendritic cells

Microglia from Creutzfeldt-Jakob disease-infected brains are infectious and show specific mRNA activation profiles

Reciprocal interference between specific CJD and scrapie agents in neural cell cultures

Cells infected with scrapie and Creutzfeldt-Jakob disease agents produce intracellular 25-nm virus-like particles

CJD and Scrapie Require Agent-Associated Nucleic Acids for Infection

Transcriptional Classification and Functional Characterization of Human Airway Macrophage and Dendritic Cell Subsets

Neuroinvasion by human respiratory coronaviruses

Viremic dissemination of mouse hepatitis virus-JHM following intranasal inoculation of mice

Contribution of monocytes and macrophages to the local tissue inflammation and cytokine storm in COVID-19: Lessons from SARS and MERS, and potential therapeutic interventions

A unifying structural and functional model of the coronavirus replication organelle: Tracking down RNA synthesis

COVID-19 infection induces readily detectable morphological and inflammation-related phenotypic changes in peripheral blood monocytes, the severity of which correlate with patient outcome

Analytical sensitivity and efficiency comparisons of SARS-CoV-2 RT-qPCR primer-probe sets

Prolonged incubation of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in a patient on rituximab therapy

Nuclease resistant circular DNAs copurify with infectivity in scrapie and CJD

All data will be findable, fully accessible and freely available under contributions: NP performed nucleocapsid RT/qRCRs and associated modifications; ML-R and JC-CH infected Vero Cells, extracted their RNA and did the GAPDH RT/qPCRs of PBMC; CH and NG did the initial PBMC CDC-72bp assay and quantitation; LM designed the project and primers, extracted PBMC RNA, did gel analyses of amplified products and band extraction for sequencing and analysis and wrote the manuscript

PBMC samples were from the Yale IMPACT Research Team

We thank Albert Ko for encouraging LM to use Yale repository samples for this study, Arnau Casanovas-Massana who was essential for identifying and gathering early admission PBMC samples, and Shelli Farhadian and Allison Nelson for helping LM with the biorepository application. We are also indebted to the YNHH nurses and technicians who obtained and prepared the PMBC, and also to the many volunteers and patients who agreed to donate to the Yale IMPACT collection. Primer syntheses and sequencing was done by the Yale Keck laboratory. This work was supported by gifts for neurodegenerative research to LM, and the 72bp NC analysis was supported by a Yale School of Public Health startup grant to NDG. We are grateful to Paula Kavathas for her helpful suggestions on the manuscript.