key: cord-1030833-yft33obm
authors: Popescu, Iulia; Snyder, Mark E.; Iasella, Carlo J.; Hannan, Stefanie J.; Koshy, Ritchie; Burke, Robin; Das, Antu; Brown, Mark J.; Lyons, Emily J.; Lieber, Sophia C.; Chen, Xiaoping; Sembrat, John C.; An, Xiaojing; Linstrum, Kelsey; Kitsios, Georgios; Konstantinidis, Ioannis; Saul, Melissa; Kass, Daniel J.; Alder, Jonathan K.; Chen, Bill B.; Lendermon, Elizabeth A.; Kilaru, Silpa; Johnson, Bruce; Morrell, Matthew R.; Pilewski, Joseph M.; Kiss, Joseph E.; Wells, Alan H.; Morris, Alison; McVerry, Bryan J.; McMahon, Deborah K.; Triulzi, Darrell J.; Chen, Kong; Sanchez, Pablo G.; McDyer, John F.
title: CD4+ T cell lymphopenia and dysfunction in severe COVID-19 disease is autocrine TNF-α/TNFRI-dependent
date: 2021-06-03
journal: bioRxiv
DOI: 10.1101/2021.06.02.446831
sha: e83fbc14becb390d613beefdbec08392dd1061a2
doc_id: 1030833
cord_uid: yft33obm

Lymphopenia is common in severe COVID-19 disease, yet the mechanisms are poorly understood. In 148 patients with severe COVID-19, we found lymphopenia was associated with worse survival. CD4+ lymphopenia predominated, with lower CD4+/CD8+ ratios in severe COVID-19 compared to recovered, mild disease (p<0.0001). In severe disease, immunodominant CD4+ T cell responses to Spike-1(S1) produced increased in vitro TNF-α, but impaired proliferation and increased susceptibility to activation-induced cell death (AICD). CD4+TNF-α+ T cell responses inversely correlated with absolute CD4+ counts from severe COVID-19 patients (n=76; R=-0.744, P<0.0001). TNF-α blockade including infliximab or anti-TNFRI antibodies strikingly rescued S1-specific CD4+ proliferation and abrogated S1-AICD in severe COVID-19 patients (P<0.001). Single-cell RNAseq demonstrated downregulation of Type-1 cytokines and NFκB signaling in S1-stimulated CD4+ cells with infliximab treatment. Lung CD4+ T cells in severe COVID-19 were reduced and produced higher TNF-α versus PBMC. Together, our findings show COVID-19-associated CD4+ lymphopenia and dysfunction is autocrine TNF-α/TNFRI-dependent and therapies targeting TNF-α may be beneficial in severe COVID-19. One Sentence Summary Autocrine TNF-α/TNFRI regulates CD4+ T cell lymphopenia and dysfunction in severe COVID-19 disease.

Severe viral pneumonia and respiratory disease due to SARS-CoV-2 infection has been the major clinical manifestation associated with patient mortality during the COVID-19 pandemic(1-3). The progression of upper respiratory symptoms to severe viral pneumonia, and at times the adult respiratory distress syndrome later in the course of infection in a subset of patients, have led many investigators to hypothesize an important role for inflammatory mediators or 'cytokine storm' in the course of disease, as multiple studies have shown increased systemic levels of cytokines during moderate and severe COVID-19 disease (4, 5) . Treatment with the corticosteroids, dexamethasone and hydrocortisone, were found to reduce mortality in severe COVID -19 pneumonia and suggested that immune modulation could impact severe disease (6, 7) . Several subsequent studies of IL-6 receptor antagonists in severe COVID pneumonia have found mixed results with some showing benefit (8) (9) (10) , while others did not find efficacy (11) (12) (13) .The recent NIH treatment guidelines recommend use of tocilizumab in severe COVID-19 with rapidly decompensating respiratory status in combination with corticosteroids https://www.covid19treatmentguidelines.nih.gov/statement-on-tocilizumab/. However, the role for immune modulators in severe COVID-19 disease remains incompletely defined, as ongoing studies have not yet been completed.

An early report correlated lymphopenia with poor outcomes in COVID-19 and disease severity (14) .

However, the mechanisms leading to lymphopenia in COVID-19 clinical syndrome remain poorly understood (15) . Recent studies of the peripheral T cells during SARS-COV-2 infection have found an activated phenotype in both CD8 + and CD4 + T cells in severe disease, including increased surface expression of CD38 + , CD95 + , HLA-DR + , Ki67 + and programmed cell death protein 1 (PD-1), compared to mild disease and noninfected normal controls (16, 17) . One study found that more profound lymphopenia was associated with increased serum levels of the inflammatory cytokines IL-6, IL-10 and TNF- (18) . Another study found that IL-5 6 levels negatively correlated with cytotoxic immune cells in severe COVID-19 disease (19) . We hypothesized that factors such as T cell activation and the inflammatory milieu together, contributed to the development of COVID-19-associated lymphopenia during severe disease. We further reasoned that other T cell responses to SARS-CoV-2 proteins would be detected in mild and severe disease, in addition to spike, as an earlier studies have demonstrated a broad response to SARS-CoV-2 epitopes across CD4 + and CD8 + T cells (20, 21) . We hypothesized that SARS-CoV-2-specific T cell responses might play an important role in the development of T cell lymphopenia, as in other viral infections such as HIV that result in activation-induced cell death (AICD) (22) .

Here, we found that COVID-19-associated lymphopenia in severe disease is disproportionately a CD4 + T cell lymphopenia and is associated with increased mortality, with significantly reduced peripheral CD4 + /CD8 + T cell ratios in severe disease compared to recently recovered mild COVID-19 patients. We further show that the immunodominant response of CD4 + T cells is S1-specific production of the pro-inflammatory Type-1 cytokine, TNF-, in severe COVID-19 disease compared to control mild COVID-19 disease patients. We observed impaired CD4 + T cell proliferation and AICD via TNFRI signaling that could be rescued in vitro with various TNF- blockade agents. Similarly, reduced CD4 + numbers and S1-specific TNF- predominant responses were detected at higher frequencies in resident lung T cells from patients with recent severe COVID-19 pneumonia.

Together, our findings show that CD4 + T cell lymphopenia and dysfunction in severe COVID-19 respiratory disease is autocrine CD4 + TNF-/TNFRI-dependent. 6 

We hypothesized that peripheral lymphopenia was associated with poor outcomes in patients with severe COVID-19 disease. We evaluated a multi-hospital cohort (n=148) within our medical system of severe patients hospitalized with documented SARS-CoV-2 infection, with demographic data shown in Table S1 . We evaluated the first absolute lymphocyte count (ALC) on hospitalization and found the median to be 700/mm 3 (Fig. 1A) . We next assessed 30-day mortality by Kaplan-Meier and observed increased mortality in those below the median ( Fig. 1B) . We performed flow cytometry on isolated PBMC from both groups (n=76 total; gating strategy Fig. S1A) and observed a significant diminution in CD4 + T cell frequencies and reduced CD4 + /CD8 + T cell ratios in patients with lymphopenia ( Fig. 1C-E) . We observed significantly reduced absolute CD4 + and CD8 + counts (though less profound) (Fig. 1F , G), based on ALC (Fig. S1B ). Together, these data indicate that COVID-19associated lymphopenia is a predominantly CD4 + T cell lymphopenia. To further evaluate T cell function and phenotype, we randomly selected a sub-cohort of 24 severe COVID-19 patients, of which n=9 had lymphopenia ( Fig. S1C) , and a control cohort (n=24) of mild COVID disease (Table 1) (Fig. 1H ), immediately after resolution of symptoms and undergoing evaluation for potential plasma donation. While blood counts were not available for mild, recovered COVID patients we observed significantly higher CD4 + /CD8 + ratios compared to our severe COVID-19 sub-cohort (Fig. 1H ).

We next assessed COVID-specific and superantigen-induced T cell effector responses (Staph. Enterotoxin B-SEB) in our mixed COVID-19 cohort (mild and severe disease; total n=48). Using pooled peptides to the major SARS-CoV-2 antigens we assessed both antigenic and effector immunodominance (major T cell response) in PBMC following a 6 h in vitro re-stimulation. Spike-1 (S1) responses (TNF- and IFN-) were found to be immunodominant in CD4 + T cells compared to the other antigens (Spike-2 (S2), viral envelope membrane protein (VEMP) and nucleocapsid (NCAP)) ( Fig. 2A-C) . A similar immunodominance pattern was found for COVIDspecific CD8 + IFN- + T cell responses (Fig. S2A) . We further compared blood S1-specific effector responses between CD4 + and CD8 + T cells in our severe COVID-19 cohort and found that S1-specific TNF- responses predominated in CD4 + T cells, in contrast to S1-specific IFN- + responses in CD8 + T cells, but with similar frequencies of cytotoxic CD107a + responses (Fig. 2D) . The overall hierarchy of T cell effector responses were found to be predominant Type-1 immune responses, with little IL-17a or IL-13 COVID S1-specific responses detected. We next determined whether there were differences adjusting for COVID-19 disease severity and found that S1-specific CD4 + responses remained overall immunodominant and with higher frequencies in severe disease ( Fig. S2B) , with CD4 + TNF- + and CD4 + CD107a + responses in particular increased among severe COVID-19 patients compared to mild disease (Fig. 2E ). One exception was S1-specific CD4 + IL-2 + frequencies which were significantly reduced among severe patients compared with mild disease patients. Notably, the hierarchy of CD4 + TNF- + > IFN- + responses in severe disease was observed with S1, and VEMP to a lesser extent, but not to other COVID antigens (Fig. S2C) . Conversely, COVID-specific CD8 + IFN- + > CD8 + TNF- + cell responses were found to be significantly increased among those with severe versus mild disease, whereas CD8 + CD107a + responses did not differ (Fig. 2F) . We next compared the multifunctionality of S1-specific CD4 + versus CD8 + effector T cell responses. We found that CD4 + T cell multifunctional responses were significantly reduced in severe COVID disease compared to CD8 + cells, and with more single + CD4 + TNF- + frequencies (Fig. 2G) . CD4 + T cells demonstrated a significant reduced multifunctionality in severe versus mild COVID-19 patients, with reduced IL-2 and increased CD4 + TNF- + and CD4 + CD107a + frequencies (Fig. 2H) , in contrast to CD8 + T cell multi-functionality, which did not differ based on COVID-19 disease severity (Fig. S2D) . We found that S1-8 specific CD4 + TNF- + responses frequencies were inversely correlated with CD4 + /CD8 + ratios in our mixed COVID-19 cohort (n=48) (Fig. 2I ). Next, we measured S1-specific CD4 + TNF- + responses in our expanded number of severe COVID patients (n=76) and found this to be an inverse immune correlate of CD4 + lymphopenia ( Fig. 2J ) and CD4 + /CD8 + ratios (Fig. S2E ). In contrast, S1-specific CD4 + IFN- + or CD4 + CD107a + responses were not as strong an immune correlate with CD4 + numbers. Taken together, COVID-specific CD4 + T cells are skewed in severe COVID-19 disease to produce high levels of TNF- that inversely correlate with absolute CD4 + counts.

Because we observed CD4 + T cell lymphopenia and reduced CD4 + /CD8 + ratios among severe COVID-19 patients, we hypothesized that COVID-specific CD4 + T cell proliferative capacities contributed to low CD4 + numbers. Therefore, we assessed S1-specific CD4 + T cell in vitro proliferation at 6 days following peptide restimulation and CFSE dilution and found significantly impaired proliferative responses in severe COVID patients compared with mild disease (Fig. 3A, B) . While S1-specific CD8 + T cells proliferative responses were also comparatively reduced in the severe disease cohort versus the mild disease cohort, impaired CD4 + proliferation was more pronounced in severe disease, including in response to SEB (Fig. S3A-D) . As we detected reduced S1specific IL-2 production from CD4 + T cells in severe COVID-19, we determined whether exogenous IL-2 could rescue impaired proliferative responses. The addition of IL-2 to cultures significantly rescued S1-specific CD4 + T cell proliferation (Fig. S3E , F) as well as enhanced S1-specific CD8 + proliferative responses (Fig S3G) , suggesting a relative IL-2-deficient state in severe COVID-19. We next determined that CD4 + proliferative responses correlated with CD4 + /CD8 + ratios in the severe/mild COVID-19 cohort and within the severe disease cohort correlated with the absolute lymphocyte count (Fig. 3C, D) . Taken together, our data support impaired CD4 + T cell proliferation as a mechanism for decreased CD4 + counts in severe COVID-19 disease. We next hypothesized that increased S1-specific TNF- production from autologous CD4 + T cells negatively impacted proliferative capacities and subsequently, CD4 + T cell numbers. To test this, we assessed the effect of TNF- blockade on S1-specific CD4 + proliferation at 6 days in PBMC that were CD8 + T cell-depleted (96.2% purity; Fig. S3H ) and found that TNF- blockade resulted in significant restoration of CD4 + proliferative responses (Fig. 3E, F) . We concomitantly measured TNF- production from monocytes in the same cultures and found low TNF- + frequencies following peptide re-stimulation (Fig. S3I, J) . Together, these data support autocrine CD4 + T cell-derived TNF- as a significant source contributing to impaired S1-specific proliferation in severe COVID-19 disease. Similarly, in the initial set of proliferation experiments in which CD8 + T cells were not depleted, we observed enhancement of S1-specific CD8 + proliferation in the presence of TNF- blockade ( Fig. S3K ). Next, we assessed TNFRI (CD120a) and TNFRII (CD120b) receptor surface expression in CD4 + T cells across our cohort and found TNFRI surface expression was strikingly up-regulated in CD4 + T cells from patients with severe disease compared to mild disease controls and normal donors (Fig. 3G, H) . Targeted blockade of TNFRI resulted in marked restoration of S1-specific CD4 + proliferative responses, similar to TNF- blockade, whereas TNFRII blockade had no significant effect (Fig. 3I, J ). Finally, we tested the impact of the anti-TNF- therapeutic, infliximab, and found it capable of significantly rescuing S1-specific CD4 + proliferation in a dose-dependent manner (Fig. 3K , L). Moreover, infliximab in vitro treatment resulted in enhanced CD4 + IFN-, IL-2, CD107a and reduced TNF- responses at 6 days following secondary S1-peptide re-stimulation, thus reversing the hierarchical dominance of TNF- in the Type-1 response (Fig. 3M, N) . Together, our findings show autocrine CD4 + TNF-/TNFRI-dependent regulation of impaired CD4 + proliferation and altered effector cytokines in severe COVID disease that can be reversed in vitro by infliximab.

We further hypothesized that high COVID-specific TNF- secretion from CD4 + T cells might contribute to activation induced cell death (AICD) and apoptosis, contributing to CD4 + T cell lymphopenia. To test this, we evaluated CD4 + T cells for Annexin V expression in short-term cultures (18 h) following re-stimulation in vitro with S1 peptides, in CD8 + -depleted cultures. We found that Annexin V induction was significantly increased in severe COVID compared to mild disease patients (Fig. 4A, B ). The addition of anti-TNF- antibodies, including anti-TNFRI antibodies significantly inhibited Annexin V induction, as well as infliximab in a dose-dependent manner, but not anti-TNFRII antibodies (Fig. 4C, D) . We also found that anti-Fas neutralizing antibodies inhibited Annexin V induction in S1-activated CD4 + T cells, but to a lesser extent than TNF- blockade ( Fig.   S4A , B). Induction of AICD in cells from patients with severe disease occurred in the setting of other markers of activation, such as CD95, CD38 and to a lesser extent PD-1 compared to mild disease or normal controls (Fig.   S4A ). We observed that surface TNFRI expression correlated with surface CD38 expression (Fig. S4B) . Thus, S1-induced CD4 + T cell AICD in severe COVID disease is a predominantly autocrine TNF-/TNFRI-dependent mechanism contributing to CD4 + lymphopenia.

We next examined the impact of TNF- blockade on CD4 + T cell gene expression by performing single cell RNA sequencing of peripheral CD4 + T cells from two patients with severe COVID stimulated in vitro with S1-peptides in the presence or absence of infliximab. Following negative selection bead isolation of CD4 + T cells, the single cell RNAseq transcriptome identified 7 distinct clusters (Fig 5A) , largely differentiated by the presence or absence of anti-TNF- blockade (Fig 5B, C) . Those CD4 + T cells exposed to S1-stimulation in the presence of infliximab demonstrated down-regulation of genes related to pro-inflammatory cytokines (IL2, IFNG, TNFa), costimulation (TNFSF14 (LIGHT) and TNFSF5 (CD40LG), NF-Kappa signaling pathway (NFKBID), antigen binding and activation (SLAMF1), and apoptosis (FASLG, MYC, BCL2A1, SELENOK, NR4A1) (Fig 5D, E) . Gene set enrichment analysis of Hallmark gene sets identified upregulation of TNF SIGNALING VIA NFKB, ALLOGRAFT REJECTION, and IL2-STAT5 SIGNALING pathways among the CD4 + T cells stimulated with S1 alone (Fig. 5D ). Together, these data show that TNF- blockade using infliximab has a profound effect on S1stimulated CD4 + T cells from severe COVID-19 patients.

We next sought to determine whether CD4 + T cells produced high TNF- in the lung in severe COVID-19 disease.

To do this, we evaluated explanted lung tissue from a 51 y/o male who underwent bilateral lung transplantation for end-stage fibrotic lung disease following severe COVID-19 infection. We assessed lung mononuclear cells from bronchoalveolar lavage (BAL) fluid and lung parenchyma (LP) cells for CD4 + /CD8 + ratios and S1-specific effector responses compared to PBMC from this patient. Further, we assessed BAL versus PBMC S1-specific responses on five lung transplant recipients with documented SARS-CoV-2 infection, all of which had severe COVID-19 respiratory disease (Table S2 ). As seen in PBMC from other severe COVID disease, the CD4 + /CD8 + ratio was 1 in the PBMC, BAL and LP (Fig. 6A ). Increased CD8 + T cell numbers are consistent with a prior report on BAL cells from severe COVID-19 patients(23). Compared to S1-specific responses from PBMC, CD4 + TNF- > IFN- responses were significantly increased in the BAL and LP (Fig. 6B , C). Lung CD4 + T cells demonstrated CD69 + CD103 +/-CD45RA -CCR7resident memory phenotype, with increased CD38 hi PD1 hi Ki67 hi activation phenotype (Fig.6D, E) . Together, these data support reduced CD4 + T cell numbers in the lung with severe COVID and increased S1-specific TNF- production from activated CD4 + resident memory T cells.

Herein, we show that severe COVID-19-associated lymphopenia is a predominant CD4 + lymphopenia and associated with an increased risk for mortality. Overall, severe COVID-19 disease was associated with a significant diminution in the peripheral CD4 + /CD8 + ratios, whereas recovered, mild COVID-19, patients demonstrated a preservation of normal CD4 + /CD8 + ratios. Based on these findings, we assessed the function of peripheral CD4 + T cells from severe COVID-19 versus mild recovered COVID-19 patients to evaluate immune mechanisms driving CD4 + lymphopenia. Indeed, we found that a disproportionate increase in TNF- production and cytotoxic function from CD4 + T cells, but not CD8 + T cells, in response to the immunodominant antigen, S1, was evident in severe COVID-19 disease. We also found that S1-specific autocrine TNF--dependent production from CD4 + T cells themselves, and enhanced TNF- responsiveness via TNFRI are key mechanisms leading to impaired CD4 + proliferation and activation induced cell death (AICD) and contribute to CD4 + lymphopenia.

Together, our findings point to a skewed pro-inflammatory CD4 + T cell response among severe COVID-19 patients that is central to CD4 + T cell dysfunction and that plausibly contributes to the immunopathogenesis of severe disease.

While our studies found that S1 is the immunodominant antigen for SARS-CoV-2-induced T cell responses, we also detected significant effector T cell responses to other viral antigens, namely S2, NCAP and VEMP. However, CD4 + T cell responses to these antigens from severe disease patients did not demonstrate a hierarchical dominance of TNF- to these antigens as the S1 response, whereas CD107a responses were similarly elevated to all antigens with severe disease. Further, S1 was also immunodominant for CD8 + T cell responses, however IFN- and CD107a were predominant in severe disease in contrast to CD4 + responses. Thus, S1-specific CD4 + T cell responses in severe disease were unique with increased levels of the pro-inflammatory effector cytokine TNF- compared to mild disease controls, other antigens, and CD8 + T cells. Lastly, we show that CD4 + TNF + responses inversely correlated with absolute CD4 + T cell counts in severe disease. Together, these 13 findings reveal that the immunodominant SARS-CoV-2 S1 protein elicits a strong TNF- response from CD4 + T cells that subverts the host T cell response.

impaired CD4 + T cell proliferation(24-26). While we cannot exclude SARS-CoV-2 replication having an impact on the reduced CD4 + proliferative responses observed in severe COVID-19 disease, we demonstrate that in vitro TNF- blockade using infliximab, anti-TNFRI or other TNF- antibodies strikingly restored these responses.

Moreover, depleting CD8 + T cells and our finding of minimal spontaneous or S1-induced TNF- from monocytes, supports the concept that autocrine CD4 + TNF- production plays a major role in driving CD4 + dysfunction. and cytokine production through TNFRI (p55 receptor) and that in vitro proliferative and Type-1 cytokine responses could be rescued using anti-TNF- therapy, consistent with our observations (35) . In contrast to our findings, other studies have found a T cell inhibitory role of TCR-dependent activation though the TNFRII and chronic TNF- exposure (36) . Additionally, chronic TNF- exposure induced T cell hyporesponsiveness was subsequently shown to lead to impaired NF-B signaling(37). Thus, differential effects of TNF- on T cell responses and T cell-mediated pathology have been observed depending on the model system and duration of cytokine exposure. Our findings suggest that severe COVID-19 infection with persistent S1-specific TNF- production from CD4 + T cells more closely models chronic TNF- exposure, resulting in T cell hyporesponsiveness and dysfunction. In addition to anti-TNF- blockade agents, we found that low dose exogenous IL-2 also rescued S1-specific proliferative responses in the setting of low IL-2 frequencies that significantly increased after 6 days in the presence of TNF- blockade. Together, these data point to a relative IL-2-deficient state in severe COVID-19 disease that is TNF--dependent.

TNF- can be a major regulatory pathway for apoptosis of various cells, as well as cell survival (27, 38) .

While the TNF-/TNRI pathway has been demonstrated to be the predominant pathway for apoptosis in T cells, there is also evidence that the TNFRII plays a role (39) . T cell apoptosis via the TNFRI and/or TNFRII has been shown to be important regulatory mechanism for the T cell response in viral infections such as lymphocytic choriomeningitis virus and influenza, as well as autoreactive and aged T cells (40) (41) (42) (43) . The TNF-/TNFR pathways play highly complex direct and indirect roles in CD4 + and CD8 + T cell apoptosis during HIV infection, that include both TNFRI and TNFRII, in addition to other mechanisms (44, 45) . Our studies show that TNF-/TNFRI-dependent apoptosis of CD4 + T cells via AICD as a major mechanism contributing to CD4 + lymphopenia in severe SARS-CoV-2 infection. We also observed that TNFRI-mediated apoptosis in CD4 + T cells in severe COVID-19 was associated with increased surface expression of other activation markers such as CD95 (Fas) and CD38, and to a lesser extent PD-1, as previously reported in HIV infection (46, 47) . We also observed that blockade of the Fas/FasL pathway, the major apoptotic pathway in HIV infection, also rescued CD4 + T cells from AICD, though to a lesser extent than TNF- (22, 48) . Our RNAseq studies also demonstrated that TNF- blockade significantly reduced expression of Type-1 cytokines including TNF- itself, NFB signaling and FasL, consistent with the significant rescue of S1 re-stimulated CD4 + T cells from AICD. Taken together, our findings support high levels of TNF-/TNFRI-dependent apoptosis of CD4 + T cells through AICD in severe COVID-19 disease that contributes to lymphopenia.

We evaluated lung resident T cells from an explanted lung (severe COVID-19) undergoing lung transplantation and found a similar diminution in the CD4 + /CD8 + ratio in lung parenchymal and BAL cells, similar to PBMC. We further assessed BAL samples from four LTRs with recent severe COVID infection and had similar findings of reduced CD4 + frequencies in the lung, along with a similar hierarchy of increased CD4 + S1specific TNF- production, that was increased in both the LP and BAL compartments compared to the PBMC.

These findings are consistent with our previous findings of enhanced CD4 + effector T cell responses in the lung during acute and chronic CMV infection, but unusual in its TNF- predominance (49) . Thus, our findings support a role for CD4 + T cell TNF- production contributing to lung inflammation in severe COVID-19 disease and reduced CD4 + numbers in the lung.

Our studies found that the TNF- blockade therapy, infliximab, a common therapy for rheumatoid arthritis and other autoimmune diseases, had a profound, dose response, in vitro effect in the restoration of S1-specific CD4 + proliferative responses and abrogation of S1-induced AICD (50) . Indeed, recent studies have shown that treatment with the IL-6R antagonist tocilizumab resulted in increased circulating lymphocytes and cytotoxic NK cells (19, 51) . Moreover, high IL-6 and TNF- levels on hospital admission correlated with mortality in severe COVID-19 disease (52) . Some observational patient data suggest a potential survival benefit from poor COVID-19 outcomes in patients already receiving TNF- blockade therapy compared to patients on other immunomodulators and a mouse SARS-CoV-2 infection model demonstrates improved survival with TNF- blockade (53) . A current NIH Phase 3 trial is evaluating infliximab along with other immunomodulators (ACTIV-1 trial) for moderate to severe COVID-19 disease; https://www.nih.gov/research-training/medical-researchinitiatives/activ/covid-19-therapeutics-prioritized-testing-clinical-trials. Together, our studies and other lines of evidence suggest a potential role for TNF- blockade therapy in severe COVID-19 respiratory disease.

In summary, we show that the CD4 + T cell immunodominant response to S1 during severe COVID induces high TNF- levels resulting in autocrine TNF-/TNFRI-dependent impaired CD4 + proliferation and susceptibility to AICD. We further found that high autologous CD4 + TNF- production correlated with a predominant CD4 + lymphopenia during severe COVID disease, which we show is associated with increased mortality. Importantly, lung resident T cells produce elevated TNF- along with reduced CD4 + T cell numbers.

Together, our studies demonstrate mechanisms leading to COVID-associated lymphopenia and provide a strong rationale for testing TNF- blockade therapy in severe COVID-19 disease.

Patients from the University of Pittsburgh Medical Center hospitals admitted with COVID-19 or screened for convalescent plasma donation were identified and provided informed written consent for participation in a UPMC Institutional Review Board-approved protocol, at the University of Pittsburgh.

Blood samples were obtained from outpatient and inpatient study subjects as above whenever possible and PBMC isolated from heparinized blood samples by density gradient centrifugation using Ficoll-Paque (Cytiva, Sweden), aliquoted and stored in freezing medium in liquid nitrogen. BAL cells were isolated from excess BAL fluid from n=6 patients (5 lung transplant recipients and 1 severe COVID-19 lung explant) by centrifugation. Lung parenchymal mononuclear cells were purified as previously described(54). Lung parenchymal and BAL cells were aliquoted and stored in freezing medium in liquid nitrogen.

In vitro 6 h stimulation and intracellular cytokine staining (ICS) for IFN-γ, TNF-α, IL-2, IL-17a, IL-13 and CD107a expression was determined using pools of overlapping 15-mer peptides of SARS-CoV-2 specific for S1, S2, VEMP and NCAP (JPT Peptide Technologies Inc., Germany). SEB or pooled COVID-19 specific peptides were used at (1μg/mL) using 10 6 cells per condition for 6 h at 37°C, 5% CO2. All re-stimulations for ICS were performed using 10 6 Boolean gating analysis, cytokine co-expression was determined using SPICE software (version 6.1), downloaded from http://exon.niaid.nih.gov/spice.

Thawed frozen PBMCs were labeled with 0.2 mM CFSE (Invitrogen, Carlsbad, CA) in PBS for 7 min. and blocked with FCS for 2 min. The cells were immediately washed and plated in complete RPMI supplemented with 10% FCS in a 96 deep well plate for 6 days. For antigen-specific stimulation, PBMCs were stimulated in the presence or absence of SARS-CoV-2 peptide pools mix (S1, S2, VEMP and NCAP; JPT, Berlin, Germany) at (1μg/mL) for 6 days. Isolated cells underwent secondary re-stimulation in vitro for 6 h with/without S1 peptide pool and assessed by flow cytometry for CFSE dilution and ICS. All gates for cytokine frequencies were set using the medium alone control and subtracted from peptide-stimulated sample frequencies. All cells were collected for flow cytometric analysis using an LSR Fortessa-cytometer with UV laser (BD Biosciences, Franklin Lakes, NJ) or Cytek Aurora Flow-cytomer (Cytek, Bethesda, MD). Data analysis and graphic representations were done with FlowJo v.10 7.2 (BD-Becton Dickinson & Company, Ashland, OR).

In select experiments (proliferation and apoptosis studies), we performed CD8 + depletion using the human kit II by EasySep™ human CD8 Kit II Cell Separation (17853-STEM Cell Technologies, Cambridge, MA) to enrich PBMC for CD4 + T cells. For single cell RNA sequencing experiments, we used peripheral CD4 + T cells purify by negative selection using the kit EasySep™ Human CD4 + T Cell Enrichment Kit Immunomagnetic negative selection (17952-STEM Cell Technologies, Cambridge, MA) to enrich PBMC for CD4 + T cells.

In select experiments we used anti-human TNF- and isotype IgG1 (Pharmingen, BD-Biosciences, San Jose, 

In apoptosis studies, cells were stained with anti-Annexin V-Brilliant Violet-450 after in vitro stimulation with S1 peptides with/without various blocking or isotype control (IgG1) Abs, for 12 hours at 37°C. In select experiments, anti-Fas IgG1 (clone-ZB4; neutralizing) at 10 μg/ml plate-bound for 12 hours at 37°C or isotype control Abs IgG1 (Millipore). All cells were collected for flow cytometric analysis using an LSR Fortessa- 

We used purified CD4 + T cells by negative selection (17952-STEM Cell Technologies, Cambridge, MA) to enrich PBMC for CD4 + T cells 3 h following S1-stimulation in the presence or absence of infliximab. For the RNAseq analysis, differential gene expression analysis between subjects was completed using CLC Genomics Workbench version 20. Enriched CD4 + T cells were divided into four samples for HTO labeling. Each sample was subjected to the Totalseq-C and cell-hashing protocol, targeting a recovery of ~5000 cells per sample(55). All HTO-tagged cells were pooled together and prepared using the 10X Genomics with 5' Gel Bead Kit V1. The prepared library was subsequently sequenced on Illumina Novaseq platform with a depth of 50K read per cell. Downstream analysis of single cell RNA was performed using Seurat v3(56) and R software(57), including permissive filtering of low-quality cells, normalization, identification of highly variable genes, and Louvain clustering. Cell-cluster identity was established using previously reported immune gene signatures (7) . Subsets of singlets were isolated based on upregulation of CD3E (expression threshold >0.5), and downregulation of CD68 (<0.5), CD19 (<0.5), and CD8A (<0.5). The remaining cells were treated as CD4 + T cells. Normalization and Louvain clustering was then performed on this purified subset of singlets. Highly variable genes between clusters of CD4 + T cells were identified based on the non-parametric Wilcoxon rank sum test. Top differentially expressed genes across clusters were visualized using the DoHeatmap function in Seurat. Feature plots were performed on key genes identified. Gene set enrichment analysis was performed using Fast Gene Set Enrichment Analysis (fgsea) https://www.biorxiv.org/content/10.1101/060012v3.full differentially expressed genes between clusters 1 and 3 among the CD4 + T cell UMAP were included as input. Genes were references to the Hallmark pathways from MSigDB(58). Pathways were included if the absolute value of the normalized enrichment score > 1.5.

Distributions of all measured variables were performed using nonparametric testing as indicated. The nonparametric tests of Wilcoxon signed-rank and Mann-Whitney U analysis were applied using GraphPad Prism 9 (La Jolla, CA). A two-tailed P < 0.05 was considered statistically significant. All correlations were determined using Spearman rho test and Spearman rank correlation test. response frequencies with CD4 + /CD8 + ratio on severe (red dots) and mild (blue dots) COVID-19 patients n=48.

(J) The inverse correlation of S1-specific COVID-19 CD4 + TNF + response frequencies with absolute CD4 + number (cells/mm 3 ) on n=76 severe COVID-19 (red dots). Analysis was performed using the Spearman rho test and correlation coefficient R and P values were calculated using Spearman rank correlation test. Representative histograms (E) and pooled data (F) depicting the Annexin V + CD4 + from COVID-19 severe patients in the presence or absence of anti-Fas neutralizing antibodies, compared to anti-TNF neutralizing antibodies vs. S1-specific + isotype control (IgG1) or medium alone. Bars represent cumulative median frequencies ± interquartile range CD4 + AnnexinV + cells from severe COVID-19 patients. P values were calculated using the Mann-Whitney test. 

Predictors of mortality for patients with COVID-19 pneumonia caused by SARS-CoV-2: a prospective cohort study

Predictors of mortality for patients with COVID-19 pneumonia caused by SARS-CoV-2

Epidemiology, clinical course, and outcomes of critically ill adults with COVID-19 in New York City: a prospective cohort study. medRxiv

Is a "Cytokine Storm" Relevant to COVID-19?

Dexamethasone in Hospitalized Patients with Covid-19

Effect of Hydrocortisone on Mortality and Organ Support in Patients With Severe COVID-19: The REMAP-CAP COVID-19 Corticosteroid Domain Randomized Clinical Trial

Interleukin-6 Receptor Antagonists in Critically Ill Patients with Covid-19

Tocilizumab in Patients Hospitalized with Covid-19 Pneumonia

Association Between Early Treatment With Tocilizumab and Mortality Among Critically Ill Patients With COVID-19

Effect of tocilizumab on clinical outcomes at 15 days in patients with severe or critical coronavirus disease 2019: randomised controlled trial

Efficacy of Tocilizumab in Patients Hospitalized with Covid-19

Tocilizumab in Hospitalized Patients with Severe Covid-19 Pneumonia

Lymphopenia predicts disease severity of COVID-19: a descriptive and predictive study

T cell responses in patients with COVID-19

Deep immune profiling of COVID-19 patients reveals distinct immunotypes with therapeutic implications

Comprehensive mapping of immune perturbations associated with severe COVID-19

Reduction and Functional Exhaustion of T Cells in Patients With Coronavirus Disease 2019 (COVID-19)

Impaired immune cell cytotoxicity in severe COVID-19 is IL-6 dependent

Targets of T Cell Responses to SARS-CoV-2 Coronavirus in Humans with COVID-19

Comprehensive analysis of T cell immunodominance and immunoprevalence of SARS-CoV-2 epitopes in COVID-19 cases. bioRxiv

Activation-induced cell death drives profound lung CD4(+) T-cell depletion in HIVassociated chronic obstructive pulmonary disease

Prevention of diabetes in nonobese diabetic mice by tumor necrosis factor (TNF): similarities between TNF-alpha and interleukin 1

Chronic tumor necrosis factor alters T cell responses by attenuating T cell receptor signaling

Chronic TNF-alpha exposure impairs TCR-signaling via TNF-RII but not TNF-RI

Genomic expression profiling of TNF-alpha-treated BDC2.5 diabetogenic CD4+ T cells

Regulation of tumour necrosis factor signalling: live or let die

TNF-RII and c-IAP1 mediate ubiquitination and degradation of TRAF2

Role of tumor necrosis factor receptors in regulating CD8 T-cell responses during acute lymphocytic choriomeningitis virus infection

Intrinsic TNF/TNFR2 interactions fine-tune the CD8 T cell response to respiratory influenza virus infection in mice

Selective death of autoreactive T cells in human diabetes by TNF or TNF receptor 2 agonism

Increased TNF-alpha-induced apoptosis in lymphocytes from aged humans: changes in TNF-alpha receptor expression and activation of caspases

Increased sensitivity of T lymphocytes to tumor necrosis factor receptor 1 (TNFR1)-and TNFR2-mediated apoptosis in HIV infection: relation to expression of Bcl-2 and active caspase-8 and caspase-3

Is HIV infection a TNF receptor signalling-driven disease?

Analysis of apoptosis in lymph nodes of HIV-infected persons. Intensity of apoptosis correlates with the general state of activation of the lymphoid tissue and not with stage of disease or viral burden

Programmed cell death in peripheral lymphocytes from HIV-infected persons: increased susceptibility to apoptosis of CD4 and CD8 T cells correlates with lymphocyte activation and with disease progression

Biochemical mechanisms of HIV induced T cell apoptosis

High-quality CMV-specific CD4+ memory is enriched in the lung allograft and is associated with mucosal viral control

Anti-TNF therapy: past, present and future

Complex Immune Dysregulation in COVID-19 Patients with Severe Respiratory Failure

An inflammatory cytokine signature predicts COVID-19 severity and survival

The Potential for Repurposing Anti-TNF as a Therapy for the Treatment of COVID-19

Representative flow cytometric plots (A) and cumulative data (B) showing S1-specific CD8 + T cell proliferation by CFSE dilution in mild (left panel-A and blue dots-B) vs. severe (right panel-A and red dots-B) COVID-19

Cumulative data showing S1-specific CD8 + T cell proliferation by CFSE dilution using exogenous IL-2 (G) in severe COVID-19 patients, (n=12) (red dots). Bars represent median values and p-values were calculated using the Mann-Whitney-Wilcoxon test. (H) Representative gating strategy and analysis of the CD8 + -depleted PBMC used in some proliferation and apoptosis experiments

Analysis was performed using the Spearman rho test and correlation coefficient (R) and (P) values were calculated using Spearman rank correlation test. (C) Representative flow cytometry plots of the COVID-19 lung explant showing CD8 + T cell surface expression of CD45RA + , CCR7 + (left panel) and CD69 + CD103 + (right panel) on PBMC (red) or Lung Parenchima (LP) (blue) and values represent cell frequencies of T

PD1 + (middle panel) and Ki67 + (right panel) overlayed LP (blue lines) on PBMC

The authors declare no competing interests.

All data are available in the main text or the supplementary materials.