The Role of Coinhibitory Receptors in B Cell Dysregulation in SARS-CoV-2–Infected Individuals with Severe Disease

COVID-19 is still a global health concern caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection (1). COVID-19 disease has already killed more than 6.9 million people with over 771M cases reported. SARS-CoV-2 has evolved in the infected population, and since 2020, various mutated variants have emerged globally (2,3). Although we are still struggling to establish effective treatments and medications against COVID-19, the total number of COVID-19 cases and COVID-related mortality have gradually decreased due to the administration of over 13 billion vaccine doses (4). Overall, the current vaccination strategies are considered very effective in preventing severe COVID-19 cases (5). Despite the availability of different vaccine candidates against SARS-CoV-2 infection, it remains a life-threatening disease in some individuals. Therefore, we still need a better understanding of how the adaptive immunity acts against SARS-CoV-2 infection and how the virus evades the immune system.

Although COVID-19 disease starts as a respiratory tract infection, it displays a systemic disease and profoundly influences the proportion and functionality of different immune cells. It is widely reported that neutrophilia, lymphopenia, thrombocytopenia, and stress erythropoiesis are the hallmarks of SARS-CoV-2 infection (6,7). Several studies have revealed that SARS-CoV-2 infection induces excessive activation of myeloid-lineage cells that aggravates inflammatory responses both at the site of infection and systemically (6, 8). Alternatively, T cell activation and skewed Th17 response in COVID-19 patients with severe disease has been reported (9). The dysregulated immune system is more evidenced by the cytokine storm, which can result in pathological alterations and multiorgan failure in COVID-19 patients with severe disease (10). Especially, plasma Galectin-9 (Gal-9) at elevated levels is considered to be a major player in the induction of cytokine storm (11,12). Activated neutrophils may act as a potential source for the elevated plasma Gal-9 (11, 13). Neutrophil extracellular trap formation is also enhanced in COVID-19 patients in both tissues and blood circulation (14). NETosis itself is an indispensable innate immune defense against various pathogens; however, the excessive NETosis extends tissue inflammation (14). Because SARS-CoV-2 infection triggers critical tissue damage accompanied with severe inflammation in the infected respiratory tract, myeloid-lineage cells are frequently exposed to activating factors such as proinflammatory cytokines, damage-associated molecular patterns, and pathogen-associated molecular patterns (15). In fact, these factors exacerbate myeloid cells activation as SARS-CoV-2-indirect aggravating players (16). Similarly, Gal-9 as a damage-associated molecular pattern by the activation of monocytes, NK cells and increased production of IL-6 and TNF-α contributes to the systemic inflammation in COVID-19 disease (11, 17). As mentioned above, lymphopenia is commonly observed in SARS-CoV-2–infected individuals and extends across all subsets (e.g., T, NK, and B cells). Apart from T and NK cells, B cells exhibit profound abnormalities in SARS-CoV-2–infected individuals, in particular in those with severe disease (18). In general, B cells possess important host defense roles against virus-associated infections including SARS-CoV-2 (19). B cell abnormalities can contribute to increased COVID-19 disease severity through non–Ag-specific or Ag-specific Ab production (20). Also, B cell deficits may result in secondary bacterial infections in COVID-19 patients (21). It is reported that hypoxia, often clinically silent, may contribute to B cell pathology in COVID-19 patients (22). Therefore, understanding the underlying mechanism of B cell defects has important clinical implications in COVID-19 patients. In this report, we show specific features of B cells in intensive care unit (ICU)–admitted COVID-19 patients who were infected with the Delta variant. We also compared B cell phenotypes in these patients versus those with milder COVID-19 disease. We observed a specific B cell phenotype characterized by increased naive, double-negative (DN), and Ab-secreting cell populations but decreased switched (SW), unswitched (USW), and memory populations in ICU-admitted COVID-19. Our in vitro functional assay revealed that exogenously stimulated COVID-19 B cells exhibited impaired effector functions such as proliferation, cytokine production, and IgG expression. As an underlying mechanism, we found that B cells in these patients were prone to stimulation-induced apoptosis. Furthermore, we found that multiple coinhibitory molecules such as PD-1, PD-L1, TIM-1, V-domain Ig suppressor of T cell activation (VISTA), Gal-9, and CTLA-4 were upregulated in the B cells of COVID-19 patients, which directly correlated with the tendency toward cell death. In particular, our metabolomic studies revealed that the alteration in tryptophan pathway contributes to the upregulation of PD-L1 in B cells.

For this study, we recruited 38 COVID-19 patients admitted to different ICUs in Edmonton, Alberta. The diagnosis of SARS-CoV-2 infection was determined by RT-PCR assay specific for viral RNA-dependent RNA polymerase and envelope transcripts using an endotracheal aspirate or nasopharyngeal swab. Similarly, 15 patients with mild COVID-19 disease were recruited. Our blood sampling was performed around 2 weeks after disease onset. For comparison, blood samples from 25 age- and sex-matched healthy controls (HCs) without exposure to SARS-CoV-2 were obtained (Supplemental Table I). These patients and HCs were SARS-CoV-2 vaccine naive. All of the study subjects were HIV, hepatitis C virus, and hepatitis B virus seronegative.

This study was approved by the Human Research Ethics Board (HREB) at the University of Alberta (Approval No. Pro00099502). Waiver of consent was obtained by the HREB for those patients admitted to the ICU. Also, the HREB approved blood collection from HCs (Approval No. Pro00063463). Written informed consent was obtained from HCs and those with mild disease.

Human PBMCs were isolated by Ficoll-Paque Premium (GE, Chicago, IL, USA) from fresh blood samples according to our previously described methods (23,24). The cells were resuspended in RPMI supplemented with 10% FBS and antibiotics (100 U/l penicillin, 100 mg/ml streptomycin). Fresh PBMCs were generally used for flow cytometry analysis and/or in vitro studies.

Cell activation mixture and 5-(and 6)-carboxyfluorescein diacetate succinimidyl ester was purchased from BioLegend (San Diego, CA, USA). Cytofix/Cytoperm kit, GolgiStop, and annexin V apoptosis, pSTAT1/pSTAT3 kits were purchased from BD Bioscience (Franklin Lakes, NJ, USA). The goat anti-human IgM, affinity purified, and recombinant human CD40L were purchased from R&D Systems (Minneapolis, MN, USA). CpG (ODN2006, Class B CpG oligonucleotide) and LPS were purchased from InvioGen (San Diego, CA, USA).

PBMCs (1.0 × 10⁶/well) were seeded into 96-well round-bottom plates in RPMI medium. The cells were treated with either anti-IgM mAb (10 μg/ml) plus CpG (5 μg/ml), anti-IgM (10 μg) plus CD40L (1 μg/ml), or CpG (5 μg/ml) at 37°C for 24 h. For B cell proliferation assay, PBMCs were stained with 5-(and 6)-carboxyfluorescein diacetate succinimidyl ester at 37°C for 30 min, and then the cells were stimulated with CpG (5 μg/ml) at 37°C for 120 h.

To investigate the effect of kynurenine (KYN) or IL-6 in B cell function, PBMCs (1 × 10⁶/well) were cultured with or without kynurenine (5, 10, 25, 50, or 100 μM) or IL-6 (100 ng/ml) at 37°C overnight. For further stimulation, the cultures were treated with anti-IgM mAb (10 μg/ml). After incubation by following each culture condition, the samples were analyzed with flow cytometry.

Total B cells were isolated from cryopreserved PBMCs (Stem Cell Technologies) as reported elsewhere (25). The recombinant biotinylated SARS-CoV-2 S1 protein (AcroBiosystems) was conjugated to streptavidin-PE (ThermoFisher). Isolated B cells were incubated with the fluorescently labeled S1 protein in PBS supplemented with 2% FBS and EDTA at 4°C for 30 min. Then the anti-IgG Ab (BD, clone G18-145) was added to B cell suspension and incubated at 4°C for 30 min. Following the binding, the cells were washed with PBS and stained for live/dead. After wash, the cell suspension was stained for surface markers such as anti–PD-1, anti–PD-L1, anti-VISTA, anti–TIM-1, and anti–CTLA-4. After 30 min at 4°C staining, the cells were washed and acquired on a flow cytometer (LSR Fortessa-SORP).

For immunophenotyping, apoptosis and proliferation assay PBMCs were subjected to flow cytometry analysis. Fluorochrome-conjugated monoclonal Abs with specificity to human cell surface Ags and cytokines were purchased mainly from BD Biosciences or Thermo Fisher Scientific. Specifically, the following Abs were used: anti-CD3 (SK7), anti-CD19 (HIB16), anti-CD11b (M1/70 or ICRF44), anti–HLA-DR (G46.6 or LN3), anti-CD38 (RPA-T8), anti-CD45 (H-130 or 2D1), anti-CD27 (G3H69), anti-IgD (IA6-2), anti-CD24 (ML5), anti-CD80 (L307.4), anti-CD86 (2331), anti–PD-1 (MIH4), anti-PD-L1 (MIH1), anti-VISTA (B7H5DS8), anti–TIM-1 (ID12), anti–CTLA-4 (BNI3), anti–IL-10 (JE53-007), anti–IL-6R (M5), anti-CD138 (MI15), anti-IgG (G18-145), anti-IgM (RMM-1), TIM-3 (7D3), TIGIT (MBSA43), and anti-IgA. The live/dead kit (Thermo Fisher Scientific, Waltham, ME, USA) was used to exclude dead cells. Intracellular cytokine staining was performed according to our routine protocols (23, 26). The stained cells were fixed with 4% paraformaldehyde, acquired on an LSR Fortessa-SORP flow cytometer (BD Bioscience), and analyzed with FlowJo software (version 10).

Frozen plasma samples stored at −80°C were thawed on ice and centrifuged for 15 min at 1,500g, and then supernatants were collected for assays. The concentrations of IL-6 and PD-L1 in the samples were measured by ELISA using the kit corresponding to the targets (R&D Biosystems).

Frozen plasma samples were thawed at room temperature and then were centrifuged at 600g for 5 min. A total of 100 μl of soluble fraction was transferred into new tube and mixed with 300 μl of liquid chromatography–mass spectrometry grade methanol. The samples were stored at −20°C for 30 min and then were centrifuged at 16,260g to let the proteins precipitate. The supernatants were transferred into new tubes and dried up completely. The dried samples were reconstituted by dissolving in 85 μl of water. The metabolomic analysis was performed by following chemical isotope labeling liquid chromatography–mass spectrometry protocol as described in our previous report (27,28).

Statistical analysis was performed using GraphPad Prism 9. Mann–Whitney U tests were used for comparison of two groups and one-way ANOVA for comparison of more than two groups. The data are presented as means ± SEM. p values < 0.05 were considered to be statistically significant.

The median age for the HC group was (51.6 ± 11.6) with 60% males. In the COVID-19 cohort, the median age was (50.9 ± 9.7) with 58% males for ICU-admitted patients and 49.9 ± 9.1 with 60% males for patients with mild disease (Fig. 1A; Supplemental Table I). To determine whether SARS-CoV-2 infection in critically ill patients influences B cell phenotype and function, we analyzed B cells in ICU-admitted and mild COVID-19 patients compared with HCs. It is widely reported that SARS-CoV-2 infection is associated with the expansion of myeloid cells (e.g., neutrophils and monocytes) at the expense of lymphoid cells (7). We have previously reported the same phenomenon in COVID-19 patients admitted to the ICU (9). In this study, we found that the percentages of CD19⁺ B cells were significantly lower in severely ill COVID-19 patients than those with mild disease and HCs (Fig. 1B, 1C; Supplemental Fig. 1A). This agrees with other reports that severe COVID-19 disease is associated with reduced B cell frequency (29). To further characterize B cell subsets, we analyzed the frequency of naive (CD27⁻IgD⁺), SW memory (CD27⁺IgD⁻), USW memory (CD27⁺IgD⁺), and double-negative B cells (CD27⁻IgD⁻) in PBMCs from COVID-19 patients and HCs using the gating strategy reported elsewhere (17). These analyses revealed that the percentages of the naive subset were significantly increased in ICU-admitted patients compared with COVID-19 patients with mild disease and HCs (Fig. 1D, 1E). In contrast, the proportion of SW and USW memory B cells were significantly reduced in critically ill COVID-19 patients, whereas the frequency of DN B cells was increased compared with HCs (Fig. 1D, 1F–H). Of note, the frequency of naive, SW, USW, and DN B cells remained unchanged in those with mild COVID-19 disease compared with HCs (Fig. 1E–H). It is worth mentioning that although the IgD⁻CD27⁺ subset is defined as memory-switched B cells (30), they may still express IgM and yet to be class-switched. The reduction in the memory B cell subpopulation was more evident when we assessed the frequency of (CD19⁺CD27⁺IgD⁻CD38⁻) cells in ICU-admitted COVID-19 patients compared with HCs (Fig. 1I, 1J). Further assessment of Ab-secreting cells revealed significant increases in the percentages of plasma cells (PCs; CD19⁺CD27⁺IgD⁻CD38⁺CD138⁺) and plasma blasts (PBs; CD19⁺CD27⁺IgD⁻CD38⁺CD138⁻) in critically ill COVID-19 patients (Fig. 1K–M) as reported elsewhere (29). In contrast, the frequency of early PBs (CD19⁺CD27⁺IgD⁻CD38⁻CD138⁻) was significantly decreased in these patients (Fig. 1K, 1N). Overall, the volcano plot summarizes the typical immunophenotype of B cells in ICU-admitted COVID-19 patients with significant increases in naive, PCs, and PBs, but decreases in SW, USW, and early PBs populations compared with HCs (Fig. 1O). These observations support phenotypical alterations in B cells from COVID-19 patients with severe disease.

We next compared the functional properties of B cells between HCs and COVID-19 patients. First, we analyzed the expression level of surface molecules such as CD80, CD86, and HLA-DR that are correlated with Ag presentation ability, as well as cellular activation status of B cells (31). We found that the expression of all of these markers were significantly downregulated in B cells from ICU-admitted COVID-19 patients compared with HCs (Fig. 2A, 2B). Given the possibility of downregulated B cell functions in COVID-19 patients, we investigated the cytokine production and proliferation capacities of B cells. When PBMCs were globally stimulated with PMA/ionomycin, B cells from ICU-admitted COVID-19 patients expressed significantly lower levels of IL-6 compared with HC cells (Fig. 2C, 2D). This observation was reproduced by BCR-mediated stimulation using anti-IgM + CpG, which again supports impaired B cell functionality in COVID-19 patients (Fig. 2C, 2D). Similarly, we found a dramatic reduction in proliferative capacity of B cells upon GpG stimulation, TLR9 agonist, in these COVID-19 patients (Fig. 2E, 2F).

We then explored the expression of IgG in B cells stimulated via either a T cell−dependent (anti-IgM + CD40L) manner or a T cell−independent (anti-IgM + GpG) manner. These studies revealed that B cells from critically ill COVID-19 patients had significantly lower percentages of IgG expression compared with their counterparts in HCs (Fig. 2G, 2H). Of note, CpG stimulation alone also induced lower percentages of IgG-expressing B cells in COVID-19 patients than HCs (Fig. 2G, 2H). These observations support the impaired effector functions of B cells in severe ill patients.

To determine why B cells from COVID-19 patients exhibited impaired effector functions, particularly when received stimulatory signals, we evaluated their apoptotic potential upon stimulation. We found that both BCR and TLR9 stimulation resulted in a significant increase in percentages of apoptotic B cells (annexin V⁺) in ICU-admitted COVID-19 patients compared with HCs (Fig. 2I, 2J). This implies stimulation-induced cell death in B cells from these patients. Moreover, we subjected B cells from COVID-19 and HCs to active caspase-3/7 assay, which confirmed significantly increased levels of active caspase-3/7 in B cells from COVID-19 patients (Fig. 2K, 2L). These observations imply that B cells from critically ill COVID-19 patients not only exhibit impaired effector functions but also are prone to stimulation-induced cell death.

To better understand the underlying mechanism of increased stimulation-induced cell death in B cells from COVID-19 patients, we performed further B cell immunophenotyping. These studies enabled us to determine the differential expression pattern of coinhibitory molecules on B cells from ICU-admitted COVID-19 patients versus HCs. It is widely accepted that persistent upregulation of coinhibitory receptors is associated with T cell exhaustion in chronic conditions (32,33). Similarly, coinhibitory molecules are widely expressed on myeloid cells (34,35) and B cells with displayed immunoregulatory properties (36–38).

Interestingly, we found that the proportion of B cells expressing PD-1, PD-L1, TIM-1, VISTA, and CTLA-4 were significantly increased in critically ill COVID-19 patients compared with HCs (Fig. 3A, 3B). We also investigated the expression of these coinhibitory molecules in B cell subpopulations (e.g., naive, SW, USW, and DN). Intriguingly, we observed a significant increase in the proportion of B cell subsets expressing all of these coinhibitory receptors, except TIM-1 in SW subset, compared with their counterparts in HCs (Supplemental Fig. 1B–F). However, despite a trend in higher TIM-3– and TIGIT-expressing B cells in ICU-admitted COVID-19 patients, we did not observe any significant difference in the proportion of PD-L2–, TIM-3–, and TIGIT-expressing cells in our cohorts (Supplemental Fig. 1G).

To determine the role of these coinhibitory receptors in regard to B cell effector functions, we subjected B cells to Ab expression analysis. We found that PD-1+ and VISTA+ B cells had significantly higher percentages of IgA expression cells among them (Fig. 3C), whereas PD-1⁺, TIM-1⁺, VISTA⁺, and CTLA-4⁺ B cells had significantly higher proportion of IgG-expressing cells compared with their negative counterparts (Fig. 3D). However, IgM expression was the same regardless of coinhibitory receptors expression in B cells from ICU-admitted COVID-19 patients (Supplemental Fig. 1H).

Considering a higher apoptotic rate in B cells from COVID-19 patients (Fig. 2I–L), we decided to investigate the association of coinhibitory receptors with apoptosis of B cells. We observed that the intensity of annexin V was significantly higher in PD-1–, PD-L1–, TIM-1–, VISTA-, and CTLA-4–expressing B cells from ICU-admitted COVID-19 patients compared with their negative counterparts (Figs. 3E, 4F). We also examined the expression of Gal-9, the role of which in apoptosis and immune cell exhaustion is well documented (32, 39,40). Interestingly, we noted significant abundance of Gal-9⁺ B cells in critically ill COVID-19 patients (Fig. 3G, 3H). As anticipated, Gal-9⁺CD19⁺ exhibited a greater apoptotic capacity than their negative counterparts (Fig. 3I, 3J). These observations suggest that the upregulation of coinhibitory receptors is indicative of an enhanced B cell activation-induced cell death in COVID-19 patients with severe disease but not in those with mild disease, because we did not find any significant difference in the proportion of B cells expressing PD-1, PD-L1, TIM-1, VISTA, and CTLA-4 in COVID-19 patients with mild disease (Supplemental Fig. 2A, 2B). Although a significant increase in the proportion of B cells expressing Gal-9 in those with mild disease was observed, it was substantially lower than those with severe COVID-19 disease (Supplemental Fig. 2A, 2C).

Given a higher apoptotic rate of B cells from critically ill COVID-19 patients upon mitogen stimulation, we aimed to determine the effects of spike protein stimulation. These studies revealed a significant increase in active caspase-3/7 in B cells from COVID-19 patients upon stimulation with the spike protein (Fig. 4A, 4B). A similar pattern was observed for annexin V expression in B cells from these patients (Fig. 4C, 4D). In these studies, CpG was used as a positive control. It is worth mentioning that we noted a dose-dependent spike protein apoptotic effects in B cells from COVID-19 patients compared with HCs (Supplemental Fig. 2D). It is worth mentioning that B cells from ICU-admitted COVID-19 patients at the baseline possessed a greater apoptosis potential compared with their counterparts from HCs. However, we noted that the spike protein exerts apoptosis effects in B cells from HCs at 2 µg/ml (Supplemental Fig. 2D). To investigate Ag-specific B cells in severely ill SARS-CoV-2–infected individuals and elucidate their function in regard to coinhibitory receptors, we focused on B cells specific for the S1 subunit of the spike protein. We first quantified the frequency of IgG⁺ S1-specific memory B cells in severely ill COVID-19 patients compared with HCs (Fig. 4E, 4F). Next, we compared the proportion of S1-specific B cells expressing different coinhibitory receptors (Fig. 4G). These analyses revealed that the majority of S1-specific B cells express PD-L1, VISTA, PD-1, CTLA-4, and TIM-1, respectively (Fig. 4G, 4H). Moreover, we assessed the frequency of IgG⁺ S1-memory B cells among these B cells expressing coinhibitory receptors. Interestingly, S1-memory B cells expressing different coinhibitory receptors had greater IgG expression than their negative counterparts (Fig. 4I, 4J). These findings suggest that Ag-specific B cells exhibit an activated phenotype illustrated by greater expression of coinhibitory receptors, which may result in activation-induced cell death.

Considering that CD19⁺CD24^hiCD38^hi cells are reported to exhibit immunosuppressive properties (41) and regulatory B cells (Bregs) express PD-L1 (42), we evaluated the nature of PD-L1–expressing B cells in our cohorts. Interestingly, we observed a significant reduction in the frequency of CD24^hiCD38^hi in COVID-19 patients (Fig. 5A, 5B). It is reported that CD19⁺CD24^hiCD38^hi mainly via IL-10 production exerts suppressive capacity (41). In contrast to their counterparts in HCs, we found that these cells upon stimulation with α-IgM⁺ CpG exhibited impaired IL-10 production capacity (Fig. 5C, 5D). To determine the mechanism underlying the upregulation of PD-L1 in COVID-19 patients, we conducted further investigations. Our studies revealed an elevation of IL-6R in B cells from COVID-19 patients (Fig. 5E, 5F). In light of the significant role of IL-6 in SARS-CoV-2 pathogenesis and cytokine storm (11), we found the elevation of IL-6 in the plasma of our COVID-19 patients (Fig. 5G) and thus examined the effects of IL-6 on PD-L1 expression in B cells. These studies revealed that IL-6 enhances PD-L1 expression in B cells, which was more pronounced in B cells from ICU-admitted COVID-19 patients than HCs (Fig. 5H, 5I). These observations suggest a role for IL-6 in altered B cell phenotype and function in COVID-19 patients.

We performed metabolomic studies on plasma samples from HCs and severely ill COVID-19 patients. Principal component analysis and partial least squares–discriminant analysis clearly demonstrated distinct metabolomic profile for COVID-19 versus HCs (Fig. 6A, 6B) as previously reported (43, 44). Pathway analysis between ICU-admitted patients and HCs revealed alteration in 29 pathways (Fig. 6C). Among these altered pathways, we focused on tryptophan metabolism because l-5-hydroxytryptophan is linked to the inhibition of PD-L1 expression (45). We found that the concentration of l-tryptophan was significantly reduced in the plasma of severely ill COVID-19 patients compared with HCs (Fig. 6D). Given that the majority (>95%) of tryptophan gets degraded by KYN, we compared its concentration in ICU-admitted COVID-19 patients versus HCs. As anticipated, we noted the elevation of KYN in the plasma of these COVID-19 patients (Fig. 6E). Further analysis of metabolites associated with tryptophan pathway revealed a significant reduction in serotonin, 5-hydroxyindoleacetate, and 4,6-dihydroxyquinoline but substantial increases in plasma metabolites such as formyl-5-hydroxy kynurenamine, 5-hydroxykynurenine, 3-hydroxyanthranilic acid, and 2-aminomuconate in ICU-admitted COVID-19 patients (Supplemental Fig. 2E). Next, we assessed the effects of KYN supplementation on the expression of PD-L1 in B cells. We found that KYN in a dose-dependent manner upregulated PDL-1 expression, which was more pronounced in B cells from ICU-admitted COVID-19 patients (Fig. 6F). Moreover, we detected a significant elevation in the concentration of PD-L1 in the plasma of severely ill COVID-19 patients (Fig. 6G), which was positively correlated with the plasma KYN concentrations (Fig. 6H). Likewise, we found a positive correlation between the KYN and IL-6 (Fig. 6I) and IL-6 with PD-L1 concentrations in the plasma of these patients (Supplemental Fig. 2F). To understand how KYN enhances PD-L1 expression in B cells, we evaluated IL-6R expression in the presence of absence of KYN. We found that KYN in a dose-dependent manner upregulates IL-6R in B cells. This was more prominent in B cells from COVID-19 patients once activated with α-IgM (Fig. 6J).

Therefore, we tried to unravel the underlying mechanism of KYN-mediated upregulation of PD-L1 expression in B cells. Given that the JAK1/JAK2–STAT-1/STAT-3–IRF1 axes are the canonical pathways in PD-L1 expression (46), we observed that treatment with KYN enhances the phosphorylation of STAT1 and STAT3 in B cells, particularly, in those from severe COVID-19 disease (Fig. 6K, 6L). Overall, these observations provide a novel regulation mechanism of PD-L1 expression in B cells through KYN and IL-6 under severe inflammatory conditions.

Taken together, our results provide insights into the profound alterations in B cell phenotype and function in severe COVID-19 patients. We found a significant decrease in the proportion of memory B cells, both the SW and USW subsets, accompanied by an increase in naive B cells in critically ill patients but not in those with mild COVID-19 disease. This distinctive immunophenotype, characterized by an expansion of plasma cells and plasma blasts, highlights the dysregulation in B cell homeostasis during severe SARS-CoV-2 infection, which is supported by other studies (18, 29).

This suggests that the critical events related to B cell depletion happen during or after class switching in COVID-19 patients. This concept is supported by a significant decrease in memory and SW B cell subpopulations in ICU-admitted COVID-19 patients. In contrast, an increase in plasma cells might be a protective mechanism to promote Ab protection. Nevertheless, disease progression in COVID-19 patients indicates that expanded plasma cells are unable to provide protective immunity. This is because the proinflammatory milieu (e.g., elevated IL-6, KYN, etc.) may modulate B cells’ effector functions.

Our further studies into the functional properties of B cells showed impaired Ag presentation ability, compromised proliferation, and reduced cytokine production capacity in severely ill COVID-19 patients. The downregulation of surface markers associated with B cell activation, coupled with decreased expression of IgG and IL-6, underscores the functional impairment of B cells in responding to SARS-CoV-2 in these patients. This altered B cell effector function may explain the inability of B cells to provide an efficient protective immunity.

In our further studies to unravel mechanism(s) associated with dysregulated B cells, we observed elevated expression of coinhibitory receptors such as PD-1, PD-L1, TIM-1, VISTA, CTLA-4, and Gal-9 not only in total B cells but also in S1 Ag-specific B cells from ICU-admitted COVID-19 patients. These molecules may play important roles in B cell functions. For instance, PD-1 is characterized as triggering a suppressive effect on B cells when it interacts with PD-L1/PD-L2 (47,48). CTLA-4 is also introduced as a negative regulator of B cell function (49). Furthermore, TIM-1 and PD-L1 are both characterized as the key functional marker of Bregs (36, 50). Thus, these molecules may inhibit/regulate the functional properties of B cells as reported for T cells in different scenarios (24, 51, 52). Although we were unable to delineate what factors are driving the upregulation of these coinhibitory molecules, we suspected that the inflammatory milieu in severely ill SARS-CoV-2–infected individuals (7, 11) may promote the expression of these coinhibitory molecules in B cells. In fact, a previous report has shown stimulation-dependent upregulation of PD-1, PD-L1, and PD-L2 in B cells (38).

Considering the immunoregulatory roles of these coinhibitory receptors/ligands in T and B cell effector functions (53–55), we found that these molecules may contribute to an increased susceptibility to stimulation-induced cell death in B cells from COVID-19 patients. In particular, we found that spike protein stimulation enhances the apoptosis of B cells from ICU-admitted COVID-19 patients. The link between these coinhibitory receptors with altered Ab expression (e.g., IgA and IgG) and increased apoptotic rates underscores the intricate interplay between B cell and SARS-CoV-2 infection. Therefore, immune dysregulation in severe COVID-19 patients may alter/impair B cell effector functions. A minimal alteration in the B cell phenotype in those with mild SARS-CoV-2 infection may, in part, explain why these patients did not end up in the ICU.

Given the reduction in CD19⁺CD24^hiCD38^hi cells and impaired IL-10 production, we argue that these expanded B cells expressing different coinhibitory receptors are distinct from Bregs. Because human Bregs are defined as CD19⁺CD24^hiCD38^hi cells primarily producing IL-10 (41, 56). These observations further underline the dysregulation of B cell–mediated immunomodulatory mechanisms in severe COVID-19 disease (29).

Our results also demonstrate a possible link between elevated IL-6 levels in the plasma of COVID-19 patients and the upregulation of PD-L1 in B cells, which has been reported in another context (57). Moreover, our metabolomic studies revealed a significant decrease in the plasma l-tryptophan and subsequently the elevation of KYN in the plasma of COVID-19 patients as reported in other cohorts (58,59). Considering that tryptophan is mainly degraded by KYN pathway (60), we found a positive correlation between plasma concentrations of KYN and PD-L1 in COVID-19 patients. PD-L1 expression via upregulation of IL-6R on B cells of COVID-19 patients provides novel insights into the intricate pathways influencing B cell dysfunction in severe COVID-19. The positive correlation between concentrations of KYN in the plasma and IL-6, as reported by others (58), implicates the importance of altered metabolomic pathway and inflammation in B cells phenotype and function. Intriguingly, we observed a significant reduction in plasma serotonin levels in critically ill COVID-19 patients, and its reduction in long COVID patients is reported to be associated with neurocognitive symptoms (61). This observation suggests that SARS-CoV-2 infection and inflammation can modulate amino acid uptake (e.g., tryptophan) (61), which subsequently may modulate adaptive immunity, as we have seen for B cells.

In particular, the association of IL-6 with PD-L1 suggests a novel underlying mechanism for the upregulation of PD-L1. This could be further supported by KYN-mediated upregulation of pSTAT1/pSTAT3. Therefore, elevated IL-6 and altered metabolites in severely ill SARS-CoV-2–infected individuals may contribute to dysregulated B cell response. Alternatively, enhanced apoptosis of B cells may explain the low frequency of B cells in severely ill COVID-19 patients. Our study indicates a role for spike protein in this process, coupled with the upregulation of coinhibitory receptors, as a potential immune evasion mechanism. It is possible to speculate that PD-L1⁺ B cells may suppress T cells through PD-L1–PD-1 interactions as reported elsewhere (54).

Taken together, the multifaceted alterations in B cell phenotype, function, and regulatory mechanisms observed in severe COVID-19 underscore the complexity of the immune response in critically ill patients. Further understanding of these dysregulations may pave the way for targeted therapeutic interventions to modulate B cell responses and improve clinical outcomes in severe cases of SARS-CoV-2 infection.

Our study has multiple limitations. Firstly, we had limited access to PBMC samples from SARS-CoV-2–infected individual with mild disease, which prevented us from performing the entire study in parallel with those with severe disease. Secondly, we were unable to compare the effects of different viral isolates on B cells. It is well accepted that SARS-CoV-2 isolates exhibit differential immunological properties (62). Therefore, such studies will enable us to understand how different SARS-CoV-2 isolates affect B cell functions. Thirdly, conducting further mechanism studies was not feasible due to the low frequency of B cells in our patients. It is worth mentioning that our studies were mainly performed on freshly collected samples from patients, and some variations in our observations might be related to performing studies on fresh versus previously frozen cells. However, our spike Ag-specific studies were performed on frozen PBMCs. It is very important to point out the sampling time because it affects B cell activation status and the frequency of transient B cell subset. Our blood sampling was performed around 2 weeks after disease onset.

The authors have no financial conflicts of interest.

We appreciate the contributions of COVID-19 patients and healthy volunteers that made our study possible.