PD-1 blockade may reduce immune problems after COVID-19 and boost the immune response against the virus

With the aim of providing a comprehensive description of the immune signature of patients who recovered from COVID-19, we first conducted an extensive immunophenotyping analysis of cells obtained from healthy controls and patients, whose main clinical characteristics are reported in Table 1. Among lymphocytes, CD19⁺ B cells were slightly decreased in patients who recovered from COVID-19 as compared with healthy controls, while an increase in CD8⁺ T cells was evident in patients who recovered from COVID-19 as compared both with those in the acute phase and with healthy controls (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.146701DS1↗). Patients who recovered from COVID-19 also showed a higher proportion of effector CD8⁺ T cells (CD45RO^–CD62L^–), while memory T cells, either effector or central (CD45RO⁺CD62L^– or CD45RO⁺CD62L⁺), were slightly altered among groups (Supplemental Figure 1, A and B). Furthermore, while the proportion of CD40L-expressing CD4⁺ T cells and that of ICOS-expressing CD8⁺ T cells was decreased in patients who recovered from COVID-19 as compared with patients with COVID-19 (Table 2), the fraction of OX40⁺ and GITR⁺ CD8⁺ T cells of patients who recovered from COVID-19 exceeded that of healthy controls, with the latter marker being the highest observed among all groups (Table 2).

In the evaluation of exhaustion markers, CD127⁺ and PD-1⁺ CD4⁺ T cell subpopulations were increased in patients who recovered from COVID-19 as compared with healthy controls, and they were at comparable levels to those observed in patients with COVID-19 (Figure 1, A–D), while 2B4⁺CD4⁺ T cell levels were similar to that of healthy controls but higher compared with patients with COVID-19 (Figure 1, E and F). Furthermore, an increased fraction of cells expressing LAG3, as well as a decreased proportion of cells marked by the expression of TIGIT, was observed in the CD4⁺ T cell population of patients who recovered from COVID-19 as compared with patients with COVID-19 and with healthy controls (Figure 1, G and H), suggesting the persistence of an abnormal exhaustion profile even after symptom remission. This was further confirmed in CD8⁺ T cells, in which several markers were upregulated in patients who recovered from COVID-19 as compared with healthy controls and/or compared with patients with COVID-19 (Table 2). Finally, transcriptomic profiling revealed an altered pattern of CD4⁺ and CD8⁺ T cell gene expression in patients who recovered from COVID-19 that particularly affected the CD4⁺ T cell subpopulation, with less dysregulation observed in CD8⁺ T cells (Figure 1, I and J). Indeed, downregulation of several proinflammatory genes, including CSF1, LAT, LTA, BTLA, CD40LG, JAK1, TNFSF14, and TNFRSF9, was evident in CD4⁺ T cells of patients who recovered from COVID-19 as compared with healthy controls (Figure 1I), with LAT and BTLA also downregulated in CD8⁺ T cells (Figure 1J). Furthermore, expression levels of genes controlling cell proliferation and apoptosis, such as TGFB1, CDK4, TNFSF10, and TNFRSF10A, were also found to be dysregulated in CD4⁺ T cells of patients who recovered from COVID-19 (Figure 1I). Overall, these results reveal unique features in the immunophenotype of patients who recovered from COVID-19.

To characterize the cytokine signature of patients who recovered from COVID-19, we assessed plasma levels of a panel of 27 cytokines in subjects of the 3 groups using a multiplex Luminex-based system (Figure 2A and Supplemental Table 1). We found that plasma levels of cytokines IL-1β, IL-1RA, IL-7, IL-8, IL-10, IFN-γ, and MIP-1α were higher in patients who recovered from COVID-19 as compared with healthy controls and were at comparable levels to those observed in patients with COVID-19 (Figure 2A), indicating a failure to return to physiological cytokine levels after COVID-19. Intriguingly, levels of IL-9, eotaxin, MIP-1β, and RANTES in patients who recovered from COVID-19 were found to be the lowest among the 3 groups (Figure 2A), further indicating that cytokine levels are dysregulated in patients with COVID-19 after clinical symptom remission. To provide mechanistic insight into the relation between systemic release of proinflammatory factors and the abnormal immunological phenotype observed in T cells of patients with COVID-19, we exposed PBMCs isolated from healthy controls to several proinflammatory cytokines, which were increased in patients’ circulation during and after COVID-19. PBMCs were cultured for 48 hours in medium supplemented with human serum and containing recombinant IL-1β, IL-1RA, IL-6, IL-8, and IP-10 added either individually or as a pool; afterward, cells were collected for exhaustion and costimulatory marker analysis by flow cytometry. We found that recombinant IP-10 administration increased the fraction of CD4⁺ T cells that were positive for the LAG3 exhaustion marker, while a lower percentage of 2B4⁺ cytotoxic CD8⁺ T cells was detected in PBMCs cultured in medium containing recombinant IL-1β or IL-6 as compared with medium alone (Figure 2B and Supplemental Figure 2, A–C). To further investigate the role of these cytokines on T cell phenotype, we also exposed PBMCs isolated from patients with COVID-19 to medium containing their own serum in the presence of blocking antibodies directed against IL-1β, IL-1RA, IL-6, IL-8, or IP-10, added either individually or as a pool. We then assessed the resultant changes on expression of T cell exhaustion and activation markers by flow cytometry. PBMCs exposed to a pool of sera obtained from patients with COVID-19 increased the proportion of several exhaustion and costimulatory markers (Figure 2C and Supplemental Figure 3, A and B). Notably, we observed an overall reversal of the COVID-19 serum–induced increase in costimulatory and exhaustion marker expression on both CD4⁺ and CD8⁺ T cells following addition of blocking antibodies to cell cultures (Figure 2C and Supplemental Figure 3, A and B). In particular, the expression of PD-1 and ICOS on CD4⁺ T cells and that of CD127 and CD40L on CD8⁺ T cells showed a marked decrease when the majority of blocking antibodies were added, both individually and as a pool, to PBMCs cultured with serum obtained from patients with COVID-19 (Figure 2C and Supplemental Figure 3, A and B). Overall, these findings suggest that high levels of serum cytokines, at least in part, account for the T cell exhaustion observed in patients who recovered from COVID-19.

Given the persistent T cell activation/exhaustion observed in patients who recovered from COVID-19, we aimed to investigate the CD4⁺ T cell–dependent response following specific or nonspecific in vitro stimulation using an ex vivo IFN-γ–based ELISpot assay. PBMCs isolated from patients who recovered from COVID-19, from patients with COVID-19, and from healthy controls were exposed in vitro to LPS, diphtheria/tetanus/pertussis vaccine (DTaP), or quadrivalent flu vaccine (FLU), and the T cell–mediated response was evaluated in terms of number of IFN-γ–secreting cells, as indicated by the number of spots observed per 1 × 10⁶ plated PBMCs (Figure 3, A–F). Cells isolated from patients who recovered from COVID-19 showed a markedly decreased response to nonspecific stimulation as compared with healthy controls; this response resembled that of patients with COVID-19 (LPS; Figure 3, A and B) and represents an additional immunological feature that distinguishes patients who have recovered from COVID-19. Given the increase in PD-1–expressing T cells observed in patients who recovered from COVID-19 and the role of the PD-1/PD-L1 axis in exhaustion onset, we sought to further confirm dysregulation of the PD-1/PD-L1 pathway in COVID-19. We thus assessed plasma levels of soluble PD-1 (sPD-1) and soluble PD-1 ligand PD-L1 (sPD-L1) — as well as the expression of PD-1– and PD-L1–targeting miRNAs miR–138-5p, miR–15a-5p, miR–16-5p, and miR–28-5p in CD4⁺ and CD8⁺ T cells — and the extent of PD1 promoter DNA methylation in CD4⁺/CD8⁺ T cells isolated from patients who recovered from COVID-19 as compared with those with COVID-19 and healthy controls (Figure 3, G–M). Interestingly, we found lower serum sPD-L1 levels in patients who recovered from COVID-19 as compared with both healthy controls and patients with COVID-19 (Figure 3H), suggesting an overstimulation of the PD-1/PD-L1 axis in these patients. This conclusion was further supported when we compared the PBMC-specific response to DTaP antigen stimulation in patients with higher (above the median) versus lower (below the median) plasma sPD-1 or sPD-L1 levels. High levels of both sPD-1 and sPD-L1 were associated with a marked decrease in the PBMC immune response to DTaP in patients who recovered from COVID-19 and patients with COVID-19 as compared with healthy controls (Figure 3, I and J). We next investigated the methylation status of a specific PD1 promoter CpG site that is reported to control PD1 gene expression in CD4⁺ and CD8⁺ T cells. Methylation-specific quantitative PCR (qPCR) results demonstrated an altered degree of PD1 promoter methylation in CD8⁺ T cells — but not in CD4⁺ T cells — of subjects who recovered from COVID-19 as compared with healthy controls, confirming dysregulation of the PD-1/PD-L1 axis in patients who recovered from COVID-19 (Figure 3L). In CD8⁺ T cells — but not in CD4⁺ T cells — the PD-1 miRNet was also found to be altered, as shown by the higher expression of miR–15a-5p in patients who recovered from COVID-19 and in patients with COVID-19 as compared with healthy controls (Figure 3M). Overall, our results suggest an abnormal T cell phenotype and function of the immune response in patients who recovered from COVID-19 (Figure 3N), and this is at least partially due to a dysregulated PD-1/PD-L1 axis in T cells. Interestingly, these immune abnormalities were found to be associated with a persistence of dyspnea and several additional COVID-19–related symptoms at the time of recruitment (Table 3).

Based on these results, we hypothesized that PD-1 blockade could counteract post–COVID-19 immune abnormalities (Figure 4A). We therefore sought to determine whether PD-1 blockade, achieved by use of a clinically relevant anti–PD-1 blocking mAb, would revert the exhaustion status of T cells and restore their functional activity in response to specific and nonspecific stimulation. In an ELISpot ex vivo assay, we examined the function of T cells in patients who recovered from COVID-19 and in controls following challenge with LPS, DTaP, and FLU in the presence of anti–PD-1 mAb. FLU-stimulated PBMCs collected from patients who recovered from COVID-19 and cultured in the presence of anti–PD-1 displayed a significantly enhanced immune response as compared with FLU-only–stimulated PBMCs (Figure 4B). A similar pattern was observed when using PBMCs from healthy controls, in which cells cultured with anti–PD-1 mAb showed an increased response to DTaP stimulation as compared with DTaP-only–treated cells (Figure 4B). Given the increased expression of costimulatory and exhaustion T cell markers observed in patients who recovered from COVID-19, we next investigated the ability of PD-1 blockade to reverse this phenotype (Figure 4, C–J, and Supplemental Figure 4, A and B). Ex vivo anti–PD-1 blockade reversed the increase in the expression of the positive costimulatory marker OX40 in the CD8⁺ T cell fraction that was observed in patients who recovered from COVID-19 (Figure 4E). Moreover, PD-1 blockade in PBMCs isolated from patients with COVID-19 resulted in a decrease in the percentage of exhausted T cells (Figure 4, C–J). Taken together, our data indicate that, in T cells isolated from patients with COVID-19 upon PD-1 immune checkpoint blockade, T cell immune response and phenotype improve. We then examined, in an ex vivo assay, the T cell response to a pool of peptides derived from SARS-CoV-2 spike and nucleocapsid proteins, with or without the administration of anti–PD-1 blocking mAb, or in the presence of an anti–human IgG used as a negative control. The number of IFN-γ–producing cells was then quantified in an ELISpot assay. A sizeable T cell response against SARS-CoV-2 antigens at baseline was mounted only when PBMCs isolated from patients who recovered from COVID-19 were used (Figure 4K). Remarkably, the addition of anti–PD-1 mAb resulted in a 2-fold increase in the number of IFN-γ–producing cells as compared with cells treated with SARS-CoV-2 peptide only or compared with the negative control (Figure 4K). This finding confirms that PD-1 blockade is able to bolster the specific T cell immune response to SARS-CoV-2 antigens, thus reinstating their functional activation. We then verified if PD-1 blockade prevents T lymphocyte–mediated neutralization of SARS-CoV-2. To this aim, we developed a SARS-CoV-2 cell-based neutralization assay to indirectly assess the CD3⁺ T cell antiviral response. First, a SARS-CoV-2 spike pseudotyped lentivirus containing a luciferase reporter gene was exposed to CD3⁺ T cells isolated from patients who recovered from COVID-19 in the presence of anti–PD-1 blocking antibody. Residual pseudoviral particles were collected and used to infect SARS-CoV-2 infection–sensitive, ACE2-overexpressing 293T cells, and T lymphocyte neutralization activity was measured by comparing luciferase activity in infected cells (Figure 4L). The sera of patients who recovered from COVID-19 containing high titers of anti–SARS-CoV-2 IgG was used as a positive control. Lymphocyte-driven pseudovirus neutralization was observed when pseudoviral particles were exposed to CD3⁺ T cells of patients who recovered from COVID-19 upon PD-1 blockade, in a pattern comparable with that which was observed when sera of patients who recovered from COVID-19 was used (Figure 4L). We thus demonstrate that PD-1 blockade enhances the anti–SARS-CoV-2–specific immune response and reinforces PBMC-mediated SARS-CoV-2 antiviral activity.

Consecutively admitted patients with COVID-19 or patients who recovered from COVID-19 infection admitted for SARS-CoV-2 acute infection at the Infectious Disease and Respiratory Division of ASST FBF-Sacco in Milan, Italy, from March 31, 2020, to October 2, 2020, were enrolled and compared with a group of healthy control subjects. SARS-CoV-2 infection of all enrolled patients was confirmed by viral PCR of nasal and pharyngeal swab specimens collected during the acute phase of the infection, according to WHO guidance. Patient baseline clinical score was determined according to a modified ordinal score based on 7 major points as previously reported (31). Baseline demographic distributions and clinical, laboratory, management, and outcome data were abstracted from patient electronic medical reports. Immunophenotyping was performed by flow cytometry using antibodies listed in Supplemental Table 2.

An ELISpot assay was used to measure the number of IFN-γ–producing cells according to the manufacturer’s protocol (BD Biosciences) as previously shown by our group (50). For testing nonspecific ex vivo T cell responses, 3 × 10⁵ PBMCs isolated from healthy control subjects, from patients with COVID-19, and from patients who recovered from COVID-19 were cultured for 48 hours in RPMI 1640 10% FBS in the presence of LPS (1 μg/mL; Merck), FLU, or a DTaP, with or without the addition of an anti–PD-1 blocking antibody (pembrolizumab, 5 μg/mL). At day 2 after stimulation, cells were collected and plated on a human IFN-γ ELISpot assay plate, and spots were counted using an ImmunoSpot Reader (CTL Europe GmbH). The specific response to SARS-CoV-2 antigens was tested by culturing 3 × 10⁵ PBMCs isolated from healthy control subjects, patients with COVID-19, and patients who recovered from COVID-19 for 48 hours in RPMI 1640 10% FBS in the presence of pooled peptides derived from SARS-CoV-2 spike and nucleocapsid proteins (PR-nCoV-1, PR-nCoV-3; Novatein Biosciences) (1 μg/mL) with or without the addition of the anti–PD-1 blocking antibody pembrolizumab (Keytruda, MSD) (5 μg/mL) or of an anti–human IgG antibody used as a negative control (clone 83.8F9, LifeSpan Bioscience) (5 μg/mL). At day 2 after stimulation, cells were collected and plated using the human IFN-γ ELISpot assay and processed as previously described.

Spike SARS-CoV-2 pseudotyped lentiviruses containing a luciferase reporter (BPS Bioscience) were used in our modified neutralization assay. Briefly, 5 μL of SARS-CoV-2 pseudotyped lentivirus were incubated with 5 × 10⁵ PBMCs from patients who recovered from COVID-19 with the addition of anti–PD-1 blocking antibody. Cells were cultured in RPMI 1640 in a 96-well white clear-bottom assay plate and incubated for 12 hours at 37°C. SARS-CoV-2 pseudotyped lentivirus incubated with serum collected from patients who recovered from COVID-19 served as a positive control. After incubation, supernatants were collected from every condition and used for infection of ACE2-overexpressing 293T cells (BPS Bioscience); they were then further incubated for 48 hours in 5% CO₂ at 37°C. Luminescence, which is correlated to the luciferase activity on 293T cells, was quantified using the One-step Luciferase Assay System as recommended by the manufacturer on a GloMAX luminometer (both from Promega).

Prism version 7.0 (GraphPad) was used for statistical analysis and graphical representation of data. The sample distribution was determined by the Shapiro-Wilk normality test. One-way ANOVA, 2-tailed t test, χ² test, or nonparametric Kruskal-Wallis and Mann-Whitney U tests were performed where appropriate and according to data distribution. All data are presented as mean ± SEM, with P values less than 0.05 considered as significant.

All samples were obtained from patients and healthy controls after provision of informed consent and in accordance with the Ethical Research Committee of the Sacco Hospital (Comitato Etico Milano Area 1), which granted the approval of the present study.

In-Press Preview

Electronic publication