Attenuated humoral responses in HIV after SARS-CoV-2 vaccination linked to B cell defects and altered immune profiles

Summary We assessed a cohort of people living with human immunodeficiency virus (PLWH) (n = 110) and HIV negative controls (n = 64) after 1, 2 or 3 SARS-CoV-2 vaccine doses. At all timepoints, PLWH had significantly lower neutralizing antibody (nAb) titers than HIV-negative controls. We also observed a delayed development of neutralization in PLWH that was underpinned by a reduced frequency of spike-specific memory B cells (MBCs). Improved neutralization breadth was seen against the Omicron variant (BA.1) after the third vaccine dose in PLWH but lower nAb responses persisted and were associated with global MBC dysfunction. In contrast, SARS-CoV-2 vaccination induced robust T cell responses that cross-recognized variants in PLWH. Strikingly, individuals with low or absent neutralization had detectable functional T cell responses. These PLWH had reduced numbers of circulating T follicular helper cells and an enriched population of CXCR3+CD127+CD8+T cells after two doses of SARS-CoV-2 vaccination.


PLWH have lower levels of neutralizing antibodies (nAbs) after SARS-CoV-2 vaccination
Delayed production of nAbs is associated with greater memory B cell (MBCs) disturbance Additional doses increase nAbs titers and breadth despite persistent MBCs perturbations SARS-CoV-2 vaccination elicits robust T cell responses in PLWH PLWH with spike-specific T cells but no nAbs have increased CXCR3+CD127+CD8 + T cells

INTRODUCTION
People living with human immunodeficiency virus (HIV) [PLWH] appear to be at a higher risk of hospitalization and worse clinical outcomes from COVID-19 disease, especially in the context of cellular immunosuppression and unsuppressed HIV viral load. 1 Although combined antiretroviral therapy (cART) has dramatically improved life expectancy in PLWH, the persistence of immune dysfunction raises concerns about the overall effectiveness and durability of vaccine responses in this potentially more vulnerable patient group, in line with other immunocompromised groups 2,3 As a result, PLWH were included in either priority group 4 (for clinically vulnerable PLWH, based on more advanced immunosuppression or co-morbidities) or 6 (all other PLWH) 4 in the UK for earlier COVID-19 vaccination than the general population. The Joint Committee on Vaccination and Immunization (JCVI) advised to invite this patient group for an additional booster dose. 4 Previously, defects have been observed in serological vaccine responses in PLWH on cART. For example, after a full course of hepatitis B 5 or influenza vaccination 6 and long-term responses to vaccination can be shorter-lived in PLWH compared to the general population. 7 We and others have previously shown a failure to mount a robust antibody response following COVID-19 vaccination in advanced HIV infection with low CD4 T cell counts below 200 cells/mL. [8][9][10][11][12] Data on vaccine efficacy and immunogenicity in PLWH remains limited (reviewed in 13 ), and although there are some conflicting results, meta-analyses 14 and recent studies 15 have shown reduced levels of seroconversion and neutralization after a second dose of viral vector vaccine dose in PLWH receiving cART, with lower CD4 T cell count/viremia and older age resulting in a more impaired response and more rapid breakthrough infection. 16 Assessment of vaccine efficacy has been continually complicated by the ongoing emergence of variants of concern (VOC), with the Alpha, Beta, Delta and Omicron variants being observed to progressively evade antibodies 17,18 raised against the original Wuhan-Hu-1 strain in most vaccines. In particular, data after three vaccine doses has been hard to assess because of the emergence of Omicron Hypertension/CVD, n -3 3 6 Renal disease, n --2 1 Liver disease, n --2 2 Respiratory disease, n -1 3 1  iScience Article Longitudinal samples from 53 PLWH and 44 controls were then evaluated to assess binding antibody responses and nAbs over time after each vaccine dose. These included samples after the first dose and for at least one additional timepoint, often including a baseline, post-second, pre-third and post-third sample ( Figure 1D). This analysis revealed two clear trajectories of the development of neutralization, firstly where nAbs were detected after a single vaccine dose, 37 defined here as ''standard neutralization'', and secondly where neutralization was not achieved until after the second dose or later, defined as ''delayed neutralization''. Most HIV-negative controls without prior SARS-CoV-2 infection show a standard neutralization profile, with only 3 individuals failing to mount a neutralizing response until after the second dose ( Figure 1D), and a similar effect was seen with binding responses (Figures S1H-S1K). In contrast, two-thirds of SARS-CoV-2 naive PLWH did not make a detectable neutralizing response until after the second dose and a substantial proportion of them lost detectable neutralizing activity before the third dose ( Figures 1A and 1E). However, both PLWH and HIV-negative controls with a history of SARS-CoV-2 infection made a standard neutralizing response ( Figures 1F and 1G). Therefore, having identified this delayed neutralization phenotype in SARS-CoV-2 naive PLWH, we have evaluated its relationship with total CD4 T cell counts, which are known to be important for SARS-CoV-2 vaccine responses in PLWH. [8][9][10]12 No significant difference was seen in median CD4 T cell count or CD4:CD8 T cell ratio between PLWH with standard or delayed neutralization profiles ( Figures 1H and 1I); or correlate either with the rapid development of neutralization (Figure S1M and S1N).
Delayed neutralization is associated with lower frequency of spike-specific MBCs and a perturbed MBC global phenotype Spike is the SARS-CoV-2 glycoprotein and is the sole antigen in most vaccines. It has been previously shown that infection and vaccination produce spike-specific MBCs in proportion to serological responses. [38][39][40][41][42] Given that the delay in neutralization observed more frequently in PLWH was not clearly associated with peripheral CD4 T cell counts, we next assessed the relationship with global MBCs and spike-reactive MBC frequency and phenotype, using a previously validated flow cytometry panel, with memory B cells defined as CD19 + CD20 + CD38 lo/-IgD-( Figure S2). This analysis was performed on available PBMC samples after the first vaccine dose, using SARS-CoV-2 naive baseline samples to determine the antigen-specific gate ( Figure 2A). We observed a significantly lower frequency of spike-specific MBCs in SARS-CoV-2 naive participants after the first dose as compared to those with a history of prior infection, regardless of HIV status ( Figure 2B). Moreover, a lower frequency of spike-specific MBCs was observed in SARS-CoV-2 naive participants who had a delayed neutralization response, although notably there were a small number of donors in the standard neutralization group ( Figure 2C). In line with this, the percentage of spike-specific MBCs showed a strong correlation with the nAb titer ( Figure 2D) in agreement with previous findings during SARS-CoV-2 convalescence. 42 Subsequent gating on CD21 and CD27 expression allowed the identification of four populations of classswitched MBCs: CD21 À CD27 À atypical MBCs (also known as tissue-like memory); CD21 À CD27 + activated MBCs; CD21 + CD27 + classical resting MBCs and CD21 + CD27 À switched naive (also known as intermediate memory) MBCs ( Figure 2E) as previously described. 42 Global defects in the balance of these MBC subsets have been identified previously in PLWH (reviewed in 30 ), including those on cART, 43 with increased numbers of activated and atypical MBCs concurrent with a decrease in resting MBCs. This phenotype is exemplified in ( Figure 2E) for a PLWH and an HIV-negative control. We have hypothesized that these inherent defects may have an impact on the quality of serological responses after SARS-CoV-2 vaccination. iScience Article Global phenotyping of the MBC response after the first vaccine dose revealed that individuals with delayed neutralization, consisting largely of PLWH, had significantly lower numbers of resting MBCs (CD21 + CD27 + ) and greater numbers of both CD21 À CD27 + activated MBCs and CD21 À CD27 À atypical MBCs compared to those with standard neutralization ( Figure 2F). Moreover, lower frequencies of resting MBCs correlated with lower nAb titers ( Figure 2G), this association is driven by participants with standard neutralization and some individuals with delayed neutralization do have reasonable numbers of resting MBCs. Similarly, higher levels of atypical MBCs significantly correlated with lower nAb titers when considering those with standard neutralization, although the strength of this association was relatively weak (r = À0.4867) (Figure 2H). Frequencies of more than 10% atypical MBCs were only observed in the delayed neutralization group. Together these findings suggest that the MBC subset perturbations seen in PLWH could account for the lower serological output.
Improved neutralization breadth after the third SARS-CoV-2 dose in PLWH but lower nAb responses persist and are associated with global, but not spike-specific, MBC dysfunction To assess the breadth of nAb responses across the cohort, samples from all timepoints were tested against an Omicron pseudovirus (BA.1 strain), which represented the dominant circulating strain at the time of the post third vaccine dose sampling. Owing to the substantial antigenic changes in the Omicron spike, 44 in participants with no prior infection, over 50% of HIV-negative controls and more than 90% of PLWH were not able to neutralize Omicron after the first vaccine dose ( Figure 3A). The second dose enabled most of the control group to mount a neutralizing response whereas only a quarter of SARS-CoV-2 naive PLWH had nAbs against Omicron. In the SARS-CoV-2 naive groups, the third dose enabled 100% of HIV-negative controls to neutralize Omicron and increased the frequency of neutralization among PLWH to over 70% ( Figure 3A). As in the analysis of WT neutralization for individuals without prior SARS-CoV-2 infection, median Omicron ID 50 titers were lower in SARS-CoV-2 naive PLWH compared to HIV-negative controls at all timepoints ( Figure 3B). In addition, there was no significant difference when individuals with complex co-morbidities were removed from the PLWH cohort at the third vaccine dose ( Figure S3C) or whether they had previously been infected with SARS-CoV-2. These data suggest that the third vaccine dose was effective in both boosting nAb titer and broadening the response to Omicron, especially in SARS-CoV-2 naive PLWH, thus rendering their responses closer to those of SARS-CoV-2 naive HIV-negative controls ( Figures 3B and 3C).
Next, we evaluated cross-sectionally the B cell phenotype after the third vaccine dose. In contrast to the first vaccine dose, there was no significant difference between the frequency of spike-specific MBCs when individuals were stratified by whether they had been previously infected with SARS-CoV-2 or not (Figure 3D) regardless of HIV status. However, the frequency of spike-specific MBCs after the third dose correlated with Omicron titers ( Figure 3E). This suggests that after three vaccine doses these individuals had mounted a specific B cell response, and that the quantity of spike-specific B cells remained linked to the improved neutralization potency and breadth observed ( Figures 3A-3C). Given that all individuals assessed after the third dose made a robust spike-specific MBC response, we wanted to evaluate further Figure 2. Neutralization titer is associated with the frequency of spike-specific MBCs after the first vaccine dose (A) Spike-specific MBCs (CD19 + CD20 + CD38 lo/mid IgD-excluding switched naive CD27 À CD21 + cell) according to dual positivity for spike-PE and spike-APC to exclude non-specific binding in a representative naive pre-vaccine sample (left) or representative post-vaccine sample (right) after the first vaccine dose.
(B) Percentage of spike-specific MBC after the first vaccine dose stratified by prior SARS-CoV-2 infection. Line represents median for each group. Statistical test: M-Whitney U test (MWU). Dotted lines represent lower limit of sensitivity of the assay (0.1% spike-specific MBCs, based on previous optimization). 42 (C) Percentage of spike-specific MBCs in SARS-CoV-2 naive donors after the first vaccine dose, stratified by delayed (magenta) or standard (gray) neutralization profile. Line represents median for each group. Statistical test: MWU. Dotted lines represent lower limit of sensitivity of the assay (0.1% spikespecific MBCs).
(D) Correlation of the percent of spike-specific MBC with WT ID 50 titers stratified by PLWH (blue) and controls (gray) after the first dose, statistical test: Spearman's rank correlation coefficient.
(F) Percentage of MBC subtypes (activated CD27 + CD21 À ; resting CD27 + CD21 + ; switched naive; switched naive CD27 À CD21 + and CD27 À CD21 À atypical) after the first vaccine dose stratified by delayed or standard neutralization profile. Line represents median for each group. Statistical test: MWU. iScience Article whether alterations in spike-specific MBC phenotype also contributed to differences in serum neutralization ( Figures 3A, 3B, 3F, S3A, and S3B). Spike-specific B cells were found to be comparable across the different MBC subsets in both PLWH and HIV-negative controls, except for a trend to fewer spike-specific resting MBCs in PLWH as compared to controls ( Figure 3G). This was the case even though the global MBC population for these post third vaccine dose samples showed classical anomalies in MBCs associated with HIV infection ( Figure 3H). These data suggest that SARS-CoV-2 serum antibody responses are lower potentially because of a global MBC disturbance thereby limiting the overall B cell response. In line with this proposal, we anticipated that underlying global MBC disturbances would also influence the efficiency of the antigen-specific B cell response in other ways, beyond limiting the number of spike-specific MBCs, for example by limiting class-switching. Indeed, this is supported by our data showing similar levels of IgG+ and IgM+ global MBCs in both groups ( Figure 3I) but a significantly lower level of spike-specific IgG+ MBCs in PLWH after the third vaccine dose as compared to controls, and conversely a higher frequency of spike-specific IgM+ MBCs ( Figure 3J).

SARS-CoV-2 vaccination induces robust T cell responses that cross-recognize variants in PLWH
To increase our understanding of the complementary role of cellular immunity after vaccination, we have examined T cell responses in our cohort, including their reactivity to SARS-CoV-2 variants. The magnitude of spikespecific T cell responses was assessed cross-sectionally by IFN-g-ELISpot using overlapping peptide (OLP) pools covering the complete sequences of the WT spike glycoprotein as previously described. 45 The majority of PLWH had detectable SARS-CoV-2-specific T cell responses at levels comparable to HIV-negative individuals following each vaccine dose ( Figures 4A-4C). A greater magnitude of spike-specific T cells was observed in individuals with prior SARS-CoV-2 infection, irrespective of HIV status (Figures 4A-4C) in keeping with previous reports. 34,46,47 There were no detectable T cell responses in a small number of PLWH with no prior exposure to SARS-CoV-2 across all timepoints. These were participants with incomplete immune reconstitution on cART and/or additional co-morbidities, such as transplant recipients on immunosuppressive therapy ( Figures 4A-4C). Next, we examined the longitudinal evolution of T cell responses in a subgroup of donors with available PBMC samples. In SARS-CoV-2 naive individuals, spike-specific T cell responses increased following the first vaccine dose, peaked after the second dose and were maintained after the third vaccine dose ( Figure 4D). In one HIV-positive, SARS-CoV-2-naïve donor with advanced immunosuppression and persistently low CD4 T cell count of 100 cells/mL on cART, a third dose (mRNA) vaccine was able to elicit a T cell response despite no evidence of neutralization ( Figure 4D). A higher proportion of PLWH without prior SARS-CoV-2 infection had detectable T cell responses at baseline compared to HIV-negative controls, which could represent the presence of cross-reactive responses to other pathogens, probably to related coronaviruses ( Figure 4D). [48][49][50][51][52] However, due to the small number of participants with detectable T cell responses at baseline, this study was     Previous work has demonstrated that T cell responses are largely retained against variants of concern (VOCs), including the highly transmissible BA.1 Omicron variant, and therefore may be important when antibody levels wane or new variants emerge that can partly escape antibody responses. To determine T cell reactivity to VOCs, we assessed T cell responses to the mutated regions, including Omicron, in our study cohort. The magnitude of T cell responses against B.1.1.529 was comparable between PLWH and HIV-negative donors regardless of prior SARS-CoV-2 infection ( Figure 4E). Notably, responses were further enhanced by a third vaccine dose in all donors, irrespective of prior SARS-CoV-2 infection or HIV status and in keeping with the beneficial effect of a third vaccine dose in boosting humoral responses ( Figure 4E). T cell reactivity to Omicron and other VOCs, including Alpha, Beta and Delta, was comparable between HIV-negative and PLWH with or without prior SARS-CoV-2 infection after three vaccine doses, and these responses were maintained against the ancestral Wuhan Hu-1 spike peptide pool, reinforcing the relative resilience of T cell responses to spike variation ( Figures S4D and S4E). We noted that three HIV-negative and five HIV-positive individuals, regardless of prior SARS-CoV-2 infection, had no detectable T cell responses to the Wuhan Hu-1 peptide pool, covering only the affected regions of spike. This could be because of the VOC mutations occurring in regions that are poorly targeted by T cell responses in some individuals. 34 Although spike-specific T cell responses were detected at similar frequencies across all groups ( Figures 4A-4C), there was variation in the magnitude of responses. To better understand the factors underlying this heterogeneity, we examined the role of various HIV parameters. 45 We have previously reported an association between the CD4:CD8 T cell ratio and total SARS-CoV-2 responses, especially against the nucleocapsid (N) and membrane (M) protein, in PLWH recovering from COVID-19 disease. 45 No correlation was observed between the CD4:CD8 T cell ratio and spike-specific T cell responses following vaccination in our cohort ( Figures S4F-S4H). However, a positive correlation was detected between the CD4 T cell count and spike-specific T cell responses after the first vaccine dose (r = 0.5153) in SARS-CoV-2 naive PLWH ( Figure 4F). This association was weaker after the second vaccine dose (r = 0.4596) and non-significant after the third dose ( Figures 4G and 4H). Together these observations suggest that an effective helper T cell response could drive the induction of cellular immunity following vaccination in individuals without prior exposure to SARS-CoV-2. However, the lack of an association between CD4 T cell counts and antibody responses further underlines the relative importance of HIV-associated B cell defects in modulating the induction of effective humoral immunity in addition to potential insufficient T cell priming.

A proportion of PLWH had low or absent nAbs (ID 50 < 1:150) but detectable T cell responses following vaccination
We examined next the relationship between humoral and cellular responses by comparing antibody responses and neutralization titers with T cell responses detected by IFN-g-ELISpot following SARS-CoV-2    Figures 5A-5C). Similar associations were observed for S1 IgG binding titers ( Figures S5A-S5C). One HIV-positive SARS-CoV-2 naive donor with a low CD4 T cell count of 40 cells/mL on cART, and one individual with relapsed lymphoma, both had no detectable humoral and cellular responses after 2 or 3 doses of mRNA vaccine. A proportion of PLWH, in particular those without prior SARS-CoV-2 infection, had low or absent nAbs (ID 50 <1:150) but detectable T cell responses following vaccination ( Figures 5A-5C). To better visualize these relationships in SARS-CoV-2 naive individuals, we ranked T cell responses after second and third doses according to the magnitude of neutralizing antibodies ( Figures 5E-5G). All of the HIV-negative donors had detectable cellular and neutralizing antibodies (Figure 5D). However, a proportion of SARS-CoV-2 naive PLWH with low or absent nAbs (n = 9 out of 10) had measurable cellular responses to the spike protein after two vaccine doses ( Figure 5E). These donors were all controlled on cART with a median CD4 T cell count of 680 cells/mL and no significant underlying comorbidity (Table S1). Although all HIV-negative individuals had both detectable nAbs and cellular responses post third dose ( Figure 2F), a small number of PLWH SARS-CoV-2 naive donors (n = 7 out of 9) had detectable T cell responses in the absence of, or only low-level, neutralization ( Figure 5G). Similarly, these donors were all well controlled on cART with a median CD4 T cell count of 492 cells/mL. One of these donors who presented with advanced HIV infection had a persistently low CD4 T cell count (100 cells/mL), and one of the donors recruited after a third vaccine dose had a previous splenectomy. These data suggest that in a small proportion of PLWH, serological non-responders or with evidence of low-level neutralization, cellular immune responses may play an important compensatory role.

PLWH with suboptimal serological responses demonstrate an expansion of CXCR3 + CD127 + CD8 + T cells after two doses of SARS-CoV-2 vaccination
The presence of detectable T cell responses in a subgroup of SARS-CoV-2 naive HIV-positive donors with low or absent nAbs after two or three vaccine doses prompted us to further evaluate the phenotype of the T cell compartment. We have compared T cell immune signatures in SARS-CoV-2 naive PLWH with potent neutralization titers (>1:150) and functional T cell responses (PLWH SARS-CoV-2 naive nAb high T + , n = 9), with SARS-CoV-2 naive PLWH with low/absent nAbs and a functional T cell responses (PLWH SARS-CoV-2-nAb À/low T + , n = 9). Both groups were age and sex matched, well controlled on cART and with a similar median CD4 T cell count (Table S1). We have used an unbiased approach and unsupervised high-dimensional analysis, global t-distributed stochastic neighbor embedding (t-SNE), followed by FlowSOM clustering, in circulating T cell populations in the two groups. Ten major CD4 and CD8 T cell subsets were examined using a combination of various activation and differentiation markers, including CD45RA, CCR7, CD127, CD25, CXCR3, CXCR5, PD-1, and CD38 ( Figures 6A, S6A, and S6B). There was no difference iScience Article in the frequencies of the main T cell subsets in the two groups ( Figures S6C and S6D). Among CD4 T cells, there was a reduction in circulating CXCR3 + CXCR5 + T follicular helper (T FH ) subsets observed in HIV-positive nAb À/low compared to nAb high donors ( Figures 6A and 6B). The reduced abundance of CXCR3 + CXCR5 + T FH in nAb À/low HIV-positive subjects was further confirmed by manual gating (Figures 6C, 6D, and S3C). CXCR3 + CXCR5 + T FH cells correlated with SARS-CoV-2 neutralization levels in HIV-positive SARS-CoV-2 naive individuals (r = 0.5294 p = 0.02388) ( Figure 6E), suggesting that reduced availability of T FH cells could influence the magnitude of vaccine-induced SARS-CoV-2 antibody responses.
We have next examined the CD8 T cell compartment in the two groups. Notably, a prominent cluster delineated by the expression of CXCR3 + CD127 + CD38 + CCR7 + CD45RA + was significantly enriched in PLWH SARS-CoV-2-nAb À/low T + (Figures 6F, 6G, and S6B). The higher abundance of CXCR3 + CD127 + CD38 + CCR7 + CD45RA + cells in PLWH SARS-CoV-2-nAb À/low T + was further confirmed by manual gating (p = 0.04) (Figures 6H, 6I, and S6C). Correlation analysis of these populations showed a positive association between their frequencies and SARS-CoV-2-specific T cell responses following two vaccine doses in PLWH with nAb À/low (Figure 6J), supporting the notion that these subsets could contribute to the observed induction of T cell responses in PLWH who lacked or generated low nAb responses. Overall, our analysis of the global T cell profile of individuals with low/absent nAbs but detectable functional T cell responses revealed that reduced availability of T FH cells could contribute to the serological defect observed in conjunction with the previously highlighted imbalance in MBCs. Moreover, we have identified a subset of CD8 T cells that is overrepresented in PLWH with low/absent nAbs and may enable stronger functional T cell responses, supported by recent findings showing that CXCR3 + CD8 T cells are polyfunctional and associated with survival in critical SARS-CoV-2 patients, and have been observed in other immunosuppressed groups. 53,54

DISCUSSION
Accumulating evidence suggests that a broad and well-coordinated immune response is required for protection against severe COVID-19 disease. The emergence of VOCs with increased ability to evade nAbs has reinforced the need for a more comprehensive assessment of adaptive immunity after vaccination, especially in more vulnerable groups including some PLWH. Our data indicate that PLWH who are well controlled on cART, elicited poorer humoral responses, in terms of magnitude and neutralizing ability compared to HIV-negative donors following first, second and third doses of SARS-CoV-2 vaccine. This was related to global B cell but not antigen-specific B cell dysfunction. This suggests that the overall disturbance in memory B cell homeostasis during HIV can limit the amount or quality of serum antibody produced indirectly, potentially by decreasing the total number of B cells available to participate in the antigen-specific response. In contrast, the observation that antigen-specific B cells in these individuals are not overtly dysfunctional suggests that lower serum titers are not the result of antigen-specific B cells failing to respond fully, as has been suggested in some chronic diseases. 27 In contrast, T cell responses were comparable in the two groups and detectable, even in a small group of PLWH with very poor serological responses, suggesting a potentially important non-redundant immunological role for functional T cells.
Overall, our data reinforce the beneficial effect of an additional vaccine dose in boosting adaptive immune responses, 20 especially against circulating VOCs in this patient group.
Weaker humoral responses were observed in PLWH compared to HIV-negative controls after each dose of vaccine when matched by prior SARS-CoV-2 status. Although the third dose largely narrowed the gap between PLWH and controls, and enabled Omicron neutralization, 13% of SARS-CoV-2 naive PLWH still had no nAbs after 3 vaccine doses. This highlights the importance of repeated vaccination in PLWH and suggests additional doses/targeted vaccines could be merited, especially given 28% of SARS-CoV-2 naive PLWH failed to neutralize Omicron after the third vaccine dose. Owing to known defects in germinal center reactions in chronic HIV infection (as reviewed in 29,30,55 recall responses following vaccination in PLWH are likely impaired resulting in lower titers and narrower neutralization breadth. As such, additional vaccinations to stimulate additional affinity maturation and diversification of the response are likely needed to achieve similar outcomes to those seen in HIV-negative individuals. In support of this, previously, it has been shown that PLWH can benefit from an additional vaccine dose and accelerated schedules during hepatitis B immunization. 56 Moreover, in other immunocompromised groups, repeated vaccination with a third vaccine dose resulted in similarly improved levels of seroconversion and breadth against VOCs. 57,58 Previous studies among similar cohorts of PLWH with undetectable HIV viral loads have produced mixed results, as previously reviewed. 13  iScience Article durability of antibody responses to HIV-negative controls 23,59 but reduced levels of seroconversion and neutralization have been reported after two doses in PLWH in a more recent study. 15 Furthermore, viral vector vaccines, lower CD4 T cell count/viremia and old age have been linked to lower serological responses and breakthrough infection. 16 In terms of mRNA vaccines, both non-significant 60,61 and significant decreases in humoral responses have been reported in PLWH. [62][63][64][65] These differences may be because of the size of cohorts examined and the range of immune reconstitution in these PLWH. In contrast to previous work, [8][9][10]12 we have found no association between the CD4 T cell count and serological outcome, which could be because of insufficient power in this study to detect differences. The recent study by 20 demonstrated a stronger humoral response after the third dose of vaccine in PLWH, regardless of their CD4 T cell count, which is consistent with our findings, 20 Thus, the lower level of nAbs observed here in PLWH could be in part owing to potential differences in boosting of memory responses to enable breadth against Omicron after three vaccine doses. Serological data correlated significantly with frequency of spike-specific MBCs. The B cell phenotyping confirmed the characteristic and persistent defects seen in global MBCs in the setting of HIV (reviewed in 30 ). Specifically, we have observed lower frequencies of resting MBCs and higher frequencies of atypical and activated MBCs. This dysregulated MBC phenotype was also associated with a delay in developing nAbs after the first dose regardless of HIV status. Further evaluation in a group of individuals after the third vaccine dose led to the interesting observation that although the global MBC landscape is still disrupted with lower levels of resting MBCs and higher levels of atypical and switched naive MBCs in PLWH, this is not reflected in the antigen-specific MBCs. Spike-specific MBCs present in PLWH had a similar memory B cell phenotype as HIV-negative controls, albeit fewer resting MBCs. However, higher levels of global atypical MBCs, also observed in PLWH with lower neutralization at the third vaccine dose, suggest that the excess atypical MBCs may be effectively exhausted, as has been described. 66 Therefore, SARS-CoV-2 serum antibody responses may be lower not because spike-specific responses are enriched within atypical MBCs and therefore unable to progress to an antibody secreting phenotype (as has been postulated for HIV/HBV 27,67 ), but rather because of global MBC disturbance. Thus, we propose that this reduced nAb to vaccination in PLWH may not be because of an alteration in the phenotype of antigen-specific cells but rather limited numbers of MBCs available to participate in the antigen-specific response via the canonical pathway.
In contrast to serological responses, SARS-CoV-2 vaccination elicited comparable T cell responses between PLWH and HIV-negative controls at all sampling points, and these responses were largely preserved against circulating VOCs, including Omicron, following three vaccine doses. These findings are in line with the recent observations showing a robust T cell response to SARS-CoV-2 after third dose in PLWH, including to known VOCs, with either homologous or heterologous combinations of SARS-CoV-2 vaccines. 19,20 Of interest, in a recent study despite the detection of significant T cell responses post a third dose mRNA vaccine in PLWH who had completed an mRNA primary course, these responses were reported to be impaired compared to the general population. 20 These results contrast our findings and could be attributed to the different study design/vaccine platforms and quantification of T cell responses using whole blood assay stimulation and IFN-g quantification via ELISA. 20 Similarly, to the scenario seen in antibody responses, prior SARS-CoV-2 infection also resulted in higher T cell responses to vaccination. 34,46,47 Of interest, detectable T cell responses were noted in a proportion of SARS-CoV-2 naive individuals at baseline, 23,45 which could represent pre-existing cross-reactive T cell cells due to past infection with other coronaviruses. 68 An association between CD4 T cell counts and the magnitude of T cell responses was observed in SARS-CoV-2 naive PLWH following vaccination, highlighting the relevance of immune cell reconstitution in producing effective immunity to vaccination, especially in people who lack memory responses elicited by natural infection. In this cohort, PLWH were well-controlled on cART and had undetectable HIV viral loads. Both PLWH with, and without, prior SARS-CoV-2 exposure had similar median CD4 T cell counts (602 and 560 cells/mL, respectively) despite different serological outcomes. However, the full impact of HIV-related immunosuppression, in addition to other factors, including age, sex and presence of co-morbidities, in dampening effective and long-lived memory responses needs to be addressed in future larger prospective studies. It is possible that different vaccine schedules, i.e., homologous versus heterologous vaccination, could also account for the observed heterogeneity in cellular immune responses. A heterologous viral vectored/mRNA vaccination has been described to lead to increased reactogenicity, combining the advantages from both vaccine classes. 69 Owing to limited numbers, it has not been possible to address the impact of different vaccine platforms in our cohort. However, previous ll OPEN ACCESS iScience 26, 105862, January 20, 2023 iScience Article work has shown that the adenovirus-based platforms induce a higher T cell response [70][71][72] whereas mRNA vaccine generates a stronger antibody response. [72][73][74] The use of heterologous boosting strategies has been shown to expand the quantity and the breadth of T cell immunity and improve the serological response in PLWH. 70,[75][76][77] Therefore, we speculate that different vaccine platforms and/or heterologous versus homogeneous combinations could lead to different magnitude T cell responses in our cohort. However, there are insufficient data to recommend the best vaccine approach to induce a more effective, resilient, and durable response in PLWH. Instead, optimization of vaccine schedules requires a randomized controlled trial to directly compare the immunogenicity of different vaccine platforms to design the most effective vaccination schedules.
Overall humoral responses correlated with the magnitude of T cell responses and our findings corroborate the importance of T FH cells supporting effective B cell responses after vaccination. Notably, in a small subgroup of patients (serological non-or low-level responders), there were detectable T cell responses characterized by a CXCR3+CD127+ CD8 T phenotype. This phenotype was not clearly related to HIV parameters or presence of co-morbidities. These T cell populations have been linked with increased survival in people infected with SARS-CoV-2 and are consistent with observations in patient groups who lack B cell responses. 53 Upregulation of CXCR3 in vaccine-induced T cells with potential to home to lung mucosa in tuberculosis 78 suggests that these CD8 T cells described herein could play a role in the protection against severe respiratory diseases such as SARS-CoV-2. One possibility is that in the absence of functional antibody responses, the increased abundance of viral antigens could drive CXCR3+ CD8 + T cell proliferation, as these cells have been shown to have an enhanced proliferative capacity 79 and improved effector differentiation. 80,81 These CXCR3+ CD8 + T cells could confer a degree of protection by localization to infected tissue compartments, 79,82 and provide site-specific responses, which are known to be important in protection against respiratory disease. 83 CD8 + T cells have also been shown to expand following vaccination in patients receiving B cell depleting therapies 84 and to contribute to vaccine mediated protection against SARS-CoV-2 in rhesus macaques. 85 However, studies in larger cohorts with breakthrough infections are necessary to clearly evaluate the contribution of these CD8 + T cell populations in vaccine-mediated protection.
Overall, our data support the benefit of a third SARS-CoV-2 dose in inducing nAbs against Omicron in PLWH, as it does in the general population. Moreover, our study provides new insights into the reasons why some PLWH fail to produce effective humoral responses via an in-depth assessment of B cell responses. Specifically, we find that global B cell dysfunction is related to lower serological output in terms of both binding and neutralizing responses. Antibody responses take longer to develop in individuals with greater global B cell dysfunction, which is most commonly seen in PLWH. Although a third SARS-CoV-2 vaccine dose improves neutralization potency and breadth for many, lower titers and MBC disturbance are still observed. Prospective longitudinal studies are now needed to assess whether global B cell disturbance fluctuates in PLWH on cART over time, what treatments/co-morbidities influence this and what level of B cell dysfunction results in inferior clinical outcomes long-term with regards to infectious diseases, particularly where vaccination has taken place. In parallel, CD8 + T cell profiles and anti-viral T cell activity should be monitored in such studies to understand whether these cells do provide the proposed immunological compensation for defects in humoral immunity.

Limitations of the study
Our study has several limitations. These include a cross-sectional analysis, which precludes the establishment of causal relationships. Our cohort is heterogeneous, with differences in sex, age and levels of immunosuppression that may contribute to the variability in the magnitude of responses. Moreover, the current analysis provides an overview of responses after up to three vaccine doses, and therefore further work is required to assess the durability and resilience of these responses against subvariants and additional vaccine doses. In addition, we could not assess the impact of breakthrough infections on humoral and cellular immune response, although some individuals were infected with SARS-CoV-2 after vaccination as noted above, because the numbers of re-infections were too low across any given timepoint for meaningful analysis.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

DECLARATION OF INTERESTS
The authors declare no competing interests. To measure total IgG levels in plasma, a 96-half-well NUNC Maxisorpä plate (Nalgene) was entirely coated overnight at 4 C with 25 mL of goat anti-human F(ab) 0 2 (1:1000). As above, plates were washed in PBS-T and blocked for 1 hat RT in assay buffer. 25 mL of serial dilutions of patient plasma (1:100 to 1:10 7 ) were added in duplicates to the plate alongside known concentrations of IgG in triplicates. As above, after 2 h of incubation at RT, plates were washed with PBS-T and 25 mL AP-conjugated goat anti-human IgG was added and then incubated for 1 hat RT. Plates were washed with PBS-T, and 25 mL of AP substrate added. ODs were measured using a Multiskanä FC plate reader at 405 nm and total IgG titers interpolated from the IgG standard curve using 4 PL regression curve-fitting on GraphPad Prism 9.

IgG purification
As the PLWH participants in this study were on cART which can interfere with the lentivirus-based pseudotype neutralization assay IgG was purified from plasma using a Pierce 96-well protein G spin plate (ThermoFischer Scientific). Plasma was incubated in wells containing protein G at RT for 30 min. The captured IgG was then eluted with 0.1M Glycine (pH = 2-3) twice into 2M Tris (pH = 7.5-9) buffer. To remove Tris/Glycine buffer from the purified IgG, the eluate was concentrated (Thermo Scientific Pierce Protein Concentrator PES, 50K MWCO, 0.5 mL) and washed thrice at 10000 rpm for 10 min before quantification by measuring absorbance of 280 nm on a NanoDrop TM (ThermoFischer, Rockford, Illinois, UK). The entire volume of purified IgG was then filtered sterile using a 0.22 mm PDVF hydrophilic membrane FiltrEXä filter plate (Corning, Corning, NY, USA) and stored at 4 C for further use.

OPEN ACCESS
iScience Article