Altered cellular and humoral immune responses following SARS-CoV-2 mRNA vaccination in patients with multiple sclerosis on anti-CD20 therapy.

SARS-CoV-2 mRNA vaccination in healthy individuals generates effective immune protection against COVID-19. Little is known, however, about the SARS-CoV-2 mRNA vaccine-induced responses in immunosuppressed patients. We investigated induction of antigen-specific antibody, B cell and T cell responses in patients with multiple sclerosis on anti-CD20 (MS-aCD20) monotherapy following SARS-CoV-2 mRNA vaccination. Treatment with aCD20 significantly reduced Spike and RBD specific antibody and memory B cell responses in most patients, an effect that was ameliorated with longer duration from last aCD20 treatment and extent of B cell reconstitution. In contrast, all MS-aCD20 patients generated antigen-specific CD4 and CD8 T-cell responses following vaccination. However, treatment with aCD20 skewed these responses compromising circulating Tfh responses and augmenting CD8 T cell induction, while largely preserving Th1 priming. These data also revealed underlying features of coordinated immune responses following mRNA vaccination. Specifically, the MS-aCD20 patients who failed to generate anti-RBD IgG had the most severe defect in cTfh cell responses and more robust CD8 T cell responses compared to those who generated anti-RBD IgG, whose T cell responses were more similar to healthy controls. These data define the nature of SARS-CoV-2 vaccine-induced immune landscape in aCD20-treated patients, and provide insights into coordinated mRNA vaccine-induced immune responses in humans. Our findings have implications for clinical decision-making, patient education and public health policy for patients treated with aCD20 and other immunosuppressed patients.


Introduction:
Coronavirus disease 19  has caused a global pandemic with profound public health and socioeconomic sequelae due to absence of protective immunity to SARS-CoV-2, the viral infectious cause of COVID-19 1,2 . Vaccine-based strategies were rapidly developed with goals of protecting individuals and achieving herd immunity that limits transmission and subsequent infection 3  in healthy individuals 4,5 . Individuals with underlying autoimmune conditions, including multiple sclerosis (MS), and those on immune-modulatory therapies were not included in these phase 3 clinical trials. As a result, the magnitude and quality of immune response to mRNA vaccination is not well characterized in these potentially vulnerable patient populations that may be at risk for higher COVID-19 associated morbidity and mortality, and more prone to infect others [6][7][8][9][10][11][12] .
Anti-CD20 antibody (aCD20) based B cell depleting strategies are implemented in hematologic malignancies 13 and across a variety of autoimmune disorders 14 , including MS 15,16 , with high rates of success. Upon antigen exposure, B cells have the ability to form memory cells or differentiate into plasmablasts and plasma cells 17 . In addition to their roles as precursors to antibody-secreting cells, B cells can function as professional antigen presenting cells, especially in the context of cognate interactions with T cells that recognize the same antigenic target 18,19 . Depletion of B cells deprives the immune system of these B cell functions to a degree that is dependent on the depth of depletion and the sensitivity of different B cell subsets to aCD20 treatment [20][21][22][23] . As a result, vaccine-specific antibody responses are diminished in patients on aCD20 therapy [23][24][25][26][27][28] .
In the case of SARS-CoV-2 mRNA vaccination, B cell depletion has been shown to result in decreased Spike-specific binding and neutralizing antibodies in patients with chronic inflammatory diseases 29 , including patients with MS 30  and CD8 T cell immunity is generated with T cell responses correlating with better outcomes in some settings [40][41][42] . In addition, robust CD8 T cell responses are associated with improved survival in patients with hematological malignancies and COVID-19, . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint including patients on therapies that deplete B cells 43 . These data provide evidence that T cells may be capable of providing protective immunity and limiting severe disease in settings where antibody responses are lacking. In healthy subjects, antigen-specific CD4 T cells are generated rapidly after the first dose of SARS-CoV-2 mRNA vaccine and these responses provide the foundation for a coordinated immune response following the second vaccine dose 44 . Moreover, T cell responses have been suggested to contribute to vaccine efficacy during the early period after the first vaccine dose [44][45][46] .
In addition, T cells are capable of recognizing mutant SARS-CoV-2 variants 47,48 that can partially escape humoral-based immunity. This T cell recognition of variants is likely based on broader epitope recognition including epitopes unrelated to antibody binding sites where antibody escape mutations occur 49 . Despite these data, the induction of T cell responses by mRNA vaccination in patients on B cell depleting therapies remains poorly understood. Defining the dynamics, magnitude and coordination of CD4 and CD8 T cell responses to SARS-CoV-2 mRNA vaccine in MS patients treated with aCD20 therapy should provide a better understanding of SARS-CoV-2 immunity in these patients and provide insights into the role of B cells in vaccine-induced T cell priming in humans.
In this study, we analyzed a cohort of patients with MS to evaluate the effect of aCD20 therapy on SARS-CoV-2 mRNA vaccine responses. Although most MS patients treated with aCD20 (MS-aCD20) made detectable Spike binding antibodies and 50% made detectable Receptor Binding Domain (RBD) antibodies, antibody titers were lower and delayed compared to healthy control subjects. All MS-aCD20 treated patients . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint developed Spike-specific CD4 T cell responses, though these responses were of lower magnitude compared to healthy controls . In contrast, CD8 T cell responses were more robust in MS-aCD20 patients especially following the second vaccine dose. Finally, comparing the subsets of MS-aCD20 patients who did and did not generate anti-RBD IgG responses revealed major differences in immune response coordination, with substantial reduction in vaccine-induced cTfh responses and reciprocal increases in CD8 T-cell responses in those who failed to generate anti-RBD antibodies. Thus, these data demonstrate the efficient induction of T cell immunity in MS-aCD20 patients following SARS-CoV-2 mRNA vaccination even in the absence of antibody induction.
Moreover, these studies provide insights into the role of B cells and humoral immunity in vaccine-induced T cell responses and shed light on immune mechanisms that accompany aCD20 therapy based on differential responses to vaccination.
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

Impact of aCD20 therapy on SARS-CoV-2 mRNA vaccine-induced antibody responses
Whereas SARS-CoV-2 mRNA-based vaccination has been successful in generating robust B cell and T cell mediated vaccine responses in healthy individuals 44,45,47,48,50,51 , less is known about responses to these vaccines in patients with autoimmune conditions, particularly those receiving immunosuppressive or immune cell-depleting agents 29,52 . To examine the impact of anti-CD20 (aCD20) monoclonal antibody therapy on responses to SARS-CoV-2 mRNA vaccination, we recruited 20 patients with multiple sclerosis (MS) treated with aCD20 (MS-aCD20) and compared their vaccine-induced immune responses to 10 healthy control (HC) subjects within the University of Pennsylvania Health System (Extended Data Table 1). Plasma and peripheral blood mononuclear cell (PBMC) samples were analyzed at 5 timepoints ( Figure 1A): prior to the 1 st vaccine dose (timepoint 1, baseline), 10-12 days following the 1 st vaccine dose (timepoint 2), prior to the 2 nd vaccine dose (timepoint 3), 10-12 days following the 2 nd vaccine dose (timepoint 4) and 25-30 days following the 2 nd vaccine dose (timepoint 5).
We evaluated the kinetics of humoral and cellular vaccine responses from induction to early memory.
Serological responses to SARS-CoV-2 Spike and RBD proteins are evidence of productive immune responses and vaccine efficacy 40,[53][54][55] . To interrogate serological outcomes in this MS-aCD20 cohort, we quantified anti-Spike and anti-RBD IgG following the first and second doses of mRNA vaccination (Figure 1B-C). All HC subjects generated both anti-Spike and anti-RBD IgG following the first dose of mRNA . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint vaccine and the level of antibody increased further after the second dose (Figure 1B-C   and Extended Data Table 2), as reported previously 51 . In contrast, these responses were considerably more variable in the MS-aCD20 patients, with 85% developing detectable anti-Spike IgG and only 50% mounting detectable anti-RBD IgG responses by timepoint 5 (Figure 1B-C and Extended Data Table 2). Among those MS-aCD20 patients with detectable IgG, the magnitude of response was generally lower, and the kinetics of the IgG response were delayed compared to the HC group. These findings extend the prior observations 29,30 that antibody responses to SARS-CoV-2 mRNA vaccine are substantially attenuated in MS patients on aCD20 therapy. Our data indicate that many of these patients can still mount detectable antibody responses, albeit with decreased magnitude, delayed kinetics and considerable heterogeneity across patients.
Since a major reason for the substantially altered antibody responses in MS-aCD20 patients was likely B cell depletion, we considered whether the heterogeneity in antibody responses (Figure 1B and C) was related to the duration between the vaccination and the last aCD20 infusion. There were trends towards increased serologic responses to both Spike (Extended Data Figure 1A) and RBD (Extended Data Figure   1B) as the duration from the last aCD20 infusion increased. To further test this idea, we quantified total CD19 + B cell numbers in the circulation of the MS-aCD20 and HC groups (Extended Data Figure 1C). Although most MS-aCD20 patients had no detectable B cells, small circulating B cell populations were detectable in some patients, and there was a clear relationship between time since last aCD20 infusion and the . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint extent of B cell reconstitution at all timepoints examined ( Figure 1D). MS-aCD20 patients with higher percentages of circulating B cells prior to the vaccine (T1) had more robust anti-Spike and anti-RBD IgG responses at T4 and T5 ( Figure 1E). The small number of MS-aCD20 patients who had circulating B cell frequencies comparable to the healthy control group achieved equivalent antibody titers following vaccination ( Figure   1E), suggesting that B cells repopulating the peripheral pool after aCD20 infusion are functionally competent. Thus, when the circulating B cell pool is repopulated in patients, as seen with increased duration from their last aCD20 administration, vaccine-induced antibody responses can approach levels observed in healthy control subjects.
Moreover, the data above demonstrate that even patients who are lacking or are severely deficient in detectable B cells in their circulation can still mount some antibody responses to mRNA vaccines, although with substantially reduced magnitude.

aCD20 effects on vaccine-induced antigen-specific memory B cells
Antigen-specific memory B cells represent a key feature of long-term immunity 56 . These cells are able to respond rapidly to subsequent infections, generate new antibody secreting cells and initiate new germinal center reactions where antibody can further improve qualitatively through somatic hypermutation and affinity maturation 51,56-59 .
Using biotinylated Spike and RBD B cell probes, we recently investigated the antigenspecific memory B cell responses following SARS-CoV-2 mRNA vaccination in healthy individuals 51 . We employed this same strategy to define the magnitude and kinetics of the memory B cell response generated in MS-aCD20 patients following SARS-CoV-2 mRNA vaccination. While circulating memory B cells specific for both Spike (Extended . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 30, 2021. ; Data Figure 1D and Figure 1F) and RBD (Extended Data Figure 1D and Figure 1G) were readily induced in all HCs, Spike-specific memory B cells were detected in only a subset of MS-aCD20 patients, where their frequencies were also substantially diminished ( Figure 1F) at all timepoints (Extended Data Table 3). Similarly, only a minority of MS-aCD20 patients had detectable RBD-specific memory B cells in the circulation at any timepoint ( Figure 1G and Extended Data Table 3). Finally, in agreement with the data above on antibody responses and total B cells, there was a strong correlation between the ability to detect antigen-specific memory B cells and a longer period of time since the last aCD20 treatment (Figure 1H-I). Thus, induction of antigen-specific memory B cell responses following SARS-CoV-2 mRNA vaccination was compromised in MS-aCD20 patients compared to HCs and this impairment was more pronounced in patients who were immunized in closer proximity to their last aCD20 infusion. There were substantially more patients with detectable antibody responses (88.9%) than patients with detectable circulating memory B cells (30%) to the Spike antigen. While this discrepancy may reflect limits of detection of the antigenspecific memory B cell assay, an alternate explanation is that early repopulation of functional B cells within lymphoid tissues enables productive antibody responses even prior to re-emergence of detectable total B cells or antigen-specific memory B cells in the circulation.

aCD20 impact on vaccine-induced CD4 T cell responses
Whereas the induction of B cell responses and antibody responses is a major goal of vaccination against SARS-CoV-2 given the key role of antibodies in protecting from . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint infection, B cells are also central to the effective priming, differentiation, proliferation and maintenance of T cell responses following immunization or infection. Yet the impact of B-cell depletion due to aCD20 treatment on T cell responses to SARS-CoV-2 mRNA vaccination is unclear. Thus, to interrogate the effect of aCD20 treatment on human vaccine-induced T cell responses, we initially implemented high-dimensional flow cytometric analysis of circulating T cell populations in the MS-aCD20 patients and HCs following SARS-CoV-2 mRNA vaccination. Specifically, we applied an unbiased approach using opt-SNE 60 dimensionality reduction followed by FlowSOM clustering 61 (Extended Data Figure 3). Examining the total CD4 + T cell landscape over time revealed dynamic changes following mRNA vaccination (Extended Data Figure 3A).
The total landscape could be mapped with key markers (Extended Data Figure 3B) and metaclusters corresponding to distinct subpopulations of CD4 T cells (Extended Data Figure 3C-D). We identified a group of small metaclusters (metaclusters 9-14) that expanded following the first vaccine dose in HC subjects and expressed high Ki67, CD38, ICOS and HLA-DR, consistent with induction of activated T cells responding to vaccination. This group of metaclusters showed less dynamic change in the MS-aCD20 group with more subtle induction at T2 and T4. No differences were observed in the abundance of these metaclusters between the MS-aCD20 and HC groups at either T2 or T4 (Extended Data Figure 3E). This high-dimensional single cell cytometry-based analysis revealed global changes in responding CD4 T cell populations in both MS-aCD20 and HC groups. We next wanted to gain deeper insights into the CD4 T cell subpopulations induced by vaccination, and identify potential differences between the MS-aCD20 and HC groups.
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint Human vaccination induces populations of Ki67 + CD38 + CD4 and CD8 T cells ~1-2 weeks post immunization and this activated, proliferating subset, contains antigenspecific T cells [62][63][64][65] . We therefore initially examined the induction and dynamics of Ki67 + CD38 + CD4 T cells (gating strategy in Extended Data Figure 2). Consistent with previous reports 44 , a population of Ki67 + CD38 + CD4 T cells was robustly induced after the first vaccine dose in HCs, peaking at T2 and then returning to baseline (Figure 2A).
In contrast, although MS-aCD20 patients had similar frequencies of activated CD4 T cells at baseline, the induction of Ki67 + CD38 + CD4 T cells following vaccination was reduced compared to the HC subjects at T2 with no increase noted after the second dose and the frequency of activated CD4 T cells in MS-aCD20 patients remained lower than HC through T5 (Figure 2A). We next applied dimensionality reduction and is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; phenotype (CM/EM1 Th1 cells). Metacluster 7 represented a combination of discrete CCR6 + T-betor CXCR3 + T-bet mid CM/EM1 CD4 T cells with high ICOS (CM/EM1 Th17like and Th1-like cells). The dynamic changes in these two metaclusters following vaccination were similar between the MS-aCD20 and HC groups ( Figure 2F). We next sought to understand the response of circulating T follicular helper (cTfh) cells given the  Figure 4B) in MS-aCD20 patients compared to HC. Thus, this analysis identified subpopulations of CD4 T cells that responded similarly to vaccination when comparing MS-aCD20 patients to HC subjects (e.g. subsets of activated Th1 cells) as well as a population of cTfh cells that had similar initial induction in the two cohorts, but poor maintenance at later timepoints in the MS-aCD20 patients.
To better understand whether these vaccine-induced changes in CD4 T cells reflected antigen-specific CD4 T cell responses, we performed peptide-dependent activationinduced marker (AIM) assays. Specifically, we stimulated PBMCs with a peptide megapool containing SARS-CoV-2 Spike epitopes optimized for presentation by MHC-II (CD4-S), as previously described 44,66,67 . AIM + CD4 + T cells were defined by co-. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  Figure 5C). There were no major differences in the distribution of AIM + CD4 T cells among these memory T cell subsets between MS-aCD20 patients and HCs, with the majority of AIM + CD4 T cells mapping to the CM and EM1 subsets ( Figure 4D) in both groups. Similarly, we used CXCR5, CXCR3 and CCR6 ( Figure 3E and Extended Data Figure 5D) to examine CD4 T helper subsets among antigenspecific AIM + CD4 T cells in the MS-aCD20 and HC cohorts. Although the distribution was largely similar between the cohorts, there was a trend towards a lower frequency of cTfh cells among the total AIM + responding CD4 T cells in the MS-aCD20 group . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint ( Figure 3F and Extended Data Figure 5E). Thus, although there were some modest reductions at particular timepoints in the MS-aCD20 patients compared to HCs, these data indicate that MS-aCD20 patients are capable of generating robust antigen-specific CD4 T cell responses to both vaccine doses despite attenuated antibody responses.

aCD20 impact on vaccine-induced CD8 T cell responses
We next examined CD8 T cell responses following vaccination in MS-aCD20 patients and HCs. We first assessed activated Ki67 + CD38 + CD8 T cells (Figure 4) using the approaches outlined above for CD4 T cells (Figure 2). Both doses of vaccination elicited increases in activated CD8 T cells in HCs and MS-aCD20 patients ( Figure 4A).

Activated CD8 T cells moderately expanded after the first vaccine dose in both cohorts,
though the magnitude of increase was more robust for HCs ( Figure 4A), possibly due to higher prevaccination (T1) CD8 T cell activation in the MS-aCD20 group. The MS-aCD20 patients, however, generated a considerably stronger response to the second vaccine dose than the HC group, suggesting more robust induction of CD8 T cell responses in the B cell depleted state. We next applied the metaclustering approach described above for CD4 T cells to interrogate the phenotype of the vaccine-responding activated Ki67 + CD38 + CD8 T cells ( Figure 4B-F and Extended Data Figure 6). The opt-SNE landscape map of activated CD8 T cells revealed differences between MS-aCD20 patients and HCs prior to the vaccine, including an abundance of CD27 + ICOS + CD38 + CD8 T cells largely lacking T-bet in MS-aCD20 patients in contrast to CD27 -T-bet + CD8 T cells in HCs ( Figure 4B). However, the activated CD8 T cell populations in both MS-aCD20 patients and HC subjects reoriented following the two . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
Metaclusters were generated to better define these vaccine-induced changes ( Figure   4D-F and Extended Data Figure 6). Specifically, metacluster 7 and 8 were the main vaccine-responding CD8 T cell populations in both groups following the first vaccine dose ( Figure 4B and Figure 4F). Metacluster 6 was the predominant population enriched after the second vaccine dose in both groups ( Figure 4B and Figure 4F). All three metaclusters were EM1 populations that expressed T-bet and PD-1, although the metaclusters responding to the primary vaccination had higher levels of CXCR3, ICOS and CD38 ( Figure 4E). Thus, these high dimensional cytometry data indicated that vaccine-induced activated CD8 T cell responses were more robust in MS-aCD20 patients compared to HC subjects after the second vaccine dose. Moreover, SARS-CoV-2 vaccination appeared to reorganize the distinct baseline landscapes of CD8 T cell activation in MS-aCD20 and HC subjects, reflecting similar vaccine-induced activation profiles.
We next performed AIM assays on CD8 T cells stimulated with a CD8-E peptide megapool as described 66,67 (Figure 5) to examine antigen-specific CD8 T cell responses. As previously reported 44 , AIM + CD8 T cell responses were detected in a subset of HC subjects following the first vaccine dose, with more individuals responding following the second vaccine dose (Figure 5A-B). A similar pattern was seen in the MS-aCD20 patients. However, following the second vaccine dose (T4), a significantly greater expansion of antigen-specific CD8 T cells was noted in the MS-aCD20 patients compared to HCs, a difference that persisted at T5. This expansion was dominated by . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint EM1 cells (Figure 5C-D) consistent with the observations above using activation markers ( Figure 4F). Of note, both groups had equivalent frequencies of total memory subsets that were largely unchanged during vaccination (Extended Data Figure 7).
Thus, although the overall distribution of memory CD8 T cell subsets was similar, SARS-CoV-2 mRNA vaccination induced a stronger antigen-specific CD8 T cell response in MS-aCD20 patients compared to HCs, in particular following the second dose of vaccine.

Subsets of MS-aCD20 patients with distinct coordination of vaccine-induced immune responses
In previous work, we demonstrated that different features of the adaptive immune response were highly coordinated following SARS-CoV-2 mRNA vaccination 44 . Thus, we next wanted to examine how variation in the extent of B cell depletion and the associated impact on both B cell and humoral immune responses in the MS-aCD20 cohort might impact other features of the vaccine-induced immune response. First, comparing antigen-specific measures across timepoints T2, T4 and T5 revealed a strong correlation between humoral and cTfh responses (Figure 6A-B). This correlation was evident earlier and to a stronger extent in the MS-aCD20 group, possibly due to the larger variance of these coordinated features within the B cell depleted cohort. In contrast, AIM + CD8 T cells showed a strong negative correlation with humoral immune features at T5 in the MS-aCD20 group (Figure 6A and 7C). AIM + Th1 cells were also no longer positively associated with some features of humoral immunity as observed in HCs ( Figure 6A). These findings prompted us to separate the MS-aCD20 patients into . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint those who made a detectable RBD-specific IgG response (RBD Ab+) and those who did not (RBD Ab-) and then to investigate other potential immune differences between these two subgroups of MS-aCD20 patients. Figure 6D shows the opt-SNE projection of Ki67 + CD38 + CD4 T cells for the three groups: HC, MS-aCD20 RBD Ab+ and MS-aCD20 RBD Ab-. The landscape of Ki67 + CD38 + CD4 T cells from RBD Ab+ MS-aCD20 patients was similar to that of HC and both RBD Ab+ MS-aCD20 and HC displayed some overlapping temporal features of change during the course of vaccination. In contrast, the RBD Ab-MS-aCD20 group displayed a distinct opt-SNE projection of Ki67 + CD38 + CD4 T cells that was notable for minimal vaccine-induced changes compared to both other groups. To quantify these differences, we used the Earth Mover's Distance (EMD) metric for all pair-wise comparisons that calculates similarities between the probability distributions within the opt-SNE maps 68,69 . EMD revealed similarity in the overall landscape of activated CD4 T cells between HC and RBD Ab+ MS-aCD20 patients, whereas the RBD Ab-MS-aCD20 group was highly dissimilar to the other two groups at all the timepoints examined (Extended Data Figure 8A). In contrast to activated CD4 T cells, vaccine-induced changes in the activated CD8 T cell compartment after the first dose (T2) were more similar in RBD Ab+ and Ab-MS-aCD20 groups, both of which resembled the HC responses ( Figure 6E). These findings were confirmed with our EMD analysis at T2 (Extended data Figure 8B). Following the second vaccine dose (T4), however, the RBD Ab+ MS-aCD20 group was different from both the HC and RBD-MS-aCD20 groups ( Figure 6E and Extended data Figure 8B) due to the larger presence of metacluster 8 (see Figure 4 for CD8 metacluster annotation). Taken together, these data show that, in the absence of a functional . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. therapy on SARS-CoV-2 mRNA vaccine responses. Therapy with aCD20 is used in many clinical settings including cancer immunotherapy, rheumatology and neurology. In MS, aCD20 treatment is commonly used as monotherapy. This offers the advantage that studying this patient population is relatively less confounded by other concurrent immune therapies.
Examining vaccine-induced immune responses in MS-aCD20 patients led to several key observations. First, MS patients on aCD20 therapy had reduced B cell functional responses that included limited antibody induction to Spike and RBD, as well as poor generation of antigen-specific memory B cells. Some MS-aCD20 patients did generate antibody to Spike and RBD and these responses correlated with detection of B cells in . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint the blood. There were, however, some patients who made antibody responses to Spike in the absence of detectable circulating B cells, perhaps pointing to early repopulation of B cells within the lymphoid tissue that are capable of contributing to serological responses. Thus, although antibody responses were compromised in patients on aCD20 therapy, some antibody generation occured in a subset of patients, perhaps related to time since last treatment with aCD20.
SARS-CoV-2 mRNA vaccines also induce T cell responses 44,70,71 . Although induction of neutralizing antibodies is likely to be important in vaccine-induced protection, precise correlates of immunity remain to be completely defined and recent evidence also points to a role of T cells in mitigating severe disease upon infection [40][41][42][43]72 . Despite poor antibody responses in most MS-aCD20 patients, all of these patients generated robust CD4 and CD8 T cell responses to SARS-CoV-2 mRNA vaccination suggesting that vaccinating such subjects on B cell immunosuppression is likely to provide some measure of immunity to SARS-CoV-2. Despite this robust T cell priming, MS-aCD20 patients had selective defects in maintaining similar frequencies of antigen-specific cTfh cells compared to HCs. The defects in cTfh responses were even more dramatic in MS-aCD20 who did not generate RBD antibody responses. While it is possible that some of these changes could reflect an impact of aCD20 on a subset of CD20 + T cells 73,74 , these data are also consistent with the idea that not only do Tfh cells provide help to B cells 75 , but that germinal center B cells also augment, or are necessary for, optimal and sustained Tfh cell responses 76 . In contrast to the cTfh responses, Th1 cell priming and maintenance were only midly impacted and CD8 T-cell responses were augmented, . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. infusion, arguing that the underlying mechanism for this effect is B cell reconstitution.
Although some MS-aCD20 patients made antibodies to Spike in the absence of detectable circulating B cells, stronger antibody responses to Spike and RBD were associated with detectable B cells in the blood. Thus, assessing re-emergence of peripheral B cells may be a better marker than time since last aCD20 treatment to . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint determine which patients will generate humoral immunity following vaccination. While requiring confirmation in larger cohorts, our results provide insights that may contribute to the development of clinical practice guidelines, including considerations around the most appropriate timing for administering SARS-CoV-2 vaccines or boosters in patients on aCD20 therapies.  77 . An important question to address in the future is the underlying mechanism of this augmented CD8 T cell response. One possibility is that, in the absence of antibody, there is an increased abundance of antigen to drive CD8 T cell activation and proliferation due to lack of clearance by vaccine-induced antibodies.
Alternatively, B cells may play a direct role in attenuating CD8 T cell responses 78,79 . For example, regulatory B cells, possibly primed by the initial dose of mRNA vaccine, may limit CD8 T cell responses following the second dose. In this scenario, distinguishing the role of B cells in fostering effective priming of CD4 T cells from negatively regulating . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint CD8 T cells may be important for optimizing future mRNA vaccination approaches in other settings. A third possibility is through effects of antibody or immune complexes via engagement of the inhibitory Fc receptor FcγRIIB on dendritic cells 80,81 or CD8 T cells 82 .
In the absence of Spike-containing immune complexes in MS-aCD20 patients, CD8 T cell responses may be augmented specifically after the second dose when such FcγRIIB-mediated inhibition could occur in the HC cohort. Future studies will be necessary to determine the contribution of these possible mechanisms. Nevertheless, the augmented CD8 T cell response in MS-aCD20 documented here may have implications for future protective immunity and provide impetus to further interrogate CD8 T cell responses in other immunocompromised populations.
Overall, these studies provide strong evidence of immune priming by SARS-CoV-2 mRNA vaccines in MS-aCD20 patients. Although most of these patients do not generate optimal antibody responses, T cell priming, especially of Th1 and CD8 T cells, remains largely intact. Treatment with aCD20 and B cell deficiency were associated with altered coordination of the immune response, however, and cTfh responses were compromised. Nevertheless, despite the intent of aCD20 treatment to remove B cell mediated immunity in patients with MS, including effects of B cells in presenting antigen to CD4 T cells, these studies reveal variable levels of residual underlying immune functionality in MS-aCD20 patients. The analysis of mRNA vaccine induced immune responses not only to measure immunity to SARS-CoV-2 but also as an "analytical vaccine" offers insights into the underlying immune health and fitness of MS-aCD20 patients. Overall, these data provide key insights about the ability to generate immune . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint responses in immunocompromised populations that will be relevant for clinical guidance in these patients and possible public health recommendations for vulnerable populations.

Study design
In this longitudinal study, healthy controls (HC, n=10) and MS patients treated with anti-CD20 (MS-αCD20, n=20, 19 patients on ocrelizumab and 1 patient on rituximab) were

Cell isolation and cryopreservation
Venous blood was collected in multiple 10ml K2 EDTA tubes (BD Vacutainer, Cat # 366643). Blood was diluted at 1:1 ratio with PBS that contains 2mM EDTA and then . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint slowly transferred to a 50ml tube that contained 15ml Ficol (GE Healthcare, Cat # CA95038-168L). Tubes were then spun at 700g at room temperature with no brake. PBMC layers were collected using a transfer pipet and then washed once with 40ml PBS+EDTA buffer before submitted for cell count. Cells were then resuspended in freezer media (Human AB serum+10% DMSO) and aliquoted into cryopreserved tubes (~20million per tube). PBMC samples were first stored in Mr. Frosty freezing containers at -80°C and then transferred to liquid nitrogen tanks for long term storage.
The tube was then stored upright at room temperature for 30min before centrifugation at 4°C for 10 at 2500g (with swinging bucket rotor). Supernatants were then collected, aliquoted and stored at -80°C until further use.

Detection of SARS-CoV-2-Specific Antibodies
Plasma samples were tested for SARS-CoV-2-specific antibody by enzyme-linked immunosorbent assay (ELISA) as previously described 83  were used for compensation. Up to 1x10 6 PBMCs were acquired per each sample. All antibodies used for high-dimensional FACS analysis can be found in Extended Data Table 4.

Detection of SARS-CoV-2 specific Memory B Cells
Antigen-specific B cells were detected using biotinylated proteins in combination with different streptavidin (SA)-fluorophore conjugates, as previously described 51  Columns 7K MWCO (Thermo Fisher) and protein was quantified with a Pierce BCA Assay (Thermo Fisher). SA-BV711 (BD Bioscience) was used as a decoy probe without biotinylated protein to gate out cells that non-specifically bind streptavidin. Antigen . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint probes for spike, RBD, and HA were prepared individually and mixed together after multimerization with 5uM free D-biotin (Avidity LLC) to minimize potential crossreactivity between probes.

Activation induced marker (AIM) assays
PBMCs were thawed by diluting with 10mL of warm RPMI supplemented with 10% FBS, 2mM L-Glutamine, 100 U/mL Penicillin, and 100 mg/mL Streptomycin (R10) and washed once in R10. Cell counts were obtained with a Countess automated cell counter (Thermo Fisher), and each sample was resuspended in fresh R10 to a density of 5x10 6 cells/mL. For each condition, duplicate wells containing 1x10 6  All data from AIM expression assays were background-subtracted using paired unstimulated control samples. For memory T cell and helper T cell subsets, the AIM+ background frequency of non-naïve T cells was subtracted independently for each subset. AIM + cells were identified from non-naïve T cell populations. AIM + CD4 T cells were defined by dual-expression of CD200 and CD40L. AIM + CD8 T cells were defined by dual-expression of 41BB and intracellular IFN-γ.

High-dimensional data analysis of flow cytometry data
Opt-SNE and FlowSOM analyses were performed using OMIQ (https://app.omiq.ai/). The Opt-SNE parameters were: . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. To group individual samples on the basis of their T cell landscape, pairwise EMD values were calculated on the opt-SNE axes for all HC and MS-aCD20 vaccinees at all timepoints collected using the emdist package in R, as previously described 68,69 .

Statistical analysis
Owing to the heterogeneity of clinical and flow cytometric data, nonparametric tests of association were preferentially used throughout this study unless otherwise specified.
Correlation coefficients between ordered features (including discrete ordinal, continuous scale, or a mixture of the two) were quantified by the Spearman rank correlation . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.     . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint shown. Groups: healthy control (HC), MS-aCD20 with anti-RBD IgG + at any timepoint (MS-aCD20 RBD Ab+) and MS-aCD20 with anti-RBD IgG -(MS-aCD20 RBD Ab-) at all timepoints examined. F) AIM + frequencies of the indicated T cell populations following mRNA vaccination at T4 and T5. Values represent the background-subtracted frequency of AIM + non-naïve T cells above paired baseline frequencies for HC (grey) and MS RBD Ab+ (orange) and MS RBD Ab-(purple) groups. Statistics were calculated using unpaired Wilcoxon test.
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint the summary statistics (median, distribution) of the EMD distances to Healthy RBD Ab+ samples on the activated CD8 T cell opt-SNE maps across all timepoints T1-T5 for the three groups: Healthy RBD IgG+, MS-aCD20 RBD Ab+ and MS-aCD20 RBD Ab-.
Pairwise comparisons of means were done with Wilcoxon test and p values are shown.
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 30, 2021. ; https://doi.org/10.1101/2021.06.23.21259389 doi: medRxiv preprint