Cellular and humoral responses to SARS-CoV-2 vaccination in immunosuppressed patients

Objective SARS-CoV-2 vaccinations have demonstrated vaccine-immunogenicity in healthy volunteers, however, efficacy in immunosuppressed patients is less well characterised. There is an urgent need to address the impact of immunosuppression on vaccine immunogenicity. Methods Serological, T-cell ELISpot, cytokines and immunophenotyping were used to assess vaccine responses (either BNT162b2 mRNA or ChAdOx1 nCoV-19) in double-vaccinated patients receiving immunosuppression for renal transplants or haematological malignancies (n = 13). Immunological responses in immunosuppressed patients (VACC-IS) were compared to immunocompetent vaccinated (VACC-IC, n = 12), unvaccinated (UNVACC, n = 11) and infection-naïve unvaccinated (HC, n = 3) cohorts. Results No significant different differences in T-cell responses were observed between VACC-IS and VACC-IC (92%) to spike-peptide (S) stimulation. UNVACC had the highest T-cell non-responders (n = 3), whereas VACC-IC and VACC-IS both had one T-cell non-responder. No significant differences in humoral responses were observed between VACC-IC and VACC-IS, with 92% (12/13) of VACC-IS patients demonstrating seropositivity. One VACC-IS failed to seroconvert, however had detectable T-cell responses. All VACC-IC participants were seropositive for anti-spike antibodies. VACC-IS and VACC-IC participants elicited strong Th1 cytokine response with immunodominance towards S-peptide. Differences in T-cell immunophenotyping were seen between VACC-IS and VACC-IC, with lower CD8+ activation and T-effector memory phenotype observed in VACC-IS. Conclusion SARS-CoV-2 vaccines are immunogenic in patients receiving immunosuppressive therapy, with responses comparable to vaccinated immunocompetent participants. Lower humoral responses were seen in patients treated with B-cell depleting therapeutics, but with preserved T-cell responses. We suggest further work to correlate both protective immunity and longevity of these responses in both healthy and immunosuppressed patients.


Introduction
In late 2019, identification of a novel coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), was described as the causative pathogen of a pneumonia outbreak, known as coronavirusinduced-disease-19 (COVID-19) [1]. What emerged as a local outbreak in Wuhan, China, rapidly progressed into a global pandemic of acute respiratory syndrome evoking mass morbidity, mortality and significant socio-economic turmoil [2]. Currently, mass vaccination programmes, utilising regulatory-approved vaccines, remains the best way to prevent viral transmission [3,4], severe disease, death [5,6] and overwhelming the already stretched healthcare services.
Currently, four vaccines have been approved by the European Medicines agency [7,8], demonstrating satisfactory safety and immunogenicity. However, these pre-authorisation trials were performed on healthy individuals and excluded immunosuppressed patients as they are poor responders to vaccines [9,10]. Consequently, ambiguity regarding vaccine efficacy in patients on immunosuppression prevails. Furthermore, it has been established immunosuppressed patients, such as kidney transplant recipients, have been considered as clinically vulnerable to SARS-CoV-2 infection [11][12][13]. This was further demonstrated by a prospective study conducted by Baek et al [14], which investigated SARS-CoV-2 infection outcomes in 871 immunosuppressed patients. The study defined patients as immunosuppressed, based on a diagnosis of malignancy, organ transplantation recipients or those receiving immunomodulatory treatments. Findings revealed immunosuppressed patients had significantly higher rate of in-hospital mortality, and increased likelihood of developing severe COVID-19 compared to immunocompetent patients.
Moreover, case reports have illustrated the lack of viral clearance in those with immunodeficient states, resulting in elevated serum cytokine profiles, progression into acute respiratory distress syndrome and succumbing to mortality. [15]. Additionally, defective SARS-CoV-2 clearance, due to impaired immune responses, have been associated with persistent SARS-CoV-2 infection and evolution of SARS-CoV-2 variants [16,17]. Choi et al [16] described an immunosuppressed patient, who experienced SARS-CoV-2 infection for 152 days. During this time, 31 substitutions and three deletions within the SARS-CoV-2 genome were identified. Importantly, twelve spike mutations were detected, of which seven were linked to immune evasion [16,18,19]. Data from these case reports also cautions against the use of convalescent plasma therapy in immunosuppressed patients [17]. It's been postulated administered antibodies may have inadequate support from cytotoxic T-cells, therefore, reducing the capacity of viral clearance, yet driving the evolution of SARS-CoV-2 variants. In view of this, characterising vaccine-induced immune responses, within the immunosuppressed, is crucial for understanding their protective immunity and formulating optimal immunisation regimes.
Both natural infection and SARS-CoV-2 vaccination induce spike protein specific antibodies with neutralising activity [8,20]. Nevertheless, the longevity and duration of such humoral protection is unclear, with several studies demonstrating waning antibody-levels over time [21]. In contrast, several findings have highlighted the role of long-term SARS-CoV-2 T-cell responses [22]. Effective cellular immune responses were attributed to mild-COVID-19 [23], alongside development of robust SARS-CoV-2 specific T-cells which were detected 6-8 months post-infection [24]. Moreover, both mRNA and adenoviral vaccines stimulated potent T-cell mediated responses in study-participants [25,26]. Furthermore, long-term duration of protective T-cell responses were identified against SARS-CoV, whereas no antigen-specific Bmemory cells or antibodies were detected 6 years post-infection [27]. As such, when assessing vaccine immunogenicity, it is critical to assess both humoral and cellular responses. Such evaluation is of greater importance in immunosuppressed cohorts, as there is an urgent need to understand the impact of immunosuppression on the efficacy of SARS-CoV-2 vaccinations.
To address this knowledge gap, we examined the impact of immunosuppressive therapies on SARS-CoV-2 vaccine responses. SARS-CoV-2 vaccine responses were assessed in adult-vaccinated kidney transplant patients, or those with haematological malignancies. Here we provide a detailed description of the cellular and humoral responses, following two doses of either mRNA or adenoviral-vector SARS-CoV-2 vaccines. Unlike current studies examining findings of early post-vaccine period, we define details of their most current response (median time: 115 days post-second dose). Based on our findings, we were able to conclude that these immunosuppressed patients produced an immunological response, to SARS-CoV-2 vaccines, which were comparable to healthy vaccinated participants. Our findings warrant further investigation to determine correlation between such observed responses with protective immunity and longevity, within this clinically vulnerable cohort. Moreover, such studies are imperative for the understanding of cellular responses towards the continual emergence of SARS-CoV-2 variants of concerns.

Study design
This study was approved by the institutional research review board of Portsmouth Hospital University NHS Trust and ethical approval was obtained from a national ethics committee (London-City and East research ethics committee, IRAS: 291009). This study was registered under National Institute of Health Research (NIHR) portfolio (CPMS ID: 48275). The study was conducted in accordance with principles of Good Clinical Practice. All enrolled participants were aged ≥18 or over. Participants were assessed for study eligibility by providing a clinical history. Before enrolment, all participants provided written informed consent.
All recruited participants were convalescent donors, except for healthy controls who were COVID-19 infection-naïve and unvaccinated. All convalescent individuals had prior positive real-time polymerase chain reaction (RT-PCR) results before study enrolment. Participants were stratified into the following cohorts: healthy unvaccinated COVID-19-infection naïve (HC, n = 3), unvaccinated (UVACC, n = 11) and vaccinated immunocompetent healthcare workers (VACC-IC, n = 12) and vaccinated immunosuppressed participants (VACC-IS, n = 13). VACC-IS participants were put forward to the study by their respective clinicians, whereas the remaining participants were enrolled through hospital communications.
Blood samples were collected upon enrolment, which were taken in heparinized, EDTA and SST-collection tubes. Samples were processed within 8 h of venepuncture. SST-collection tubes were centrifuged at 2000g for 10 min for collection of serum. Collected serum was stored at − 20 • C until SARS-CoV-2 serological assays were performed. EDTA tubes were used to perform a full-blood count using the DxH Haematology Analysers (Beckman Coulter). Heparinized tubes were processed for peripheral blood mononuclear cells (PBMCs) collection as described below for ELISpot analysis.

T-cell ELISpot
SARS-CoV-2-specific T-cell responses were identified using the T-Spot Discovery SARS-CoV-2 (Oxford Immunotec) according to manufacturer's instructions. In brief, leucosep tubes (Oxford Immunotec) were used to isolate PBMCs from lithium-heparinised whole blood. A total of 2.6 × 10 6 PBMCs were plated into each individual well of T-spot plate. Each well is coated with one of the four different SARS-CoV-2 structural peptides; Spike (S1) protein, nucleocapsid (NC), membrane (MN) protein, and homology (segments of similar sequences which were eliminated from NC and MN panel). Negative and positive controls (phytohaemagglutinin) were used to control for cellular contamination and functionality, respectively. PBMCs were incubated overnight (37 • C, 5% CO2) for 20 h and IFNϒ-secreting SARS-CoV-2 specific T-cells were detected by using an automated plate reader (Autoimmun Diagnostika-ASK JM). IFNϒ secreting SARS-CoV-2 T-cells were reported as spot forming units (SFU) per well.

Representation of high-dimensional flow cytometry
Flow cytometric t-distributed stochastic neighbor embedding (tSNE) and FlowSOM analysis were performed using Cytobank (https://p remium.cytobank.org). For surface T-cell activation marker expression, analysis was performed using the above outlined markers. CD3 + gated events from individuals within each cohort were collected and concatenated into a single file. Data from 103,311 CD3+ gated events, from all cohorts per peptide, were exported as flow cytometry standard (FCS) files using Kaluza 2.1 software (Beckman Coulter, Miami, FL). Then, data from 4453 CD3+ events, with the following settings: 1,000 iterations, perplexity 30, and theta 0.5, subsampling equal each cohort was used to generate tiSNE analysis. FlowSOM was performed using above outlined markers, which were performed individually with gated CD4 + and CD8 + populations from tiSNE analysis per cohort. The following parameters were used to conduct FlowSOM analysis: number of clusters: 225; number of metaclusters: 15; iterations: 10; and hierarchical consensus clustering method was used.
For T-cell subset, analysis was conducted using above outlined markers. S-peptide stimulated CD4 + and CD8 + gated events from each individual were concatenated into a single file per cohort; VACC-IC CD4: 83,612 events; VACC-IC CD8: 35,721; VACC-IS CD4: 60,932; VACC-IS CD8: 16,397 events. All CD4 and CD8 events, per cohort, were used to conduct tiSNE analysis with aforementioned parameters. For both T-cell activation and T-subset analysis, heat maps were used to report statistical phenotypic changes in marker expression within CD4 and CD8 populations per cohort.

Serological testing
Serum was tested for antibodies to Spike (S) protein using the Binding site Anti-spike IgG/A/M ELISA assay according to manufacturer's instructions. Result outcomes are reported as positive or negative with a threshold index-value of ≥1.0. Samples with optical density greater than top-standard of curve were reported as >4.00 index value.

Cytokine profiling
Cytokine responses (IL-6, TNFα, IL-1β, IL-10) were measured in supernatant derived from PBMC stimulation with SARS-CoV-2 peptides within ELISpot plate. 20 µl of supernatant was collected and stored in − 80 • C until analysis was conducted. IFN-γ was not tested as supernatant was derived from ELISpot plate which captures IFN-γ secretion. Cytokine responses were measured using Multiplex assays as performed by the Clinical Immunology laboratory at Addenbrooke's Hospital, Cambridge.

Statistical analysis
Statistical analysis was conducted using Prism V9.0 (GraphPad Software, San Diego, California, USA). Unless otherwise stated, all data are reported as median with IQR. Where appropriate, Kruskal-Wallis test with Dunn's post-hoc comparison test was performed to assess differences between >2 groups. Two-sided Mann-Whitney was used to assess perform differences between 2 groups. P < 0.05 unless otherwise stated. Other details, if any, for each investigation are provided within relevant figure legends. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Study participant characteristics
A total of 39 participants were recruited into the study and stratified into appropriate cohorts based on their clinical characteristics. In unvaccinated cohort (UNVACC, supplemental table S1), the median age of participants was 37 years (IQR: 31-47), male-to-female ratio was 3:8, with 81.8% of participants from white-British ethnicity. Co-morbidities of hypothyroidism (n = 1), stroke (n = 1) and sleep apnoea (n = 2) were reported in this cohort. The reported time between positive RT-PCR result and study enrolment was 160 days (IQR: 145-165). Nine UNVACC participants (81.8%) were classified as ambulatory mild disease [28], based on reported signs and symptoms during active SARS-CoV-2 infection. Two participants were hospitalised requiring oxygen therapy (non-invasive ventilation) and treated with dexamethasone, which as shown by the RECOVERY trial (NCT: NCT04381936), lowered mortality in hospitalised adult COVID-19 patients.
Twelve participants were stratified as vaccinated immunocompetent (VACC-IC, supplementary table 2) with a median age of 45 years (IQR: 30-53) and male-to-female ratio of 1:5. Five VACC-IC participants reported co-morbidities of depression (n = 2) and mild asthma (n = 3); none of these participants, alongside remainder of VACC-IC cohort, were treated with immunosuppressive therapies. Time reported between positive RT-PCR result and study enrolment was 175 days (IQR: 143-431), with all twelve participants classified as having mild COVID-19 disease, and double-vaccinated with BNT162b2 vaccine. Median time between receiving the second vaccine dose and study enrolment was 112.5 days (IQR: 87.2-153).
As UNVACC, VACC-IC and VACC-IS cohorts comprised of convalescent individuals, NC and MN responses were detected (Fig. 1b, 1c), albeit without any significant differences between all cohorts. Nevertheless, UNVACC and VACC-IS had equal non-responders to NC (n = 3) and MN (n = 4), whilst non-responders against NC (n = 2) and MN (n = 1) were also seen in VACC-IC. Moreover, VACC-IC S-peptide responses were significantly higher compared to NC (p = 0.005) and MN (p = 0.001) by 11.00 and 12.00 SFU, respectively (Fig. 2a). Similar trend was observed in VACC-IS (Fig. 2b), where S-peptide responses were significantly higher by 14.50 SFU compared to both NC (p = 0.005) and MN (p = 0.003). Whilst UNVACC S-peptide responses were higher, there were no significant differences between NC and MN (Fig. 2c). Together, this demonstrated that both vaccination cohorts induced immunodominance towards S-peptide, whereas no significant precedence to either SARS-CoV-2 peptides were seen with natural infection.

Humoral response
Humoral responses were evaluated using total Ig anti-spike ELISA immunoassay. As expected, convalescent unvaccinated and vaccinated cohorts had significantly higher serological responses compared to HC individuals (Fig. 3a, p < 0.0001). UNVACC cohort demonstrated the lowest serological response (median: 2.92 index-value, IQR: 2.28-3.18), with 10/11 participants displaying seropositivity. Seronegative responses were seen in one UNVACC participant, who also displayed absent SARS-CoV-2 T-cell response. All 12 VACC-IC participants were seropositive (median: 4.40 index-value, IQR: 4.15-4.40), with 10/12 participants generating a serological response which was greater than top standard of assay (4.00 index-value). Moreover, whilst humoral responses were higher in both vaccinated cohorts compared to UNVACC; only VACC-IC demonstrated a significantly higher humoral response compared to UNVACC (p = 0.002).

Characterisation of CD4 + and CD8 + T-cell activation marker expression
To assess phenotypic changes in T-cell activation marker expression post-SARS-CoV-2 peptide stimulation, we performed an unsupervised analysis which evaluates the entire complex scenario depicted by CD4 + and CD8 + T-cells. Initially, we conducted a dimensionality reduction analysis, flow cytometric and combined t-distribution stochastic neighbour embedding (tSNE), to acquire a phenotypic landscape of CD45 + CD3 + CD4 + and CD8 + lymphocytes in all cohorts (Fig S1). We then explored CD4 + and CD8 + T-cell panel by unsupervised analysis using FlowSOM [29]. Such analysis conducts multivariate clustering of cells based on self-organised map (SOM) algorithm, enabling cells to be stratified into specific meta-clusters based on HLA-DR and CD38 expression [30]. Heat maps were used to statistically report differences in phenotypic expression between cohorts.
A similar trend was seen in CD8 + subsets, where T n and T e were higher in VACC-IC, whilst T em populations higher in VACC-IS (Fig. 5b). Moreover, both VACC-IC and VACC-IS demonstrated a 15.19% and 10.14% increase in CD8 + T e , respectively, compared to CD4 + T e . These observations were supported with elevated senescent-terminally differentiated CD8 + (CD8 + CD57 + ) levels compared to CD4 + . Overall, with exception of T n , CD4 + and CD8 + T em, displayed highest percentage values across both cohorts. Subsequently, Tem was identified as the predominant memory T-cell subset. Furthermore, no exhausted CD4 + and 8 + T-cells (CD4 + CD57 + PD1 + ) were seen in both cohorts using manual gating strategy.
Subsequently, we used the same unsupervised analysis, as used for activation-markers, where tiSNE analysis identified CD4 + and CD8 + TCS, whose percentages are represented alongside the heatmap; which portrays TCS marker expression in both cohorts (Fig. 5c-d). Immediately, it can be recognised VACC-IS had absent CD4 + T e subset (Fig. 5d), whereas a low percentage (1.55%) were identified for VACC-IC (Fig. 5c). Unlike manual gating strategy, tSNE analysis demonstrated CD4 + T cm were dominant for VACC-IC (19.29%) and VACC-IS (18.39%); which were similar between both cohorts. Whereas, VACC-IS evoked a 7.91% increase in CD4 + T em compared to VACC-IC. Furthermore, as highlighted in Fig. 5d, VACC-IS CD4 + T n exhibited considerably weaker CD27 and CD28 expression, compared to VACC-IC. Furthermore, VACC-IS CD4 + T n , along with T cm , T em and CD8 + T em exhibited weak expression of exhausted T-cells (CD57 PD1, dark green on heatmap); whereas such CD57 PD1 phenotypes were absent in these subsets in VACC-IC. These subtle variations were not detected from use of manual gating strategy.
Both VACC-IS CD8 + T cm and T em were greater by 7.12% and 14.94% (Fig. 5d), respectively, compared to VACC-IC (Fig. 5c). Whereas VACC-IC depicted a 6.50% increase in CD8 + T e compared to VACC-IS. Overall, tSNE analysis demonstrated CD4 + T cm as the dominant memory T-cell subset in both VACC-IC and VACC-IS. Whereas CD8 + T e and T em were the dominant subset in VACC-IC and VACC-IS, respectively, post Speptide stimulation.

Ex vivo production of pro-inflammatory cytokines
Multiplex cytokine analysis (IL-6, TNFα, IL-1β and IL-10) was performed on study cohorts after antigen-specific stimulation in PBMCs with SARS-CoV-2 S, NC, and MN peptides (Fig. 6a-c). We recognise IFNγ secretion as a key cytokine signature in viral infections [31], however as PBMCs were harvested within an IFNγ capture ELISpot plate this cytokine was excluded.

Discussion
Immunisation represents the most effective intervention against infectious diseases, such as SARS-CoV-2; as evident by success of mass global vaccination programmes reducing viral spread and preventing severe disease [32]. Nevertheless, there are very few studies exploring the immunogenicity of COVID-19 vaccines in immunocompromised patients, such as solid-organ transplant recipients (SOTs) and haematological malignancies. Moreover, reduced vaccine-induced immune responses have been associated in SOTs, or in general, in patients on active immunosuppressive therapies [33]. To address this, we explored the immunogenicity of two SARS-CoV-2-19 licenced vaccines (either BNT162b2 mRNA or ChAdOx1 nCoV-19 adenoviral-vector) in doublevaccinated adult renal transplant recipients and those diagnosed with haematological malignancies. Unlike previous studies where vaccine immunogenicity was limited to early post-vaccine period [34], we enrolled immunosuppressed patients with a median time of 115 days post second-dose, thus, providing an up-to-date snapshot of their Fig. 4. Unsupervised analysis of CD4 + and CD8 + T-cells post S-peptide stimulation. A Representation of CD4 + phenotypic landscape, by coupling tSNE dimensionalreductional analysis with FlowSOM which was used to identify specific CD4 + T-cell metaclusters based on HLA-DR and CD38 expression for each cohort. B Heat map representing the different CD4 + metaclusters identified by FlowSOM for each cohort, where the colours in the heatmap represent the median acrsinh ratio for HLA-DR and CD38 expression of each metacluster. Heatmap colours vary from black for lower expression, to yellow for higher expression of each surface marker (HLA-DR, CD38). C The same unsupervised analysis was used to define the CD8 + phenotypic landscape, coupled with FlowSOM, for identification of CD8 + metaclusters between cohorts. D Heat map representing the median arcsinh ratio of HLA-DR and CD38 observed for each cohort. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) immune response to SARS-CoV-2 vaccination. We compared the humoral and cellular responses of this immunocompromised group (VACC-IS) to healthy vaccinated (VACC-IC), unvaccinated (UNVACC) and infection-naïve (HC) cohorts. Our data demonstrates that VACC-IS patients responded to the vaccine by producing comparable cellular and humoral responses to VACC-IC. However, findings from large prospective studies [35] are required to correlate such vaccine-induced response with protective immunity.
Recent reports have highlighted diminished T-cell responses against COVID-19 vaccines in renal transplants patients receiving T-cell directed therapies [34] and in haematological cancer patients [36]. In response to these studies, we examined vaccine-induced SARS-CoV-2 specific Tcell responses in these patients through using an IFNγ release assay. Two samples from UNVACC (n = 1) and VACC-IC (n = 1) were excluded due to laboratory technical error. Individual data points are illustrated as individual scatter plots for each cytokine, expressed as pg/ml, with median (centre bar) and IQR (upper and lower bars). Statistical analyses were determined using nonparametric Kruskal-Wallis test with Dunn's post-hoc test for multiple comparisons. *P < 0.05, **P < 0.01. HC, Healthy infection-naïve; UNVACC, unvaccinated convalescent; VACC-IC, vaccinated immunocompetent; VACC-IS, vaccinated immunosuppressed. compared to NC an MN peptides; findings that are consistent with both vaccine clinical trials [25,26].
Both VACC-IC and VACC-IS had one T-cell non-responder each; however, both these participants demonstrated positive serology for anti-spike SARS-CoV-2 antibodies. Moreover, 3 VACC-IS patients had received B-cell depleting therapy (Rituximab) 2 months following their second SARS-CoV-2 vaccine dose. Whilst 2/3 of these patients, were serologically positive, their anti-spike antibody levels were lower in comparison to those receiving T-cell targeted therapies. Such correlation between diminished vaccine specific humoral responses and B-cell depleting therapies have been reported in prior studies [37]. Nevertheless, all 3 patients receiving B-cell therapy, including the patient who failed to seroconvert, elicited T-cell responses to S-peptide stimulation. These results highlight that B-cell negative patients, due to primary or therapy-induced aetiologies, can still reap benefit from T-cell compartment of vaccination.
A plausible explanation for comparable T-cell responses observed in our data, which were not seen in a study conducted by Prendecki et al [34], could be that our immunosuppressed patients had prior natural infection, subsequently it could represent an augmented response to second-dose vaccine ("third" challenge in these convalescent patients).
In fact, the same study reported a 54% increase in T-cell response in their immunosuppressed patients following second-dose vaccination [34]. Such findings support additional vaccine doses could provide an immunogenic "top-up" in immunosuppressed patients. Going forward, we propose a comparative evaluation of assessing vaccine immunogenicity between convalescent and infection-naïve vaccinated immunosuppressed patients. Findings from these studies could provide evidencebased data for optimal vaccine type and dosing schedule in these patients.
All study participants, except HC-infection naïve, had prior natural infection, where unvaccinated (UNVACC) had the highest T-cell nonresponders (n = 3) to S-peptide stimulation. Interestingly, one UNVACC T-cell non-responder was also seronegative for anti-spike antibodies. This participant represented a house-hold case of COVID-19; with positive real-time polymerase chain-reaction nasopharyngeal result, and no significant medical history.We speculate one of two reasonings; firstly, this could represent a case of natural waning immunity, or, secondly, a false-positive result. We believe the latter is unlikely, as the house-hold contact was tested in our study and had detectable serology and T-cell response. Furthermore, all but one HC participant had no detectable T-cell responses. One HC participant had a weak response of 3 SFU to S-peptide. We favour two hypothetical models which could explain this. Firstly, this result could represent a crossreaction with other six human pathogenic coronaviruses [38]. Secondly, as these were healthcare workers, both occupational and household exposure could evoke very low concentration of SARS-CoV-2, which may be insufficient to elicit a B-cell response but may induce a T-cell response.
Investigating the CD4 + and CD8 + vaccine-induced landscape highlighted key differences between VACC-IC and VACC-IS. Firstly, whilst no significant differences in CD4 + surface activation markers (CD38 and HLA-DR) were observed between VACC-IC and VACC-IS, the abundance of the dominant metacluster population were reduced in VACC-IS. Similarly, reduction of metacluster abundance were identified in VACC-IS CD8 + , however, with notable differences in T-cell activation marker expression. Over 40% of VACC-IC CD8 + metaclusters depicted dual HLA-DR + CD38 + expression with elevated levels of CD8 + T e (CD45RA + CD197 -CD27 -CD28 -) cells post-S-peptide stimulation. Such finding is consistent with prior studies which have highlighted terminal effector T-cells overexpress the activation markers CD38 and HLA-DR [39]. However, the same metaclusters were identified as HLA-DR + CD38 wk in VACC-IS. Moreover, VACC-IS demonstrated a greater increase in CD8 + T em (CD45RA -CD197 -CD27 +/-CD28 -CD57 + ) subsets compared to VACC-IC. Such findings may explain the increased levels of IFNγ secreting SARS-CoV-2 specific T-cells observed in our ELISpot data; as CD8 + T em have shown to secrete the greatest IFNγ levels compared to other T-cell memory subsets [40].
Similar findings have been reported in a recent study investigating vaccine-induced response in multiple sclerosis patients on anti-CD20 therapy [41]. No differences in T-cell activation were seen in CD4 + compartments post-vaccination in both healthy and MS-patient cohorts. However, CD8 + HLA-DR + CD38 wk metaclusters were seen in MS patients, which were predominantly of the T em subset in line with our findings in VACC-IS cohort. The authors concluded such findings in MSpatients are indicative of a robust CD8 + T-cell response compared to healthy controls. However, we hypothesise the lack of CD4 + T-follicular helper cells and vaccine-induced antibodies could have preferentially driven and augmented CD8 + T-cell responses. Whilst these findings are encouraging, we believe extensive deep-immune profiling comprising a broader range of immunosuppressed patients are required to achieve a definitive illustration of vaccine-induced T-cell responses. We propose undertaking activation-induced marker (AIM) assays on CD4 + and CD8 + antigenic specific cells. Such experimental design may provide in-depth information surrounding CD4 + T-cell priming by examining coexpression of CD200 and CD40L. Functional CD8 + T-cell responses can be investigated through examining IFNγ, TNFα, IL-2 and granzyme B expression. Such functionality could be correlated with polyfunctional status of SARS-CoV-2 specific T-cells as it remains unclear whether mono-or polyfunctional T-cells are of greater protective value [25].
Our study has some limitations. We are aware of the small sample size in this study, which makes it challenging to draw firm conclusions. Moreover, demographical risk factors for COVID-19, such as age, BMI and ethnicity [42] were not controlled for. This was mainly due to patient enrolment was restricted only within Portsmouth NHS trust; therefore, limited patients were available for enrolment who satisfied the requirements of being both convalescence and vaccinated. We suggest a multi-site recruitment from a diverse set of clinical specialities, which would enhance sample power and lead to fewer demographic differences between cohorts. Regardless the small sample size, we have demonstrated vaccine responses and immunophenotypic landscape can be successfully derived using the methodologies within this study and provides a blueprint for larger-scale multi-site studies. Secondly, we were only able to re-bleed a small proportion of our VACC-IC and VACC-IS study participants for T-cell subset analysis. Going forward, we propose to extend the T-cell subset panel along with drop-in markers of activation and proliferation (such as Ki-67) to validate these findings. This would provide a more detailed phenotypic landscape of T-cell memory subsets found in vaccinated healthy and immunosuppressed cohorts.
Overall, our data confirms an immunological response to SARS-CoV-2 vaccines in immunosuppressed patients, when assessed by combination of cellular and serological assays. The observed vaccine-induced responses within this immunosuppressed cohort were comparable to healthy vaccinated participants. Furthermore, our data highlights the robust and broad capacities of SARS-CoV-2 specific T-cells. Further work is required to decipher these responses with the continual emergence of global SARS-CoV-2 variants of concerns. Our findings warrant further work correlating the observed immunological responses with protective immunity and evaluate if longevity of these responses is comparable to healthy individuals. Such information may aid development of a standardised immunisation schedule required to optimise the vaccineinduced responses observed in this clinically vulnerable patient group.

Funding
Oxford Immunotec and Beckman Coulter provided the study with free of cost evaluation kits. Both Oxford Immunotec and Beckman Coulter had no other direct involvement in this study. This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.

Institutional Review Board Statement
Study proposal was evaluated and approved by Portsmouth NHS Trust Research Ethics Committee (Study reference: 21/BPR10029) and national Health Research Authority ethics committee (London-City and East research ethics committee, IRAS: 291009).

Data Availability Statement
Data available upon request.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.