Persons with HIV Develop Spike-Specific Lymph Node Germinal Center Responses following SARS-CoV-2 Vaccination

This article is featured in Top Reads, p. 855

The COVID-19 pandemic caused by SARS-CoV-2 has led to >630 million cases and 6.5 million deaths worldwide. Vaccination has been critical in limiting the mortality of COVID-19 (1–3). Areas of the world such as sub-Saharan Africa that have been hit hardest by the HIV pandemic have experienced hurdles in acquiring doses of SARS-CoV-2 vaccines. Indeed, the cold chain required to deploy SARS-CoV-2 mRNA vaccines is not always available in many worldwide locations with high burdens of HIV infections.

Persons with HIV (PWH) are at high risk for severe COVID-19, in part because HIV directly infects and alters the immune system, and because PWH often have associated HIV-unrelated comorbidities (4–9). It is established that PWH on antiretroviral therapy (ART) who have achieved HIV viral suppression and CD4⁺ T cell recovery can continue to have persistent immune dysfunction that, in the case of some vaccines, leads to impaired or short-lived vaccine-induced immune responses (10–14). The cause of these reduced vaccine responses is likely multifactorial. Studies have shown that there is persistent HIV replication and immune dysfunction within lymphoid tissue (15–18), that T follicular helper (T_FH) cells serve as an ongoing reservoir for latent HIV infection (19–21), and that lymph node (LN) architecture and cellularity are disrupted in HIV infection (21–25).

It is critical to understand how PWH respond to SARS-CoV-2 vaccination given their increased risk for severe COVID-19 and their risk for persistent immune dysfunction despite ART. Although some studies have shown that PWH can generate SARS-CoV-2 vaccine–induced immune responses following a two-dose vaccine series that are comparable to persons without HIV (9, 26–32), others have observed that these vaccine responses can be reduced and/or variable in PWH compared with persons without HIV (32–35). Furthermore, a recent study showed that vaccinated PWH had a higher risk of COVID-19 breakthrough infections compared with vaccinated persons without HIV (36). A separate study performed in South Africa showed that the effectiveness of the single-dose Ad26.COV2.S vaccine against COVID-19–related death was reduced in healthcare workers with HIV compared with those without HIV (37).

We have previously shown that SARS-CoV-2 mRNA vaccination of otherwise healthy individuals without HIV infection induces robust and long-lived spike-specific germinal center (GC) B cell responses (38, 39) and T_FH cellular responses (40) in the LN of all 15 subjects who underwent LN fine needle aspiration (FNA) sampling. Sampling the draining LN of PWH following SARS-CoV-2 vaccination is an important question because GCs are necessary for the development of long-lived memory B cells, long-lived plasma cells, and the production of affinity-matured Abs. In this study, we longitudinally sample the blood and LNs of PWH on ART following SARS-CoV-2 vaccination to better understand the magnitude and heterogeneity of vaccine-induced immune responses.

This is a prospective observational cohort study of PWH already on ART with established viral suppression who were receiving SARS-CoV-2 vaccination as standard-of-care treatment at the time of the study. Single-dose Ad26.COV2.S was the primary SARS-CoV-2 vaccine administered as standard-of-care to PWH by the Washington University in St. Louis Infectious Diseases Clinic at the time of the study. All vaccines were administered as directed by the manufacturer, after the collection of prevaccination blood and tissue samples. The two mRNA vaccine recipients had prevaccination tissue samples collected ∼2 wk prior to receiving the first vaccine dose and received the second mRNA vaccine dose according to the manufacturer’s instructions; that is, 21 d after the first dose for the BNT162b2 vaccinee, and 28 d after the first dose for the mRNA-1273 vaccinee. Outside of the Washington University in St. Louis Infectious Diseases Clinic, mRNA vaccines represented most vaccines administered at the time of the study, limiting the availability of a control group who received a single dose of the Ad26.COV2.S vaccine and did not have HIV infection. This study was approved by the Institutional Review Board of Washington University in St. Louis (approval no. 202103088). Written informed consent was obtained from all participants.

Blood samples from study participants were collected into EDTA-anticoagulated tubes and prepared into PBMCs and plasma using Ficoll density gradient purification. Plasma was frozen and stored at −80°C. RBCs contaminating the resultant PBMC preparations were lysed using ACK (ammonium-chloride-potassium) lysis buffer (41), and then PBMCs were washed, counted, and resuspended in 90% FBS with 10% DMSO and stored in liquid nitrogen until analysis.

Ultrasound-guided FNA of axillary LNs was performed as previously described (38, 42). Briefly, dominant lateral axillary LNs on the ipsilateral arm to the vaccination site were localized with ultrasound by a radiologist or a qualified physician’s assistant working under the supervision of a radiologist. Six passes targeting the LN cortex were made with 25G needles, each of which was then flushed with 3 ml of RPMI 1640 containing l-glutamine supplemented with 10% FBS and 100 U/ml penicillin/streptomycin (R10), followed by three 1-ml rinses. RBCs were lysed with ACK lysis buffer, washed twice with PBS supplemented with 2% FBS and 2 mM EDTA (P2), and then cells were counted and cryopreserved in 10% DMSO/90% FBS and stored in liquid nitrogen until analysis.

Assays were performed in MaxiSorp 96-well plates (Thermo Fisher Scientific) coated with 100 µl of recombinant SARS-CoV-2 WA1/2020 spike protein or BSA diluted to 1 µg/ml in 1× PBS and incubated at 4°C overnight. Plates were then blocked with 10% FBS and 0.05% Tween 20 in 1× PBS (blocking buffer). Participant plasma was tested at 1:15 starting dilution followed by seven or more 3-fold serial dilutions using blocking buffer as the diluent.

Plates were incubated for 90 min at 25°C and then washed three times with 1× PBS with 0.05% Tween 20. Secondary Ab (HRP-conjugated goat anti-human IgG (H+L) Ab, Jackson ImmunoResearch) was diluted 1:2500 in blocking buffer before adding it to the wells and incubating for 60 min at 25°C. Plates were washed three times with 1× PBS with 0.05% Tween 20, followed by three washes with 1× PBS before the addition of o-phenylenediamine dihydrochloride peroxidase substrate (MilliporeSigma). Reactions were stopped by the addition of 1 M HCl. OD measurements were taken at 490 nm using a microplate reader (ELx800, BioTek).

Recombinant soluble SARS-CoV-2 spike protein and monobiotinylated Avi-tagged spike protein were generated as previously described (39, 43). A mammalian cell codon-optimized nucleotide sequence coding for the soluble ectodomain of the spike protein of SARS-CoV-2 (GenBank: MN908947.3, aa 1–1213) including a C-terminal thrombin cleavage site, T4 foldon trimerization domain, and hexahistidine tag was cloned into mammalian expression vector pCAGGS. The spike protein sequence was modified to remove the polybasic cleavage site (RRAR to A), and two prefusion stabilizing proline mutations were introduced (K986P and V987P, wild-type numbering). Recombinant spike protein was produced in Expi293F cells (Thermo Fisher Scientific) by transfection with purified DNA using the ExpiFectamine 293 transfection kit (Thermo Fisher Scientific). Supernatants from transfected cells were harvested at 3 d posttransfection, and recombinant proteins were purified using Ni-NTA agarose (Thermo Fisher Scientific), then buffer exchanged into 1× PBS and concentrated using Amicon Ultracel centrifugal filters (EMD Millipore). To biotinylate Avi-tagged spike protein, the spike-AviTag substrate was diluted to 40 µM and incubated for 1 h at 30°C with 15 µg/ml BirA enzyme (Avidity) in 0.05 M bicine buffer at pH 8.3 supplemented with 10 mM ATP, 10 mM MgOAc, and 50 µM biotin. The protein was then concentrated and buffer exchanged with PBS using a 100-kDa Amicon Ultra centrifugal filter (EMD Millipore).

For flow cytometry of PBMCs, recombinant spike protein was labeled with DyLight 405-NHS ester, AF647-NHS ester, or biotinylated using the EZ-Link micro NHS-PEG4-biotinylation kit (Thermo Fisher Scientific); excess DyLight 405 was removed using 40-kDa Zeba desalting columns; excess Alexa Fluor 647 and biotin were removed using 7-kDa Zeba desalting columns (Pierce). For flow cytometry of LNs, spike probes were generated by incubating BirA-biotinylated recombinant spike protein with a 1:1 molar ratio of either BV421- or BV605-conjugated streptavidin (BioLegend) on ice. Of note, the spike protein was added to the streptavidin in three equal additions 15 min apart, after which 1.5 molar excess biotin was added to block excess unoccupied streptavidin sites.

Serial dilutions of sera (starting at 1:20) were incubated with 10² focus-forming units of WA1/2020 D614G, B.1.351 (World Health Organization Beta variant), or B.1.617.2 (World Health Organization Delta variant) for 1 h at 37°C. Ab–virus complexes were added to Vero-TMPRSS2 cell monolayers in 96-well plates and incubated at 37°C for 1 h. Subsequently, cells were overlaid with 1% (w/v) methylcellulose in MEM. Plates were harvested 30 h (WA1/2020 D614G) or 70 h (B.1.351 and B.1.617.2) later by removing overlays and fixed with 4% paraformaldehyde in PBS for 20 min at room temperature. Plates were washed and sequentially incubated with a pool (SARS2-02, -08, -09, -10, -11, -13, -14, -17, -20, -26, -27, -28, -31, -38, -41, -42, -44, -49, -57, -62, -64, -65, -67, and -71) of anti-spike murine Abs (44) (including cross-reactive mAbs to SARS-CoV) and HRP-conjugated goat anti-mouse IgG (Sigma-Aldrich, catalog no. A8924, RRID: AB_258426) in 1× PBS supplemented with 0.1% saponin and 0.1% BSA. SARS-CoV-2–infected cell foci were visualized using TrueBlue peroxidase substrate (KPL) and quantitated on an ImmunoSpot microanalyzer (Cellular Technology).

Frozen PBMC samples were washed twice in R10. Cells were counted on a Cellometer Auto 2000 (Nexcelom), and 1.5 × 10⁶ cells were added to round-bottom 96-well plates and washed in P2.

For PBMC T cell tetramer staining experiments, cells were stained in P2 for 20 min on ice with PE-labeled HLA-DPB1*04:01 spike 167–180 (S_167–180) tetramer and allophycocyanin-labeled HLA-DPB1*04:01 spike 816–830 (S_816–830) tetramer (40, 45, 46). Then, without removing the tetramer, a master mix was added to the cells with pretitrated volumes of the following reagents for 30–45 min on ice: CCR7 BV785 (G043H7, BioLegend), CXCR3 BV650 (G025H7, BioLegend), CD45 AF532 (H130, Invitrogen), CD45RO BV510 (UCHL1, BioLegend), CXCR5 PE Dazzle 594 (J252D4, BioLegend), HLA-DR BV605 (L243, BioLegend), CD27 PE Fire 810 (O323, BioLegend), CD4 PerCP (SK3, BioLegend), PD-1 BB515 (EH12.1, BD Horizon), CD19 BV750 (HIB19, BioLegend), CD3 AF700 (HIT3a, BioLegend), CD8 BV570 (RPA-T8, BioLegend), IgD PE-Cy7 (IA6-2, BioLegend), DyLight 405–conjugated recombinant soluble spike protein, ICOS BV421 (C398.4A, BioLegend), CD20 allophycocyanin Fire 750 (2H7, BioLegend), CD38 BB700 (HIT2, BD Horizon), CD14 PE-Cy5 (M5E2, BioLegend), Alexa Fluor 647–conjugated recombinant soluble spike protein, and Zombie NIR (BioLegend) diluted in Brilliant Staining buffer (50 µl per test, BD Horizon).

For PBMC B cell spike staining experiments, cells were stained in P2 for 30 min on ice with biotinylated and Alexa Fluor 647–conjugated recombinant soluble spike proteins. Following the spike staining, cells were washed twice in P2. Cells were resuspended and stained in a master mix with pretitrated volumes of the following reagents for 30–45 min on ice: IgG BV480 (goat polyclonal, Jackson ImmunoResearch), CD4 BV570 (OKT4, BioLegend), IgD SB702 (IA6-2, Invitrogen), CD45 AF532 (H130, Invitrogen), CXCR5 PE Dazzle (J252D4, BioLegend), CD14 PerCP (HLD14, BioLegend), CD3 allophycocyanin Fire 810 (SK7, BioLegend), CD24 BV421 (ML5, BioLegend), IgM BV605 (MHM-88, BioLegend), CD19 BV750 (HIB19, BioLegend), FcRL5 PE (509f6, BioLegend), CD27 PE Fire 810 (O323, BioLegend), CD71 PE-Cy7 (CY1G4, BioLegend), CD20 allophycocyanin Fire 750 (2H7, BioLegend), streptavidin Qdot 655 (Invitrogen), CD38 BB700 (HIT2, BD Horizon), IgA FITC (M24A, Millipore), CD62L BV785 (DREG-56, BioLegend), CD44 AF700 (C44Mab-5, BioLegend), and Zombie Aqua (BioLegend) diluted in Brilliant Staining buffer (50 µl per test, BD Horizon).

Following the staining, cells for both PBMC T cell and B cell staining experiments were washed twice with P2 and then fixed with 2% paraformaldehyde (Electron Microscopy Sciences) for 20 min on ice. Cells were washed once in P2 and then resuspended in P2 for analysis (see below).

Frozen LN FNA samples were washed twice in R10. The entire FNA sample (1.4 × 10⁵ to 2.0 × 10⁶ cells) was added to round-bottom 96-well plates and washed in P2. For LN staining experiments, cells were stained in P2 for 45–60 min on ice with allophycocyanin-labeled HLA-DPB1*04:01 S_167–180 tetramer, BV421-labeled recombinant spike protein, and BV605-labeled recombinant spike protein. Following tetramer/spike staining, LN cells were washed twice in P2. Cells were resuspended and stained in a master mix with pretitrated volumes of the following reagents for 30 min on ice: IgG BV480 (goat polyclonal, Jackson ImmunoResearch), CD20 Pacific Blue (2H7, BioLegend), CD4 Spark Violet 538 (SK3, BioLegend), CD45RO (UCHL1, BioLegend), CD19 BV750 (HIB19, BioLegend), CCR7 BV785 (G043H7, BioLegend), PD-1 BB515 (EH12.1, BD Horizon), IgA FITC (M24A, Millipore), CD8 AF532 (RPA-T8, BioLegend), CXCR5 PE Dazzle 594 (J252D4, BioLegend), IgD PE-Cy5 (IA6-2, BioLegend), CD14 PerCP (HLD14, BioLegend), CD38 BB700 (HIT2, BD Horizon), CD71 PE-Cy7 (CY1G4, BioLegend), CD27 PE Fire 810 (O323, BioLegend), IgM AF700 (MHM-88, BioLegend), CD3 allophycocyanin Fire 810 (SK7, BioLegend), and Zombie NIR (BioLegend) diluted in Brilliant Staining buffer (50 µl per test, BD Horizon) and P2. Following staining, LN cells were washed twice with P2 and then fixed and permeabilized using a True-Nuclear fixation kit (BioLegend). Briefly, cells were fixed with 1× True-Nuclear fixation buffer for 1 h at 25°C, then washed three times with 1× True-Nuclear permeabilization buffer, and then stained for 1 h at 25°C with pretitrated volumes of FOXP3 Spark NIR 685 (206D, BioLegend), Ki67 BV711 (Ki-67, BioLegend), and Bcl6 PE (L112-91, BioLegend) diluted in 1× True-Nuclear permeabilization buffer. Following staining, LN cells were washed twice with 1× True-Nuclear permeabilization buffer, then resuspended in P2 for analysis (see below).

All PBMC and LN flow cytometry samples were run on an Aurora spectral flow cytometer using SpectroFlo v2.2 software (Cytek). Flow cytometry data were analyzed using FlowJo v10 (Tree Star).

Genomic DNA was extracted from PBMCs using the AllPrep DNA/RNA kit (Qiagen). HLA-DPB1 genotyping was performed on genomic DNA using the reverse sequence–specific oligonucleotide method (RSO2PT, One Lambda) on a LABScan3D instrument (LABSCNXS4, One Lambda). The data were analyzed with Fusion software (v4.6) to determine HLA-DPB1 alleles at G group resolution.

Graphs showing longitudinal data are labeled on the x-axis according to the time since each participant’s completion of his or her respective vaccine series. Time 0 represents completion of the vaccine series (for Ad26.COV2.S this is the date of the single dose, for BNT162b2 [participant 01] this is 3 wk after starting the vaccine series, for mRNA-1273 [participant 12] this is 4 wk after starting the vaccine series).

For T and B cell longitudinal graphical analyses from the blood or LN, the x-axis was set based on a hypothetical limit of detection (LOD). First, for each individual sample from either the blood or LN we calculated the hypothetical frequency of only a single event of interest being collected given the actual total number of parent events collected. In the case of CD4⁺ T cells, this was calculated as one event/total CD4⁺CD3⁺ live single lymphocyte events collected in the sample; in the case of CD19⁺ B cells, this was calculated as one event/total CD19⁺CD3⁻ live single lymphocyte events collected in the sample. We then calculated the mean and population SD for this hypothetical single-event frequency using all available blood samples, and the LOD was set as this mean + 2 SD.

For ELISA experiments, half-maximal binding was calculated using nonlinear regression in Prism v9 (GraphPad Software). The LOD was set at 15 (based on starting serum dilution). For a given sample, when the Prism software could not confidently calculate a half-maximal dilution or a 95% confidence interval for the half-maximal dilution, then that sample was set to the LOD. For the focus reduction neutralization test experiments, half-maximal neutralization was calculated using nonlinear regression in Prism v9 (GraphPad Software). The LOD was set at 20 (based on starting serum dilution). Details regarding the statistical analyses included in the supplemental figures are provided in the legend for the respective figure.

A total of 15 PWH were enrolled. Eleven of 15 participants (8 men, 3 women) were included in the final analysis because 4 participants did not have postvaccination blood or LN samples collected (Table I). The average age of included participants was 44 y (range, 25–62 y). All participants were on ART during the study with an average CD4⁺ T cell count of 669 cells/μl (range, 85–1451 cells/μl). Most participants had been virally suppressed for at least 2 y prior to the study, with notable exceptions including participant 13 and participant 14 (Table I). One participant received the two-dose BNT162b2 vaccine, one participant received the two-dose mRNA-1273 vaccine, and all other participants received the single-dose Ad26.COV2.S vaccine. No participants reported previous SARS-CoV-2 vaccination. Three participants had documented SARS-CoV-2 infection with a positive clinical PCR test prior to vaccination, and another two participants had detectable serum SARS-CoV-2 nucleocapsid Abs prior to vaccination (Table I).

We longitudinally studied the immune responses generated in PWH following SARS-CoV-2 vaccination. We first measured SARS-CoV-2 spike-specific Abs in the blood using ELISA against wild-type WA1/2020 SARS-CoV-2 spike protein. Three of four participants who had evidence of SARS-CoV-2 infection prior to vaccination and had an available prevaccination sample exhibited detectable spike-specific serum Abs prior to vaccination (Fig. 1A). The highest spike-specific Ab titers postvaccination were observed in participants who received two doses of an mRNA SARS-CoV-2 vaccine (01 and 12), and participants with evidence of SARS-CoV-2 infection prior to vaccination (03, 08, 09, 11 and 13) (Fig. 1A). Vaccination boosted spike-specific titers in every participant with a prevaccination blood draw available for comparison. The highest overall spike-specific titer was observed in participant 04 at 3 mo after vaccination, which likely reflected a breakthrough infection as discussed in more detail below.

We also performed SARS-CoV-2 neutralization assays on the participant’s serum against SARS-CoV-2 variants of concern circulating in human populations during the study interval on blood samples from participants who received the Ad26.COV2.S vaccine. We detected neutralizing Abs in some, but not all, study participants following vaccination. Overall, neutralization against WA1/2020 D614G, B.1.351 (Beta) and B.1.617.2 (Delta) was associated with the magnitude of the total spike-specific Ab titer (Fig. 1B).

To characterize the cellular immune responses in blood following SARS-CoV-2 vaccination, we used flow cytometry to identify spike-specific B cells using biotinylated and fluorescently labeled spike protein probes and spike-specific T cells using two distinct HLA-DPB1*04-restricted spike-specific HLA class II tetramers (S_167–180 and S_816–830, Supplemental Fig. 1) (40, 45, 46). We detected spike-specific plasmablasts (CD19⁺IgD⁻CD20⁻CD38⁺) and memory B cells (CD19⁺IgD⁻CD20⁺CD38⁻) in both participants who received an mRNA-based vaccine series (01 and 12). We also detected a high frequency of spike-specific plasmablasts in all participants with evidence of SARS-CoV-2 infection prior to vaccination (03, 08, 09, 11, and 13). There was a range of spike-specific plasmablast and memory B cell responses within the SARS-CoV-2 naive participants (Fig. 2). However, participants 07 and 14, who exhibited the lowest spike-specific Ab titers, did not produce detectable spike-specific B cell responses in blood.

Participant 04 had a sharp increase in serum Ab titer and re-emergence of circulating spike-specific plasmablasts in the blood draw obtained 3 mo after vaccination (Fig. 2A). This re-emergence of Ag-specific plasmablasts, a cell type that typically only circulates in blood 5–10 d after Ag exposure, is highly suggestive of recent breakthrough infection in this participant. Indeed, participant 04 exhibited SARS-CoV-2 nucleocapsid protein binding at the 3 mo time point in a clinical assay but did not exhibit nucleocapsid binding in the tested prevaccination sample (data not shown and Table I). Furthermore, the time frame between this participant’s 1-mo and 3-mo sample collection dates correlated with the peak of the B.1.617.2 (Delta) variant surge in the local region.

We used tetramers to interrogate samples from participants with an HLA-DPB1*04 allele (Table II) for the emergence of CD4⁺ T cells targeting the HLA-DPB1*04-restricted immunodominant S_167–180 epitope as well as the subdominant S_816–830 epitope (40, 45, 46). We identified tetramer-specific CD4⁺ T cell responses in these participants and observed that tetramer-binding CD4⁺ T cells acquired a memory phenotype characterized by increasing CD45RO expression over time following vaccination (Fig. 3A). Consistent with previous observations that the S_167–180 epitope is immunodominant, we observed that S_167–180-specific CD4⁺ T cell responses were at a higher frequency than the corresponding S_816–830-specific responses (Fig. 3B, 3C).

The CD4/CD8 T cell ratio is often used to evaluate the immune status in PWH and to assess for immune recovery (47–49). We did not observe any appreciable relationship between the circulating spike-specific B cell responses or the spike-specific CD4⁺ T cell responses and the peripheral blood CD4/CD8 T cell ratio (Supplemental Fig. 2). Indeed, the two participants who received mRNA vaccines had some of the highest spike-specific Ab titers and highest frequencies of S_167–180-specific T cells, despite having CD4/CD8 T cell ratios <1 (Supplemental Fig. 2).

Eight participants underwent longitudinal LN FNA following vaccination, and six of eight participants provided a prevaccination baseline FNA sample. We used flow cytometry to identify spike-specific B and T cell populations within the draining LN both before and after vaccination (Supplemental Fig. 3).

Given the importance of GCs in the selection of B cells that produce high-affinity Abs, we first assayed for spike-specific GC B cells (CD19⁺IgD⁻Bcl6⁺CD38^int) within the draining LN (Fig. 4A, top two panels). Of the six participants with a prevaccination FNA, only one participant (14) had any positive staining with our spike probe prior to vaccination but was also notable for a very large baseline total GC B cell population in the LN sample (55.9% of CD19⁺ B cells) of unclear etiology.

Following vaccination (Fig. 4a, bottom two panels), both participants who received a two-dose mRNA vaccine series (01 and 12) developed robust spike-specific GCs in the draining LN. Participant 01 had a high-frequency and long-lived spike-specific GC B cell response 3 mo after vaccination (Fig. 4B). However, only one of six participants who received the Ad26.COV2.S vaccine (04) had a detectable increase in spike-specific GC B cells following vaccination (Fig. 4). None of the participants with evidence of SARS-CoV-2 infection prior to vaccination (08, 09, and 13) had spike-specific GC B cells in postvaccination FNA samples, despite a high frequency of circulating spike-specific B cell responses in the blood. To help determine whether the lack of detectable spike-specific GC B cells postvaccination is universally observed following Ad26.COV2.S vaccination, we obtained an FNA sample from an HIV-negative person who received the Ad26.COV2.S vaccine, and we were able to identify a spike-specific GC B cell response in this LN sample postvaccination (Fig. 4A). We further characterized the spike-specific GC B cell response in all LN FNA samples and demonstrated that Ag-specific GC B cells also express CD71, Ki67, and CXCR5 (data not shown). In addition to GC B cells, we also assessed for activated memory B cells (CD19⁺IgD⁻CD20⁺CD38⁻Bcl6⁻Ki67⁺) within the draining LN postvaccination. We identified activated memory B cells only in participants with spike-specific GC B cells (Fig. 5).

T_FH cells (CD4⁺CXCR5⁺Bcl6⁺PD-1⁺) are necessary for the formation and maturation of GC B cell responses (50). T_FH cells in the LN can be directly infected by HIV and indirectly affected by structural changes in the LN follicle caused by HIV infection. Therefore, we measured the size of the T_FH cell population in the LNs of PWH following SARS-CoV-2 vaccination. Consistent with previously published observations in otherwise healthy vaccinees (40), the total T_FH cell population frequency postvaccination in this PWH cohort correlated with the presence of detectable spike-specific GC B cell responses in the LN. Compared to the prevaccination baseline samples, we found larger T_FH cell populations in the two mRNA vaccine recipients as well as the Ad26.COV2.S vaccine recipient who had a spike-specific GC B cell response (Fig. 6). We also observed a large total T_FH cell population in the HIV-negative Ad26.COV2.S vaccine recipient postvaccination (Fig. 6). We observed a large prevaccination T_FH cell population (∼21% of total CD4⁺ T cells) in participant 14 (Fig. 6A, top panel), which suggests a linkage between the nonspecifically enlarged GC B cell population and the T_FH cell population in this individual who did not demonstrate an increase in spike-specific GC B cell staining in the draining LN following vaccination.

Some of the participants who underwent LN FNA carried the HLA-DPB1*04 allele (Table II). Utilizing the HLA-DPB1*04:01 tetramer to the immunodominant S_167–180 epitope, we found that PWH who expressed HLA-DPB1*04 and generated a robust total T_FH cell response and a spike-specific GC B cell response following vaccination (01, 12, and 04) also made an S_167–180-specific T_FH cell response that peaked 1–2 wk after completion of the respective vaccine series (Fig. 6). Similar to postvaccination spike-specific B cell populations in the LN, S_167–180-specific T_FH cells expressed high levels of Ki67 following vaccination (Fig. 7).

Our study examined whether PWH generate Ag-specific GC B cell and T_FH cell responses in the draining LN following SARS-CoV-2 vaccination. Our longitudinal analyses show that PWH can produce spike-specific GCs that are detectable in the draining LN via FNA sampling following vaccination with either a two-dose mRNA-based vaccine series (BNT162b2 or mRNA-1273) or a single-dose adenovirus-based vaccine series (Ad26.COV2.S).

The two participants who received either the BNT162b2 or mRNA-1273 vaccines (01 and 12) had a high frequency of spike-specific GC B cells and spike-specific T_FH cell responses in the draining LN following vaccination. Within the participants who received the Ad26.COV2.S vaccine, only one of six participants with available postvaccination LN samples had detectable vaccine-induced spike-specific GC B cell and spike-specific T_FH cell responses. It is unclear why vaccine-induced Ag-specific GC responses were not detected in the LN FNA samples from the other five Ad26.COV2.S vaccine recipients, and this study’s PWH cohort was too small, too diverse, and lacked a complete HIV-negative control group for us to draw any conclusions regarding causation.

One potential explanation is limited sampling of the entire LN follicular zone during the FNA procedure. Although sampling issues can never be excluded, in our previous work exploring the GC response following BNT162b2 vaccination of healthy individuals, 15 of 15 sampled subjects had detectable spike-specific GCs in FNA samples following vaccination (38–40), which is consistent with our current detection of Ag-specific GC responses in both mRNA-vaccinated PWH included in this study, thus making sampling error an unlikely explanation here.

Another possible explanation for the poor vaccine-induced LN immune responses we observed within the Ad26.COV2.S vaccine recipients is the overall modest immunogenicity of the single-dose Ad26.COV2.S vaccine, which has lower efficacy than the BNT162b2 and mRNA-1273 two-dose mRNA-based vaccine series (51–53). Reduced vaccine immunogenicity may affect our ability to detect GC responses. Previously, we identified SARS-CoV-2–specific GC responses in 15 of 15 BNT162b2 vaccine recipients, but we were only able to identify influenza-specific GC responses in 3 of 7 participants who received the less efficacious 2018–2019 seasonal influenza vaccine (38–40, 42, 54). However, due to the small cohort size of the current study, we cannot make any conclusions regarding the effectiveness of a single-dose Ad26.COV2.S vaccine series in PWH. Similarly, this study’s small cohort prevents any comparison between the effectiveness of a single-dose Ad26.COV2.S vaccine series and a two-dose mRNA-based vaccine series in PWH.

The participant’s underlying HIV infection could also be contributing to the lack of a Ad26.COV2.S-induced LN immune response. Indeed, previous studies have suggested that HIV may contribute to reduced SARS-CoV-2 vaccine–induced immune responses (32–37). However, there are limitations with the current study that prevent an in-depth investigation into how underlying HIV may affect these Ag-specific vaccine responses. First, this study lacked an HIV-negative Ad26.COV2.S vaccine control group. Most otherwise healthy individuals in the local population at the time of the study were receiving mRNA vaccines, limiting our recruitment of this unique population. We did include a single HIV-negative participant who received an Ad26.COV2.S vaccine in our study, and this individual did make a spike-specific GC response in the draining LN following vaccination. Nevertheless, we cannot make any conclusions from a single individual regarding the role of underlying HIV infection on vaccine-induced LN vaccine responses. Second, there was heterogeneity within this PWH cohort regarding HIV status. For example, some participants had CD4⁺ T cell counts <250 cells/μl, some participants had recent transient viral load blips, and one participant had a relatively new HIV diagnosis. Although we did not observe any clear relationship between the CD4/CD8 T cell ratio and vaccine-induced immune responses, this cohort was not powered to address how HIV status may affect the vaccine response.

A history of SARS-CoV-2 infection prior to vaccination may affect vaccine-induced LN immune responses, regardless of HIV status. The three participants who had both longitudinal LN sampling and evidence of SARS-CoV-2 infection prior to vaccination (08, 09, and 13) did not generate detectable Ag-specific LN immune responses following vaccination. However, these participants demonstrated the highest frequencies of circulating spike-specific plasmablasts and high spike-specific Ab titers, which suggests that the vaccine did induce a peripheral Ag-specific immune response. Previous studies have shown that persons with a history of SARS-CoV-2 infection prior to vaccination have increased Ag-specific peripheral immune responses compared with naive persons after vaccination (55, 56). However, these studies have not examined the postvaccine Ag-specific LN immune responses. Our data raise the possibility that a previous SARS-CoV-2 infection may boost the vaccine-specific peripheral immune response in the absence of a vaccine-specific immune response in the draining LN, perhaps through the presence of extrafollicular memory B cells. Although preliminary, this observation warrants additional study.

The other two participants (07 and 14) in whom we did not detect vaccine-induced immune responses in LN FNA samples following Ad26.COV2.S vaccination did not have evidence of SARS-CoV-2 infection prior to vaccination. Participant 07 had the lowest spike-specific Ab titer of the entire cohort and did not exhibit detectable SARS-CoV-2 neutralization above the LOD, suggesting a weak immune response to the vaccine in this individual. Participant 14 had detectable Ab and neutralizing titers to the vaccine and some degree of a cellular spike-specific B cell response in blood, but these responses were lower than those observed in the participants with evidence of SARS-CoV-2 infection prior to vaccination.

Participant 14 also demonstrated high-frequency GC B cell and T_FH cell populations in the draining LN prior to vaccination that modestly decreased following vaccination, which did not fit the typical pattern we previously observed where the GC B cell and T_FH cell populations both increase following vaccination (38–40, 42). The etiology of these cell populations is not clear; there was no documented history of infection other than HIV infection or hematologic malignancies prior to or within the time period surrounding this study. The only other vaccine that this participant received was an influenza vaccine ∼12 mo prior to vaccination. It is notable that this participant had transient HIV viremia ∼20 mo prior to vaccination. Previous studies have shown that HIV replicates in the LN and can persistently replicate in this location even after ART initiation and viral suppression in the peripheral blood (15, 16, 57). HIV can also lead to LN follicular hyperplasia with distorted GC architecture in some individuals (22–25). Indeed, PWH who are not on ART can have higher frequencies of T_FH cells and GC B cells within their LNs, and this increase can remain even after initiation of ART (17). Because participant 14 did not clearly develop a detectable Ag-specific GC response postvaccination, we speculate that dysregulated LN T_FH cell and GC B cell responses associated with underlying HIV infection might have contributed to the poor or absent spike-specific GC response in this individual.

In summary, we demonstrate that PWH can mount spike-specific GC and T_FH cell responses in the draining LN following SARS-CoV-2 vaccination. Our data demonstrating the inconsistent generation of a detectable Ag-specific GC in PWH vaccinated with a single dose of the Ad26.COV2.S vaccine support other studies (36, 37) suggesting that single-dose Ad26.COV2.S-based vaccine series may not be the optimal option for PWH. However, the small and diverse nature of the present cohort makes these conclusions preliminary and restricts our ability to extrapolate our results to all PWH.

The Ellebedy laboratory and Infectious Disease Clinical Research Unit received funding under sponsored research agreements from Moderna unrelated to the data presented in the current study. The Ellebedy laboratory received funding from Emergent BioSolutions and Abbvie that are unrelated to the data presented in the current study. A.H.E. has received consulting and speaking fees from InBios International, Fimbrion Therapeutics, RGAX, Mubadala Investment Company, Moderna, Pfizer, GSK, Danaher, Third Rock Ventures, Goldman Sachs, and Morgan Stanley. A.H.E. is the founder of ImmuneBio Consulting. W.B.A., A.H.E., and J.S.T. are recipients of a licensing agreement with Abbvie that is unrelated to the data presented in the current study. M.S.D. is a consultant for InBios International, Vir Biotechnology, Senda Biosciences, Moderna, and Immunome. The Diamond laboratory has received unrelated funding support in sponsored research agreements from Vir Biotechnology, Emergent BioSolutions, and Moderna. The other authors have no financial conflicts of interest.

We thank all of the participants who enrolled in this study for their dedication to the intensive study protocol. HLA class II tetramers were provided by the Washington University in St. Louis Bursky Center for Human Immunology and Immunotherapy Programs Immunomonitoring Laboratory tetramer core facility with the assistance of Dr. Likui Yang. We also thank Christopher W. Farnsworth and Brittany Roemmich for supervising and performing the nucleocapsid protein binding assay.