B-Cell Subset Representation Predicts SARS-CoV-2 Vaccine Response in Solid Organ Transplant Recipients

Abstract

Background

Solid organ transplant recipients (SOTRs) suffer increased morbidity and mortality due, in part, to chronic immunosuppression. The determination of an individual's immune competence is currently difficult but would improve risk assessment and inform medical decisions. We reasoned that correlating qualitative and quantitative measures of the B-cell compartment with serologic responses to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccination would reveal novel B-cell–based predictors of immune competence.

Methods

We performed an integrated analysis of B-cell phenotypes, serology, and antibody repertoires in heart, lung, liver, kidney, and multiorgan transplant recipients and healthcare worker (HCW) controls (62 individuals total). We utilized K-means clustering and correlation analyses to identify B-cell features that correlated with vaccine serology.

Results

K-means clustering identified 3 distinct B-cell compartment–based groups in SOTRs, which correlated with serum responses to SARS-CoV-2 vaccination. Group 1 SOTRs had a naive-dominant circulating B-cell pool and serologic responses closest to HCWs. Group 2 SOTRs had reduced naive but hyperexpanded memory B cells (MBCs) and variable vaccine responses that segregated by immunosuppression. Group 3 SOTRs had lymphopenia across B-cell subsets and poor serologic responses. Antibody repertoire analysis showed reduced clonal diversity across SOTRs, regardless of MBC numbers. Even in SOTRs with the largest immune responses, vaccine-specific B cells showed evidence of reduced maturation and clonal diversity.

Conclusions

These findings reveal a hierarchy of B-cell impairment in SOTRs that can be measured rapidly, with implications for immune monitoring and intervention in immunocompromised individuals.

Graphical Abstract

Graphical Abstract.

This graphical abstract is also available at Tidbit: https://tidbitapp.io/tidbits/b-cell-subset-representation-predicts-vaccine-response-in-solid-organ-transplant-recipients?utm_campaign=tidbitlinkshare%26utm_source=IO

The authors investigated humoral measures of immune competence in diverse solid organ transplant recipients (SOTRs) and identified 3 distinct circulating B-cell compartment types, discernable by flow cytometry, that correlate with SARS-CoV-2 vaccination responsiveness in previously uninfected SOTRs.

Most solid organ transplant recipients (SOTRs) require sustained immunosuppressive therapy (IST) to prevent transplant rejection. Unfortunately, chronic IST also predisposes to an increased risk of infections, lymphoproliferative disorders, and malignancy, significantly impacting quality of life. Knowledge of an individual's immune competence, and therefore infection risk, would inform both patient lifestyle choices and physicians’ data-driven decision-making regarding vaccination, preventive therapies, and IST. However, clear correlates of immune competence remain enigmatic, and the mechanisms underlying humoral immunodeficiency across clinically diverse SOTRs are poorly defined.

Immune competence evaluation in SOTRs is complicated by variability in transplant type, IST regimen, comorbidities, and overall health. The generation of vaccine-induced antibodies represents the gold standard test to determine immune competence, but it has several limitations. Vaccination risks stimulating allogenic responses and is thus a potential safety concern, although non-live vaccines are generally considered safe in SOTRs [1]. Vaccine responses also require significant time, as antibody titers typically peak around 4 weeks postvaccination in healthy individuals and are further delayed in SOTRs, who often require multiple doses to generate a detectable response [2]. Finally, the relationship between vaccine response and general immune competence can be difficult to interpret. Since adults typically exhibit memory responses to commonly offered vaccines, preexisting antibodies and the potential for immune imprinting to limit new responses to antigenically similar pathogens lead to confounding effects [3–5].

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic provided an opportunity to evaluate immune responses to a new antigen. We reasoned that correlating measures of an individual's immune landscape with SARS-CoV-2 vaccine–specific immunoglobulin G (IgG) antibody responses could reveal novel cell-based predictors of immune competence, while avoiding confounding effects of immune memory. We focused on circulating B cells since their representation is remarkably stable over time [6], they are less impacted by IST than T cells, and they produce antibodies, the common correlates of protection. Prior studies supported our rationale by demonstrating a range of impaired SARS-CoV-2 vaccine–specific B-cell immunity [7, 8] and associating vaccine responsiveness with frequencies of transitional or switched memory B cells (MBCs) in kidney transplant recipients [9, 10]. In contrast, we performed an integrated analysis of humoral immunity across a broad spectrum of transplant types.

We recruited a clinically diverse cohort of SOTRs and healthcare worker (HCW) controls and analyzed B-cell immunophenotypes, immune repertoires, and serology to generate a multimodal view of each individual's humoral immune system. We identified 3 B-cell–based groups that revealed different classes of humoral immunity in SOTRs and correlated with SARS-CoV-2 vaccine responses. This study provides insight into how SOTRs’ humoral immunity is perturbed and suggests that B-cell classification may be useful for risk stratification and immunocompetence evaluation.

MATERIALS AND METHODS

Study Approval

This work was conducted under institutional review board–approved protocols (SOTRs: 814870; HCWs: 843812) at the University of Pennsylvania Perelman School of Medicine, in accordance with the Helsinki Declaration ethical standards. All participants provided written informed consent.

Human Subjects

Kidney, liver, heart, lung, and multiorgan transplant recipients at ≥18 years of age with a functional graft >1 month posttransplant and planned administration of SARS-CoV-2 messenger RNA (mRNA) vaccines were enrolled (n = 45; Supplementary Table 1). SOTRs and HCWs (n = 17) had venipuncture performed at points before and following vaccination. By vaccination, all but 4 SOTRs were >6 months posttransplant (mean, 6.3 years [range, 0.25–21 years]). Transplant IST protocols from the Hospital of the University of Pennsylvania are provided in Supplementary Table 2.

SOTR exclusion criteria included rejection <3 months before vaccination; prior SARS-CoV-2 immunization or infection; enrollment in a coronavirus disease 2019 treatment/prevention study <28 days before study entry; receipt of blood/plasma products or intravenous immunoglobulin <60 days before vaccination; history of serious allergy to prior immunization; human immunodeficiency virus infection; active cytomegalovirus, BK polyomavirus, or Epstein-Barr virus infection; or malignancy (except limited skin cancer). B-cell subset data, repertoire sequencing, and clustering analysis are derived from either prevaccination or 4 weeks post–dose 2 blood draws, as B-cell subset representation was similar between time points (data not shown). Patients were monitored for SARS-CoV-2 infection symptoms and tested as necessary via polymerase chain reaction (PCR) throughout the study period, and samples after SARS-CoV-2 infection were excluded.

Sample Processing

SOTRs had blood drawn prevaccination and 4 weeks following dose 2. Samples from SOTRs who received a third dose and had a planned blood draw were included to assess the vaccine-specific antibody repertoire. Peripheral blood mononuclear cells (PBMCs) and plasma were processed and stored using standard procedures in the Human Immunology Core Facility at the University of Pennsylvania. For B-cell subset analysis and sorting, frozen PBMCs were thawed and washed in RPMI medium, filtered (70 μm), and rested in R10 media (RPMI, 10% fetal bovine serum [FBS], 1% penicillin/streptomycin, 1% L-glutamine) at 37°C for 2 hours before staining.

SARS-CoV-2 Enzyme-Linked Immunosorbent Assay

Plasma was tested at 1:100 dilution for IgG binding to the receptor-binding domain (RBD) by enzyme-linked immunosorbent assay, as described previously [11].

B-Cell Immunophenotyping, Sorting, and Analysis

PBMCs were prepared for flow cytometry and sorting by washing in phosphate-buffered saline (PBS), staining with the live/dead fixable Aqua Dead Cell Stain Kit (Invitrogen, catalog number L34957) for 15 minutes in the dark at room temperature (RT), and staining with a surface antibody cocktail in fluorescent activated cell sorting (FACS) buffer (PBS, 5% FBS, 0.5 mM ethylenediaminetetraacetic acid, 0.1% sodium azide) for 25 minutes in the dark at RT. Cells were washed with FACS buffer, resuspended in PBS with 1% paraformaldehyde (flow cytometry) or FACS buffer (sorting), and stored at 4°C in the dark until analysis. An S1 protein probe (Sino Biological, catalog number 40150-V08B1-B) conjugated to BV421 or AF647 was used to identify dual-staining SARS-CoV-2 spike–specific B cells, similar to previous studies [12]. Naive, total isotype-switched (immunoglobulin D negative and immunoglobulin M negative, IgD-IgM⁻) memory and switched S1⁺ memory populations were sorted and collected separately. Flow cytometry was performed on a BD LSR II (BD Biosciences) and sorting on a FACS Aria II (BD Biosciences) in the Penn Cytomics and Cell Sorting Shared Resource facility. All flow cytometry data were analyzed in FlowJo (BD Biosciences). B-cell subset frequencies for each individual were compiled, normalized by subset as z-scores, and clustered into 3 groups via K-means clustering using Morpheus (https://software.broadinstitute.org/morpheus/). The K-means calculation metric was 1 minus cosine similarity with default settings. Antibodies were purchased from BioLegend: IgM (clone: MHM-88; catalog number 314523), CD27 (O323; 302827), CD19 (HIB19; 302239), CD3 (UCHT1; 300425), CD21 (Bu32; 354911), CD38 (HIT2, 303528); from BD Biosciences: IgD (IA6–2; 562540), IgG (G18–145; 561296), CD14 (MΦP9, 557831), CD16 (3G8, 561726); from Thermo-Fisher: immunoglobulin A (IgA) (polyclonal; AHI0108), CD10 (eBioCB-CALLA, 15-0106-42), CD11c (3.9; 35-0116-42).

Antibody Heavy Chain Gene Rearrangement Sequencing

DNA was extracted from sorted cells using a Gentra Puregene Cell kit (Qiagen, catalog number 158767). Immunoglobulin heavy chain variable gene (VH) family–specific PCRs were performed using FR1 and JH primers [13, 14]. Two biological replicates were run on all sequenced samples except S1-binding MBCs (low cell numbers). Sequencing was performed using an Illumina 2× 300-bp paired-end kit (Illumina MiSeq Reagent Kit v3, 600-cycle, Illumina MS-102-3003).

Immunoglobulin H Sequence Analysis

Sequence reads were filtered, annotated, and grouped into clones as described previously [14]. pRESTO v0.6.0 [15] was used to align paired-end reads, remove short and low-quality reads, and mask low-quality bases. Quality-filtered sequences were aligned, annotated with IgBLAST v1.17.0 [16], and imported into ImmuneDB v0.29.10 [17, 18]. For clonal inference, each donor's sequences with the same IGHV gene, IGHJ gene, complementarity-determining region 3 (CDR3) length, and ≥85% CDR3 amino acid sequence identity were grouped as clones. Clones with 1 sequence copy at the subject level were filtered out and immunoglobulin H rearrangements were analyzed separately for CDR3 length, VH gene usage, somatic hypermutation (SHM), and clonal diversity measures. For S1-binding clonal lineages, data were further filtered on a minimum of 10 sequence copies per node to minimize effects of sequencing error.

Data Availability

Sequencing data are available on GenBank Sequence Read Archive under project number PRJNA1193547 and processed data will be made available via the Adaptive Immune Receptor Repertoire Data Commons upon publication.

Statistical Analysis

Graphs and statistics were generated with Prism software (GraphPad). Two-sided nonparametric statistical tests were used in all cases, including Mann-Whitney U test, Kruskal-Wallis test with multiple comparisons, Wilcoxon test, and Spearman correlation. P ≤.05 was considered significant. The mean is shown in all graphs.

RESULTS

Study Overview and Cohort

The study aim was to define B-cell correlates of serologic responses in SOTRs. We analyzed orthogonal measures including cell phenotype, subset, B-cell receptor repertoire, and antibody production to empirically define B-cell compartment types. To capture clinical diversity, we recruited SOTRs (n = 45) with different organ transplants and ISTs and compared them to immunocompetent HCW controls (n = 17). Demographic characteristics of SOTR and HCW cohorts are presented in Table 1. The number of individuals per experiment varied (Supplementary Table 3).

Table 1.

Baseline Demographic Characteristics

Characteristic	SOTRs (n = 45)	HCWs (n = 17a)
Age, y, mean (range)	57 (25–77)	45 (24–61)
Sex
Female	21 (47)	12 (75)
Male	24 (53)	4 (25)
Race
White	23 (51)	11 (69)
African American	20 (44)	5 (31)
Asian	2 (4)	…
Organ transplanted
Heart	5 (11)	NA
Kidney	22 (49)
Liver	9 (20)
Lung	6 (13)
Liver/Kidney	1 (2)
Kidney/Pancreas	1 (2)
Heart/Kidney	1 (2)
Time, transplant to first dose
<6 mo	4 (9)	NA
>6 mo	41 (91)
Comorbidity		NA
Diabetes	18 (40)
Chronic kidney disease	29 (64)
Hypertension	3 (7)
Rheumatoid arthritis	2 (4)
Psoriasis	2 (4)
IST regimen		NA
Myco, Pred, Tacro	19 (42)
Myco, Pred, Bela	1 (2)
Myco, Pred, Cyclo	1 (2)
Myco, Tacro	5 (11)
Aza, Pred, Tacro	3 (7)
Aza, Tacro	2 (4)
Pred, Tacro	6 (13)
Siro, Tacro	1 (2)
Tacro	7 (16)b

Data are presented as No. (%) unless otherwise indicated.

Abbreviations: Aza, azathioprine; Bela, belatacept; Cyclo, cyclosporine; HCW, healthcare worker; IST, immunosuppressive therapy; Myco, mycophenolate mofetil; NA, not applicable; Pred, prednisone; Siro, sirolimus; SOTR, solid organ transplant recipient; Tacro, tacrolimus.

^aDemographic information was only available for 16 of 17 HCWs (not HCW27). See Supplementary Table 1 for information by donor or Supplementary Table 5 for information by B-cell group.

^bOne individual on Tacro was also taking daratumumab.

Preimmune B-Cell Deficiency in SOTRs

We analyzed PBMC B-cell subsets via flow cytometry. SOTRs had marked B-cell lymphopenia compared to HCWs (Figure 1A), similar to previous findings [9]. Within each transplant and IST group, B-cell frequencies were variable and reduced compared to HCWs (Supplementary Figure 1A). To determine if specific B-cell subsets were altered, we identified the major preimmune and antigen-experienced populations and calculated the frequency of each as a fraction of total live PBMCs (Figure 1B). Transitional and naive B cells were significantly decreased in SOTRs (Figure 1C). Both lung SOTRs and individuals taking tacrolimus with either azathioprine or prednisone tended to have lower frequencies of these subsets, but there was substantial variability within most transplant and IST groups (Supplementary Figure 1B and 1C). Since novel immune responses are derived from naive B cells, which develop from transitional B cells, these findings suggest insufficient precursor cell numbers limit humoral immunity in SOTRs.

Figure 1.

Peripheral blood B-cell frequencies and B-cell receptor repertoire diversity in solid organ transplant recipients (SOTRs) and healthcare workers (HCWs). A, Frequency of peripheral blood B cells in HCWs (n = 11) and SOTRs (n = 45). The SOTR group in A and C includes heart (n = 5), lung (n = 6), liver (n = 9), kidney (n = 22), and multiple transplants (n = 3). B, B-cell subset gating scheme from a representative HCW. Subsets are defined as follows: transitional, IgM⁺IgD^+/−CD38⁺CD10⁺; naive, IgD⁺IgM^+/−CD38^lowCD10⁻CD27⁻CD21⁺; IgM⁺ memory, IgM⁺IgD^+/−CD38^lowCD10⁻CD27⁺; IgG⁺ memory, CD38^lowIgD⁻IgM⁻IgG⁺; IgA⁺ memory, CD38^lowIgD⁻IgM⁻IgA⁺; plasmablast (PB), CD38⁺⁺CD27⁺; and CD38^lowIgD⁻IgM⁻ memory subsets including CD21⁻CD11c⁻, CD21⁻CD11c⁺, CD21⁺CD10⁺, CD21⁺CD10⁻CD27⁺, and CD21⁺CD10⁻CD27⁻. C, B-cell subset frequencies within total peripheral blood mononuclear cells. D, Jaccard index and cosine similarity clonal overlap analyses between independently amplified immunoglobulin VH gene sequencing replicates from sorted naive and switched memory B cells (MBCs) in HCWs and SOTRs. Clones were subsampled to 1500 for overlap analysis. HCW naive, n = 5; SOTR naive, n = 18; HCW MBCs, n = 7; SOTR MBCs, n = 22. Statistics in A, C, and D performed using Mann-Whitney test. **P < .01, ***P < .001, ****P < .0001. Values for .1 > P > .05 are shown.

We next examined MBC subset frequencies. IgM⁺ MBCs, comprised of both T-cell–independent and early T-cell–dependent memory, were significantly reduced in SOTRs (Figure 1C). Mean IgA⁺ and IgG⁺ MBC frequencies were comparable between SOTRs and HCWs (Figure 1C); however, frequencies were very low in some SOTRs and elevated well above HCWs in others. This MBC variability was poorly resolved by transplant type or IST (Supplementary Figure 1D–F). We also assessed class-switched MBC subsets using cellular markers to delineate activation (CD21, CD11c) and maturation (CD27, CD10) (Figure 1B). Again, most subsets showed a wide range of frequencies in SOTRs and thus were not significantly different versus HCWs (Supplementary Figure 1G). Hence, a variety of B-cell compartment types exist in SOTRs that do not track clearly by transplant type or IST regimen.

Reduced Naive and MBC Diversity in SOTRs

Since SOTRs displayed quantitative reductions in naive cells and varying MBC levels, we asked if IST was also associated with changes in B-cell receptor repertoire diversity. We isolated naive B cells and total isotype-switched MBCs and amplified antibody heavy chain gene rearrangements from each population (Supplementary Table 4). Since clone counts varied by cell acquisition (Supplementary Figure 2A), we estimated clonal diversity by calculating clonal overlap between each sorted sample's independently amplified sequencing libraries (biological replicates). Naive libraries showed increased overlap in SOTRs versus HCWs by the Jaccard index, but not by cosine similarity (Figure 1D). Naive overlap values were reduced compared to MBCs, consistent with smaller, more diverse naive B-cell clones [13]. Among MBCs, we observed increased overlap in SOTRs by both the Jaccard index and cosine similarity (Figure 1D), suggesting reduced clonal diversity. Overlap values segregated poorly by transplant type and IST regimen (Supplementary Figure 2B and 2C), except that lung SOTRs showed higher overlap. Combined with our subset analyses, these data suggest that limited naive B-cell production and clonal diversity in SOTRs result in the generation of fewer MBCs. To maintain MBC levels, we hypothesize that MBCs undergo increased clonal expansion in SOTRs, leading to a narrower MBC repertoire for antigen recognition.

K-Means Clustering Reveals Distinct B-Cell Phenotype–Based Groups

Having observed significant variability in B-cell frequencies, we asked whether B-cell phenotypes themselves could classify SOTRs. We used unbiased K-means clustering by B-cell subset representation to separate HCWs and SOTRs into 3 groups (Figure 2A; Supplementary Table 5). Group 1 contained most of the HCWs plus some liver and kidney transplant recipients, whereas groups 2 and 3 had SOTRs from all transplant types (Figure 2B). Interestingly, K-means groups did not associate with particular IST regimens (Figure 2B). Recently transplanted SOTRs (<1 year) tended to fall into group 3, while those further from transplant were distributed across the 3 groups (Supplementary Figure 2D). Total B-cell frequency was comparable between group 1 and 2 SOTRs and lowest in group 3 (Supplementary Figure 2E).

Figure 2.

B-cell subset–based K-means clustering of solid organ transplant recipients (SOTRs) and healthcare workers (HCWs). A, K-means clustering of HCWs (n = 11) and SOTRs (n = 45) into 3 groups, based upon B-cell subset frequencies. Heat map represents z-scores by row. B, Fraction of HCWs and SOTRs, by both transplant and immunosuppressive therapy type (Aza, azathioprine; Myco, mycophenolate mofetil; Pred, prednisone; Tacro, tacrolimus), binned by K-means B-cell groups (G1–G3). C, B-cell subset frequencies in HCWs and SOTRs by B-cell group. D, Summary of B-cell subset representation by group. Representation was defined as low (−), intermediate (−/+), normal (+; ie, comparable to HCWs), or elevated (++). Statistics in C were performed using Kruskal-Wallis test with multiple comparisons. *P < .05, **P < .01, ***P < .001, ****P < .0001. Values for .1 > P > .05 are shown.

K-means groups were associated with clear differences in B-cell subset representation. Group 1 SOTRs had similar levels of naive, IgG⁺, and IgA⁺ MBCs compared to HCWs and intermediate levels of transitional B cells (Figure 2C and 2D). Group 2 SOTRs had reduced naive and transitional B-cell frequencies but elevated levels of IgA⁺ and IgG⁺ MBCs that surpassed those in HCWs (Figure 2C and 2D). Group 3 had the lowest frequencies for all subsets (Figure 2C and 2D). The estimates of MBC clonal diversity (Figure 1D) did not segregate by K-means group, suggesting that reduced diversity is broadly characteristic of SOTRs (Supplementary Figure 2F).

B-Cell Groups Correlate With SARS-CoV-2 Vaccine Response in SOTRs

To test the predictive value of K-means B-cell groups in SOTRs, we examined serologic responses to RBD from HCWs (n = 11) and SOTRs (n = 32) following 2 SARS-CoV-2 mRNA lipid nanoparticle vaccine doses. The mean time from dose 2 to postdose blood draw was 29 days for SOTRs and 15 days for HCWs (Supplementary Table 6). The increased time before sampling in SOTRs allowed for their slower immune response kinetics [19] and is conservative since it would be expected to reduce differences between HCWs and SOTRs.

Despite their extra time for antibody development, RBD IgG responses were variable and significantly reduced in SOTRs compared to HCWs (Figure 3A), similar to previous observations [19]. Notably, SOTR RBD IgG levels did not correlate with timing of the post–dose 2 blood draw (Supplementary Figure 3A). Instead, we found that the K-means B-cell groupings were associated with vaccine responsiveness: Group 1 SOTRs had the highest RBD levels, with 5 of 5 positive responses (optical density >0.75), while 15 of 17 group 3 SOTRs failed to generate a positive response (Figure 3A). Group 2 SOTRs had intermediate RBD responses with large variability between donors (Figure 3A). However, prednisone use could distinguish group 2 nonresponders (all on 5-mg dose) from responders (no prednisone) (Figure 3B).

Figure 3.

Post–dose 2 severe acute respiratory syndrome coronavirus 2 vaccine responses in solid organ transplant recipients (SOTRs) by B-cell group. A, Receptor-binding domain (RBD) immunoglobulin G (IgG) optical density (OD) values for healthcare workers (HCWs; n = 11) and SOTRs (n = 32), by group (G1–G3). G1, n = 6; G2, n = 9; G3, n = 17. B, RBD IgG OD values in group 2 SOTRs by prednisone use. C, Frequency of isotype-switched (IgD^–IgM^–) S1-binding B cells in HCWs (n = 11) and SOTRs (n = 31), split by group. G1, n = 6; G2, n = 9; G3, n = 16. D, S1-binding B-cell frequency in group 2 SOTRs by prednisone use. Statistics in A and C were performed using Kruskal-Wallis test with multiple comparisons (G1–G3 comparison) and Mann-Whitney test (HCWs vs SOTRs). Statistics in B and D were performed using Mann-Whitney test. *P < .05, **P < .01, ****P < .0001. Values for .1 > P > .05 are shown.

Optimal vaccination robustly induces MBCs, which are activated during pathogen reencounter. We therefore investigated circulating SARS-CoV-2 spike–specific MBCs after dose 2, combining flow cytometry and immune repertoire profiling. We used the S1 domain of the spike protein, which contains the RBD, as a fluorophore-conjugated probe to identify antigen-binding B cells (Supplementary Figure 3B), similar to previous studies [12]. IgD⁻IgM⁻ S1-binding MBCs were significantly decreased in SOTRs, and cell frequencies tracked by K-means group: Few group 3 SOTRs had detectable responses (zero S1⁺ B cells in >1.2 million cytometer events), whereas responses in most group 1 and 2 SOTRs were detectable (Figure 3C). Again, prednisone use was associated in group 2 SOTRs with reduced S1-specific MBCs (Figure 3D). Together, our findings suggest that the B-cell compartment–derived groups described herein are predictive of humoral responsiveness to novel immunogens in SOTRs.

Reduced Affinity Maturation and Clonal Diversity in S1-Binding B Cells From SOTRs

Since many group 1 and 2 SOTRs responded to SARS-CoV-2 vaccination, we investigated the quality of their S1-binding MBC repertoire. We evaluated S1-binding cells after dose 3 since clinical recommendations for additional vaccine boosts were made to increase protection in SOTRs [20]. IgD⁻IgM⁻S1⁺ B cells were sorted from 11 SOTRs and 5 HCWs that received a third dose, remained SARS-CoV-2-negative, and had a subsequent blood draw during the study period. Of these individuals, 4 HCWs and 3 SOTRs with the highest post–dose 3 B-cell responses were chosen for antibody gene rearrangement sequencing (Supplementary Figure 4A and 4B; Supplementary Table 7). Unsurprisingly, no group 3 SOTRs yielded sufficient cell numbers. SOTRs and HCWs averaged 86 days and 35 days between dose 3 and blood draw, respectively (Supplementary Table 6).

S1-binding B cells in HCWs and SOTRs had similar immunoglobulin VH gene usage and CDR3 length (Supplementary Figure 4C–E). We also measured VH gene SHM, which corresponds to the degree of affinity maturation [21] and increases in spike-binding clones over time following SARS-CoV-2 infection or vaccination [22–24]. SOTRs had significantly fewer S1-binding clones with mutation (Figure 4A) and reduced SHM per mutated clone (Figure 4B), suggesting that SOTR clones had undergone fewer rounds of affinity maturation and selection.

Figure 4.

Immune repertoire of isotype-switched S1-binding B cells in severe acute respiratory syndrome coronavirus 2–vaccinated solid organ transplant recipients (SOTRs) and healthcare workers (HCWs) after dose 3. A, Frequency of S1-binding clones showing >2% mutation in the VH gene. A binomial test shows the 2 groups differ at P < .001. B, Distribution of S1-binding clones binned by percent VH gene mutation. HCWs and SOTRs have different somatic hypermutation (SHM) distributions at P < .001 via Mann-Whitney test. C, Clone size of top 100 copy S1-binding B-cell clones. Circle size is proportional to copy number (clone size); circle color denotes subject. Bubble plot visualization was performed in Tableau Desktop 2023.2. D, Sample S1-binding B-cell clonal lineages are shown for HCWs (blue) and SOTRs (red), n = 5 for each. GL indicates nearest germline VH gene sequence. Numbers indicate the number of somatic hypermutations in the VH gene sequence compared to preceding vertical node. Black circles are computationally inferred nodes, and colored circles represent sequencing data with size proportional to copy number fraction.

To control for sampling differences, we limited clone size analysis to the 100 most abundant clones. S1-binding clone sizes were larger in SOTRs than HCWs (Figure 4C), which may reflect lymphopenia-associated clonal expansion. Clonal lineage analysis in HCWs revealed narrow trees with few branches and minimal diversification. HCW clones were highly mutated, consistent with a repertoire composed of diverse, smaller clones that have undergone significant affinity maturation (Figure 4D). In contrast, the larger SOTR clonal lineages exhibited reduced SHM with increased branching and diversification (Figure 4D), including more nodes per lineage (Supplementary Figure 4F). Thus, SOTRs had a smaller clonal pool with reduced maturation that may be less effective at mounting adequate or antigenically broad recall responses. These findings highlight qualitative deficiencies in the generation of novel humoral responses even among the most immunocompetent SOTRs.

DISCUSSION

Herein, we studied the B-cell compartment in SOTRs and identified features that can predict vaccine responsiveness to a novel antigen. SOTRs with sufficient naive cells and normal MBC frequencies (group 1) exhibited the most robust humoral response to the SARS-CoV-2 spike protein. SOTRs with a smaller naive pool and significantly expanded MBCs (group 2) exhibited a range of serologic and cellular responses that segregated by prednisone usage. SOTRs with globally reduced B cells (group 3) exhibited very poor vaccine responses and are unlikely to be protected from infection. These 3 B-cell groups are discernable using a straightforward immunophenotyping analysis of peripheral blood with a lineage marker for B cells (CD19) and markers for naive versus memory status (CD27, IgM, IgD). Since the patients’ B-cell group could be determined by a clinical laboratory within a day, as opposed to waiting weeks or longer to test for a successful vaccine response, this analysis could streamline prediction of a patient's immune competence. Moreover, the considerable stability of B cells in adult peripheral blood allows sampling for B-cell group determination to be performed at any time before vaccination [6, 25, 26]. B-cell–based grouping may also be preferable to test immune function since it avoids the potential danger of vaccine-induced allo-activation, although this risk appears to be low [27]. External validation of this method in larger SOTR cohorts will be required to fully establish its efficacy and utility before integration into the clinic, particularly in relation to non-SARS-CoV-2 vaccines and other immune responses.

In our study, B-cell groups predicted vaccine responsiveness regardless of transplant type and could be used to guide lifestyle choices, evaluate depth of IST, and inform decisions regarding vaccination or other immunomodulatory interventions. However, these applications would require confirmation in larger longitudinal studies. SOTRs with sufficient naive B cells (group 1) may benefit from vaccine regimens with higher doses, additional doses, or adjuvants promoting T-cell–independent humoral responses. SOTRs with fewer naive cells but many MBCs (group 2) may respond best to vaccination strategies that efficiently engage MBCs and promote additional maturation within germinal centers. Here, a critical challenge will be avoiding original antigenic sin–type responses in cases involving significantly drifted antigens. The reduced responses observed in group 2 SOTRs on prednisone suggest that IST interruption preceding vaccination may also specifically benefit these individuals. Finally, in individuals deficient in all circulating B-cell subsets (group 3), active immunization may not be viable. Indeed, other studies have shown that some SOTRs failed to mount a SARS-CoV-2 vaccine response after 3 or more doses [28], or remained susceptible to infection after 5 doses [29]. While strategies to elicit preimmune B-cell production would likely be beneficial, passive forms of immunity and prophylaxis with minimization of infectious exposures may be necessary for protection in group 3 SOTRs.

While our cohort's diversity enabled us to draw conclusions that are broadly applicable across transplant and therapy types, the relatively small subject number limited age-matching and prevented us from fully defining the relationship between B-cell classification and specific clinical categories such as IST regimen. A larger cohort also would be useful for determining if a multivariate method combining B-cell compartment measurements with clinical information enhances predictivity. Future studies should integrate additional immunological measurements, including total IgG levels, absolute lymphocyte counts, and T-cell phenotyping, which could further enhance SOTR classification. Additionally, future studies should longitudinally assess B-cell group classification dynamics, since the tendency for recently transplanted individuals (<1 year) to fall into group 3 suggests that classification can change as the effects of induction immunosuppression wane. While our study focused on SOTRs, we hypothesize that B-cell–based classification may be useful for risk stratification in other immunocompromised populations, including patients with autoimmunity or cancer. This method may also have utility in predicting recall responses to vaccination or protection from a broader array of pathogens, but these possibilities require careful testing in relevant cohorts.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Acknowledgments. The authors thank the SOTRs and HCWs for contributing samples. E. A. B. and I. L. acknowledge and thank the Gift of Life Transplant Foundation (Philadelphia, Pennsylvania) for providing critical funding. Immunophenotyping was performed in the Cytomics and Cell Sorting Shared Resource facility (RRID:SCR_022376) at the University of Pennsylvania. We also gratefully acknowledge the support of the clinical research coordinators from the Division of Transplant Surgery: Dorina Domi, Lexi Tumbelty, and Mary Shaw.

Author contributions. J. J. K., E. T. L. P., and I. L. wrote the manuscript. I. L., E. A. B., A. M. S., and K. M. O. designed the clinical study and recruited patients. J. J. K., W. M., and E. T. L. P. designed the experiments. J. J. K. and W. M. performed the experiments. J. J. K., A. M. R., W. M., F. L., and E. T. L. P. analyzed the data. C. K. and U. D. provided HCW samples. E. A. B., I. L., E. T. L. P., U. D., and A. S. secured funding. All authors reviewed and revised the manuscript. Co-first authorship was decided based on experimental design, organization, and analysis (J. J. K.) and clinical study design, data acquisition, and interpretation (I. L.).

Financial support. This work was supported by the Gift of Life Transplant Foundation (Philadelphia, Pennsylvania) to cover costs associated with SOTR recruitment, sample processing, and immunology assays. E. T. L. P. and U. D. received funding from the Department of Pathology and Laboratory Medicine and from the Penn Center for Precision Medicine, both at the Perelman School of Medicine of the University of Pennsylvania for recruitment and processing of HCW samples. Enzyme-linked immunosorbent assays and immune repertoire profiling work were performed in the Human Immunology Core (RRID:SCR_022380) at the University of Pennsylvania, which is supported in part by the National Institutes of Health (grant numbers P30 AI045008 and P30 CA016520).