For this study, we used samples obtained in the context of a prospective, open-label, randomized, controlled trial as part of the Immune Response study of the RECOVAC (Dutch REnal patients COVID-19 VACcination) consortium, described elsewhere [6]. The studies involving human participants were reviewed and approved by the Erasmus MC ethics committee. The participants provided written informed consent to participate in the study. Ethical approval for our study was obtained from the Erasmus MC ethics committee (MEC-2025-0115). Serum and peripheral blood mononuclear cells (PBMCs) were collected in March and October 2021. Selection of participants for our study was based on availability of materials, which resulted in the inclusion of 64 KTR and 37 individuals without kidney disease (controls). For the PBMC analysis, we randomly selected 15 KTR and 14 controls with PBMCs available at both timepoints.

To determine humoral RSV-specific immune responses, we measured binding IgG and neutralizing antibody titers. Binding antibody titers were measured against the G protein of RSV-A (Ga) and pre-F protein using enzyme-linked immunosorbent assay (ELISA). Briefly, 96-well high binding plates (Corning Incorporated, ME, USA) were coated with 100 ng stabilized pre-F (DS-Cav1, NIH, USA) or Ga (RSV A, rsb1734, Sino Biological, Beijing, China) protein in phosphate-buffered saline (PBS, in-house) per well and incubated overnight at 4 °C or for 1 h at 37 °C. The plates were then blocked for 1 h at 37 °C with 100 µL blocking buffer (Blocker^TM BLOTTO in tris-buffered saline (TBS, Thermo Scientific, MA, USA) diluted 1:1 in PBS). After blocking, sera were added in 3-fold dilutions in blocking buffer starting at 1:40 (50 µL/well) and incubated for 1 h at 37 °C. This was followed by adding 1:6,000 polyclonal rabbit anti-human IgG/HRP (DAKO, CA, USA) (50 µL/well) and incubation for 1 h at 37 °C. Subsequently, KPL TMB peroxidase substrate diluted 1:1 with TMB diluent (SeraCare, MA, USA) was added and incubated for 10 min at room temperature (RT). Finally, the reaction was stopped by adding 100 µL sulphuric acid and plates were read out at 450 nm with 620 nm background in a plate reader (Tecan, Männedorf, Switzerland). The absorbance values were used to determine the 50% inhibitory concentration (IC50), considered the binding antibody titer, by fitting a log-logistic function on the S-curves using the R package drc (v3.0-1).

Recombinant RSV containing EGFP as an additional transcription unit [rHRSV^A11EGFP(5)] was kindly gifted by Dr. W.P. Duprex [14]. Virus stocks were generated in and titrated on HEp-2 cells as previously described [15]. Titers were expressed in tissue culture infectious dose-50 (TCID50)/mL as calculated using the Reed & Muench method [16]. Neutralizing antibody titers were determined by virus neutralization (VN): in brief, a 2 or 4-fold dilution series, starting at a 1:32 serum dilution, was incubated in duplicate with ∼1,000 TCID50 rHRSV^A11EGFP(5). This virus-serum mixture was then added to 4–5-day old monolayers of Vero cells and incubated for 2 days. Plates were scanned on a Typhoon laser-scanning platform (Cytiva Life Sciences, DE, USA) and fluorescent foci were counted using ImageQuant TL v8.2.0.0. Counts were used to determine the 50% plaque reduction neutralization titer (PRNT50), considered the neutralizing antibody titer, by calculating the proportionate distance as described previously [17].

RSV-specific cellular immune responses were detected by stimulation of PBMCs with RSV-derived peptide pools. First, PBMCs were thawed and rested for approximately 6 h in a 96-well plate (Greiner, Frickenhausen, Germany). Cells (1 million/well) were subsequently stimulated for 21 h with 1 μg/mL RSV peptide pools (Supplementary Table 1, JPT Peptide Technologies, Berlin, Germany). These peptide pools consist of 15-mer peptides with an 11-amino acid overlap encompassing the surface proteins F from both RSV strains (Fa, Fb) or Ga, the internal proteins M or N, or a selection of known immunogenic peptides (Select). The solvent DMSO (0.2%, VWR international, Fortenay-sous-Bois, France) served as negative control, whereas Concanavalin A (1:200, ConA, Sigma-Aldrich, MO, USA) served as positive control. Stimulations with DMSO, ConA and Select were always performed. Cell availability determined the number of stimulations that could be included per sample: for each additional 1 million cells available, an extra stimulation was included in the following order: Fa, Ga, Fb, N and M. After PBMC stimulation, we determined the relative frequency of RSV-specific T cells and measured cytokine production in the supernatant.

To assess the relative frequency of RSV-specific T cells, we performed a flow cytometry-based activation induced marker (AIM) assay. Flow cytometry staining was performed by incubating cells with a Live/Dead marker for 15 min at RT in the dark. After washing, 50 µL antibody mix (see Supplementary Table 2) was added and incubated for 30 min at RT in the dark. Data acquisition was performed using the FACSymphony A3 Flow Cytometer (BD Biosciences, NJ, USA). RSV-specific memory CD4⁺ T cells (CD45RA⁻CCR7^-, CD45RA⁺CCR7^-, or CD45RA⁻CCR7⁺) were defined through Boolean OR gating. This method allowed us to include all memory CD4⁺ T cells which were positive for at least any two of the activation markers OX40⁺, CD137⁺, CD69⁺ and CD25⁺. This gate is thus a summation of the different double-positive gates, with events occurring in multiple gates only counted once [18]. RSV-specific memory CD8⁺ T cells were defined as CD69⁺CD137⁺. Data were processed using FlowJo software v10.10.0 and OMIQ software (Dotmatics).

Next, we measured cytokines produced in supernatant of PBMCs stimulated with the Select, Fa or Ga peptide pools, with the DMSO condition serving as negative control, using a multiplex human Th cytokine panel kit (LEGENDplex, Biolegend, CA, USA) as described previously [19].

For baseline characteristics of our cohort, continuous data are presented as median and inter-quartile range (IQR) for non-normally distributed variables, and differences between groups were tested using Wilcoxon test. Categorical data are presented as count and percentage, and groups were compared using the Chi-squared test. To compare binding and neutralizing antibody titers between groups and timepoints, Wilcoxon signed rank test with Bonferroni correction for multiple testing was performed, resulting in a significant p-value of <0.025. To correlate binding and neutralizing antibody titers, Spearman rank correlation was used to calculate ρ and coefficient.

To obtain the relative frequencies of antigen-specific T cells, the AIM-positive background percentage (as measured in the DMSO condition) was subtracted. Frequencies below 0.01% after adjustment for DMSO were set at 0.01%. The groups were compared using the Kruskal-Wallis test.

The clustering of the cytokine analysis comprised of multiple steps. First, concentrations were adjusted for the DMSO background; negative concentrations were set on 0.01 pg/mL. To avoid false skewing of the analysis, we excluded cytokines for further analysis if more than 50% of observations was negative after adjusting for background, or if more than 20% of observations was under the limit of detection. Next, we calculated the z-scores of the log₁₀-transformed values. Guided by the agglomerative coefficient of different distance metrics and clustering methods, we continued with the Manhattan distance and Ward method. The optimal number of clusters was chosen based on the largest dissimilarities in the tree diagram. Next, a heatmap was plotted based on the unsupervised hierarchical clustering using the R package pheatmap (v1.0.13). Selected clinical characteristics were compared between the three groups using the Kruskal-Wallis test, and further post-hoc analyses between individual groups using the Wilcoxon test and the Bonferroni correction for multiple testing resulting in a significant p-value of <0.017. All analyses were performed in R v4.5.2 and a p-value of <0.05 was considered statistically significant, unless stated otherwise. Figures were created in R v4.5.2 and with BioRender Graphing. R scripts are available on request to the corresponding author.

We included a total of 64 KTR and 37 controls in our study. The baseline characteristics of the two groups are presented in Table 1. The median time since last transplantation was 6.8 years and the majority of KTR (73.4%) used two immunosuppressant agents. Tacrolimus use (95.3%) and mycophenolate mofetil use (70.3%) were most prevalent in our patient group. There was no significant difference in age or sex between the two groups.

We measured pre-F- and Ga-specific IgG and neutralizing antibodies in serum samples collected in March and October 2021; 60 KTR and 28 controls could be measured at both timepoints. Due to limited serum availability, 4 KTR and 7 controls only had a March sample available, and 2 controls only an October sample. Pre-F-specific binding IgG titers were comparable between KTR and controls and were stable over time (Figure 1A). Ga-specific binding IgG titers were comparable between KTR and controls in March, but were significantly higher in KTR than in controls in October (p < 0.05; Figure 1B).

Neutralizing antibody titers were also comparable between KTR and controls and stable over time (Figure 1C). We subsequently assessed the relationship between binding and neutralizing antibody titers (Figure 1D) and found a moderate but significant correlation in both March and October samples between pre-F-specific binding antibody titers and neutralizing antibody titers for KTR (rho = 0.55, p < 0.001 and rho = 0.46, p < 0.001, respectively) and controls (rho = 0.59, p < 0.001 and rho = 0.66, p < 0.001, respectively). Similarly, Ga-specific antibody titers correlated with neutralizing antibody titers for both groups and timepoints, albeit less pronounced (rho = 0.41, p < 0.001 and rho = 0.42, p < 0.001 for KTR, and rho = 0.5, p = 0.003 and rho = 0.21, p = 0.27 for controls; Supplementary Figure S1).

To measure RSV-specific T cell responses, we determined the relative frequency of RSV-specific memory T cells in 15 KTR and 14 controls which were randomly selected. Lymphocyte counts were previously determined in whole blood [20] and were significantly lower in KTR compared to controls at both timepoints (p < 0.001; Figure 2A). Next, we quantified RSV-specific CD4⁺ and CD8⁺ T cells as relative frequencies of total CD4⁺ or CD8⁺ memory T cells using the gating strategy depicted in Figure 2B. No significant differences in RSV-specific CD4⁺ T cell frequencies were observed for any of the stimuli at either timepoint, except for stimulation with the Fa peptide pool in March, which resulted in higher frequencies observed in KTR (p < 0.05; Figure 2C). As expected, the highest frequencies of RSV-specific T cells were detected after stimulation with the Select pool that contained immunogenic peptides, followed by the overlapping F peptide pool from either RSV strain. Similarly, RSV-specific CD8⁺ T cell frequencies were not different between KTR and controls at either timepoint, but were in general lower than RSV-specific CD4⁺ T cell frequencies (Figure 2D).

Next, we investigated the memory phenotype of RSV-specific T cells. Within the CD4⁺ T cell compartment, we distinguished central memory (CM, CCR7⁺CD45RA⁻) and effector memory (EM, CCR7⁻CD45RA⁻) populations. To determine the RSV-specific CD4⁺ T cell memory phenotype, we pooled all samples from each group per timepoint and stimulus. Aspecific ConA stimulation indicated a predominance of CM in both groups (Supplementary Figure S2A), whereas RSV-specific T cells showed a balanced distribution of CM and EM populations (Figure 2E). In the CD8⁺ T cell compartment, an additional memory population was defined: terminally differentiated EM re-expressing CD45RA (EMRA, CCR7⁻CD45RA⁺). Due to the low frequency of RSV-specific CD8⁺ T cells, data from both timepoints were pooled per cohort to determine the relative contributions of CM, EM and EMRA populations. ConA stimulation revealed balanced CM, EM, and EMRA proportions in controls but reduced CM and EM proportions in KTR (Supplementary Figure S2B), whereas RSV-specific T cells were predominantly phenotyped as EMRA cells (Figure 2F).

Lastly, we investigated the targets of the RSV-specific CD4⁺ T cell repertoire. To this end, we included all samples that could be tested for all peptide pools, calculated the total RSV-specific CD4⁺ memory relative frequencies across the mutually exclusive stimulations (Fa, Ga, M and N) and determined the proportion attributable to each target. Most individuals exhibited CD4⁺ memory T cell responses against all four proteins, without a clear dominance of any single antigen (Supplementary Figure S2C). KTR showed a distribution comparable to that of controls, but the relative contributions of each protein target varied between individuals.

We measured 12 cytokines in the supernatant of PBMCs obtained in March 2021 stimulated with Select, Fa and Ga peptide pools. IFN-γ and IL-2 levels trended higher in KTR compared to controls, though this was not statistically significant (Figures 3A,B). We observed no significant differences in concentrations of other cytokines. Next, we performed unsupervised hierarchical clustering on cytokine signatures after Select peptide pool stimulation, as this resulted overall in the highest cytokine concentrations. Based on our criteria, 7 cytokines were included for further analysis: IL-2, IL-6, IL-10, IFN-γ, TNF-α, IL17-F and IL-22 (Figures 3A,B; Supplementary Figure S3) and 5 cytokines were excluded as per pre-defined exclusion criteria (materials and methods): IL-4, IL-5, IL-9, IL-13 and IL-17A. The clustering analysis resulted in three cytokine clusters with a distinct profile (Figure 3C). KTR and controls were equally distributed throughout these clusters. Cluster 1 was Th₁-skewed, cluster 2 showed a balanced profile, and cluster 3 was characterized by a non-Th₁ profile. Of note, individuals in cluster 1 had significantly lower pre-F-specific binding antibody titers compared to cluster 2, indicating a reduced humoral response (p < 0.01, Figure 3D). In contrast, individuals in cluster 3 had significantly lower CD4⁺ and CD8⁺ RSV-specific memory T cells compared to cluster 2, indicating a reduced cellular response (p < 0.01 and p < 0.01, Figures 3E,F). Individuals in cluster 3 were significantly older than individuals in cluster 2 (p < 0.01, Figure 3G).