Background

Amtagvi (lifileucel) melanoma tumor-infiltrating lymphocytes (TIL) was the first adoptive cellular therapy (ACT) against solid tumors approved by the FDA.1 Initial studies from the National Cancer Institute (NCI) had shown an overall response rate (ORR) of ~50%, and durable complete response (CR) of 20%.2–4 More recent reports have shown ORR of 31.4% and CR of 3.7% when patients received TIL after progressing through immune checkpoint inhibition (ICI).5 Currently, anti-PD-1 immunotherapy is the first line of treatment in patients with metastatic melanoma, 50% of which are at risk of progressive disease following treatment.6 Per current standard of care, Amtagvi is administered post-ICI therapy failure, at which point the function of tumor-infiltrating T-cells is likely compromised. Therefore, strategies to improve the expansion and quality of TIL manufactured post-ICI may enhance the clinical benefit of this therapy. Even when administered as an earlier line of therapy, as reported by Rohaan et al, OR and CR remained at 49% and 20%, respectively,7 providing additional rationale for its improvement.

TIL therapy is a personalized autologous treatment that involves ex vivo expansion of tumor-infiltrating T-cells, which are then reinfused. The goal is to enhance the performance of tumor-reactive T-cells by removing them from the immunosuppressive microenvironment.8 Here, we investigated whether ex vivo stimulation of TIL-B using a CD40 agonist can serve as a new approach to enhance TIL generation. CD40 is expressed on antigen-presenting cells (APCs), meanwhile CD40L is primarily expressed by activated T-cells. The CD40-CD40L interaction between APCs and T-cells provides bidirectional signaling that is critical for the activation, maturation, and effector function of both cell types.9 10 In vivo targeting of the CD40-CD40L axis has been tested using agonistic CD40 monoclonal antibodies showing modest outcomes with dose-limiting toxicity, restricting their clinical application.11–13 Here, we hypothesized that the activation of the CD40-CD40L axis ex vivo can lead to an improvement in the expansion and quality of TIL. We now report the results obtained in human melanoma and lung TIL of CD40L-enhanced TIL expansion and quality and provide mechanistic insight through analysis of B-cell subsets and gene expression changes.

MethodsTumor samples

Metastatic melanoma and non-small cell lung cancer (NSCLC) samples were obtained from consenting patients treated at Moffitt Cancer Center. Tumor tissues were minced with scalpels and then mechanically and enzymatically digested with RPMI 1640 with L-glutamine containing 30 U/mL DNAse type IV, 100 ug/mL hyaluronidase type V, 1 mg/mL collagenase type IV, 50 ug/mL gentamicin, 100 IU/mL penicillin, 100 ug/mL streptomycin, and 62.5 mg/mL amphotericin-B. Single cell suspensions were counted, followed by cryopreservation for later use.

RNA-seq data analysis

RNA sequencing was performed on formalin-fixed paraffin-embedded (FFPE) melanoma samples from patients enrolled in Moffitt Cancer Center’s TIL clinical trials (NCT01005745, NCT01659151, NCT01701674, and NCT02652455) at Moffitt’s Molecular Genomics Core. Further details are provided in the online supplemental methods. Gene expression data were then normalized, and differential expression between responders and non-responders (NRs) was evaluated using DEseq2. In silico estimation of cell abundance obtained was performed using xCell Analysis software, a robust signatures-based method for deconvolution for 64 cell types using ranked values within each dataset, not absolute values.14 The Tertiary Lymphoid Structure (TLS) Score was calculated by using the first principal component (PC1) from a principal component analysis model based on the 12-chemokine genes expressed by tumors resected from responders and NRs to TIL therapy.15–17 Demographic information from patients is displayed in online supplemental table 1.

TIL culture

For the initial TIL expansion phase (pre-REP), TILs were expanded ex vivo from frozen or fresh melanoma tumor digests or frozen lung digests in 24-well plates at a density of 1e6 cells/well under two different conditions, 6000 IU/mL IL-2 alone (Proleukin, Novartis, Emeryville, California, USA), or in 6000 IU/mL IL-2 plus 60 ng/mL human recombinant CD40L (R&D Systems, 6420-CL-025) and 0.5 ug/mL α-HA-tag (R&D Systems, MAB060) for 3–5 weeks. Demographic information from patients is provided in online supplemental table 2. For certain analyses, TILs were expanded from lung tumor fragments, approximately 1–2 mm2, in 24-well plates under two different conditions, 6000 IU/mL IL-2 alone (Proleukin), or in 6000 IU/mL IL-2 plus 1000 ng/mL human recombinant CD40L (Miltenyi Biotec, 130-098-776) for 3–4 weeks. In both cases, CD40L was added only once, on the first day of expansion. Media used for TIL expansion consisted of RPMI 1640 with L-glutamine, 10% heat-inactivated human AB serum, 10 mM HEPES, 100 IU/mL penicillin, 100 ug/mL streptomycin, 50 ug/mL gentamicin, 55 uM 2-mercaptoethanol, and 62.5 mg/mL amphotericin-B. Half of the media was replaced at least every 3–4 days, and the wells were subcultured when 80% confluent. For the rapid expansion protocol (REP), TIL were stimulated with 30 ng/mL anti-human CD3 antibody (Clone OKT3, Biolegend, 317325) in the presence of 3000 IU/mL of IL-2 and allogeneic irradiated (5000 rad) PBMC from three different donors. Media was replaced as needed. Cells were harvested on days 12–15.

Single-cell RNA sequencing

Single-cell RNA-sequencing was performed on tumor digests cultured for 48 hours in standard TIL expansion media or supplemented with CD40L+α-HA-tag from eight frozen melanoma digests, from patients enrolled in Moffitt Cancer Center’s TIL clinical trials (NCT01005745, NCT01659151, NCT01701674, and NCT02652455). Sequencing was performed at Moffitt’s Molecular Genomics Core, using the 10X Genomics platform, with 5’ amplification. Further details are provided in the online supplemental methods. A two-component computational tool (ISCVA, Interactive Single Cell Visual Analytics18) was used to perform scRNA-seq analyses including quality control, clustering, cell type classification, curation, and visualization. One of the patients was excluded from further analyses because very few cells passed QC. Cells identified as doublets were also removed from further analysis.

Identification of tumor antigen-specific TCR clonotypes

Neoantigen-specific TCRs were identified using the modified MANAFEST assay as previously described.19 20 The productive frequency of those TCR identified as known neoantigen-reactive T-cell clonotype sequences was compared between Control and CD40L conditions. TCR repertoire analysis from post-REP TIL was performed by immunosequencing of CDR3 regions of human TCR-beta chains using immunoSEQ assay (human TCR-beta deep resolution service—Adaptive Biotechnologies, Seattle, Washington, USA). Samples were obtained from consenting patients treated at Moffitt Cancer Center. Further details are provided in the online supplemental methods.

Statistical analysis

The Wilcoxon signed-rank test and Mann-Whitney U test were used for paired and for unpaired analyses, respectively. Normality was tested by the Shapiro-Wilk test. If the data were considered normal, paired, or unpaired Student’s t-test was used. The TIL expansion was tested by the χ2 test. To compare effects of CD40L stimulation (CD40L vs control) and cell population difference (B-cells vs myeloid cells) on HLA-DR, DQ, DP expression, and percentage of cells double positive for CD80 and CD86, a two-way analysis of variance model (with the interaction term included) and Tukey’s multiple comparisons test were employed. The Mann-Whitney U test was performed to compare the xCell scores, estimated using the RNA-seq data, between the responders and NRs to TIL therapy using the R software. Due to small sample size, the results with marginal significance (p≤0.07) were reported without adjusting for multiple comparison. Similarly, the Mann-Whitney U test was used to compare the TLS scores between the responders and NRs. Where indicated, outliers were excluded for the analysis based on Grubbs’ test.

ResultsResponse to TIL therapy is associated with greater abundance of class-switched memory B-cells, dendritic cells, and TLSs in melanoma

Based on prior research showing an association between tumor-infiltrating B-cells and improved clinical responses to ICI therapy,21 22 we hypothesized that their presence could also impact TIL therapy response. To test this hypothesis, we performed RNA-seq of FFPE samples from a cohort of patients who received TIL therapy. Melanoma samples collected for TIL expansion were evaluated for immune cell content in 9 responders (R) and 11 NRs to TIL therapy using the xCell algorithm (figure 1 and online supplemental figure 1). As shown in figure 1A, although the overall Bcell population was similar between R and NR, a greater abundance of class-switched memory B-cells was identified in R (p=0.07). Within the myeloid compartment, dendritic cell (DC) abundance was similar between R and NR. However, a greater abundance of plasmacytoid (p=0.03), immature (p=0.07), and activated (p=0.07) DC was observed in R compared with NR (figure 1B). Within the T-cell compartment, represented in online supplemental figure 1, effector memory CD8+T cells (CD8+Tem; p=0.07) and γδ T-cells (p=0.04) were greater in R compared with NR. In addition to individual cell types, we analyzed if the presence of organized TLSs containing active germinal centers were associated with response to TIL, as has been described for ICI.21 22 Using a 12-chemokine gene expression signature associated with the presence of TLS,15–17 we queried the same gene expression dataset and found that the tumors from responders had significantly greater 12-CK score than those of NRs (p=0.03, figure 1C). Because TLS contain B- and T-cells in active interaction, this observation suggests that a cross-talk between these (and potentially other) immune cells may condition the quality of the resulting TIL products and the clinical outcome, prompting us to further analyze the effects of induced interactions between B-cell and T-cell subsets.23

Figure 1

Immune cell content analysis in responders (R) and non-responders (NR) to melanoma TIL therapy. (A, B) Estimation of cell abundance in metastatic melanomas resected for TIL manufacture. Patients grouped based on clinical response status by RECIST criteria (total n=20; R=9 and NR=11) following TIL therapy. Boxplots are shown along with outliers’ individual data points. Gene expression was quantified by bulk RNA sequencing of FFPE sections. Cell content was estimated using the xCell deconvolution algorithm. Statistical analyses were performed by Wilcoxon signed-rank test. (C) The tertiary lymphoid structure (TLS) score was calculated based on expression of 12-chemokine genes combined, using principal component analysis for tumors resected. Boxplots are shown along with outliers’ individual data points. Statistical analyses were performed by Mann-Whitney U test. FFPE, formalin-fixed paraffin-embedded; TIL, tumor-infiltrating lymphocyte.

Stimulation with a CD40 agonist enhances melanoma TIL expansion

We next tested the hypothesis that immune cells present in the tumor microenvironment may serve as actionable targets to deliver (co)stimulatory signals to T-cells, leading to TIL enhancement during manufacturing. We used soluble CD40L as a CD40 agonist, aiming to broadly activate B-cells and DCs ex vivo. We cultured melanoma digests in standard TIL expansion media containing IL-2 (control), or in media supplemented with CD40L (CD40L), and compared the resulting TIL after 4 weeks. Of 21 patient samples analyzed, among which 19 were frozen tumor digests, TIL expansion was achieved in 7 (33%) in the control condition and in 14 (67%) in the CD40L condition (figure 2A). In addition, the total number of TIL at the end of the expansion was significantly higher in the CD40L-treated condition (p≤0.01, figure 2B). A significant increase in the number of both CD4+ and CD8+ TIL was also observed (p≤0.001 and p=0.04, respectively).

Figure 2

CD40L enhances the expansion of melanoma TIL. Melanoma single-cell suspensions were cultured in standard TIL expansion media (control) or media supplemented with an agonist of CD40 (CD40L) for 3–4 weeks. (A) TIL expansion success rate (n=21 patients) using control or CD40L-supplemented media. *p≤0.05 (χ2). (B) Cell counts of total viable cells, CD4+ and CD8+ T cells after 4 weeks of expansion. Individual values are plotted with lines connecting the paired samples, corresponding to individual patients. The dashed line represents one patient with no expansion in the control condition. This patient was not considered for the statistical analysis (n=14). *p≤0.05, **p≤0.01, ***p≤0.001 (Wilcoxon signed-rank test). (C) Differentiation phenotype analyzed by flow cytometry in pre-REP TIL (Effector Memory: CCR7− CD45RA−; Central Memory: CCR7+ CD45RA−; Naive: CCR7+ CD45RA+; CD45+Effector Memory: CCR7− CD45RA+). Individual values are plotted with lines connecting the paired samples, corresponding to individual patients. The dashed line represents one patient with no expansion in the control condition. This patient was not considered for the statistical analyses (n=14). *p≤0.05 (Wilcoxon signed-rank test). Scatter plot of a representative example is shown in the right panel. (D) Expression of CD39 and CD69 in TIL (gated on lymphoid, single, live, CD3+ cells) after 4 weeks of ex vivo expansion. Individual values are plotted with lines connecting the paired samples. The dashed lines represent patients with no expansion in the Control condition. For two of the patients a control sample expanded from fragments was used for comparison (black line). In the third patient, a value of zero was attributed to the control condition with no expansion (gray line). Those patients were not considered for the statistical analyses (n=11). *p≤0.05 (Wilcoxon signed-rank test). One outlier removed based on Grubb’s test. (E) Analysis of reactivity of melanoma TIL. Expanded T-cells were cultured with autologous tumor single-cell suspensions for 24 hours, at an E:T ratio of 1:1. Tumor reactivity was assessed by IFN-γ levels in the supernatant. Each graph represents an individual patient (n=4). TILs were expanded from tumor fragments (40355, 40348 and 40372) or tumor digest (40298). ‘Pool’ indicates TIL co-cultures generated from more than one fragment pooled together. ‘F’ indicates data from a single fragment. The black bar represents TIL co-cultured with autologous tumor cells (Digest). The gray bar represents TIL co-cultured with autologous tumor cells (Digest) preincubated with anti-HLA class I blocking antibody (W6/32). Reactivity was defined by secretion of >50 pg/mL of IFN-γ, inhibited by at least 20% when the HLA-I antibody was used. TILs, tumor-infiltrating lymphocytes.

Having established the quantitative impact of CD40 agonism on TIL expansion, we next tested whether CD40L treatment affected their phenotype. We evaluated classical markers of differentiation, CCR7 and CD45RA, as well as a T-cell subset defined by Krishna et al24 as a stem-like, characterized by the negative expression of CD39 and CD69 markers (online supplemental figure 2). As shown in figure 2C, CD40L treatment was associated with a significant increase in effector memory TIL (CD45RA−CCR7−, p=0.02) and an increase in stem-like TIL (CD39−CD69−, p=0.04) (figure 2D). After polyclonal stimulation using REP, we observed no phenotypic differences between the CD40L-treated TIL and Control, indicating that the pre-REP boost in TIL numbers did not result in premature exhaustion or terminal differentiation (online supplemental figure 3). Importantly, we did not observe any CD40L-induced expansion of CD4+CD25+FOXP3+GITR+CTLA-4+ regulatory T-cells (online supplemental figure 4). We next tested whether CD40L stimulation had an impact on tumor reactivity. To that end, we co-cultured TIL with autologous tumor digests and monitored IFN-γ secretion (>50 pg/mL) as a proxy for T-cell activation. An anti-HLA class I-blocking antibody (W6/32) was used as a negative control. TIL from four different patients (40355, 40298, 40348, and 40372) were analyzed, based on availability of autologous tumor digests (used as target). Patients 40355, 40348 and 40372 were expanded from fragments. 40355 control TIL did not expand, while CD40L TIL expanded from two individual fragments. These were pooled together for analysis of reactivity, due to low availability of tumor digest to be used as target. For patients 40348 and 40372, TIL cultures grown from individual fragments analyzed separately (indicated as F#) for greater resolution. For patient 40298, TILs were expanded as a single culture of either control or CD40L-enhanced TIL, starting from a frozen tumor digest. Figure 2E shows that whenever tumor reactivity was detected by IFN-γ secretion, that reactivity was not compromised by CD40L stimulation. In some cases, such as sample 40355, CD40L rescued the expansion of tumor-reactive TIL, which did not grow under control conditions. For patient 40348, three tumor-reactive cultures were obtained using CD40L, whereas only one culture was obtained in the control condition. Finally, for patient 40372, two cultures showing IFN-γ secretion above 50 pg/mL were obtained using CD40L, while none of the control cultures showed specific reactivity. Further studies will be required to determine if CD40L enhancement can consistently increase tumor reactivity, but our data indicate that CD40L-enhanced TIL do not lose reactive clones due to amplification of bystander T-cells.

To confirm the feasibility of this approach, we performed four validation runs at Moffitt’s Cell Therapy and Gene Engineering Facility, starting with either frozen (n=2) or fresh (n=2) melanoma single cell suspensions (online supplemental table 4). Validation runs were performed at limited scale production numbers. By the end of the REP phase, all patient samples with the inclusion of CD40L showed a fold expansion higher than 4000, and varying percentages of CD4/CD8 TIL. Overall, the TIL expansion for all four patient samples was successful in reaching a projected number of cells meeting the clinical doses of more than 50e9. TIL doses higher than 50e9 cells achieved in these validation runs fall within the range of 1e9 to 150e9 TIL doses that were described in a recently published phase II TIL trial in advanced melanoma patients.25 In conclusion, these feasibility results further support the use of CD40L supplementation in TIL clinical treatment settings.

Melanoma infiltrating B-cells express CD40 and upregulate costimulatory markers on CD40L stimulation

We next sought to delineate the cell populations responsible for the CD40L-induced improvement in TIL expansion. First, we evaluated the presence of leukocytes, T-cells and B-cells in tumor digests at the start of TIL expansions. Figure 3A shows a higher percentage of CD45+ cells in the samples that successfully expanded TIL compared with the samples that failed to do so (p≤0.05). Although the T-cell percentage present in the tumor digest had no impact on TIL expansion (figure 3B), higher numbers of B-cells in the tumor digest were significantly associated with TIL expansion (p≤0.05; figure 3C). Consistently, a higher percentage of non-B and non-T cells (CD45+CD3−CD19− cells), which includes DCs, was observed in the samples that did not expand TIL (p≤0.05, figure 3D).

Figure 3

Melanoma infiltrating B-cells upregulate costimulatory ligands in response to CD40L. (A–D) Percentage of CD45+cells (A), T-cells (CD45+CD3+CD19−) (B), B-cells (CD45+CD3−CD19+) (C), and non-B/non-T cells (CD45+CD3−CD19−) (D) present in melanoma single-cell suspensions, grouped based on their ability to yield TIL expansion. Individual values corresponding to independent patients are plotted with mean±SEM (n=20). *p≤0.05 (Mann-Whitney U test). (E) Relative abundance of different B-cell subsets (Switched memory: CD27+ IgD−; Unswitched memory: CD27+ IgD+; Naive: CD27− IgD+; Double-Negative: CD27− IgD−) present in melanoma tumor digests based on their ability to yield TIL expansion. Individual values are plotted with mean±SEM (n=24). *p≤0.05, ***p≤0.001 (Mann-Whitney U test). (F) Flow cytometry analysis of CD40 expression in myeloid cells (CD45+CD3−CD19−HLA-DR+) and B-cells (CD45+CD3−CD19+) present in melanoma tumor digests. The individual values are plotted with lines connecting the paired samples corresponding to the same patient (n=11). ***p≤0.001 (paired Wilcoxon signed-rank test). A representative histogram of CD40 expression is also shown. (G, H) Flow cytometry analysis of HLA-DR, DQ, DP expression and percentage of CD80/CD86 double positive cells within myeloid cells (CD45+CD3−CD19−HLA-DR+) and B-cells (CD45+CD3−CD19+) present in melanoma tumor digests. Tumor digests were cultured for 48 hours in standard TIL expansion media (control) or supplemented with a CD40 agonist (CD40L). Individual values corresponding to independent patients are plotted with mean±SEM (n=10). ***p≤0.001, ****p≤0.0001 (two-way ANOVA and Tukey’s multiple comparison test). ANOVA, analysis of variance.

Based on these observations, we further investigated the association of different B-cell subsets with TIL expansion. We classified B-cells based on CD27 and IgD expression as described by Sanz et al,26 as follows: switched memory (CD27+IgD-), unswitched memory (CD27+IgD+), naïve (CD27-IgD+), and double negative (CD27-IgD-) B-cells. Figure 3E shows a higher percentage of switched memory B-cells in the samples that successfully expanded TIL compared with the samples that did not, approximately 50% vs 38% (p≤0.05). In contrast, double negative B-cells were more abundant in samples that did not yield TIL (21%) than in the samples that did (12%, p≤0.001). The presence of CD27−IgD− B-cells, especially those that were CD38−CD24−, was associated with a failure in TIL expansion (p≤0.05). Of note, CD40L addition during TIL culture was able to successfully rescue the expansion of a fraction of samples containing this double negative population, again highlighting the benefit of CD40L supplementation.

Having established an association between B cells and TIL expansion, we next analyzed other CD40-expressing leukocytes. We compared the expression of CD40 between B-cells (CD45+CD3-CD19+) and myeloid cells (CD45+CD3−CD19−HLA-DR,DQ,DP+) cells (see also online supplemental figure 5). As shown in figure 3F, both B-cells and myeloid cells expressed CD40, although expression in B-cells was higher (p≤0.001). Since both could respond to the CD40 agonism, we then evaluated their response to CD40L stimulation. Tumor digests were cultured for 48 hours in control or CD40L media, and the expression of molecules involved in antigen presentation and costimulation of T-cells were analyzed by flow cytometry. In basal condition, B-cells presented higher expression of MHC class II compared with myeloid cells (p≤0.001). The addition of CD40L further increased the expression of MHC class II on B-cells (p≤0.0001), while no significant change was observed in myeloid cells (figure 3G). Interestingly, as shown in figure 3H, only B-cells were able to upregulate the costimulatory markers CD80 and CD86 on CD40L stimulation. The percentage of CD80+CD86+ B-cells increased from 0.71% to an average of 25% (p<0.0001). As observed in figure 3H, although myeloid cells displayed basal expression of CD86, this expression was not further increased on stimulation with CD40L, nor did CD80 expression. These findings suggest that B-cells are the main mediators of the effects of CD40L in the context of ex vivo TIL expansion.

CD40 agonism induces transcriptional changes in melanoma infiltrating B-cells

To better define the effects of CD40L at a cellular level, we analyzed the gene expression changes induced by this treatment. Melanoma single-cell suspensions from seven patients were cultured for 48 hours with control or CD40L media. Individual cells were then encapsulated, and gene expression was analyzed by next-generation sequencing. As shown in figure 4A, tumor digests contained melanocyte/melanoma cells, T-cells (CD4+ and CD8+), NK cells, γδ T-cells, B-cells, and myeloid cells, subdivided into DCs, monocytes (Mo), and macrophages (MΦ). Interestingly, B-cells segregated into two well-defined subclusters, defined by the culture conditions (figure 4B). In contrast, no clear partition was observed for other cell types, reinforcing the notion that B-cells are the most affected by CD40L stimulation. The subclustering of B-cells was driven by the largest number of differentially expressed genes (DEGs) between CD40L and control expansion. DEGs were defined as those having an average log2FC ≥1 and an adjusted p≤0.05. B-cells had 126 DEG, notably more than DCs (10 DEGs), macrophages (12 DEGs) and monocytes (0 DEG). For a detailed list of DEGs, please see online supplemental table 5. Within the B-cell clusters, all BCR/antibody isotypes were represented, based on expression of the genes IGHA, IGHG1, IGHD, IGHG2, IGHE, IGHG3, IGHM, and IGHG4 (figure 4C). Using the Seurat algorithm, we identified 28 different B-cell subclusters (figure 4D). The top genes defining each cluster are shown in online supplemental figure 6. To classify each B-cell cluster based on their memory phenotype, we performed a gene set enrichment analysis employing the Immunologic Signatures (C7), from the Molecular Signatures Database27 (figure 4E). This gene set enrichment analysis identified clusters 4, 10, 12, 13, 16, and 18 from the CD40L-treated B-cells as naïve, and clusters 3, 8, 23, 24, and 28 as memory B-cells. Cluster 21 was identified as germinal center B-cells and cluster 27 as plasma cells. For the control condition, clusters 1, 2, 6, 9, 14, 17, 19 and 20 were identified as naïve, and clusters 5, 7, 11, 15, 22, 25 and 26 as memory B-cells.

Figure 4

Immune cell composition of melanoma tumors and response to CD40L. Melanoma digests were cultured in standard TIL expansion media (control) or supplemented with an agonist of CD40 (CD40L) for 48 hours (n=14 samples, 7 patients) and analyzed by single-cell RNA sequencing. (A) UMAP projection of different cell populations identified in the tumor digests. (B) UMAP projection of the tumor digests based on culture conditions (CD40L vs control). (C) UMAP projection of B-cells, showing expression of IgA, IgD, IgE, IgG1, IgG2, IgG3, IgG4 and IgM antibody/BCR isotypes. (D) UMAP projection of B-cells showing subclusters identified by the Seurat algorithm. Clusters corresponding to CD40L-treated samples are indicated with the red square. (E) Cross-referencing of B-cell clusters with previously described gene sets corresponding to naive, memory, germinal center (GC) and plasma cells. TIL, tumor-infiltrating lymphocyte.

We next analyzed the top DEGs between Control and CD40L conditions. As shown in figure 5A,B, CD40L-treated B-cells upregulated CD83 (avg log2FC=0.95, adjusted p≤0.0001), a marker of activation. CCL22 was the most upregulated gene in the CD40L-treated B-cells (avg log2FC=3.6, adjusted p≤0.0001), with the highest expression associated with the clusters identified as memory B-cells, clusters 3, 8, and 24 (figures 4D,E and 5A,C). Figures 4D,E and 5D show the expression of FcRL4 in the memory B-cell clusters on CD40L stimulation, which is a marker associated with tissue resident memory B-cell phenotype.26 CCL22 and FcRL4 inductions were further confirmed by flow cytometry (figure 5E,F; p≤0.0001, p≤0.01, respectively).

Figure 5

CD40L induces gene expression changes in melanoma infiltrating B-cells. Melanoma digests were cultured in standard TIL (control) or CD40L-supplemented media for 48 hours (n=14 samples, 7 patients). (A) Volcano plot displaying differentially expressed genes in CD40L-treated versus control B-cells assessed by scRNA-seq (Log2FC=0.95; p≤0.05). Upregulated genes are shown in red. Downregulated genes are shown in blue. (B–D) UMAP projections of the B-cell compartment, showing expression of differentially expressed genes CD86, CCL22, and FcRL4. (E) Flow cytometry analysis of CCL22 expression in melanoma infiltrating B-cells. Individual values are plotted with lines connecting the paired samples corresponding to individual patients (n=22). ****p≤0.0001 (Wilcoxon signed-rank test). Flow cytometry scatter plot of a representative example shown in the right panel. (F) Flow cytometry analysis of FcRL4 expression in melanoma infiltrating B-cells. Individual values are plotted with lines connecting the paired samples (n=22). ***p≤0.001 (paired Wilcoxon signed-rank test). Representative example shown in the right panel. (G) UMAP projection of B-cells showing expression of CD58. (H) Flow cytometry analysis of CD58 expression in B-cells cultured in standard (control) or CD40L-supplemented media for 48 hours. Individual values are plotted with lines connecting the paired samples corresponding to individual patients (n=22). ****p≤0.0001 (paired Wilcoxon signed-rank test). (I) Percentage of CD58 positive B-cells in samples grouped based on their ability to yield TIL expansion (successful expansion, n=13; unsuccessful expansion, n=9). ns=not statistically significant, ***p≤0.01 (paired Wilcoxon signed-rank test). TIL, tumor-infiltrating lymphocyte.

Within the T-cell clusters, the impact of CD40L was minimal. Both CD4+ and CD8+ T-cells had only one DEG (online supplemental table 5 and figure 7). Of interest, although the study was not designed or statistically powered to compare R (n=2) vs NR (n=5), we observed an apparent clustering of T-cells based on patient clinical response (online supplemental figure 8A,B). A higher frequency of both T follicular helper (Tfh) and CD8+ T effector memory (CD8 Tem) cells was detected in R, while NR presented a higher frequency of naïve CD4+ T-cells and other CD4+ T-cell subsets (online supplemental figure 8B). The gene signatures that define this clustering are shown in online supplemental figure 8C.

CD58 upregulation in B-cells is associated with TIL expansion

CD58 was significantly upregulated in CD40L-treated B-cells (avg log2FC=1.1, adjusted p≤0.0001, figure 5A), with positive expression across all CD40L-treated clusters (figure 5G). Due to its importance for B-cell and T-cell interaction,28 we further investigated the expression of CD58 on the surface of TIL-B, in response to CD40L stimulation. Melanoma digests were cultured for 48 hours in the presence or absence of CD40L. As shown in figure 5H, CD40L-treated B-cells had greater expression of CD58 compared with control B-cells (p≤0.0001), validating the scRNA-seq results. A further analysis of CD58 expression in the different subsets of B-cells defined by Sanz et al26 showed comparable expression across all four subsets (online supplemental figure 9). When we classified the samples shown in figure 5H based on whether they yielded TIL expansion or not, we found that CD58 upregulation by TIL-B remained statistically significant in tumors with successful TIL expansion (p≤0.001), but not in samples that failed to expand TIL in at least one condition (figure 5I). These observations suggest a potential role of CD58 in the effects of CD40L on TIL expansion.

CD40L enhances lung TIL expansion and increases the frequency of neoantigen-reactive TIL

To determine if TIL enhancement by CD40L is exclusive to melanoma, we next tested the expansion of TIL from metastatic NSCLC. Tumor fragments were cultured in control media, as previously described,20 or with CD40L supplementation. From 12 independent samples analyzed, TIL expansion was achieved in 71 fragments (72.5%) in the control condition and 82 fragments (85.4%) in the CD40L condition (figure 6A; p≤0.05). Recapitulating the results obtained with melanoma samples, the total number of TIL was higher in the CD40L than in the control cultures (p≤0.01, figure 6B). The same was observed when comparing the total number of T-cells expanded per individual fragment (figure 6C; p≤0.0001). Interestingly, the addition of CD40L led to an expansion of 13 individual cultures by day 20; meanwhile, the same number of expanded cultures from individual fragments was only achieved by day 28 using the control culture condition (figure 6D; p≤0.0001). This result shows that CD40L addition can reduce by about 1 week the TIL expansion process, shortening the culture time needed for patient infusion. As observed in melanoma samples, CD40L treatment was associated with a significant increase in CD39-CD69- TIL (