Host-specific differences in top-expanded TCR clonotypes correlate with divergent outcomes of anti-PD-L1 treatment in responders versus non-responders

Abstract

Immune checkpoint inhibitors (ICIs) have revolutionized cancer treatment; however, the responses to ICI treatment are highly variable in different individuals and the underlying mechanisms remain poorly understood. Here, we employed a mouse squamous cell carcinoma (SCC) model where tumor-bearing recipients diverged into responders (R) versus non-responders (NR) upon anti-PD-L1 treatment. We performed in-depth TCRβ sequencing with immunoSEQ platform to delineate the differences in CD8 tumor-infiltrating lymphocytes (TILs). We found that R and NR CD8 TILs both exhibited evidence of clonal expansion, suggesting activation regardless of response status. We detected no differences in clonal expansion or clonal diversity indexes between R vs. NR. However, the top expanded (>1%) TCRβ clonotypes appeared to be mutually exclusive between R and NR CD8 TILs, showing a preferential expansion of distinct TCRβ clonotypes in response to the same SCC tumor in R vs. NR. Notably, the mutual exclusivity of TCR clonotypes in R vs. NR was only observed when top TCRβ clonotypes were counted, because such top-expanded clonotypes are present in the opposite outcome group at a much lower frequency. Many TCRβ sequences were detected in only one recipient at a high frequency, implicating highly individualized anti-tumor immune responses. We conclude that differences in the clonal frequency of top TCR clonotypes between R and NR CD8 TILs may be one of the factors underlying differential anti-PD-L1 responses. This notion may offer a novel explanation for variable ICI responses in different individuals, which may substantially impact the development of new strategies for personalized cancer immunotherapy.

Keywords: T cell receptor sequencing; TCR repertoire; head and neck squamous cell carcinoma (HNSCC); immune checkpoint inhibitor (ICI); individualized anti-tumor immune responses.

Figure 1

Clonal expansion in responder and non-responder CD8 TILs. (A) Tumor growth curves of R (blue) and NR (red) mice. CD8 T cells were isolated from the tumors of R (n=3, 1R, 2R, 3R) and NR (n=4, 4NR, 5NR, 6NR, 7NR) mice after anti-PD-L1 treatment on day 18 after tumor inoculation followed by gDNA extraction for ImmunoSEQ TCRβ sequencing. (B) TCRβ clonal expansion in 7 sequenced TIL samples. Each pie chart shows the percent of top 20 TCRβ clonotypes in each sample. A single-colored pie slice represents the percentage of cells containing the same TCRβ clonotype in the entire sample, and the gray slice represents the combined percentage of remaining TCRβ clonotypes in each sample. (C) Relative abundance of TCRβ clonotypes in 7 sequenced CD8 TIL samples. The relative abundance of individual TCRβ clonotypes was calculated using repClonality function in immunarch package and TCR clonotypes were grouped accordingly as small, medium, large, and hyperexpanded. (D) Average of clonal frequency of top-20 TCRs in R (n=3 mice, N=60 clonotypes in total) vs. NR (n=4 mice, N=80 clonotypes in total). (E) Clonal expansion index between R vs NR CD8 TILs. (F-H) No differences in TCR diversity indexes between R vs. NR CD8 TILs, including Simpson clonality (F), Simpson’s dominance index (G), and normalized TCR richness (H).

Figure 2

Top TCRβ clonotypes appear to be mutually exclusive between responder (R) and non-responder (NR) CD8 TILs. (A) Heatmap of top TCRβ clonotypes in all samples. Top TCRβ clonotypes (abundance >1% of a given sample) defined as having the same TCRβ CDR3 a.a. sequences were sorted by average abundance in R vs. average abundance in NR. (B) Statistical analysis for the mutual exclusivity of R vs. NR top TCRβ clonotypes. Top TCRβ clonotypes were assigned to R (n=30) or NR (n=32) group according to their overall abundance in either group. 26 of 30 R clonotypes were observed only in R and 28 of 32 NR clonotypes were observed in NR only, when the abundance of shared clonotypes is set to be >1% for the opposite outcome group. Statistical significance was calculated by Fisher’s Exact test (****, P<0.0001). (C) Differences between two groups were calculated by Fisher’s Exact test to evaluate the exclusivity of TCRβ clonotypes in R or NR. Top: 23 R clonotypes were observed only in R, whereas 7 R clonotypes were also observed in NR. 28 NR clonotypes were observed in NR only, whereas 4 NR clonotypes were also observed in R, when the abundance of shared clonotypes is set to be >0.5% in the entire repertoire of the opposite outcome group (****, P<0.0001). Middle: 14 R clonotypes were observed only in R, whereas 16 R clonotypes were also observed in NR. 19 NR clonotypes were observed in NR only, whereas 13 NR clonotypes were also observed in R, when the abundance of shared clonotypes is set to be >0.1% in the entire repertoire of the opposite outcome group (N.S., no significance). Bottom: 11 R clonotypes were observed only in R, whereas 19 R clonotypes were also observed in NR; 15 NR clonotypes were observed in NR only, whereas 17 NR clonotype was also observed in R, when all CDR3β sequences were counted for the opposite outcome group including clones less than< 0.1%.