3.1 Comb potentiates the anti-tumour efficacy of reduced anti-PD-1 mAb dosage against CRC
C57BL/6 mice were inoculated s.c. with MC38 cells and treated with anti-PD-1 mAb at 100 and 200 µg (Fig. 1a). A significant anti-tumour effect was observed in mice treated with 200 µg/injection anti-PD-1 mAb at 32 days post-tumour inoculation compared to that in untreated mice (P < 0.05) (Fig. 1b). Meanwhile, mice treated with 100 µg/injection anti-PD-1 mAb showed no significant difference in tumour size compared to that in untreated mice (P < 0.05) (Fig. 1b). Mice inoculated with MC38 had significantly prolonged survival when treated with 100 and 200 µg/injection anti-PD-1 mAb compared to that of untreated mice. Mice treated with 200 µg/injection anti-PD-1 mAb had significantly prolonged mouse survival compared to those treated with 100 µg/injection (P < 0.05) (Fig. 1c). As for irAE, no significant differences in body weight were observed among the three groups (Figure S1a). These findings indicate that a reduced dosage of anti-PD-1 mAb decreased anti-tumour efficacy against CRC.
C57BL/6 mice were inoculated s.c. with MC38 cells and subsequently treated with RD-αPD-1 mAb monotherapy, Comb monotherapy, and combined treatment of both (Fig. 1d). Compared to untreated mice, only RD-αPD-1 mAb with Comb inhibited tumour growth at 28 days after tumour inoculation, and a significantly prolonged mouse survival was observed in mice treated with RD-αPD-1 mAb with Comb compared to that of untreated mice and those treated with each monotherapy (P < 0.05) (Figs. 1e, f). No significant differences in body weight were observed among the four groups at 28 days (Figure S1b). These findings indicate that the addition of Comb to RD-αPD-1 mAb substantially improved the therapeutic efficacy against MC38, overcoming the limited anti-tumour efficacy of RD-αPD-1 mAb monotherapy.
3.2 RD-αPD-1 mAb with Comb induces high numbers of CD8 + T cells, and Comb modulates myeloid cells in the TME
C57BL/6 mice were inoculated s.c. with MC38 cells and treated with RD-αPD-1 mAb monotherapy, Comb monotherapy, and RD-αPD-1 mAb with Comb. CD45+ immune cells were harvested from tumour tissue (Fig. 2a). The proportion of CD3+ TILs in CD45+ cells was significantly higher in mice treated with RD-αPD-1 mAb with Comb compared to that in untreated mice (P < 0.05) (Fig. 2b). The proportion of CD8+ TILs in total viable cells was significantly higher in mice treated with Comb monotherapy or RD-αPD-1 mAb with Comb compared to that in untreated mice, although the difference was insignificant when the RD-αPD-1 mAb monotherapy group was compared to the untreated mice (P < 0.05) (Fig. 2b). Total CD11b+ myeloid cells were significantly reduced in the RD-αPD-1 mAb with Comb group compared to that in the untreated group (P < 0.05) (Fig. 2c). Moreover, the relative proportion of a Gr-1− MHCII+ subset in CD11b+ CD11c− myeloid cells was significantly reduced in mice treated with RD-αPD-1 mAb with Comb compared to that in untreated mice (P < 0.05). The proportion of a Gr-1+ MHCII− subset in CD11b+ CD11c− myeloid cells was significantly increased in mice treated with Comb monotherapy compared to that in untreated mice (P < 0.05). These findings indicated that the Comb treatment specifically altered the infiltration of T and myeloid cells into the TME. No significant difference was observed in the expression of CD40L, TIGIT, and TIM3 on CD4+ and CD8+ T cells (Figure S2).
3.3 RD-αPD-1 mAb impacts the exhausted profile of CD8 + TILs when used in combination with Comb
For detailed analysis of T and myeloid cells in the TME, scRNA-seq analysis was performed by employing the same mouse model used for flow cytometry analysis (Fig. 3a). Figure S3 shows the flowchart of the uniform manifold and approximation projection (UMAP) plots (Figure S3). CD8+ T cells were clustered into four distinct clusters according to the expression of canonical markers for each cell type:35,36 effector/exhausted T cells (the antigen expression-related genes, CD44 and pdcd1, along with several exhaustion [Harcv2, Lag3, Tigit, Ctla4, and Il2rb] and cytotoxic-related genes [Ifng, Gzmb, Gzme, and Prf1]), effector memory T cells (Sell and cytotoxic-related genes: Gzma, Gzmb, Gzmc, and Prf1), resident memory T cells (CD69 and cytotoxic-related genes: Gzma, Gzmc, Gzme, and Prf1), and central memory T cells (Sell and Tcf7, without activation and exhaustion-related genes) (Figs. 3b–e). Considering the distinct distribution of effector/exhausted T cells between groups observed in the UMAP analysis, we re-clustered these clusters and identified four subgroups within the CD8+ effector/exhausted T cells: precursor effector group (CD69, with low expression of cytotoxic and exhaustion-related genes), effector group (cytotoxic-related genes: Gzma, Gzmb, and Gzmk, with low expression of exhaustion-related genes), exhausted group (cytotoxic-related genes: Gzma and Ifng; and exhaustion-related genes: Lag3, Tigit, and Ctla4, excluding Havcr2), and terminally exhausted group (cytotoxic-related genes: Gzmb, Gzmc, Prf1, and Ifng; and exhaustion-related genes, including Havcr2) (Figs. 3f–i).37 Almost all CD8+ T cells in mice untreated or treated with RD-αPD-1 mAb monotherapy exhibited an exhausted group. Conversely, mice treated with Comb monotherapy and RD-αPD-1 mAb with Comb exhibited a higher proportion of precursor and terminally exhausted T cells, whereas most effector T cells belonged to mice treated with RD-αPD-1 mAb with Comb (Fig. 3j). Collectively, these results indicate that combining RD-αPD-1 mAb with Comb markedly transformed the exhaustion landscape of CD8+ TILs.
3.4 Comb remodels the TME
Re-clustering myeloid cells (itgam+) based on gene expression using an unsupervised inference analysis identified 11 clusters of myeloid cells through UMAP dimension reduction (Figure S4a). For simplicity, we organized these clusters into five meta-clusters based on their hierarchical relationships (Figure S4b). Based on the expression of canonical markers, five meta-clusters were observed: macrophages (CD68; MHC class II-related genes: H2-Aa and H2-Ab1; Trem2, C1qb, and Apoe), conventional DCs (cDC; itgax and MHC class II-related genes: Sirpa and CD83), plasmacytoid DCs (pDC; Ly6c, Bst2, and Irf8), neutrophils (Gsr, low expression of MHC class II-related genes, and Arg1), and MDSCs (Gsr, Arg1, and low expression of MHC class II-related genes) (Figs. 4a, b).38 In Comb monotherapy and RD-αPD-1 mAb with Comb, the proportion of macrophages decreased, whereas that of pDCs increased (Fig. 4c).
Upon further analysis of macrophages through re-clustering, seven clusters were identified and visualized via UMAP dimension reduction (Figure S4c), which were organized into five meta-clusters following their hierarchical ordering based on gene expression (Figure S4d). Five distinct clusters were defined by the expression of canonical markers, including M1-like macrophages, three TAM subpopulations, and others (Figs. 4d, e).39 The three TAM subpopulations included TAM-M2 (Arg1, Tgfb1, and Fn1), TAM-1 (CD74, Apoe, and Ctsz), and TAM-2 (Ctsz and Spp1). M1-like macrophages had high expression of M1-related genes (Tlr2, Adgre1, Tnf, and Il1b) and low expression of M2-related and immunosuppressive genes. Almost all macrophages in mice treated with Comb were M1-like macrophages, whereas TAM-M2, TAM-1, and TAM-2 were observed in mice untreated or treated with RD-αPD-1 mAb monotherapy (Fig. 4f). Hence, Comb may remodel the TME by shifting macrophage polarization from M2-like to M1-like states. Additionally, the proportion of PD-L1+ TAM was lower in mice treated with Comb and RD-αPD-1 mAb with Comb compared to those untreated or treated with RD-αPD-1 mAb monotherapy (Figs. 4g, h).
In the analysis of pDCs, further re-clustering identified five distinct DC clusters based on the expression of canonical markers (Figs. 4i, j). These included two pDC subtypes with antigen-presenting capacity (pDC-AP, characterized by MHC class I and class II-related genes: H2-K1, H2-Q4, H2-Q6, H2-Ab1, H2-Eb1, H2-Aa, CD83, IRF7, and Tnf) and three pDC subtypes without antigen-presenting capacity.40 Among the pDC-AP subtypes, pDC-AP1 was distinguished by high expression of costimulatory molecule (CD86) and T cell recruitment chemokine-related genes (Cxcl9 and Cxcl10). Conversely, among the pDCs without antigen-presenting capacity, pDC3 exhibited high expression of immune suppression-related genes (IRF5 and Tgfb1) and low expression of Type I interferon-related genes (IRF7 and Tnf). Notably, most pDC-AP cells were observed in mice treated with Comb and RD-αPD-1 mAb with Comb, whereas nearly all pDC3 cells were found in mice either untreated or treated with RD-αPD-1 mAb monotherapy (Fig. 4k).
3.5 RD-αPD-1 mAb with Comb induces oligoclonal expansion of CD8 + TILs exhibiting various memory, activation, and exhaustion profiles
We generated droplet-based 5′-scRNA-seq and -scTCR-seq libraries to perform a combined analysis of CD8+ TIL gene expression and clonal alterations. The RD-αPD-1 mAb with Comb group exhibited an increase in the number of TCR clones that comprise over five cells (24 clones) when compared to that of the untreated (9 clones), RD-αPD-1 mAb monotherapy (11 clones), and Comb monotherapy (9 clones) groups (Fig. 5a and Table 2). We assessed the diversity of the TCR repertoire. The diversity index in the RD-αPD-1 mAb monotherapy-treated mice nearly doubled compared with that in untreated mice (24.70 vs. 62.71) (Fig. 5b). Meanwhile, that of mice treated with Comb monotherapy and RD-αPD-1 mAb with Comb showed a gradual increase compared to that of untreated mice (30.84 and 37.67, respectively).
We further analysed the gene expression profile of CD8+ T cells with the same TCR clones that comprised over five cells in each group. Based on the seven clusters used in Fig. 3, the exhausted profile constituted a large proportion of each TCR clone in mice untreated and treated with RD-αPD-1 mAb monotherapy (Fig. 5c). Conversely, CD8+ TILs of each TCR clone in the Comb monotherapy group mainly consisted of precursor effector and terminally exhausted or effector memory profiles, whereas those in the RD-αPD-1 mAb with Comb group exhibited a diverse range of profiles, including precursor effector, effector, terminally exhausted, and effector memory profiles. Finally, we identified “TCR specificity groups”— clusters of distinct TCR sequences that likely recognise the same or similar antigens— through shared motifs in the CDR3 sequence, using the GLIPH algorithm on TCR clones (Table 2). Three specific TCR groups among the seven abundant TCR β CDR3 sequences were exclusively observed in mice treated with RD-αPD-1 mAb with Comb.
Table 2 Amino acid sequence of the T cell receptor repertoire of CD8+ T cells in each expanded clone
|
Untreated
|
RD-αPD-1 mAb
|
Comb
|
RD-αPD-1 mAb + Comb
|
Clonotype
|
Clone size
|
Clonotype
|
Clone size
|
Clonotype
|
Clone size
|
Clonotype
|
Clone size
|
1
|
CASSRTGGNQDTQYF**
|
40
|
CASSQTPGTGGYEQYF
|
10
|
CASSQVQGSAETLYF§
|
26
|
CTCSAGPANSDYTF♱
|
47
|
2
|
CASSLELGGREQYF‡
|
35
|
CAWSRQGAHTEVFF
|
9
|
CASSRLGGRGDYAEQFF
|
21
|
CASSPGQANSDYTF
|
47
|
3
|
CASSRTGGEQDTQYF;
CASSRTGGNQDTQYF**
|
17
|
CASRRGQEGTLYF
|
8
|
CASSRTGGNQDTQYF**
|
18
|
CAWSLGGQNTLYF
|
45
|
4
|
CASSQEQGGAETLYF
|
7
|
CASSWDWGVNYAEQFF
|
8
|
CASSRLGPSAETLYF¶
|
13
|
CASSDARAGNTLYF
|
40
|
5
|
CASSRTGGNQDTQYF**
|
6
|
CASGKGWGISAETLYF
|
7
|
CASSQEGANTEVFF
|
9
|
CASSLRENTEVFF
|
31
|
6
|
CASSINWRAETLYF
|
6
|
CAWSPGYNSPLYF
|
7
|
CASSQDWGTSAETLYF
|
7
|
CAWSLGGNSPLYF
|
26
|
7
|
CTCSAGQANSDYTF♱
|
6
|
CASSQVQGVGNTLYF§
|
6
|
CGATGQGSGNTLYF
|
7
|
CASSDAQWSTLYF
|
26
|
8
|
CAWSPQGAGTGQLYF
|
5
|
CASSRRTGVNSDYTF
|
6
|
CASSLELGGPEQYF‡
|
6
|
CASSQVQGGQDTQYF§
|
22
|
9
|
CASSPVGGRQDTQYF
|
5
|
CASSLNPGGTYEQYF
|
6
|
CASSERLGGRQNTLYF
|
6
|
CASSLELGGREQYF‡
|
21
|
10
|
|
|
CASSRLGPSAETLYF¶
|
5
|
|
|
CASSDGTGSTGQLYF
|
16
|
11
|
|
|
CASSIAAQGASGNTLYF
|
5
|
|
|
CASSQPGQNTEVFF
|
13
|
12
|
|
|
CASSQTPGTGGYEQYF
|
10
|
|
|
CASSWDWGGDTQYF
|
11
|
13
|
|
|
CAWSRQGAHTEVFF
|
9
|
|
|
CASSQALGGDTQYF
|
11
|
14
|
|
|
|
|
|
|
CASSQVQGSAETLYF§
|
10
|
15
|
|
|
|
|
|
|
CASSLTGGANQAPLF
|
10
|
16
|
|
|
|
|
|
|
CASSTRGREQYF
|
8
|
17
|
|
|
|
|
|
|
CASSLELGGLEQYF;
CASSPGGASAETLYF‡
|
7
|
18
|
|
|
|
|
|
|
CASSLELGGLEQYF‡
|
7
|
19
|
|
|
|
|
|
|
CAWSRGYNSPLYF
|
6
|
20
|
|
|
|
|
|
|
CASSIRGRWDTQYF
|
6
|
21
|
|
|
|
|
|
|
CTCSAGQANSDYTF♱
|
6
|
22
|
|
|
|
|
|
|
CTCSVTGGMATGQLYF
|
6
|
23
|
|
|
|
|
|
|
CTCSAGRDRAGERLFF
|
5
|
24
|
|
|
|
|
|
|
CASGEPRDFYEQYF
|
5
|
Amino acid sequences of the T cell receptor β-chain with a clone size >5 are shown. Those of the TCR that recognize the same or similar antigens evaluated using GLIPH2 are shown in the same colour fill, and red indicates sites where the amino acid sequences are matched by GLIPH2. Abbreviations: RD-αPD-1 mAb, reduced dose of anti-PD-1 monoclonal antibody; Comb, novel immune drug combination; **, CASSRTGGNQDTQYF; ¶, CASSRLGPSAETLYF; ‡, SLELGG*E; ♱, SAG*ANSD; §, QVQG.