3.1. DFB suppress rapid growth of CRC in DIO mice by restricting terminal Tex
DFB, comprised of five commonly used herbs: Radix et Rhizoma Rhei (root or rhizome of perennial herbaceous plant Rheum palmatum L. Gansu, China), Radix Aconiti Lateralis Praeparata (root of perennial herbaceous plant Acontium carmichaeli Dehx. Sichuan, China), Herba Asari (root of Asarum sieboldii Miq. Liaoning, China), Semen Coicis (nut of Coix lachrhryma-jobi L.var.mayuen(Roman)Stapf. Guizhou, China), Herba Patriniae (entire plants of Patrinia scabiosaefolia Fisch. ex Link. and P. villose Juss. Hubei, China). The formula and amount of the prescription were listed in Table S2.
The raw herbs for DFB were purchased from Beijing Tongrentang Co. Ltd., China. The voucher specimens were deposited in the storage cabinet of Chinese traditional medicine of School of Traditional Chinese Medicine, Southern Medical University. A voucher specimen was deposited in the Herbarium of school of traditional Chinese medicine. These were mixed in the ratio of 12:6:30:15:3 (dry weight). Aqueous extracts of DFB were extracted at 80℃ by stirring it for 1 h using 10 volumes of distilled water (v/m). Then, we centrifuged the extract at 1,500×g at room temperature. To obtain the semisolid DFB solution, the supernatant was collected and subjected to condensation under reduced pressure of 70℃. The quality of DFB was controlled by HPLC analysis (Figure S2, Table S3). It was performed on Shimadzu LC-20A high-performance liquid chromatographic (HPLC) system (Shimadzu Co. Kyoto, Japan) by using a C18 column Shimadzu VP ODS (250 × 4.6 mm; particle size 5 µm, Japan). The mobile phases comprised eluent A (water) and eluent B (methanol). The gradient flow was as follows: 0.00–125.00 min, 20–90% B. The analysis was performed at a flow rate of 1.0 mL/min with PDA detection at 254 nm. The injection volume was 20 µL. Determination of Rhein(Pubchem CID:10168), Fuziline(Pubchem CID:14163819), Asarinin(Pubchem CID:11869417), Coixol(Pubchem CID:10772), Quercetin(Pubchem CID:5280343) in DFB sample(Figure S1): as the five active compounds in Radix et Rhizoma Rhei, Radix Aconiti Lateralis Praeparata, Herba Asari, Semen Coicis and Herba Patriniae respectively.
To explore the suitable dosage of DFB, the size of murine colon adenocarcinoma (MC38-luciferase) (Supplementary Figure 3)xenografts was recorded every other day༈Supplementary Figure 4A,C༉. Both middle dosage and high dosage significantly restricted tumor growth, whereas the middle dosage showed better performance in regressing growth kinetics༈Supplementary Figure 4D,E༉. Hence, DFB middle dosage was chosen for subsequent research because of the best curative efficacy among the three different doses.
High fat diet (HFD)-fed obese mice closely mimicks the natural development of human obesity(Supplementary Table 1). We then monitored tumor growth in diet-induced obese mice (DIO, 60%-fat diet) versus control-diet mice (control, standard chow diet, SCD ). DIO mice had a obvious increase of body weight at the end of 8 weeks༈Figure 1B༉. MC38 adenocarcinoma grew significantly faster in DIO mice as compared with control counterparts༈Figure 1A༉. DFB enormously limited the growth of MC38 tumors as demonstrated by luciferase-based biochemiluminescence imaging༈Figure 1C༉, suggested that DFB is highly effective in suppressing tumor growth even under a protumoral obese microenvironment.
We next studied the function of DFB on T cell phenotype in the obese microenvironment(Supplementary Figure 6A,B). Obesity slightly reduced the ratio of tumor-infiltrating CD8+ T cells (CD8+ TILs) in the tumor bed, whereas DFB significantly increased the recruitment of CD8+ TILs༈Figure 1D,F༉. Specifically, HFD resulted in a markedly higher frequency PD-1hiTim3+ CD8+ TILs concomitant with an obvious decrease of PD-1intTim3− subset as compared to control mice༈Figure 1E,G,H༉.PD-1hiTim3+subset is generally defined as terminally differentiated exhausted T cells whereas intermediate expression of PD-1 (PD-1int) and co-existence of TCF-1 confer Tex with stem cell-like properties [9, 12]. DFB increased the number of PD-1int subset and lowered PD-1hiTim3+ cells in the tumor. Taken together, DFB suppresses HFD-accelerated tumor progression by reserving PD-1int subset and decreasing Tex counts.
3.2 TOX and TCF1 control Tex differentiation status
Terminally differentiated Tex is the driving force to form immunosuppressive microenvironment, whereas progenitor Tex subset is critical for generating response to PD-1 blockade [13–15]. To characterize the differentiation state of the two subsets, we sorted and analyzed transcription activity of TCF1 and thymocyte selection-associated high mobility group box (TOX)(Supplementary Figure 6C). PD-1intTIM3− cells are characterized by strong expression of TCF1 which is absent in PD-1hiTim3+ cells༈Figure 2A,C༉. These TCF1+PD-1int progenitor cells can expand and yield terminal exhausted PD-1hiTim3+ cells. We observed a highly expression of TOX in the PD-1hiTim3+ subset༈Figure 2A,D༉.TOX transcriptionally activates co-inhibitory receptor gene Pdcd1 and stimulates the transcription of Tcf7 which is the encoding gene for TCF1. We also noticed a overlap of TOX and TCF1 expression in TCF1+PD-1int subset, suggested that TOX may act upstream of TCF1 to drive the differentiation from progenitor to terminal Tex.
As Tex progressively lose its effector function in a hierarchical pattern, we examined the alterations of effector cytokines with an ex vivo experiment. The sorted TCF1+PD-1int and PD-1hiTim3+ cells were stimulated with CD3 and CD28 antibody (Supplementary Figure 6C). TCF1+PD-1int cells displayed robust ability to produce TNFɑ and IFNγ. However, PD-1hiTim3+ exhibited defects in generating IFN-γ while showed residue competence to produce TNFɑ༈Figure 2E,F༉. Loss of IFNγ has been considered occurring at the late stage of exhaustion [12], our results suggested that the cytotoxity defect in PD-1hiTim3+ cells is corresponding to the terminally exhausted state༈Figure 2E,G༉. These distinct fingerprints of Tex suggested that TOX and TCF1 drive the evolution between progenitor and terminal Tex which determine the exhausted status and their response to checkpoint blockade.
3.3 DFB reserves progenitor Tex to elicit immune response to αPD-1
As PD-1intTCF-1+ progenitor TILs are poised to respond to PD-1 blockade, we reasoned that DFB might coordinate anti-PD-1 (αPD-1) response by modulating PD-1 expression in CD8+ T cells. Ob/ob mice carry a spontaneous mutation at the leptin locus which lead to obesity, hyperglycemia, and elevated plasma insulin. Ob/ob mice exhibited visible obese with elevated body weight and abdominal circumference(Supplementary Figure 7A,B). Mirroring the protumorigenic activity of DIO, ob/ob phenotype drastically accelerated MC38 tumor growth as compared to its corresponding C57BL/6J background༈Supplementary Figure 6A,C༉. DFB not only moderately attenuated tumor growth, but displayed prominent cooperative efficacy in limiting tumor size in ob/ob mice༈Supplementary Figure 3B,C༉.
To investigate whether reduced growth rates of tumors by DFB was due to control by T cells, we re-genotyped the T cells in the tumor with PD-1 and TIM3(Supplementary Figure 6A,B). DFB augmented the number of CD8+ T cells and synergized the effects of αPD-1༈Figure 3D,F༉. As expected, ob/ob mice increased the percentage of PD-1hiTim3+ cells, corresponding to an reduction of of PD-1int cells༈Figure 3E,G,H༉. DFB substantially decreased the ratio of terminal PD-1hiTim3+ Tex to ameliorate the immunosuppressive condition. It also increased the number of PD-1int T cells that serve to respond to αPD-1 immunotherapy. Together, DFB curbs the tumor growth by limiting PD-1hiTim3+ subset and amplifying PD-1int population to enhance the efficacy of PD-1 checkpoint blockade.
3.4 DFB reduces tumor interstitial and serum CCL2 production
Obesity has been demonstrated to induce a chronic, low-grade inflammatory status with increased adipokines, such as IL-6, TGF-β, leptin, TNF-α, as well as elevated levels of CCL2. Ob/ob genotype elevated interstitial IL-6 production, while had no obvious impact on TGF-β secretion as compared to age-matched wild type mice(Figure 4A,D). DFB substantially depressed tumor interstitial IL-6 content whereas displayed no synergistic effects with αPD-1 on IL-6 level༈Figure 4B,E༉.
CCL2, which is symbolically presented at high levels in obesity, gives rise to immune evasion through PD-1 signaling [16]. We first accessed the effects of DFB on CCL2 in ob/ob mice. DFB reversed obesity-raised stromal CCL2 content. Although, αPD-1 alone can not lowered CCL2 secretion, DFB helped αPD-1 inhibitor to reduce stromal CCL2 level(Figure 4C,F). We next evaluated the impact of DFB in DIO mice and observed a remarkably similar pattern. DFB attenuated interstitial CCL2 content in HFD feeding obese mice. Additionally, we also measured the alteration of serum CCL2 contents. As expected, either DFB and αPD-1, and the two together decreased serum CCL2 elevation which is induce by obese background. Collectively, DFB dampens obesity-driven tumor interstitial and serum CCL2 production both in DIO and transgenic obese mice.
3.5 CCR2 inhibitor blocks shift from progenitor Tex to terminal Tex
As elevated CCL2 expression drives PD-1 inhibitor-resistance [17] and CCR2 antagonism enhances tumor response to αPD-1 monotherapy [18], we then tested if CCL2-CCR2 axis contributes to Tex differentiation with a specific inhibitor CCR2-RA-[R]. As we saw earlier, MC38 tumor growth was accelerated in ob/ob mice compared with control C57BL6/J mice(Figure 5A,B). CCR2 inhibitor restrained tumor growth and increased CD8+ T cell infiltration༈Figure 5C,E༉.
We also observed an reduction of progenitor PD-1int number in ob/ob mice as compared to their wild type counterparts. Importantly, CCR2 inhibitor enhanced the percentage of progenitor PD-1int cells, and also expanded this subset when combined with αPD-1 immunotherapy(Figure 5D,F,G).Corresponding to its protumor phenotype, ob/ob mice also increased the ratio of terminal PD-1hiTIM3+ CD8+ TILs, whereas CCR2 inhibitor reduced the counts of this subset. Hence, obese enhance tumor interstitial CCL2 to facilitate the differentiation from progenitor to terminal Tex. DFB exerts it antitumor activity through disturbing CCL2/CCR2 mediated terminal T cell exhaustion [18]. Altogether, DFB decreases obesity-driven CCL2 secretion and reverses the Tex differentiation shift to regress CRC progression.