Modulation of Gut Microbiota: A Novel Paradigm of Enhancing the Efficacy of Programmed Death-1 and Programmed Death Ligand-1 Blockade Therapy

Blockade of programmed death 1 (PD-1) protein and its ligand programmed death ligand 1 (PD-L1) has been used as cancer immunotherapy in recent years, with the blockade of PD-1 being more widely used than blockade of PD-L1. PD-1 and PD-L1 blockade therapy showed benefits in patients with various types of cancer; however, such beneficial effects were seen only in a subgroup of patients. Improving the efficacy of PD-1 and PD-L1 blockade therapy is clearly needed. In this review, we summarize the recent studies on the effects of gut microbiota on PD-1 and PD-L1 blockade and discuss the new perspectives on improving efficacy of PD-1 and PD-L1 blockade therapy in cancer treatment through modulating gut microbiota. We also discuss the possibility that chronic infections or inflammation may impact on PD-1 and PD-L1 blockade therapy.


PD-1 AND PD-L1 BLOCKADe iN CANCeR iMMUNOTHeRAPY
In anti-tumor immune responses, the tumor antigens generated by gene mutations, are recognized by the immune system and specific CD8+ CTLs targeting tumor antigens are generated (22). These specific effector CTLs recognize the target tumor cells and induce tumor cell apoptosis.
The interaction of PD-L1 with PD-1 in the tumor microenviro nment enables the tumor cells to resist the endogenous anti-tumor activities from the immune system. PD-L1 expressed in tumor tissues interacting with PD-1 expressed on the activated T cells leads to the dysfunction of the effector T cells, via multiple mechanisms, such as promoting T cell apoptosis, anergy, and exhaustion (6,7,9,(16)(17)(18)(19)(20)(21). More recently, it was found that interaction of PD-L1 with PD-1 expressed on tumor-associated macrophages inhibits the phagocytic potency of macrophages against tumor cells (33). The importance of PD-L1 and PD-1 interaction in tumor cell evasion has led scientists to explore the use of these molecules as therapeutic targets in cancer immunotherapy (Figure 1). Gut Microbiota and PD-1/PD-L1 Blockade Frontiers in Immunology | www.frontiersin.org March 2018 | Volume 9 | Article 374 Dong et al. showed that PD-L1 positive human tumor cells induced apoptosis of co-cultured activated effector T cells and this effect was blocked by an anti-human PD-L1 monoclonal antibody (mAb). They also showed that the growth of PD-L1 positive murine tumors in syngeneic mice was suppressed by an anti-murine PD-L1 mAb (9). Other researchers later reported similar findings in examination of different types of cancer cells using mice models (24,(34)(35)(36). These important laboratory observations led to numerous clinical trials of using monoclonal antibodies targeting PD-1 or PD-L1 in cancer immunotherapy for a variety of cancers. In addition to affecting the immunological pathways, PD-L1 and PD-1 blockade may also work in part by disrupting autologous PD-1 and PD-L1 signaling within tumors (37,38).
To date, the U.S. Food and Drug Administration (FDA) has approved the use of five monoclonal antibodies targeting PD-L1 or PD-1 in cancer treatment. The details of the clinical trials of these five monoclonal antibodies are summarized in Table 1.
Despite the clear benefits of PD-L1 and PD-1 blockade in treating some cancer patients, not all cases responded to treatment ( Table 1). Given this, strategies to improve the efficacy of cancer immunotherapy are needed. Emerging evidence suggests that modulation of the gut microbiota is a promising approach.

MODULATiON OF GUT MiCROBiOTA eNHANCeS THe ANTi-TUMOR eFFiCACY OF PD-1 AND PD-L1 BLOCKADe THeRAPY
A very interesting study by Sivan et al. provided strong evidence that the efficacy of PD-L1 blockage therapy can be improved by the modulation of gut microbiota (70). In this study, Sivan et al. examined the subcutaneous growth of B16.SIY melanoma in genetically similar C57BL/6 mice raised in the Jackson Laboratory (JAX) and Taconic Farms (TAC), and found that the tumor growth was more aggressive in TAC mice as compared to that in JAX mice and that TAC mice had a significantly lower intratumoral CD8+ T cell accumulation. They then conducted various experiments, which demonstrated that gut microbiota contributed to this difference.
They first showed that prophylactic transfer of fecal material from JAX mice to TAC mice was sufficient to delay tumor growth. To examine whether microbial community alone was effective as a therapy, they administered fecal material from JAX mice alone or in combination with anti-PD-L1 mAbs to TAC mice. These experiments showed that fecal material alone was sufficient to significantly inhibit tumor growth and that the combination treatment further improved tumor control. To identify the responsible bacterial species, they used 16S ribosomal RNA (16S rRNA) sequencing and identified Bifidobacterium species, particularly Bifidobacterium breve, Bifidobacterium longum, and Bifidobacterium adolescentis as the candidate species. The role of these Bifidobacterium species in enhancing protective immunity against tumors were further investigated by administering TAC mice bearing established tumors with a cocktail of Bifidobacterium species containing B. breve and B. longum by oral gavage. This experiment resulted in Bifidobacterium-treated mice having significantly improved tumor control as compared to mice that did not receive Bifidobacterium. Sivan et al. also showed that the possible mechanisms by which Bifidobacterium species inhibited tumor growth were through activating DCs, which in turn, improves the effector function of tumor-specific CD8+ T cells. Given that the enhanced anti-melanoma effect from Bifidobacterium species had occurred at the innate immunity level, the authors anticipated that Bifidobacterium species also provide anti-tumor beneficial effects to other types of tumors. However, the mechanisms by which Bifidobacterium species activated DCs improved the effects of anti-tumor CD8+ cells still need to be clarified.
The findings by Sivan et al. using mice models suggest that it is possible to enhance the anti-tumor efficacy of PD-L1 blockade therapy in treating cancer patients by modulating their gut microbiota and their findings are summarized in Figure 2.
Interestingly, a very recent study by Matson et al. examining the stool samples collected from patients with metastatic melanoma before anti-PD-1 immunotherapy found that B. longum, Collinsella aerofaciens, and Enterococcus faecium were more abundant in the anti-PD-1 immunotherapy responders, supporting the anti-tumor effects of Bifidobacterium species (71).
Several additional studies also compared the gut microbiota in patients with metastatic melanoma receiving anti-PD-1 therapy. A recent study by Frankel et al. using metagenomic shotgun sequencing method showed that melanoma patients who responded to immune checkpoint inhibitors were enriched with Bacteroides caccae (72). Furthermore, they showed that the bacteria that are enriched within responders are most likely to be antibody dependent. Patients who responded to nivolumab (PD-1 antibody) were enriched with Fecalibacterium prausnitzii, Bacteroides thetaiotamicron, and Holdemania filiformis, whereas patients who responded to pembrolizumab (another PD-1 antibody), their gut microbiota enriched with Dorea formicogenerans. However, the mechanisms responsible for these changes are not clear. Studies comparing the gut microbiota changes prior to and following anti-PD-1 therapy of individual patients are required, which will provide information regarding whether anti-PD-1 antibodies directly affect gut bacterial species.
A study by Wargo et al. examined the human gut microbiota and metabolites of metastatic melanoma patients who received anti-PD-1 therapy using 16S rRNA and whole genome shotgun sequencing (73). They found that bacterial diversity and composition in patients that responded to the therapy were significantly different from that in patients who did not respond to the therapy. The responding patients had a higher diversity of bacteria and a higher abundance of Clostridiales, and the non-responders had a higher abundance of Bacteroidales. In a very recent study with multiple first authors and J. A. Wargo being the responding author, they further compared the gut microbiota of patients with metastatic melanoma receiving anti-PD-1 therapy (74). They found that patients who responded to anti-PD-1 therapy were associated with a significantly higher bacterial diversity and abundance of bacteria from the Ruminococcaceae family, which belongs to the Clostridiales order, as compared to patients who did not respond to the therapy. Furthermore, they performed fecal microbiota transplantation experiments in germ-free mice,    in which they showed that germ-free mice transplanted with stool samples from patients who responded to anti-PD-1 and anti-PD-L1 therapy had a significantly reduced tumor growth and improved responses to anti-PD-1 and anti-PD-L1 therapy, coupled with a higher density of CD8+ T cells. However, it is not clear which bacterial species in the Ruminococcaceae family has played the role in enhancing the PD-1 blockade therapy. Another recent study by Routy et al. investigated the effects of gut microbiota in PD-1 blockade therapy (75). In their study, data from 140 patients with advanced non-small-cell-lung cancer, 67 patients with renal cell carcinoma, and 42 patients with urothelial carcinoma were collected, and they found that 69 patients who took antibiotics before or soon after starting the PD-1 blockade therapy had shorter progression-free survival and overall survival. They then explored the composition of the gut microbiota using shotgun sequencing, which showed that Akkermansia muciniphila was enriched in patients who responded to anti-PD-1 therapy. This suggests that A. muciniphila may enhance patient response to PD-1 blockade therapy. They verified this observation by transplanting the patients stool samples in specific pathogen-free mice or germ-free mice and observed tumor growth in these mice. They also found that A. muciniphila alone was able to restore the anti-tumor effects of PD-1 blockade that was inhibited by antibiotics. However, the mechanism by which A. muciniphila enhancing PD-1 blockade therapy is not known.
Bacterial species that are positively associated with PD-1 and PD-L1 blockade therapy are summarized in Table 2. Some bacterial species have also been demonstrated to affect CTLA-4 blockade immunotherapy, which were not reviewed here (76,77).

POTeNTiAL MeCHANiSMS OF GUT MiCROBeS ON iMPROviNG THe eFFiCACY OF PD-1 AND PD-L1 BLOCKADe THeRAPY
Despite the exciting findings in this research field, the underlying molecular mechanisms by which the identified gut bacterial species in the above studies enhance PD-1 and PD-L1 blockade therapy remain largely unknown.
Recently, unmethylated CpG oligodeoxynucleotides, which are abundant in bacterial DNA, were found to enhance CD8+ T cell anti-tumor immunity by downregulating PD-1 expression via the IL-12 pathway, suggesting that gut bacterial species that

Akkermansia muciniphila Mouse Human
Metagenomic shotgun sequencing Fecal microbiota transplantation Immunohistochemistry Flow cytometry Cytokines assay • 27% cancer patients b who took antibiotics before or soon after starting the PD-1 blockade therapy had shorter progression-free survival and overall survival • A. muciniphila was found enriched in those patients who respond to anti-PD-1 therapy • A. muciniphila alone was able to restore the anti-tumor effects of PD-1 blockade that was inhibited by antibiotics.

Modulation of gut microbiota
Explore the mechanisms of the anti-tumor effects of Bifidobacterium species Identify the gut bacteria that are positively associated with PD-1 and PD-L1 blockade therapy in humans at species and strain level and understand their anti-tumor mechanisms Examine the impact of chronic inflammation on PD-1 and PD-L1 blockade therapy and develop treatment strategies accordingly Enhance the efficacy of PD-1 and PD-L1 blockade therapy are positively associated with PD-1 and PD-L1 blockade therapy may release components that directly downregulate PD-1 or PD-L1 expression (78,79).
It is also possible that the gut bacterial species indirectly affect PD-1 and PD-L1 expression through locally or systematically regulating immune responses, thereby affecting the efficacy of PD-1 and PD-L1 blockade therapy. Gut microbiota has been shown to impact on both innate and adaptive immune cells. Germ-free animals had a reduced number of intestinal DCs and administration of Escherichia coli in these animals was able to recruit sufficient DCs to the intestines (80,81). In Germ-free pigs, systemic circulating macrophages were also reduced and their functions were compromised (82). Germ-free mice had markedly decreased presence of lamina propria CD4+ T cells and absence of lymphocyte zones in spleens and mesenteric lymph nodes (83,84). Polysaccharide A from Bacteroides fragilis was found to induce the Th1 response (83). Reduction of commensal microbiota in mice by using broad-spectrum antibiotics resulted in defective T and B cell responses against influenza infection (85). The findings that gut microbes can affect the immune functions, both locally and systematically suggest that bacterial species positively associated with PD-1 and PD-L1 blockade may enhance PD-1 and PD-L1 immunotherapy through regulation of the immune response. The previous study by Sivan et al. showed that Bifidobacterium species that inhibited tumor growth activated DCs, further supporting this view (70).

THe POSSiBLe iMPACT OF CHRONiC iNFeCTiONS AND iNFLAMMATiON ON PD-1 AND PD-L1 BLOCKADe THeRAPY
Several microbes cause chronic infections in humans, some of which are known to increase host PD-1 and PD-L1 expression (86)(87)(88)(89)(90)(91)(92)(93)(94). However, studies have not examined whether existing chronic infections in patients with cancer affect the efficacy of PD-1 and PD-L1 blockade therapy.
An example of a chronic infection is Helicobacter pylori infection. H. pylori are a Gram-negative bacterium that colonizes the stomach of more than 50% of the world population. While most of the individuals colonized with H. pylori are asymptomatic, some may develop chronic gastritis and peptic ulcers, and H. pylori colonization is also a risk factor for gastric cancer (95). Previous studies have shown that patients with H. pylori infection have a significantly higher level of pro-inflammatory cytokines, such as TNF-α (96)(97)(98). Das et al. showed that H. pylori increased the gastric epithelial expression of PD-L1 using a gastric epithelial cell line model (86). Furthermore, they showed that gastric epithelial cells exposed to H. pylori inhibited the proliferation of CD4+ T cells isolated from blood and the inhibitory effect can be blocked using antibodies PD-L1. Similarly, Wu et al. observed increased PD-L1 expression in gastric biopsies of individuals infected with H. pylori, and co-culture of H. pylori infected primary gastric epithelial cells with T cells isolated from blood induced T cell apoptosis (87). These results suggest that H. pylori infection may cause the non-specific inhibition of circulating T cells, including tumor-specific T cells. In addition to H. pylori, several viruses, such as the hepatitis B virus, hepatitis C virus, human papillomavirus, and Epstein-Barr virus are also able to establish chronic infections in humans and increase host PD-1 or PD-L1 expression (88)(89)(90)(91)(92)(93)(94). Future studies are needed to examine whether chronic infections or inflammation impact on the efficacy of PD-1 and PD-L1 blockade. A recent study by Kottke et al. using a mouse model showed that pro-inflammatory cytokine TNF-α promoted tumor recurrence, while TNF-α blockade prevented tumor recurrence (99)(100)(101)(102). Some bacterial species that are known to reduce chronic inflammation after administration orally may be examined to see whether they can improve cancer treatment (103-108). If chronic infections or inflammation reduce the efficacy of PD-1 and PD-L1 blockade, it would be through mechanisms other than the induction of the PD-1 and PD-L1 expression in the tumor tissues, as previous studies observed better responses to PD-1 blockade in patients with higher expression of PD-L1 in tumor tissues (51).

FUTURe DiReCTiONS
As discussed, despite the clear benefits of PD-1 and PD-L1 blockade in treating some cancer patients, the efficacy and the recurrence of tumor are issues that remain to be tackled. Emerging evidence suggests that modulation of the gut microbiota is a promising approach for improving PD-L1 and PD-1 blockade therapy. However, future studies are needed to further develop this research area.
The Bifidobacterium species, particularly B. longum, increased anti-PD-L1 efficacy in mice models and was positively associated with anti-PD-1 efficacy in metastatic melanoma patients. Future studies are needed to understand the molecular mechanisms of these Bifidobacterium species in enhancing PD-1 and PD-L1 blockade therapy. In addition to the Bifidobacterium species, various studies reported positive associations of gut microbes with PD-1 and PD-L1 blockade therapy at genus level. These microbes need to be identified at species and strain level and their potential anti-tumor mechanisms require further investigation.
Several bacterial and viral pathogens are known to cause chronic human infections and the pro-inflammatory cytokines are known to induce host PD-1 and PD-L1 expression. In addition, some of these pathogens are known to directly attach immune cells. Whether chronic infections caused by different pathogens impact on PD-1 and PD-L1 blockade therapy should be investigated, and appropriate strategies to enhance PD-1 and PD-L1 blockade therapy in these patients can then be developed accordingly. A suggested course of action is outlined in Table 3.

AUTHOR CONTRiBUTiONS
Wrote the paper: YW, LZ. Figures and tables: YW. Revised the paper: YW, LZ, RM, FL, SL. All authors have approved the final version of the manuscript.