Co-inhibitory T cell receptor KLRG1: human cancer expression and efficacy of neutralization in murine cancer models

Background KLRG1 is a lymphocyte co-inhibitory, or immune checkpoint, receptor expressed predominantly on late-differentiated effector and effector memory CD8+ T and NK cells. Targeting of KLRG1 neutralization in murine cancer models has not previously been reported. Methods We studied KLRG1 expression in human blood and tumor samples from available genomic datasets. Anti-KLRG1 neutralizing antibody was studied in the murine 4T1 breast cancer as monotherapy, and in the MC38 colon cancer and B16F10 melanoma models as combination therapy with anti-PD-1 antibody. Results In human blood and tumor samples, KLRG1 expression is aligned with cytotoxic T and NK cell differentiation, and upregulated in human tumor samples after a variety of therapies, potentially contributing to adaptive resistance. In in vivo murine models, anti-KLRG1 antibody monotherapy in the 4T1 breast cancer model reduced lung metastases (decreased lung weights p=0.04; decreased nodule count p=0.002), while anti-KLRG1 + anti-PD-1 combination therapy in the MC38 colon cancer and B16F10 melanoma models produced synergistic benefit greater than anti-PD-1 alone for tumor volume (MC38 p=0.01; B16F10 p=0.007) and survival (MC38 p=0.02; B16F10 p=0.002). Conclusions These studies provide the first evidence that inhibition of the KLRG1 pathway enhances immune control of cancer in murine models, and provide target validation for KLRG1 targeting of human cancer. The mechanism of efficacy of KLRG1 blockade in murine models remains to be determined.


INTRODUCTION
Killer cell lectin-like receptor G1 (KLRG1) is a co-inhibitory, or immune checkpoint, receptor inhibiting the activity of T and NK cells. It's ligands are E-cadherin and N-cadherin with similar affinities of 7-12 μM [1], respective markers of epithelial and mesenchymal cells [2]. Whereas targeting of other co-inhibitory receptors for applications in oncology has gained widespread interest [3][4][5], including multiple FDA approvals for targeting CTLA-4, PD-1, and its ligand PD-L1, less attention has been focused on the therapeutic potential of KLRG1 modulation. Unlike the obvious enhanced immune activation present in CTLA-4 and PD-1 gene knockout mice [6,7], KLRG1 knockout mice initially were found to have no abnormal features [8], though were subsequently found to have enhanced immunity in a tuberculosis challenge model [9]. No anti-mouse KLRG1 antibody with characterized neutralizing function has been available as a reagent for murine model studies, and no www.oncotarget.com Oncotarget, 2019, Vol. 10, (No. 14), pp: 1399-1406 Research Paper www.oncotarget.com murine cancer studies targeting KLRG1 neutralization have been published. The characterization of KLRG1 as a "senescent" marker, but all other co-inhibitory receptors as "exhaustion" markers [10][11][12], may have contributed to relatively fewer studies of this molecule.
Nevertheless, previous studies have demonstrated that an anti-KLRG1 antibody can increase ex vivo human NK cell interferon-gamma secretion [13] and that anti-Ecadherin antibodies can result in enhanced ex vivo human CD8 T cell proliferation and NK cell cytotoxicity [14][15][16]. Because E-cadherin is also a ligand for the T cell receptor αEβ7 integrin, the effects of anti-E-cadherin antibodies leave uncertain the role of KLRG1 in human CD8 T cell activation. Here, we report on translational studies of human KLRG1 expression and the in vivo activity of an anti-mouse KLRG1 neutralizing antibody in murine cancer models.

KLRG1 is preferentially expressed on effector and effector memory CD8 T cells and NK cells and differentially expressed than PD-1
We mined available gene expression datasets and publications (Supplementary Table 1) to compare human co-inhibitory receptor expression by various blood lymphocyte populations from healthy people. KLRG1 is differentially expressed from CTLA-4 and PD-1, with predominant expression on cytotoxic CD8 T and NK cells over CD4 T cells. Within the CD8+ T cell population, KLRG1 expression, unlike CTLA-4 and PD-1 expression, is linked to greater antigen-driven differentiation states, with increased expression on CD45RO+CCR7-T effector memory (TEM) and CD45RA+CCR7-T effector memory RA (TEMRA) cells compared to CD45RA+CCR7+ naïve T cells (TN) and CD45RO+CCR7+ central memory T cells (TCM) ( Figure 1A, 1B). The cytotoxic potential of CD8+ T cells, as assessed by the presence of cytokine and cytotoxic molecules IFNγ, TNFα, perforin and granzyme B, is aligned with KLRG1, but not CTLA-4 or PD-1, expression ( Figure 1C, 1D).
KLRG1 has been little studied in human tumor samples. Together with additional datasets containing single cell RNA-seq gene expression data from human cancer biopsies, KLRG1+ TILS accounted for 16-48% of CD8+ TILS, a frequency similar to that of PD-1+ TILS, in renal cell carcinoma, hepatocellular carcinoma, melanoma, ovarian cancer, HNSCC, and astrocytoma ( Figure 1E, 1F). A distinct population of PD-1−KLRG1+ infiltrating CD8 T cells accounted for 13-26% of CD8+ TILS across a range of cancer types.
We also studied the expression of the KLRG1 ligands E-cadherin and N-cadherin in tumor sample data. Their transcripts were highly expressed in single cell RNA-seq data of melanoma, prostate, breast, HNSCC, and colorectal cancer cells with expression levels substantially higher than the PD-1 ligand PD-L1 ( Figure  1G-1I). In bulk RNA data across 6,358 cancer samples from 19 different cancer types, E-cadherin and N-cadherin expression were similarly over-expressed compared to PD-L1 ( Figure 1J).

Inhibition of metastasis in the 4T1 breast cancer model with monotherapy
We confirmed that anti-KLRG1 antibody inhibited binding of mouse E-cadherin to KLRG1 (Supplementary Figure 1) and tested its effect on preventing metastasis in the 4T1 metastatic breast cancer model. 4T-1 cells express high levels of E-cadherin (Supplementary Figure 2). Although there was no effect of anti-KLRG1 antibody on primary tumor growth, anti-KLRG1 antibody significantly reduced lung metastases, measured by lung weight (p=0.04) and lung nodule count (p=0.002) (Figure 2A, 2B).

Inhibition of primary tumor growth and improved survival in the MC38 colon cancer and B16F10 models with combination therapy
We tested anti-KLRG1 antibody as a combination therapy with anti-PD-1 antibody ( Figure Figure 2E). Anti-KLRG1 and anti-PD-1 antibody monotherapies had similar efficacy not significantly different from control antibody. Tumor growth inhibition by combination therapy was significantly greater than anti-PD-1 monotherapy (MC38 p=0.01; B16F10 p=0.007) or control antibody treatment (MC38 p=0.002; B16F10 p=0.0001) alone. Combination therapy resulted in a survival benefit compared to anti-PD-1 monotherapy (MC38 p=0.02 and B16F10 p=0.002) or compared to control antibody (MC38 p=0.008 and B16F10 p<0.0001) (Figures 2D, 2F). A tumor-free durable response (alive with complete regression of visible tumor at Day 35) was seen in 10% of combination therapy treated mice and 0% of all other cohorts in both models.

KLRG1 expression is upregulated in human tumor after treatment with therapies that result in T cell proliferation
As KLRG1 expression increases as T cells differentiate in response to antigen stimulation [17], and a variety of cancer therapies are predicted to increase T cell differentiation, either directly (e.g., immunotherapy) or indirectly (e.g., immune activation after tumor cell death in chemotherapy and radiation therapy), we searched for all publicly available gene expression data from paired pre-and post-treatment human tumor samples, identifying 21 datasets (Supplementary Table 2). KLRG1 expression was numerically increased post-treatment in 20/21 (95%) of datasets, statistically significant in 10/21 (48%), in response to a range of treatments including radiotherapy,

DISCUSSION
Extensive preclinical murine cancer models and clinical development efforts have been undertaken for a number of co-inhibitory, or immune checkpoint, receptors that have been characterized as identifying exhausted T cells, including CTLA-4, PD-1, LAG-3, TIM-3, and TIGIT [3][4][5]. In contrast, the co-inhibitory receptor KLRG1 has been viewed as a marker of senescent T cells [10][11][12]. Very little human cancer data examining KLRG1 expression has been published, and no murine cancer models involving neutralization of KLRG1 have been reported.
Here, we illustrate through analyses of public domain data that KLRG1 marks a highly cytotoxic population of CD8 T and NK cells. Within the CD8 T cell population, KLRG1 expression is tied to antigen experience and differentiation status, an observation that has been previously emphasized [18], and aligned with cytotoxic potential, so that KLRG1 marks cells with the greatest cytotoxic potential. Although KLRG1 has infrequently been studied among TILS [18][19][20][21], single cell RNA-seq data indicates abundant KLRG1-expressing TILS across a range of cancer types, and significant numbers (13-26%) of CD8 T cells that do not express PD-1 but do express KLRG1.
Although KLRG1 signaling involves intracellular immunoreceptor tyrosine-based inhibitory motif (ITIM) domains and inhibition of Akt phosphorylation [15] resulting in vitro in T and NK cell inhibition, the in vivo function of KLRG1 has been unknown. The current in vivo KLRG1 neutralization studies confirm that KLRG1 restrains the immune system from cancer defense.
In the MC38 colon cancer and B16F10 melanoma models, monotherapy with KLRG1 was not significantly different than anti-PD-1 therapy, but combination therapy showed significant efficacy, including 10% of mice showing tumor regression and durable cure. Because cancer checkpoint immunotherapies induce T cell proliferation, they are predicted to expand the population of KLRG1+ cells, resulting in a homeostatic checkpoint brake on efficacy contributing to adaptive resistance. In animal models, this expansion of KLRG1+ cells has been observed with anti-CTLA-4 and anti-PD-1 [22], anti-4-1BB [23,24], and HPV vaccine [25] therapies. The increase in KLRG1+ CD8 T cells with anti-PD-1 therapy, a phenomenon also seen in people treated with anti-PD-1 therapy, perhaps contributed to limited efficacy of anti-PD-1 monotherapy treatment in the MC38 model. Further studies using CD8 and NK cell depleted mice could elucidate the mechanism of anti-KLRG1 and anti-PD-1 combination efficacy.
More generally, here we have examined publicly available human cancer gene expression datasets and found that across a wide range of human tumors and therapies, including radiation, endocrine therapy, chemotherapy, and immunotherapy, KLRG1 is upregulated in posttreatment compared to pre-treatment tumor biopsies. These data suggest that the upregulation of the inhibitory checkpoint receptor KLRG1 could contribute to limited efficacy and adaptive resistance that develops with current immunotherapies [26], and suggests KLRG1 blockade may work efficaciously as a neo-adjuvant therapy.
Lastly, in vitro studies have demonstrated that KLRG1 is less inhibitory in mice than in humans, due to different amino acids at position 62 of its stalk region resulting in formation of KLRG1 monomers and oligomers in mice, but only dimers in humans [27]. This observation suggests that the inhibitory role of human KLRG1 in restraining anti-tumor responses is underestimated in mouse studies, and that anti-KLRG1 neutralizing antibody therapy could demonstrate substantial efficacy in people.

Genomic datasets
All identified published flow cytometry data of CTLA-4, PD-1, and KLRG1 expression in human normal blood was compiled from publication tables and figures after literature searches were used to attempt to identify all available data sources (Supplementary Table  1). Bulk blood and tumor gene expression datasets were searched for KLRG1 and other relevant gene expression data, identifying data from the European Bioinformatics Institute ArrayExpress and the National Institutes of Health Gene Expression Omnibus (GEO) databases (Supplementary Table 2

4T1, MC38, and B16F10 models
For the 4T1 model, 20 female 6-8 week old BALB/c mice were inoculated subcutaneously with 1×10 5 4T1 cells suspended in 50μL RPMI 1640. Ten animals each were assigned into either of 2 groups, anti-KLRG1 or control antibody 10 mg/kg intraperitoneal injections on Days 1,4,7, and 11. For the MC38 and B16F10 models, 80 female 6-8 week old C57BL/6 mice were inoculated subcutaneously with 1×10 6 MC38 cells (N=40 mice) or 1×10 6 B16F10 cells (N=40 mice) suspended in 100μL DMEM. In each tumor model, ten animals each were assigned into 4 groups, control antibody (10 mg/kg), anti-KLRG1 (10 mg/kg), anti-PD-1 (5 mg/kg) (suboptimal dosage chosen to detect a combination effect), or anti-KLRG1 + anti-PD-1 on Day 6, 9, 12, 16 in the MC38 model and Day 5,8,11, and 15 in the B16F10 model. Randomized block designs based upon body weight and order of inoculation were used. Animals were sacrificed within 1 hour of when examined and found to have tumor volume of at least 2000 mm 3 and, for 4T1, lungs examined for metastases, and for the MC38 anti-PD-1 treated cohort, blood sampled for flow cytometry detection of KLRG1 expression using ABC-m01. Animals were monitored daily and for tumor growth measured 3 times per week.

Author contributions
SAG, ET, and SVG analyzed and interpreted all data. SK performed bioinformatics analyses. All authors read and approved the final manuscript

DECLARATIONS Ethics approval and consent to participate
Animal studies have been conducted in accordance with the ChemPartner Institutional Animal Care and Use Committee (work performed by ChemPartner, Shanghai, China).

Availability of data and material
The datasets analysed during the current study are available in the GEO repository with accession numbers indicated in the Supplementary Table 2.

CONFLICTS OF INTEREST
ET and SVG are employees of Abcuro, Inc. SAG is a consultant to Abcuro, Inc. SAG, ET, and SVG are inventors on intellectual property owned and managed by Brigham and Women's Hospital, and own equity in Abcuro, Inc.

FUNDING
No grant support to any authors; funding for murine models from Abcuro, Inc.

Editorial note
This paper has been accepted based in part on peerreview conducted by another journal and the authors' response and revisions as well as expedited peer-review in Oncotarget.