mDKN-01, a Novel Anti-DKK1 mAb, Enhances Innate Immune Responses in the Tumor Microenvironment

The Wnt signaling pathway is involved in embryonic development and adult tissue homeostasis as well as the regulation of stem cell maintenance, cell proliferation, and cell survival (1–4). Wnt signaling is typically identified by two main pathways, one β-catenin–dependent (canonical) and the other β-catenin–independent (noncanonical; refs. 2, 5). These pathways are active in a variety of tumor types (6, 7). In select cases, inhibition of components of the Wnt pathway results in antitumor effects in preclinical studies and has shown promise in early-stage clinical trials (8). An inhibitor of β-catenin–dependent Wnt signaling, Dickkopf-1 (DKK1), binds low-density lipoprotein receptor–related protein 5/6 (LRP5/6) receptors and blocks Wnt binding, thereby resulting in the degradation of β-catenin (9–12). While stabilization of β-catenin can result in transcription of many oncogenes, contributing to carcinogenesis and tumor progression, several studies have demonstrated that DKK1 in itself can function to promote tumor growth and metastasis, and that targeting DKK1 can have clinical benefit (13–15). In addition, in several tumor types, overexpression of DKK1 is associated with worse clinical outcomes (13). Recently, DKK1 has been shown to promote tumor growth through a cytoskeleton-associated protein 4 (CKAP4)-AKT signaling pathway further supporting a model where DKK1 has oncogenic activity (16).

DKK1 has immunomodulatory properties. Reductions in β-catenin expression in myeloid cells drive myeloid-derived suppressor cell (MDSC) accumulation in bone marrow, spleen, and tumors (17). In lung and melanoma tumor models, DKK1, through inhibition of β-catenin signaling in MDSCs, increased MDSC suppressive activity on T-cell activation and proliferation, effects that were reversed with an anti-DKK1 antibody (18). In studies with human lung and breast cancer xenograft models, human latency-competent cancer cells with tumor-initiating capabilities and metastatic potential downregulated NK-activating ligands in a DKK1-dependent fashion, thereby avoiding NK-mediated clearance (19). Furthermore, the DKK1 homolog DKK2, has immunomodulatory activity by suppressing activation of cytotoxic NK and CD8 cells, and DKK2 neutralization reduces tumor growth in the MC38 colon cancer model (20).

A trial of DKN-01 plus pembrolizumab has recently been completed in patients with gastroesophageal junction (GEJ) cancer with promising results (NCT02013154; ref. 21). To better understand the DKN-01 mechanism of action (MOA), we have engineered a murine framework version of DKN-01, mDKN-01, and examined its efficacy in mouse models of melanoma and breast cancer. We show here that targeting DKK1 suppresses B16F0 melanoma growth in an NK-dependent manner and inhibits pulmonary metastases in the 4T1 breast cancer model. Finally, the mDKN-01/anti-PD-1 combination was more effective at inhibiting melanoma growth than mDKN-01 alone.

Use of mice in this investigation conforms with the guidelines of the Committee on Animals of SmartLabs as well as with the Guide for the Care and Use of Laboratory Animals published by the U.S. NIH (Publication No. 85-23). Female C57BL/6J (IMSR, catalog no. JAX:000664, RRID:IMSR_JAX:000664), RAG1^−/− (B6.129S7-Rag1^tm1Mom/J; IMSR catalog no. JAX:002216, RRID:IMSR_JAX: 002216), NSG (NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ; IMSR, catalog no. JAX:005557, RRID:IMSR_JAX:005557), and BALBc/J mice (IMSR, catalog no. JAX:000651, RRID:IMSR_JAX:000651; 6–8 weeks of age) were purchased from the Jackson Laboratory. B16F0 (ATCC, catalog no. CRL-6322, RRID:CVCL_0604) and 4T1 (ATCC, catalog no. CRL-2539, RRID:CVCL_0125) cells were obtained from ATCC and cultured as recommended. Upon receipt from ATCC, all cell lines were expanded and frozen in aliquots without further Mycoplasma testing. A fresh aliquot of each frozen cell line was thawed and expanded (passage number ≤ 5) for each individual experiment. For all experiments, mice were inoculated subcutaneously in the rear flank with B16F0 mouse melanoma or 4T1 cells on day 0. The following day, twice weekly intraperitoneal treatment of mDKN-01 (Absolute Antibody) or isotype (IgG2a) control (Bio X Cell, catalog no. BE0085, RRID:AB_1107771) was initiated at 10 mg/kg. Anti-mouse PD-1 mAb was injected twice weekly i.p. at 12.5 mg/kg. Tumor measurements were performed every 2 or 3 days with calipers and volume (V) was calculated using the following formula: V = ½[length (mm) × width (mm)²].

mDKN-01 was produced by grafting complementary determining regions of DKN-01 (IgG4/kappa), with minor modifications, onto a murine IgG2a/kappa construct with a D265A substitution to abrogate FcR interactions (22). Murine IgG2a was used as the isotype control (Bio X Cell, catalog no. BE0085, RRID:AB_1107771). The anti-mouse PD-1 (Bio X Cell, catalog no. BE0273, RRID:AB_2687796) and anti-mouse NK1.1 (Bio X Cell, catalog no. BE0036, RRID:AB_1107737) antibodies were purchased from Bio X Cell along with the suggested isotype controls (Bio X Cell, catalog no. BE0089, RRID:AB_1107769 and Bio X Cell, catalog no. BE0085, RRID:AB_1107771, respectively).

Mice were sacrificed and tumors and spleen were removed. Spleens were disaggregated by rubbing between the ends of two frosted glass slides. Cells were passed through a 70 μm cell strainer (Falcon) and transferred to 50 mL tubes with PBS +5% FBS and washed. Red blood cells were lysed with red blood cell lysing solution (Sigma-Aldrich). Tumors were minced and disaggregated in a solution of 3 mg/mL collagenase A (Roche) and 50 U/mL DNAse1 (Sigma-Aldrich) in serum-free media for 60 minutes at 37°C. Cells were passed through 40 μm cell strainer (Falcon) with PBS +5% FBS and washed. Red blood cells were lysed with red blood cell lysing solution (Sigma-Aldrich). Cells were washed with PBS + 5% FBS. Whole blood was collected into Microtainer Blood Collection Tubes with K2 EDTA (BD Biosciences). Whole blood (500 μL) was then placed into 15 mL polypropylene tubes and 2 mL of ACK Lysis Buffer (Thermo Fisher Scientific) was added and incubated for 3 minutes at room temperature. PBS +5% FBS (10 mL) was added and samples spun at 300 × g for 5 minutes. Cells were resuspended in PBS + 5% FBS and stained with the following anti-mouse antibodies: Allophycocyanin (APC)-conjugated anti-Gr-1 (BioLegend, catalog no. 108419, RRID:AB_493480), Phycoerythrin: Cy-5.5 Tandem Conjugate (PE Cy5.5) anti-CD11b (BioLegend, catalog no. 101207, RRID:AB_312790), Phycoerythrin (PE)-conjugated anti-Ki-67 (BioLegend, catalog no. 652403, RRID:AB_2561524), Allophycocyanin-Cyanine7 Tandem Conjugate (APC Cy7) anti-CD8α (BioLegend, catalog no. 100711, RRID:AB_312750), FITC-conjugated anti-CD4 (BioLegend, catalog no. 100405, RRID:AB_312690), PE-conjugated Granzyme B (GZMB; BioLegend, catalog no. 372207, RRID:AB_2687031), V500 CD45 (BD Biosciences, catalog no. 561487, RRID:AB_10697046), BUV395 anti-NK1.1 (BD Biosciences, catalog no. 564144, RRID:AB_2738618), and Brilliant Violet 421 anti-PD-L1 (BD Biosciences, catalog no. 564716, RRID:AB_2738911). Flow cytometry was performed on a BD LSR II or a Beckman Coulter CytoFLEX. Analysis was performed using FlowJo software (FlowJo, RRID:SCR_008520).

For NK-cell depletion, anti-mouse NK1.1 was injected intraperitoneally on day (−3) and day (−2) prior to inoculation with B16F0 cells. Inoculation with B16F0 cells subcutaneously on day 0 was followed by isotype control and mDKN-01 injection intraperitoneally on day 1. NK depletion was assessed by flow cytometric analysis of CD49b⁺NKp46⁺ cells in whole blood and spleen.

DKK1 levels from B16F0 cell culture supernatant or mouse serum were measured by ELISA according to the manufacturer's instructions (Abcam, ab197746).

Tumor lysates were normalized to protein concentration using the Bio-Rad BCA Protein Assay according to the manufacturer's instructions (Bio-Rad). Protein lysates were resolved by SDS-PAGE using 4% to 15% Mini-PROTEAN TGX precast protein gels (Bio-Rad) and then transferred to an Immobilon-FL polyvinylidene difluoride (PVDF) membrane (Millipore Sigma). PVDF membranes were then blocked with Intercept (TBS) blocking buffer (LI-COR) and probed with non-phospho (Active) β-catenin (Ser45) XP rabbit mAb that specifically recognizes β-catenin that is non-phosphorylated at Ser45 (Cell Signaling Technology, catalog no. 19807, RRID:AB_2650576). Primary antibody was recognized by a goat anti-rabbit IRDye 800CW secondary antibody (LI-COR Biosciences, catalog no. 926-32211, RRID:AB_621843) and developed and quantitated using the Odyssey CLx Imaging System (LI-COR).

Quantitation of intratumoral protein levels of IL15, IL33, macrophage inflammatory protein 1 beta (MIP-1β), macrophage inflammatory protein 3 alpha (MIP-3α), and Monocyte Chemoattractant Protein 1 (MCP-1/CCL2) were performed using a Meso Scale Discovery (MSD) U-PLEX Biomarker Group 1 (Mouse) assay plate which utilized a linker technology and biotinylated capture antibodies (MSD). Cytokines were then detected using SULFO-TAG–labeled antibodies to generate the electrochemiluminescence readout and read using the MESO QuickPlex SQ 120 imager (MSD).

Cell proliferation was measured using CellTiter-Glo Luminescent Cell Viability Assay according to the manufacturer's instructions (Promega). B16F0 cells were plated at 2,500 cells per well in a 96-well plate (Costar). mDKN-01 or isotype control was added to a final concentration of 10.0, 1.0, or 0.1 μg/mL in 200 mL. Cells only and media only controls were included on each plate. Cells were incubated for 48 hours at 37°C, 5% CO₂. After 48 hours, media were removed and 100 mL of media were added to all wells. ATP standard was prepared and 100 mL of each concentration was added in triplicate. CellTiter-Glo Reagent (100 mL) was added to sample wells and plates were placed on rocker for 2 minutes. Cells were then read on Envision plate reader for luminescence at 1 second per well (PerkinElmer).

RNA was isolated using TRIzol reagent (Invitrogen, 15596018) and Phasemaker tubes (Invitrogen, A33248), following the manufacture's protocols. cDNA synthesis was conducted with a QuantiTect Reverse Transcription Kit (Qiagen, 205311) and TaqMan Real-Time PCR was run with Taqman Fast Advanced Master Mix (Applied Biosystems, 4444963) according to the manufacture's protocols with the following Taqman assays: Axin2 (Mm00443610_m1), Tcf4 (Mm00443210_m1), Gapdh (Mm99999915_g1), and Rn18s (Mm03928990_g1). Real-Time PCR was run on a LightCycler 480 Instrument (Roche Life Science). All assays were done in triplicate in 384-well plates. Mean threshold cycle (C_t) values of the triplicates were calculated and normalized to the Gapdh and Rn18s endogenous control genes utilizing the LightCycler software.

RNA was isolated using TRIzol reagent (Invitrogen, 15596018) and Phasemaker tubes (Invitrogen, A33248) and subjected to RNeasy MinElute Cleanup Kit (Qiagen, 74204) with on column DNase digestion following the manufacture's protocols (Qiagen, 79254). RNA quality was assessed by Bioanalyzer (Agilent 2100) prior to analysis on the SPRINT system (NanoString Technologies) with the PanCancer Immune Profiling Panel containing a custom set of 30 Wnt signaling-related genes. Data were quality controlled and analyzed with nCounter Advanced Analysis 2.0 software (NanoString Technologies).

Mouse tumors treated with either mDKN-01 or isotype control were removed and fixed overnight in 10% neutral buffered formalin. Fixed tumors were then washed with PBS and stored in 70% ethanol until processing. After paraffin embedding, 5 μmol/L sections were cut and mounted onto Superfrost Plus slides (Thermo Fisher Scientific) and underwent IHC staining using IntelliPath FLX detection reagents (MenaPath, A. Menarini Diagnostics). Heat-antigen retrieval was performed in Access Revelation (pH 6.4), followed by peroxidase quenching and background blocking with casein. Tissue sections were incubated with antibodies raised against human β-catenin (Cell Signaling Technology, catalog no. 9562, RRID:AB_331149) and appropriate isotype control antibody followed by MenaPath HRP polymer or Universal Probe and HRP-polymer, respectively. Slides were subsequently developed with 3,3′Diaminobenzidine (DAB) and counterstained with hematoxylin.

Data are presented as means ± SEM. Comparisons between two groups only were performed using a Student t test and comparisons between three or more groups were performed using a one-way ANOVA with the Tukey–Kramer post hoc test using GraphPad Prism v9.0.0 software (GraphPad Prism, RRID:SCR_002798). Differences were considered statistically significant at P < 0.05.

Prior to initiating in vivo tumor growth inhibition (TGI) studies, the binding kinetics of mDKN-01 was evaluated and the K_D of mDKN-01 for murine DKK1 was determined to be approximately 20 pmol/L, similar to the affinity of the clinical humanized DKN-01 antibody for human DKK1 (Supplementary Fig. S1). Thus, the mDKN-01 reagent is an appropriate surrogate for evaluating DKN-01 MOA. A dose-finding and preliminary efficacy study determined that 10 mg/kg was the optimal treatment dose in the B16F10 melanoma syngeneic model (Supplementary Fig. S2). Efficacy of mDKN-01 was verified in the B16F0 melanoma syngeneic mouse model using a larger cohort of animals (Fig. 1A and B). Tumor-bearing mice treated with mDKN-01 showed significant TGI when compared with isotype-treated controls (148 ± 16 vs. 245 ± 23 mm³, respectively, P < 0.005; Fig. 1A and B).

To address the possibility of a direct antiproliferative effect of mDKN-01 on B16F0 tumors, we treated B16F0 cells in culture with a range of mDKN-01 concentrations. No effect on tumor cell growth was observed (Supplementary Fig. S3). B16F0 cell supernatants produced nearly undetectable DKK1 protein levels, confirming lack of DKK1 expression in this tumor (Supplementary Fig. S4A). While DKK1 is present in the serum of nontumor-bearing mice, it was not significantly different from isotype-treated tumor-bearing mice. However, its level is modestly increased in mDKN-01–treated tumor-bearing mice, likely due to reduced clearance of DKK1 when complexed with mDKN-01 (Supplementary Fig. S4A and S4B). In addition, at day 11 after inoculation, tumoral DKK1 mRNA expression was analyzed and remained near background expression levels irrespective of isotype or mDKN-01 treatment (Supplementary Fig. S4D). Lack of activated β-catenin–dependent Wnt signaling in these mDKN-01 tumors (Supplementary Fig. S5A–S5D) suggests that mDKN-01 is not inducing a canonical WNT signaling cascade.

To determine whether immune cells are required for mDKN-01 efficacy, we challenged immune-deficient NSG mice with B16F0 melanoma cells and treated the mice with mDKN-01 (Fig. 1C and D). We did not observe any reduction in tumor growth as compared with isotype controls (535 ± 54 vs. 565 ± 51 mm³, respectively). These data suggest that an immune cell type absent in NSG mice is required for mDKN-01 efficacy in the B16F0 model. DKK1 was present in the serum of both the isotype and mDKN-01–treated NSG mice and levels increased as was seen in C57BL/6J mice following mDKN-01 treatment, demonstrating target engagement (Supplementary Fig. S4C). Interestingly, tumor-bearing RAG1^−/− mice (deficient in T and B cells) treated with mDKN-01 also showed significant TGI when compared with isotype-treated controls (241 ± 32 vs. 394 ± 37 mm³, respectively; Fig. 1E and F) similar to what was observed in C57BL6/J mice (Fig. 1A and B). Taken together, these data suggest that T and B cells are not essential for TGI activity, and that an innate immune cell absent or functionally compromised in NSG mice may be required for the full mDKN-01 effect.

To further define the innate immune cell types critical for the TGI seen in the B16F0 model, we treated mice prior to tumor inoculation with an anti-NK1.1 mAb (Supplementary Fig. S6). Following depletion, C57BL/6J mice were inoculated with B16F0 melanoma cells and treatment with either mDKN-01 or an isotype control was initiated. In the non-NK1.1⁺-depleted mice, mDKN-01 treatment resulted in a significant reduction in tumor size compared with isotype controls, as expected (250 ± 23 vs. 454 ± 76 mm³, respectively; Fig. 2A and B). However, in the NK1.1⁺ cell–depleted cohort mDKN-01 treatment had no effect on tumor size compared with isotype controls (472 ± 55 vs. 467 ± 67 mm³, respectively; Fig. 2A and B) indicating that NK1.1⁺ cells are required for the mDKN-01 TGI effect. While there was significant depletion of CD49b⁺NKp46⁺ (i.e., NK cells) in both whole blood and spleen as well as CD49b⁺CD3⁺ cells (i.e., NKT cells) in whole blood, we did not observe any significant changes in other immune cells types tested including CD45⁺, CD3⁺, and CD8⁺ cells in whole blood or spleen (Supplementary Fig. S6). Taken together, these data suggest that mDKN-01 requires a functioning immune system and that an innate and not an adaptive component may be an essential arm of the mDKN-01 MOA.

To assess mDKN-01 immune-mediated changes, flow cytometry, proteomic, and gene expression analysis of the TME was conducted. We observed an increase in CD45⁺ cells by both flow cytometry and NanoString that was not associated with a CD45⁺ proliferative response (Fig. 3A–C). Consistent with the efficacy of mDKN-01 in the RAG1^−/− model, no significant alterations in the CD4⁺ or CD8⁺ cells in the TME were observed by flow cytometry (Fig. 3D and E). However, there was activation of NK1.1⁺ cells, as evidenced by an increase in GZMB expression in mDKN-01–treated mice as compared with isotype controls, (Fig. 3F; Supplementary Fig. S7A). Given that neutralization of DKK1 has been reported to reduce MDSC levels in the TME (18), we assessed the effect of mDKN-01 on MDSC infiltrates. We observed a decline in both splenic and intratumoral MDSCs (Fig. 3G–I; Supplementary Fig. S7B). PD-L1 expression was also increased on intratumoral MDSCs (Fig. 3J).

Given the increase in CD45⁺ cells in tumors of mice treated with mDKN-01 and lack of a significant proliferative response in this population (Fig. 3B), we hypothesized there might be a mDKN-01–driven increase in the production of chemokines. Transcriptional changes showed, in addition to confirming an increase in Cd45 expression (Ptprc), a significant upregulation of seven chemokines, most prominently the C-X-C motif chemokine ligand 10 (Cxcl10; Fig. 4; Supplemental File 1). Two other changes of note were an increase in Il1β and a decrease in Tgfβ1. To evaluate mDKN-01–related changes at the protein level, we analyzed tumor lysates and observed consistent upregulation in several cytokines following mDKN-01 treatment, including MIP-1β, MIP-3α, and MCP-1 (CCL2; Fig. 5A–C). The NK cell–activating cytokines IL15 and IL33 were also upregulated, consistent with the increase in GZMB expression seen in intratumoral NK1.1⁺ cells (Fig. 5D and E and Fig. 3F, respectively).

Having demonstrated mDKN-01–mediated TGI in the B16F0 melanoma model as well as recently in a human prostate cancer cell line, PC3, in collabration with Wise and colleagues (15), we asked whether mDKN-01 has similar efficacy in a tumor model with metastatic potential. To do so, we inoculated BALB/cJ mice SC with 4T1 cells and evaluated mDKN-01 efficacy on TGI as well as its ability to affect the metastatic burden in the lungs. Tumor-bearing mice treated with mDKN-01 showed a modest but significant TGI at day 30 when compared with isotype-treated controls (1,067 ± 74 vs. 1,384 ± 62 mm³, respectively, P < 0.005; Fig. 6A and B). The effect of mDKN-01 treatment resulted in a marked reduction of lung metastases at day 30 (Fig. 6C and D).

Prompted by the innate immune dependency of the mDKN-01 TGI, and the elevated expression of both Cxcl10 and PD-L1, we undertook combination studies with an anti-PD-1 antibody. We treated B16F0 tumor-bearing mice with mDKN-01 or anti-PD-1 alone or in combination. Mice treated with mDKN-01 or anti-PD-1 showed 33% or 45% TGI, respectively, compared with isotype controls (Fig. 7A and B). The mDKN-01/anti-PD-1 combination treatment was statistically superior to mDKN-01 alone, reducing mean tumor volumes by an additional 55% (Supplementary Table S1). While the combination treatment resulted in an additional 45% TGI as compared with mice administered anti-PD-1 alone, this difference was not statistically significant likely due to the biological variance within the anti-PD-1 cohort.

In this study, we demonstrate that a novel high-affinity antibody targeting DKK1 in a syngeneic B16F0 melanoma model leads to TGI that is NK dependent and provides enhanced efficacy when combined with a PD-1 inhibitor. mDKN-01 monotherapy studies both in vitro and in vivo suggest that in B16F0 melanoma TGI is not mediated by direct effects on tumor growth. Three lines of evidence support this conclusion. First, in vitro, mDKN-01 has no effects on B16F0 proliferation. Second, there is no evidence of β-catenin–dependent (canonical) Wnt signaling changes in B16F0 tumors. Finally, NSG mice, as well as NK1.1⁺ cell–depleted mice, show no TGI effects with mDKN-01 treatment. Tumor growth inhibition is associated with important favorable changes in the TME, including reductions in percentages of Gr-1⁺CD11b⁺ cells, upregulation of PD-L1, and upregulation of GZMB expression in NK1.1⁺ cells. In addition, we report upregulation in the TME of the NK-activating cytokines IL15 and IL33, as well as a diverse panel of chemokines, most prominently, CXCL10, and at the protein level: MIP-1β, MIP-3α, and MCP-1 (CCL2). However, an important question that remains to be answered is what cell types within the TME are responsible for these mDKN-01–induced cytokines changes observed here. Interestingly, in NK1.1⁺-depleted animals, mDKN-01 stimulated increases of several of the cytokines were diminished, including IL33, MCP-1 (CCL2), and MIP-3α. This suggests that some of the mDKN-01–mediated cytokine secretion within the TME is NK1.1 dependent. Further studies, including single-cell RNA-seq, will be critical to further interrogate the cellular source of these mDKN-01–induced cytokine changes.

The results presented here demonstrate that mDKN-01 treatment has efficacy not only in a primary tumor setting with the B16F0 model, but also in a metastatic model with the 4T1 breast cancer cell line. In addition, we have recently extended these observations with DKN-01 to a PC3 prostate xenograft model. In Wise and colleagues, (15), DKN-01 mediates TGI in tumor-bearing BALB/c scid mice, is inactive in NSG mice, and is lost on predepletion of asialo-GM1⁺ cells in these BALB/c scid mice. Hence in a prostate xenograft model, efficacy is also innate immune dependent and likely depends on NK cells. Transcriptional analyses of biopsies from patient metastatic castration-resistant prostate tumor tissues show an inverse relationship between tumor DKK1 mRNA levels and activated NK cells (15).

The relationship between DKK1 and NK activity is of interest considering results in human xenograft breast and lung cancer metastasis models where DKK1 expression protected tumors from NK-mediated lysis by the downregulation of NK-activating ligands and cell death receptors (19). There, DKK1-depleted cells show increased susceptibility to NK-mediated cytotoxic killing. In related studies, DKK2, a homolog of DKK1, suppressed the IL15-mediated activation of NK cells, and in an MC38 model, an anti-DKK2 antibody enhanced GZMB expression in NK1.1 and CD8 T cells (20).

Limitations of this study focus on the characterization of the NK1.1 cell subtype(s) whose activation is associated with TGI, and whose depletion is responsible for the loss of mDKN-01 activity. As the TGFβ-driven conversion of NK cells into innate lymphoid cells (ILC) group 1 is a mechanism by which tumors may escape surveillance by the innate immune system (23), we looked for such changes in the TME. Other than an increase in NK1.1⁺ cell levels of GZMB expression in association with elevated expression of IL15 and IL33, we were unable to document a change in an ILC/NK phenotypes within the TME using several markers of NK ILC subtypes, including NKp46, CD103, CD49a, and CD49b. Nevertheless, the elevated expression of IL33 and its ability to activate ILC2 cells in an orthotopic pancreatic model (24) suggest that ILC phenotypic changes in this model deserve further investigation. As shown previously (18), DKK1 in circulation in the B16 model is predominantly derived from bone. Thus, evaluation of the bone marrow environment or the peripheral circulating pool for mDKN-01-mediated changes in NK-ILC phenotype should be a subject for future studies.

The mechanism by which mDKN-01 enhances anti-PD-1 activity is not fully understood. The conditions in other studies that mediate PD-1 unresponsiveness are high levels of TGFβ in the TME (25) and the expansion of intratumoral MDSCs (26). That DKK1, by suppressing β-catenin in MDSCs, can promote their expansion and function is of interest in the context of DKK1 immunomodulation (18). In addition to microsatellite instability status, tumor mutational burden and other characteristics of PD-1–responsive tumors; the expression of PD-L1 on tumor-infiltrating myeloid cells may also be a predictive biomarker of response to anti-PD-1 therapy (27–29). Our finding that Cxcl10 mRNA was highly expressed following mDKN-01 treatment is of interest considering the recent observations that the CXCL9-CXCL10-CXCR3 axis may be important in determining the outcome of anti-PD-1 therapy (30). However, at early stages of tumor development, we did not observe an increased infiltrate of CD4 or CD8 T cells, nor did we see any enhanced expression of Cxcr3 mRNA. Interestingly, in a recent attempt to classify gastric tumors by pretreatment TME phenotypes, TME subtypes that are NK^hi showed a better overall survival than NK^lo (31). If confirmed, it will be of interest to explore how NK^lo phenotypes correlate with high DKK1 levels, allowing further prospective definition of a responsive patient population. Studies with DKN-01 and a PD-1 inhibitor in gastric/GEJ cancer are underway (NCT04363801).

M.H. Kagey reports a pending patent for “Methods of Treating Cancer Using DKK-1 Inhibitors” and is a stock option holder. J.B. Rottman is a paid consultant for Leap Therapeutics. M.S. Haas, H. Heath, and W. Newman are stock option holders. No disclosures were reported by the other author.

M.S. Haas: Conceptualization, data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. M.H. Kagey: Conceptualization, data curation, formal analysis, investigation, methodology, writing–review and editing. H. Heath: Resources, data curation, software, formal analysis, investigation, methodology, writing–review and editing. F. Schuerpf: Resources, data curation, software, investigation, methodology. J.B. Rottman: Data curation, formal analysis, supervision, visualization, methodology. W. Newman: Conceptualization, data curation, formal analysis, supervision, writing–original draft, project administration, writing–review and editing.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.