Pharmacologic Inhibition of FGFR Modulates the Metastatic Immune Microenvironment and Promotes Response to Immune Checkpoint Blockade

Immune checkpoint blockade (ICB) therapy is approved for the treatment of certain patients with metastatic breast cancer, but the clinical benefits of such therapy are limited in comparison with its effectiveness against other cancer types (1). A major rate-limiting aspect of ICB therapies such as PD-L1–targeted antibodies is that breast cancer metastases are poorly infiltrated with inflammatory lymphocytes (2, 3). Indeed, several studies indicate that as compared with matched primary mammary tumors, breast cancer metastases have decreased immune cell infiltration (4, 5). The drivers of these immune excluded phenotypes in metastatic disease remain to be definitively determined. Therefore, there is a clear clinical need to develop therapeutic interventions capable of altering the composition of the metastatic immune microenvironment to enhance the success of ICB therapies.

Dissemination of tumor cells can occur years before the detection and treatment of primary tumors. These disseminated cells can be held in an asymptomatic state by a number of mechanisms including the establishment of a balance with the immune system (6). Here, we report that the systemically dormant phenotype of the 4T07 cell model only manifested in immune competent BALB/c mice and was dependent on the function CD8⁺ immune cells (7). Emergent lesions that overcame dormancy were devoid of CD8⁺ cells. Establishing the immune dependence of this dormancy phenotype allowed us to interrogate surrogate therapeutics that enhanced immune cell infiltration into pulmonary lesions.

Fibroblast growth factor receptors (FGFR) are a family of receptor tyrosine kinases that facilitate cancer cell proliferation, migration, and resistance to currently used therapeutics when functioning aberrantly (8, 9). Overexpression and activation of FGFR1 is a marker of epithelial–mesenchymal transition (EMT) and metastasis (10, 11). In urothelial carcinoma, mutationally activated FGFR3 is a biomarker for treatment with the FDA-approved FGFR inhibitor erdafitinib (12). In breast cancer, the FGFR1 gene locus is amplified in 13% of primary tumors and is de novo amplified in metastatic tumors as compared with the primary tumor from which they were derived (13, 14). Whether or not FGFR1 amplification can effectively determine breast cancer response to erdafitinib remains to be determined (15). In addition to erdafitinib, several additional selective and nonselective inhibitors of FGFR kinase activity are being developed and demonstrate efficacy against in vitro cell lines as well as in preclinical mouse models (16). In addition to tumor cell–intrinsic effects, systemic inhibition of FGFR impacts various aspects of the primary tumor microenvironment such as angiogenesis and infiltration by various immune cells (17–20). For instance, application of the FGFR inhibitor BGJ-398 reduces G-CSF levels, which reduces mobilization of myeloid-derived suppressor cells (MDSC; ref. 21). However, it is not well understood how systemic inhibition of FGFR signaling during the treatment of pulmonary metastasis may change the immune landscape and potentiate a response to immunotherapy. Here, we sought to address the hypothesis that modulation of the metastatic immune microenvironment via systemic inhibition of FGFR would sensitize metastatic tumors to ICB therapy.

All mouse experiments were performed in accordance with, and with the approval of Purdue University IACUC. Five- to 6-week-old female BALB/c and athymic Nude (NU/J) mice were purchased from The Jackson Laboratory. Female NRG (NOD-Rag1^null IL2rg^null, NOD ^{rag gamma}) were purchased from Biological Evaluation Shared Resource facility at Purdue University (West Lafayette, IN).

The 4T1, 4T07, and D2.A1 cell lines were obtained from the lab of Fred Miller (Wayne State University, Detroit, MI) in 2008. These cells were cultured in DMEM media (Thermo Fisher Scientific) supplemented with 10% FBS (Nucleus Biologics) and penicillin/streptomycin (100 U/mL; Life Technologies). Stable expression of firefly luciferase was established via stable transfection under Zeocin selection as described previously (22, 23). Jurkat (clone e6–1) cells were obtained from Kazemian lab (Purdue University) in 2018. The Jurkat cells were cultured in RPMI1640 media supplemented with 10% FBS (Nucleus Biologics) and penicillin/streptomycin (100 U/mL; Life Technologies). The D2.A1 cells are a BALB/c-derived cell model that produces pulmonary tumors in their syngeneic host upon tail injection. These cells express readily detectable levels of FGFR1 and proliferate in response to FGF2 stimulation (22). Depletion of FGFR1 in the D2.A1 cells was achieved via stable shRNA expression as described previously (11). All cell lines were routinely checked for Mycoplasma using the MycoAlert Detection Kit from Lonza. Cell lines were authenticated by IDEXX RADIL Diagnostic labs using the CellCheck 9-marker STR profile. Authentication was conducted in July of 2018 and all cell lines were used within 10 passages after authentication.

The development and synthesis of FIIN4, the covalent FGFR inhibitor, were described previously (10). This irreversible inhibitor demonstrates an IC₅₀ of 2.6, 2.6, 5.6, and 9.2 nmol/L against FGFR1–4, respectively. The t_1/2 of FIIN4 is 2.4 hours resulting in a max serum concentration of 290 ng/mL, 30 minutes after oral administration of 10 mg/kg.

4T07 cells (1 × 10⁴) were plated into a nonadherent round bottom spheroid plate (4520; Corning) in 200 μL of DMEM containing 10% FBS and penicillin/streptomycin as described above in the cell lines section. After 48 hours, these spheroids were transferred via a 200 μL micropipette to a 50 μL bed of Cultrex basement membrane matrix (3432–001–01; Trevigen) in a white-walled 96-well dish with 150 mL of full growth media containing 10% Cultrex. This was done in the presence or absence of 50 or 100 nmol/L FIIN4 or 100 nmol/L AZD4547 (Sellekchem), or 100 nmol/L erdafitinib (Selleckchem). Spheroid growth was tracked by bioluminescence as determined by adding 30 ng of D-luciferin dissolved in 2 μL of PBS immediately following transfer of the spheroids to the Cultrex and 6 days thereafter. Luminescence was read on a Glo-max Discover (Promega) plate reader.

4T07 (5 × 10⁵) and D2.A1 (5 × 10⁵) cells were injected via the lateral tail vein and pulmonary tumor growth was visualized and quantified using bioluminescent imaging (Spectral Instruments) at the indicated time point. In orthotopic experiments, 4T07 cells (1 × 10⁵) were injected into the nipple of BALB/c or NU/J mice and tumor growth was traced via bioluminescence. Fourteen days after orthotopic injection, primary tumors were surgically removed, and subsequent metastases were tracked by bioluminescence. Where indicated, mice were treated via oral gavage with 25, 50, or 100 mg/kg of FIIN4, AZD4547, or erdafitinib. For these in vivo drug treatments, 50 μL of 50 mg/mL DMSO stocks of the compounds were further diluted in 950 μL methylcellulose and the proper amount was administered to the animals via daily oral gavage. For CD8 depletion experiments, mice were administered 200 μg of anti-CD8 (BP0117; Bio X Cell) via intraperitoneal injection. For ICB experiments, mice were administered 200 μg of anti–PD-L1 (BE0101; Bio X Cell) via intraperitoneal injection.

Single-cell suspensions were prepared from isolated lung tumors and spleens. Briefly, collected tumor-bearing lung tissues were minced and digested for 1 hour at 37°C in HBSS media containing Gibco collagenase type II (100 U/mL; Thermo Fisher Scientific #17101015). Isolated spleens were mechanically disrupted by grinding. Samples were treated with ACK buffer to lyse red blood cells before being filtered with 70-μm sterile cell strainers. To characterize the immune microenvironment, the resultant single-cell suspensions were incubated with TruStain anti-CD16/32 (BioLegend #101320) and Zombie violet (BioLegend #423114). Cells were subsequently stained with CD45-PerCP (103130), CD11b-PE/Cy7 (101216), CD8-PacificBlue (100725), ly6C-FITC (128006), ly6G-APC/Cy7(127624), F4/80-BV 605 (123133), and PD-L1-PE (124308; all the antibodies were purchased from BioLegend). A BD Fortessa LSR flow cytometry cell analyzer was used to perform the experiments and the resulting data were analyzed using Flowjo software.

Immunoblot assays were performed on cell lysates prepared by lysing the samples with modified RIPA lysis buffer containing 50 mmol/L Tris, 150 mmol/L NaCl, 0.25% sodium deoxycholate, 1.0% NP40 and 0.1% SDS supplemented with protease inhibitor cocktails (Sigma), 10 mmol/L sodium orthovanadate, 40 mmol/L b-glycerolphosphate, and 20 mmol/L sodium fluoride. Following SDS PAGE and transfer, PVDF membranes were probed with antibodies specific for FGFR1 (9740; Cell Signaling Technology) and β-tubulin (E7-s; Developmental Studies Hybridoma Bank). For IHC, formalin-fixed and paraffin-embedded tissue sections were deparaffinized, rehydrated, and boiled in 10 mmol/L sodium citrate buffer (pH 6.0) to retrieve antigens. These processed sections were stained with antibodies specific for CD4 (#25229; Cell Signaling Technology), CD8 (#70306; Cell Signaling Technology), and CXCL16 (#AF503; R&D Systems). Staining was detected using the anti-rabbit (#111–065–003; The Jackson Laboratory) or anti-mouse (#115–065–062; The Jackson Laboratory) biotinylated secondary antibodies in combination with ABC (Vector) and DAB reagents (Vector). Sections were counter stained with hematoxylin (Fisher). At least two people who were blinded to the experimental groups counted the positively stained cells in multiple fields of the stained sections using a Nikon TS-100 microscope.

4T1 cells were treated with either 500 nmol/L FIIN4 or DMSO for 24 hours and total RNA was isolated using an RNeasy Plus Kit (Qiagen). cDNA libraries were synthesized using iScriptcDNA Synthesis System (Bio-Rad) using 100 ng of total RNA. Semiquantitative real-time PCR was performed using iQ SYBR Green (Bio-Rad) as reported previously (24). The unique primers used for CXCL16 were forward: 5′-CTCTGCAGGTTTGCAGCTCT-3′ and reverse: 5′-CACTGATGGAGACGAGCCTG-3′.

Jurkat cells (1 × 10⁴) were cultured in RPMI1640 media lacking FBS but containing penicillin/streptomycin (100 U/mL; Life Technologies) and the indicated doses of FIIN4 for 2 hours in 24-well plates. Next, these cells were washed with sterile PBS twice and incubated with 5 μg/mL anti-CD3 (#317301; BioLegend) for the indicated amounts of time. Following incubation, cells were isolated via centrifugation at 800 × g for 30 seconds and lysed and these lysates were probed with antibodies specific for phospho-LCK (#2751), phospho-PLCγ1(#14008), phospho-SLP-76 (#14745), phospho-ZAP70 (#2717), and phospho-LAT (#3584; all antibodies from Cell Signaling Technology) as described below in the immunoblot section. 4T07 cells at 50% confluency were cultured in DMEM in the absence of serum for 18 hours prior to stimulation with 20 ng/mL of FGF2 (R&D Systems). These cells were lysed as described above in the immunoassays section. These lysates were probed with antibodies for phospho-ERK1/2 (#4370), phospho-Akt (#4070), and total ERK1/2 (#9194; all antibodies Cell Signaling Technology).

The impact of FIIN4 on the cytolytic capacity of murine immune cells was assessed through a luciferase release assay. A total of 5 × 10⁴ 4T07 cells expressing firefly luciferase were cocultured in a 12-well dish for 6 hours with freshly isolated splenocytes (5 × 10⁵) derived from 4T07 immunized mice. This was done in the presence or absence of 1 μmol/L of FIIN4 in a total volume of 1 mL of serum free DMEM. Following this incubation, 200 μL of supernatant was collected, mixed with 30 ng of d-luciferin (Gold Biotechnology), dissolved in 2 μL of PBS, and assayed for luminescence in a 96-well white-walled dish using a Glo-Max Discover plate reader (Promega) as a measure of tumor cell lysis. 4T07 cells alone in the presence or absence of 1 μmol/L FIIN4 served as background controls.

MGHU3 cells were treated with 100 nmol/L of the FGFR inhibitor PD173074 for 48 hours prior to gene expression analyses via microarray (25). These data were deposited on GEO as GSE52452. The raw expression values for CXCL16 in control and PD173074-treated samples were extracted from the series matrix file and plotted. For off-target comparisons of FIIN4 and erdafitinib, the previously conducted kinome analyses were rank ordered and compared (10, 26).

Bonferroni-corrected threshold (significance level 0.05) was used to compare the survival of multiple groups in the combination studies. A log-rank study was performed to identify statistically significant differences in animal survival between control and treatment groups. All the performed statistical analyses were conducted using PRIZM version 7 and no statistical corrections were used. All associated P values are indicated in the respective figure legends.

In the absence of mutational activation of the receptor, additional molecular factors are required for FGFR to effectively signal and act as an oncogenic driver. The 4T1 and 4T07 cells are isogenic models derived from BALB/c mice. These cells not only express FGFR1, but also express integrins and other cofactors required to respond to exogenous FGF ligands, and their growth was decreased by small-molecule inhibitors of FGFR (Supplementary Fig. S1; ref. 10). Therefore, these model systems are well-suited to evaluate the impact of FGFR inhibitors on the metastatic tumor immune microenvironment. We recently reported that pulmonary growth of 4T1 cells following tail vein injection can be reduced by FIIN4, a covalent FGFR kinase inhibitor (10). Tail vein injection of the 4T07 cells similarly results in robust pulmonary tumor formation and lethality within 3 weeks of tumor cell inoculation (23). To further investigate the ability of FGFR inhibitors to block pulmonary tumor growth, we treated BALB/c mice bearing 4T07 pulmonary tumors with FIIN4 and two additional FGFR-targeted compounds, AZD4547 and erdafitinib (Supplementary Fig. S2). At the same mg/kg treatment dose, all three compounds effectively blocked pulmonary tumor growth, and all three compounds caused reductions in body weight (Supplementary Fig. S2). Given these findings, together with our previous studies establishing FIIN4 as an effective in vivo probe molecule for understanding FGFR biology, FIIN4 was chosen for the subsequent studies herein (10, 27).

We injected both syngeneic BALB/c and immunodeficient NRG mice with 4T07 cells and FIIN4 treatment was initiated 48 hours later (Fig. 1A). A 7-day treatment course with FIIN4 significantly delayed pulmonary tumor growth in NRG mice, suggesting tumor cell–intrinsic effects of FGFR inhibition. However, this same treatment led to transient regression of tumor burden in BALB/c animals (Fig. 1B and C). During this 7-day treatment, we observed a reduction in BALB/c body weight that did not exceed 20% of the original weight. In contrast, NRG mice did not experience any reduction in body weight due to FIIN4 treatment (Fig. 1D). Upon cessation of FIIN4 treatment, BALB/c mice experienced a significant survival benefit, whereas the NRG mice did not (Fig. 1E). To more accurately recapitulate the clinical course of breast cancer metastasis, we also orthotopically implanted 4T1 cells onto the mammary fat pad of syngeneic BALB/c mice and treated with FIIN4 at varying times following surgical removal of the primary tumor (Supplementary Figs. S3 and S4). Using this approach, we were able to demonstrate a robust inhibition of pulmonary and extrapulmonary metastatic tumor growth upon oral administration of FIIN4 (Supplementary Figs. S3 and S4). These data demonstrate that FGFR inhibition can extend survival in the metastatic setting and suggest that the efficacy of systemic FGFR inhibition is enhanced by the presence of an intact immune system.

To understand how FIIN4 might be utilized in combination with immune-directed therapeutics, we first sought to better characterize the role of the immune system in regulating the metastatic potential of the 4T07 cells. Therefore, we established orthotropic primary tumors in the mammary fat pad of immunodeficient NU/NU and immunocompetent BALB/c mice (Fig. 2A). In contrast to the robust pulmonary growth of 4T07 cells upon tail vein injection, we observed nearly complete tumor rejection when 1 × 10⁶ 4T07 cells were engrafted onto the mammary fat pad of BALB/c mice as compared with NU/NU mice (Fig. 2A–C). Bioluminescent tracking of metastasis following primary tumor removal demonstrated that 4T07 cells can form macroscopic pulmonary metastases in NU/NU mice but not in BALB/c mice (Fig. 2D and E). These data suggest that the previously reported systemically dormant phenotype of 4T07 cells is dependent on adaptive immune function (28).

To further explore the immunogenicity of the 4T07 tumor model, we conducted primary tumor immunization experiments (Fig. 3A). Further optimization of 4T07 engraftment procedures indicated that intraductal injection of 1 × 10⁵ cells subverted an immune response and resulted in robust primary tumor growth on the mammary fat pad (Fig. 3B). Following this primary tumor exposure and a prolonged period of recovery, firefly luciferase expressing 4T07 cells were delivered via the tail vein to the primary tumor exposed mice and their age matched primary tumor naïve littermates (Fig. 3C). Bioluminescent imaging demonstrated pulmonary tumor growth typical of the 4T07 model in the primary tumor-naïve animals (Fig. 3C–E). In contrast, animals that were previously exposed to primary tumors failed to form pulmonary tumors upon tail vein injection of the 4T07 cells (Fig. 3C–E). These findings suggest that orthotopic 4T07 tumors elicit a systemic immune response capable of preventing 4T07 cellular outgrowth even when the cells are delivered directly to an alternative organ system (7).

To more specifically characterize the role of the adaptive immune system in regulating 4T07 tumor growth, orthotopic and tail vein–injected pulmonary tumors were grown in BALB/c mice for 14 days. These analyses demonstrated that either route of tumor cell engraftment resulted in a similar systemic immune response, as noted by splenic enlargement (Supplementary Fig. S5A). Furthermore, total numbers of CD8⁺ cells and myeloid populations were similar in the primary tumors and tumor-bearing lungs (Supplementary Fig. S5B–S5F). However, CD8⁺CD11b⁺ cells, which are recently activated T cells, were more prevalent within orthotopic mammary fat pad tumors as compared with pulmonary lesions (Fig. 4A–C; ref. 29). To establish the role of CD8⁺ immune cells in mediating the systemically dormant phenotype of 4T07 cells, these cells were engrafted onto the mammary fat pad of BALB/c mice and allowed to develop and disseminate for a period of 4 weeks (Fig. 4D). Following surgical removal of the primary tumor, mice were treated with either a CD8-depleting antibody or with an isotype control IgG (Fig. 4D). Bioluminescent imaging indicated that mice depleted of CD8⁺ cells demonstrated significantly more frequent incidence of metastasis compared with those that received control IgG (Fig. 4E and F). These findings demonstrate that restricted infiltration and activation of CD8⁺ immune cells are critical for the metastatic progression of the 4T07 tumor model.

Given the impact of the adaptive immune system on both 4T07 metastasis and response to FGFR inhibition, we sought to further characterize the effect of FGFR on immune function (19–21). Mice bearing pulmonary 4T07 tumors were split into four separate cohorts and treated with increasing doses of FIIN4 (Fig. 5A). Consistent with our data in Fig. 1, FIIN4 treatment led to a dose-dependent inhibition of pulmonary tumor growth (Fig. 5A and B). Histologic examination of the pulmonary tumors demonstrated that those in control mice who received DMSO were nearly devoid of CD8⁺ cytolytic lymphocytes (Fig. 5A). In contrast, FIIN4 treatment produced a dose-dependent increase in the number of CD8⁺ cells present in the pulmonary lesions (Fig. 5C). To investigate a mechanism of FIIN4-mediated recruitment of CD8⁺ cells, we examined the dataset GSE52452. These gene expression data were derived from urothelial carcinoma cells treated with an alternative FGFR inhibitor, PD173074 (30). Examination of this dataset for differential expression of T-cell recruitment molecules revealed a significant increase in expression of the chemokine CXCL16 upon inhibition of FGFR (Fig. 5D; ref. 31). Consistent with these findings, CXCL16 mRNA was also increased in pulmonary lesions when mice bearing pulmonary 4T07 tumors were treated with FIIN4, and IHC analyses of the 4T07 pulmonary lesions demonstrated a marked increase in CXCL16 staining upon administration of FIIN4 (Fig. 5E and F). To examine whether these drug-induced immune phenotypes could be genetically recapitulated, we utilized shRNA-mediated depletion of FGFR1 in D2.A1 cells. The D2.A1 cell model was derived from spontaneous mammary tumors, which originated from a D2 hyperplastic alveolar nodule (HAN) line in BALB/c mice (32). These cells grow aggressively in the lungs upon tail vein injection into their syngeneic host, express high levels of FGFR1, and readily respond to exogenous stimulation with FGF2 (10, 11, 24). Previous studies from our laboratory demonstrate that shRNA-mediated depletion of FGFR1 inhibits pulmonary tumor growth by the D2.A1 cells. However, these studies did not examine modulation of the immune microenvironment (11). Here, we found that like FIIN4 treatment, shRNA-mediated depletion of FGFR1 in D2.A1 cells led to an increased presence of CD8⁺ cells in pulmonary tumors following tail vein injection (Fig. 5G and H).

In addition to enhanced recruitment of CD8⁺ lymphocytes, FGFR inhibitors also modulate the immune microenvironment via recruitment of MDSCs and tumor-associated macrophages (21). As a result, we sought to define how FIIN4 treatment affects these immune regulatory cells in pulmonary 4T07 tumors. Flow cytometry analyses revealed that treatment of mice bearing 4T07 pulmonary tumors with FIIN4 reduced the number of MDSCs in both the lungs and spleens (Fig. 6A and B; Supplementary Fig. S6). In contrast, FIIN4 did not change the number of macrophages (Fig. 6C and D). Further characterization of the MDSCs that were present in pulmonary tumors revealed enhanced expression of the immunosuppressive molecule PD-L1 upon FIIN4 treatment (Fig. 6E and F). Taken together, these data suggest that FGFR signaling actively contributes to an immunosuppressive microenvironment in pulmonary metastases via active exclusion of CD8⁺ lymphocytes. Systemic application of an FGFR inhibitor allows for CD8⁺ lymphocyte infiltration and decreases peripheral mobilization of MDSCs.

To translate the observations above, we initiated combination therapy approaches using both FIIN4 and ICB antibodies. Concomitant addition of FIIN4 and either PD-1 or PD-L1–blocking antibodies failed to show any improvement in FIIN4-mediated survival of mice bearing either 4T07 or the 4T1 pulmonary tumors (Fig. 7A; Supplementary Fig. S7). Previous studies developing FIIN4 and erdafitinib established compound binding against a panel 450 kinase using the KINOMEscan platform (10, 26). Comparison of these analyses indicated the TCR-proximal kinase, LCK as a predicted shared off-target for both FIIN4 and erdafitinib (Supplementary Fig. S8). Furthermore, previous studies have found that FGFRs can participate in T-cell receptor (TCR) signaling and in the activation of a T-cell response (33). Therefore, we hypothesized that pharmacologic inhibition of FGFR signaling may inhibit the antitumor activity of cytolytic lymphocytes. To test this hypothesis, we isolated splenocytes from the enlarged spleens of 4T07 primary tumor bearing mice, cocultured these immune cells with 4T07 cells, and assayed for tumor cell lysis (Fig. 7B). As predicted, the splenocytes from in vivo primed, tumor-bearing animals efficiently killed the 4T07 cells in this ex vivo assay (Fig. 7B). However, this immune-mediated killing was nullified in the presence of FIIN4 (Fig. 7B). To further investigate the inhibition of T-cell function by FIIN4 we stimulated Jurkat cells in vitro with CD3-specific antibodies in the presence or absence of FIIN4 and assayed for differential phosphorylation of downstream signaling molecules. This approach clearly indicated that FIIN4 is capable of inhibiting TCR proximal signaling events (Fig. 7C). Given these inhibitory effects of FIIN4 on T-cell function, we reasoned that sequential combination of FIIN4 and anti–PD-L1 would first abrogate the immunosuppressive microenvironment followed by enhanced T-cell activity via ICB. Indeed, sequential dosing of FIIN4 followed by anti–PD-L1 significantly limited pulmonary tumor growth and increased overall survival as compared with either monotherapy (Fig. 7D and E).

Previous studies have characterized the 4T07 tumor model as a highly immunogenic tumor model capable of dissemination but not macrometastatic outgrowth (7). Our findings expand upon these conclusions by demonstrating that CD8⁺ cells are required for maintenance of 4T07 systemic tumor dormancy following surgical resection of the primary tumor. These data establish the 4T07 cells as a model of immune-mediated tumor mass dormancy. This opens up opportunities for the use of this postsurgical system for the study of various pharmacologic, dietary, and other environmental influences on metastatic tumor relapse. In addition, we demonstrate that upon tail vein injection of 4T07 cells, the resulting pulmonary tumors are able to establish an immune exclusion phenotype. These findings recapitulate breast cancer patient data in which metastases are less infiltrated with immune cells as compared with matched primary tumors (34). These data point to the importance of the tumor microenvironment in dictating response to kinase inhibitors as well as immunotherapy. In contrast to the 4T07 cells, the highly metastatic 4T1 model contains a high number of T cells in both the primary tumor and the resulting metastases. This raises the interesting question of whether metastases that manifest quickly following primary tumor interventions evolve more active modes of immunosuppression, whereas metastases that only emerge following prolonged periods of remission rely on immune exclusion to first maintain tumor cell survival, eventually leading to metastatic outgrowth. Continued understanding of these various modes of immune evasion will continue to refine and optimize how we apply ICB therapy or other immune therapies in the metastatic breast cancer setting.

FGFR recently has been established as a key modulator of the immune microenvironment in several primary cancers (19–21). Our work expands upon these studies by establishing a combined approach to facilitate the efficacy of ICB therapy in immune excluded pulmonary metastases. The concept of combining kinase inhibitors with ICB is clearly not without precedent. Recent studies have begun to identify important mechanisms by which kinase inhibitors, originally designed to target pathways within tumor cells, can influence immune cell recruitment and/or function (35). Although some of these mechanisms can be readily targeted in direct combination with ICB therapy, our study is likely reflective of a common occurrence in that either on- or off-target kinase inhibition can block T-cell function, thus negating the antitumor activity of the ICB therapy. These results point to the importance of a complete understanding of the systemic effects of small molecules before they can effectively be combined with ICB therapy or other immune therapies.

FIIN4 covalently modifies FGFR, whereas erdafitinib is a competitive inhibitor and the first FGFR-targeted small molecule to be approved by the FDA. However, there are numerous other competitive and covalent FGFR-targeted molecules in various stages of clinical trials being tested alone and in combination with ICB therapy (e.g., NCT03473756 and NCT04024436). Studies reported herein indicate that in addition to toxicity and on-target potency and specificity, analysis of these molecules should include potential interference with T-cell function. Consideration of these parameters is clearly required for an optimal combination therapy. The precise mechanisms by which FGFR-targeted compounds inhibit TCR function remain to be definitively determined. Our comparison of the KINOMEscan profiling for both FIIN4 and erdafitinib suggests potential off-target inhibition of LCK, a critical mediator of TCR signaling (10, 26). However, other studies suggest that FGFR interacts with the TCR and participates in TCR-mediated signaling events (33).

Overall, the data presented herein support the conclusion that systemic inhibition of FGFR reduces MDSC mobilization from bone marrow. Concurrently, genetic depletion studies suggest that FGFR signaling actively inhibits chemokine expression, limiting lymphocyte recruitment into a developing metastatic lesion. Continued understanding of how FGFR signaling and systemic administration of FGFR-targeted agents modulate immune function will improve our ability to enhance immunotherapy in metastatic breast cancer.

No disclosures were reported.

S.S. Akhand: Conceptualization, resources, data curation, formal analysis, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. Z. Liu: Resources, data curation, methodology. S.C. Purdy: Conceptualization, data curation, validation. A. Abdullah: Conceptualization, data curation, formal analysis. H. Lin: Conceptualization, data curation, formal analysis. G.M. Cresswell: Conceptualization, resources, data curation, software, formal analysis. T.L. Ratliff: Resources, supervision, writing–review and editing. M. Wendt: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, writing–original draft, project administration, writing–review and editing.

This research was supported in part by the American Cancer Society (RSG-CSM130259; to M. Wendt), the NIH (R01CA207751; to M. Wendt), and the Purdue Center for Cancer Research via an NIH NCI grant (P30CA023168).

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.