Targeting the Src pathway enhances the efficacy of selective FGFR inhibitors in cancers with FGFR3 alterations

Selective FGFR inhibitors such as infigratinib (BGJ398) and erdafitinib (JNJ-42756493) have been evaluated in clinical trials for cancers with FGFR3 molecular alterations, particularly in urothelial carcinoma patients. However, a substantial proportion of these patients (up to 50%) display intrinsic resistance to these drugs and receive minimal clinical benefit. There is thus an unmet need for alternative therapeutic strategies to overcome primary resistance to selective FGFR inhibitors. In this study, we demonstrate that cells expressing cancer-associated activating FGFR3 mutants and the FGFR3-TACC3 fusion showed primary resistance to infigratinib in long-term colony formation assays in both NIH-3T3 and urothelial carcinoma models. We find that expression of these FGFR3 molecular alterations resulted in elevated constitutive Src activation compared to wildtype FGFR3 and that cells co-opted this pathway as a means to achieve intrinsic resistance to infigratinib. Targeting the Src pathway with low doses of the kinase inhibitor dasatinib synergistically sensitized multiple urothelial carcinoma lines harbouring endogenous FGFR3 alterations to infigratinib. Our preclinical data provides evidence that supports the use of dasatinib in combination with selective FGFR inhibitors as a means to overcome intrinsic drug resistance in the salvage therapy setting in cancer patients with FGFR3 molecular alterations.


Introduction
Activating molecular alterations in the Fibroblast Growth Factor Receptor 3 (FGFR3) gene have been identified through large-scale next generation genomic sequencing efforts across a range of tumour types including urothelial carcinoma, lung squamous cell carcinoma, glioblastoma and myeloma (1)(2)(3)(4). These molecular alterations primarily comprise of point mutations that span across the gene (e.g. S249C, R248C and K652E) and gene fusions (e.g. FGFR3-TACC3 fusion). The cancer type with the highest proportion of FGFR3 alterations is urothelial carcinoma with a frequency of 60%-80% in the non-muscle-invasive form and 15-20% in the muscle-invasive form harbouring these molecular aberrations (5)(6)(7)(8).
The most common FGFR3 point mutation found across cancer types is S249C in which the extracellular domain cysteine substitution leads to the formation of a disulphide bond between receptor monomers resulting in ligand-independent constitutive dimerization and activation (2,(9)(10)(11). In contrast, the FGFR3-TACC3 fusion has been described at low frequencies (2-8%) in urothelial cancers and glioblastoma and mediates it oncogenic activity through constitutive dimerization of FGFR3 driven by TACC3 (1,2,4,12,13).  (14). Similarly, the BLC2001 phase II single-arm trial of erdafitinib in FGFR-altered urothelial carcinomas showed clinical responses in 40% of patients (15). This result has led to the accelerated approval of erdafitinib in 2019 by the US Food and Drug Administration (FDA) for locally advanced or metastatic urothelial carcinomas with FGFR2 and FGFR3 alterations. Despite these promising results, the BLC2001 trial showed that only 49% and 16% of patients with point FGFR mutations and fusions, respectively, achieved disease response with erdafitinib.
Similarly, Pal et al., reported that only 42.9% of patients with documented FGFR3 activating point mutations responded to infigratinib (14,15). These results indicate that notwithstanding the presence of a FGFR3 molecular alteration, there remains a significant proportion of patients that harbour intrinsic resistance to selective FGFR TKIs and do not respond to these agents. For these patients, there are limited therapeutic alternatives available and there is thus an urgent unmet need to identify effective ways to overcome primary resistance.
In this study, we show that cells expressing a subset of activating FGFR3 mutants and the FGFR3-TACC3 fusion are intrinsically resistant to selective FGFR TKI inhibition in long-term assays. We further demonstrate that expression of these FGFR3 mutants and FGFR3-TACC3 results in elevated constitutive Src activation levels compared to wildtype (WT) FGFR3, which can be exploited for therapy with the multi-target TKI dasatinib. We find that targeting the Src pathway with low dose dasatinib sensitizes urothelial carcinoma cell lines harbouring endogenous FGFR3 molecular alterations to selective FGFR TKIs including infigratinib. Our preclinical data suggests that patients with FGFR3 molecular alterations may benefit from treatment with FGFR and Src inhibitors in combination, which can overcome intrinsic resistance associated with selective FGFR TKI monotherapy.

FGFR3-TACC3 gene fusion and activating FGFR3 point mutants are intrinsically resistant to infigratinib in the NIH-3T3 cell line model
We first evaluated the effects of infigratinib in NIH-3T3 cells expressing a selection of the most prevalent activating FGFR3 molecular alterations in cancer. These included the most common FGFR3 fusion (FGFR3-TACC3) and extracellular domain (S249C) and kinase domain (K652E) point mutations. In addition, we examined N542K, which has previously been described as a congenital mutation in hypochondroplasia (16), due to its reported high kinase activity in in vitro assays (17). Expression of the FGFR3-TACC3 gene fusion and FGFR3 mutations S249C, N542K and K652E led to constitutive phosphorylation of Src when compared to the WT FGFR3 control with no differences observed in Erk1/2 and AKT phosphorylation levels ( Figure 1A). Consistent with previous reports, the transforming potential of these FGFR3 aberrations was demonstrated by an increase in the ability to grow under anchorage-independent conditions versus WT FGFR3 ( Figure 1B-C) (17)(18)(19). Shortterm (72 hours) cell viability experiments finds that as previously shown (17,(20)(21)(22), cells expressing FGFR3-TACC3 and the three FGFR3 mutants were sensitive to treatment with infigratinib with IC 50 values in the nanomolar range ( Figure 1D-E). It has been demonstrated that FGFR3 inhibition with FGFR inhibitors leads to selective downregulation of the Erk1/2 pathway (22)(23)(24). We show that unlike the WT FGFR3 cells, treatment with infigratinib decreased phosphoErk1/2 levels in the three mutant FGFR3 lines and abolished Erk1/2 phosphorylation in a dose-dependent manner in the FGFR3-TACC3 cells, confirming that this pathway is a direct downstream target of the FGFR3 mutants and FGFR3-TACC3 fusion. Interestingly, although constitutive activation of Src was associated with the expression of the FGFR3 mutants and FGFR3-TACC3 fusion in the NIH-3T3 cells ( Figure   1A), the phosphorylation levels of Src were not downregulated upon FGFR3 kinase inhibition by infigratinib ( Figure 1F). Further assessment of the efficacy of infigratinib in long-term (2 weeks) colony formation assays finds that despite the potent effects of this drug in the FGFR3 mutant and fusion expressing cells in short-term cell viability assays; there was an unexpected observation of a high number of resistant colonies persisting after 2 weeks of drug treatment ( Figure 1G-H). At 1µM drug treatment, there was no significant decrease in colony formation in any of the FGFR3 point mutant expressing cells, with only a small decrease (~20%) in the FGFR3-TACC3 expressing cells ( Figure 1H). Importantly, these intrinsic resistant colonies persist even at very high infigratinib doses of 2uM which models the primary resistance observed in a subset of patients in clinical trials.

FGFR3-TACC3 and S249C expression confers resistance to dasatinib
To interrogate the key signalling dependencies in the panel of FGFR3 expressing NIH-3T3 cells, a targeted small molecule inhibitor screen was undertaken. This screen was comprised of 32 small molecule inhibitors that target major kinase and non-kinase oncogenic signalling pathways in cells. These inhibitors include broad-spectrum kinase inhibitors such as imatinib, dasatinib and foretinib as well as selective kinase inhibitors such as infigratinib (FGFR), binimetinib (MEK), AZD5363 (AKT), NVP-BEZ235 (PI3K/mTOR) and MK8776 (CHK1). The screen also has a small number of non-kinase inhibitors including NVP-AUY922 (HSP90), GSK126 (enhancer of zeste homolog 2 (EZH2)) and JQ1 (bromodomain and extra-terminal (BET)) (See Table S1 for list of compounds used in the screen and key targets).
As a positive control for the assay, we show that as expected, FGFR3 point mutant and fusion expressing cells were sensitive to both multi-target (ponatinib, foretinib, lenvatinib, cediranib) and selective (infigratinib and AZD4546) FGFR TKIs ( Figure 2A). Interestingly, the expression of some FGFR3 molecular alterations conferred a survival advantage (of >1.2 fold) to a small number of compounds compared to WT FGFR3. These included BEZ235 (N542K and N652E), JQ1 (FGFR-TACC3, S249C and N652E), MK8776 (S249C) and dasatinib (FGFR3-TACC3 and S249C). Given that the expression of FGFR3 molecular alterations led to the constitutive activation of Src ( Figure 1A), dasatinib, a broad-spectrum TKI that potently inhibits Src as one of its targets (25,26), was taken forward for further investigation.
Full dose response assessment of dasatinib in short-term viability assays showed that FGFR3-TACC3 and S249C conferred resistance to this drug with an IC 50 of 3-4µM compared to the WT FGFR3 and the N542K and K652E mutations (<1µM) ( Figure 2B-C).
These results were recapitulated in the long-term colony formation assays with the FGFR3-  Figure 2F).
We hypothesized that the constitutive dimerisation in these FGFR3 molecular alterations may be responsible for the observed resistance to dasatinib. Since disruption of the FGFR3-TACC3 fusion through genetic means is non-trivial, we focused our efforts on the S249C point mutation. It has previously been demonstrated that the cysteine substitution in S249 within the extracellular domain of FGFR3 leads to constitutive receptor dimerisation and activation through the formation of an intermolecular disulphide bond (2,(9)(10)(11). To investigate the role of this cysteine residue in driving resistance to dasatinib, we engineered NIH3T3 cell lines expressing an FGFR3 mutant where the cysteine has been substituted to an alanine residue (S249A). Evaluation in non-reducing PAGE showed that the alanine substitution led to the loss of constitutive FGFR3 dimers both in the presence and absence of infigratinib ( Figure S1A). While cells expressing S249C were resistant to dasatinib in both short-term viability assays and long-term colony formation assays, the substitution to alanine resulted in pronounced sensitivity to this drug with levels comparable to WT FGFR3 expressing cells ( Figure S1B-D). Similar results were seen in cells engineered to express another common FGFR3 cysteine mutant found in urothelial carcinoma (R248C) and its glycine substituted mutant (R248G) (Figure S1E-F). Taken together, our data demonstrates that while treatment with dasatinib resulted in a decrease in Src phosphorylation levels in all FGFR3 expressing cell lines examined, the expression of FGFR3-TACC3 and S249C unexpectedly confers resistance to this drug. Furthermore, we show that in two of the most common FGFR3 extracellular cysteine mutants in urothelial carcinoma (S249C and R248C), the constitutive receptor dimerisation mediated by the cysteine residue is necessary for driving dasatinib resistance.

Dasatinib sensitizes NIH-3T3 cells with FGFR3 molecular alterations to infigratinib treatment
The use of combination therapy as a means to suppress multiple signalling pathways to overcome drug resistance is a well-established concept in cancer biology (27,28). Given that infigratinib and dasatinib act through parallel signalling pathways of Erk1/2 and Src respectively ( Figure 1F and 2F), we posited that the addition of dasatinib may enhance the activity of infigratinib through the dampening of the Src pathway to overcome the intrinsic resistance observed in long-term colony formation assays ( Figure 1G). Co-treatment of infigratinib with low dose (100-200nM) of dasatinib resulted in an almost complete elimination of the resistant colonies that were previously observed in the treatment with single agent infigratinib or dasatinib ( Figure 3A-B). Notably, this effect appeared to be more potent in cells expressing FGFR3-TACC3 and the FGFR3 point mutants with some resistant colonies remaining in the WT FGFR3 expressing cells even at high drug doses ( Figure 3B).
This result was recapitulated in the short-term cell viability assays with the FGFR3-TACC3 and S249C expressing cells having a more pronounced sensitivity to the combination than WT FGFR3 expressing cells ( Figure 3C).
To evaluate the signalling consequence of combination therapy with infigratinib and dasatinib, an assessment of Erk1/2 and Src activation levels after treatment with both drugs was performed by immunoblotting. As expected, cells bearing FGFR3-TACC3, S294C, N542K and K652E showed a decrease in the phosphorylation levels of both Src and Erk1/2 upon treatment with the combination, which is the result of the independent effects of dasatinib and infigratinib on these pathways respectively as single agents ( Figure 3D). In the S249C and K652E expressing cells, there was a more potent inhibition of phosphoErk1/2 in the drug combination compared to single agent treatment with infigratinib alone ( Figure   1F), suggesting that in a subset of FGFR3 mutants, dasatinib may enhance the effects of infigratinib by increased suppression of the Erk1/2 pathway.

Targeting the Src pathway enhances the efficacy of selective FGFR inhibitors in urothelial cancer cell lines with FGFR3 molecular alterations
We sought to establish if the drug responses observed in the NIH-3T3 model system is also present in cancer cell lines harbouring endogenous FGFR3 molecular alterations. To that end, we evaluated the effects of two structurally unrelated selective FGFR inhibitors (infigratinib and PD173074) and dasatinib as single agents or in combination in a panel of urothelial cancer cell lines. The panel included RT112M (FGFR3-TACC3) and 639V (R248C) as well as BFTC905 which expresses WT FGFR3 and was used as a control (29)(30)(31). Longterm colony formation assays show that in line with previous reports, RT112M was partially sensitive to infigratinib and PD173074 as single agents compared to BFTC905 and 639V ( Figure 4A) (23). It should be noted that despite the comparatively higher sensitivity observed in the RT112M cells, there were still a large proportion of residual colonies upon drug treatment, indicating that a significant fraction of cells displayed intrinsic resistance to selective FGFR inhibitors. As observed in the NIH-3T3 panel, WT FGFR3 expressing BFTC905 cells were sensitive to dasatinib treatment as a single agent while RT112M and 639V cells with FGFR3 molecular alterations were partially resistant to this drug. The addition of dasatinib at low dose enhanced the efficacy of both selective FGFR inhibitors in the RT112M and 639V cells resulting in a decrease in colony formation in the combination arm compared to single agent FGFR inhibitor treatment ( Figure 4A-B). We further evaluated if the dasatinib and infigratinib combination was synergistic in dose response cell viability assays ( Figure 4C). Assessment of the combination index (CI) showed that the combination of dasatinib and infigratinib led to a synergistic decrease in the cell viability (CI <1) in the RT112M and 639V cells across all drug doses tested up to 1µM ( Figure 4D).
To assess the signalling alterations associated with the observed phenotypes in the urothelial cancer cell lines upon drug treatment, immunoblotting of Erk1/2 and Src phosphorylation levels was undertaken. Infigratinib and PD173074 treatment did not affect phosphoErk1/2 levels in the BFTC905 and 639V cells but abolished phosphoErk1/2 levels in the RT112M cell line ( Figure 4E). Dasatinib treatment suppressed Src phosphorylation in all three cell lines which was accompanied by an increase of total Src levels. As observed in the NIH3T3 cells, the combination of a selective FGFR inhibitor with dasatinib led to a reduction in the phosphorylation levels of both Erk1/2 and Src in the two cell lines harbouring FGFR molecular alterations, RT112M and 639V. The Epidermal Growth Factor Receptor (EGFR) has previously been shown to be a compensatory bypass mechanism which confers intrinsic resistance to PD173074 treatment in the RT112M cell line (23). Since Src is known to be regulated by EGFR, we set out to ascertain if the observed Src phosphorylation in the urothelial cancer cell lines is influenced by EGFR activity. Treatment of the cell line panel with the EGFR inhibitor gefitinib either as a single agent or in combination with the selective FGFR inhibitors did not lead to a reduction in Src phosphorylation levels ( Figure S2A), confirming that Src is not an effector of EGFR activation but rather a parallel resistance pathway in urothelial cancer cell lines.
Given that dasatinib is a broad spectrum kinase inhibitor that blocks multiple kinases including Src (25,26), we sought to demonstrate that Src is the primary target of this drug and the causative driver of the observed intrinsic resistance to selective FGFR inhibitors.
Specifically, we evaluated the ability of the gatekeeper mutation in the avian Src gene (T338I), which blocks dasatinib binding to its kinase domain, to rescue the phenotypic effects of dasatinib (32). As a comparator, the WT Src gene and an empty vector (EV) control plasmid were used. Engineered urothelial cancer cell lines expressing each of the Src variants or the EV control were subjected to long-term colony formation assays in the presence of infigratinib and dasatinib as single agents or in combination. Expression of the gatekeeper mutant rescued the effects of dasatinib, conferring robust resistance to this drug in all three cell lines when compared to the WT Src and EV cells ( Figure S2B). Collectively, these experiments provide evidence that low dose dasatinib is highly effective in overcoming intrinsic resistance to selective FGFR inhibitor treatment in urothelial cancer cell lines expressing FGFR3 molecular alterations and that dasatinib mediates this effect primarily through blockade of the Src pathway.

Discussion
Despite some clinical success with selective FGFR inhibitors in cancers with FGFR3 molecular alterations, these therapies are not curative and there remains a substantial patient population that harbour intrinsic resistance to these drugs (14,15). There is a need to consider new treatment strategies to tackle intrinsic drug resistance in order to improve longterm outcomes in patients. In this study, we use a combination of NIH-3T3 and urothelial cancer cells to demonstrate that cells expressing a subset of FGFR3 molecular alterations co-opt the Src pathway to promote intrinsic resistance to selective FGFR inhibitors. While cells expressing the FGFR3-TACC3 fusion and extracellular cysteine mutants were resistant to the Src inhibitor dasatinib and selective FGFR inhibitor infigratinib as single agents, the addition of low dose dasatinib profoundly sensitized these cells to selective FGFR inhibitors.
Our data provides evidence that supports the use of this drug combination as a means to either overcome intrinsic drug resistance to FGFR inhibitors in the salvage therapy setting or as an upfront strategy to achieve durable drug responses in cancer patients with FGFR3  (35,36). In the paradigm of intrinsic resistance, Herrera-Abreu et al., has shown that EGFR activation is an escape mechanism to PD173074 treatment and that a combination of gefitinib and PD173074 could overcome drug resistance in urothelial cancer lines harbouring FGFR3 molecular alterations (23). Our study defines Src as a new bypass pathway by which urothelial cancer cells achieve intrinsic resistance to selective FGFR inhibitors. We further demonstrate that in the urothelial cancer models examined, Src is not an effector of the EGFR-mediated mechanism previously described by Herrara-Abreu et al., but rather a parallel resistance pathway that is independent of EGFR signalling. To our knowledge, our report is the first description of the use of Src inhibitors such as dasatinib as a means to overcome intrinsic resistance to selective FGFR inhibitors in cancers with FGFR3 molecular alterations.
Dasatinib has been previously evaluated as a monotherapy in urothelial carcinoma. RT4 cells, which expresses a different form of the FGFR3-TACC3 fusion employed in this study, are sensitive to treatment with dasatinib both in vitro and in vivo (37). Interestingly, it has been shown in another study that genetic silencing of Src in RT4 cells promotes cancer cell migration and invasion (38). Dasatinib has been clinically evaluated in the neoadjuvant setting in muscle-invasive urothelial carcinoma (39). This phase II trial showed that while treatment with this drug led to an on-target decrease in the phosphorylation of Src, there was no clinical benefit as reflected by the lack significant changes in proliferation and apoptosis markers in pre-and post-treatment tumours (39). Our data shows that expression of FGFR3-TACC3 and S249C results in a paradoxical resistance to dasatinib despite the constitutive activation of the Src pathway, reinforcing the idea that this drug as a monotherapy is likely to be ineffective in cancers with these FGFR3 molecular alterations. One of the primary ways to address intrinsic drug resistance is the use of combination therapies that abrogate compensatory survival mechanisms (27,40). We find that co-targeting FGFR3 and Src with low nanomolar concentrations of infigratinib and dasatinib respectively is highly effective in reducing cell viability in urothelial cancer cell lines. This drug combination simultaneously abrogates Erk1/2 and Src phosphorylation leading to durable efficacy in long term-colony formation assays, suggesting that simultaneous blockade of these parallel survival pathways is necessary to kill these cells. The use of Src inhibitors such as dasatinib to overcome resistance to kinase inhibitors has previously been investigated in the context of mutant EGFR lung cancer (41,42). A phase I clinical trial of dasatinib in combination with the EGFR inhibitor erlotinib in advanced non-small-cell lung cancer showed that the combination is tolerable with disease control in a subset of patients (43). The clinical experience in lung cancer suggests that it is feasible to evaluate the efficacy of dasatinib in combination with selective FGFR inhibitors in patients with FGFR3 molecular alterations.
There remain several outstanding questions. For instance, the mechanisms by which Src blockade sensitizes cells to FGFR3 inhibition are poorly understood. Since expression of the FGFR3-TACC3 fusion and the substitution of extracellular cysteines in FGFR3 confers resistance to dasatinib as a single agent, it is tempting to speculate a mechanism whereby dasatinib treatment alone has no therapeutic effect, while the addition of a selective FGFR inhibitor induces a new dependency to the Src pathway leading to vulnerability to cotreatment with dasatinib. Such induced collateral sensitivity has been observed with other tyrosine kinase inhibitors (44, 45). There is also a need to investigate the efficacy of this combination beyond the small panel of FGFR3 molecular alterations examined in this study in order to establish the generality of these findings and expand its utility to patients harbouring other FGFR3 mutations. Furthermore, it remains unclear why some FGFR3 aberrations (FGFR3-TACC3 and cysteine mutants) induce resistance to single agent dasatinib while others (N542K and K652E) do not. Finally, this study did not investigate the relevance of Src as a bypass compensatory pathway in the paradigm of acquired resistance to selective FGFR inhibitors. Future work will seek to determine if blockade of the Src pathway is able to re-sensitise cells in this acquired resistance context and achieve durable drug responses In summary, we show that consistent with the clinical experience, cells bearing a subset of FGFR3 molecular alterations are intrinsically resistant to selective FGFR inhibitors.
Expression of FGFR3-TACC3 and FGFR3 cysteine point mutants led to constitutive Src activation which can be exploited for therapy with the Src inhibitor dasatinib. Treatment with dasatinib in combination with selective FGFR inhibitors was highly effective in overcoming intrinsic drug resistance in urothelial cancer cells via the dual blockade of the Src and Erk1/2 pathways. This study provides proof-of-principle evidence that supports co-targeting Src and FGFR3 as a therapeutic approach to tackle drug resistance in the treatment of cancers with FGFR3 molecular alterations.

Anchorage independent cell growth in soft agar.
To evaluate growth in soft agar, cells were seeded at a density of 20,000 cells/well (in 6-well plates), in a solution containing 0.35% (w/v) Noble Agar (Sigma-Aldrich) in DMEM supplemented with 10% FBS. This was layered over a 0.3% (w/v) agar solution in DMEM also supplemented with 10% FBS. Plates were incubated for 3 weeks with regular media change. Colonies were stained with Giemsa Stain Modified Solution (48900, Sigma-Aldrich) with 1 part Giemsa to 5 parts glycerol:methanol (5:24 parts). Colonies with a diameter greater than 10 pixels were analysed on ImageJ software.

Colony formation assay
Cell growth in the presence of selected inhibitors was evaluated by colony formation assays, were cells were seeded at a density of 1000 cells/well (in 6-well plates) for 24 h, followed by addition of relevant drug treatment at the indicated concentrations. Fresh medium with drug was added twice a week for 14 days. Colonies were fixed with Carnoy's solution (1 acetic acid: 3 methanol) and stained with 1% (w/v) crystal violet (Sigma-Aldrich). The percentage of colonies covering the plates were calculated with ImageJ software.

Statistical Analysis
Statistical analysis was performed on GraphPad Prism 8 statistical software with one-way or two-way ANOVA with multiple comparisons tests. Data are presented as mean ± standard deviation (SD). The threshold for statistical significance was set to * p < 0.05, ** p < 0.01, or *** p < 0.001.