Dual Targeting of FGFR3 and ERBB3 Enhances The Ecacy of FGFR Inhibitors in FGFR3 Fusion-Driven Bladder Cancer

Mutations and fusions in Fibroblast Growth Factor Receptor 3 (FGFR3) occur in 10-20% of metastatic urothelial carcinomas and confers sensitivity to FGFR inhibitors. However, responses to these agents are often short-lived due to the development of acquired resistance. The objective of this study was to identify mechanisms of resistance to FGFR inhibitors in two previously uncharacterised bladder cancer cell lines harbouring FGFR3 fusions and assess rational combination therapies to enhance sensitivity to these agents. pERBB3 is a key mechanism of adaptive resistance to FGFR inhibitors in FGFR3-fusion driven bladder cancers, and that this also occurs rapidly following FGFR inhibitor treatment. Our ndings demonstrate that resistance can be overcome by combination treatment with a pan-ERBB inhibitor and suggest that upfront combination treatment with FGFR and pan-ERBB inhibitors warrants further investigation for FGFR3-fusion harbouring bladder cancers.


Introduction
Urothelial bladder cancer is responsible for approximately 150,000 deaths per year worldwide, and the median survival of patients with metastatic disease is approximately 18 months (1,2). Fibroblast Growth Factor Receptor 3 (FGFR3) is an attractive therapeutic target in bladder cancer given the 10-30% prevalence of FGFR3 aberrations (activating mutations or aberrant gene fusions) in these tumours, and their preclinical sensitivity to FGFR-targeted therapy (3).
A smaller proportion of bladder cancers (3-6%) have FGFR3 chromosomal translocations which generate oncogenic FGFR3 fusion proteins (4,16,17). These fusion proteins comprise of amino acids 1-760 of FGFR3 (which include the kinase domain) fused in-frame to either transforming acid coiled-coil 3 (TACC3) or BAI-Associated Protein 2-Like-1 (BAIAP2L1) (5), and form overexpressed, permanently dimerised inclusion bodies in the cytosol that do not undergo lysosomal degradation, and are not susceptible to feedback inhibition (17). The FGFR3 component of the fusion gene is identical, with a conserved breakpoint lacking only the nal exon (exon 19). Expression of these fusion proteins in normal human urothelial cells has been shown to induce mitogenic activation of the MAPK pathway (18).
Preclinical studies have demonstrated that human bladder cancer cell lines with FGFR3 mutations and fusions are sensitive to FGFR inhibitors such as BGJ398 (In gratinib, Novartis), PD173074 (P zer) and erda tinib (Balversa, Janssen) (17,(19)(20)(21)(22), thereby forming the basis for clinical trials of FGFR inhibitors in patients with metastatic urothelial cancer. Results from phase I and II trials of BGJ398 and erda tinib in this population reported response rates of 25-40%, and based on these ndings, erda tinib was FDA approved for FGFR2/3 aberrant bladder cancers in 2019 (23). Despite this success, the e cacy of single agent FGFR inhibitors is limited by the short duration of e cacy with a median progression-free survival of 3.7-5.5 months (24,25). Several studies have investigated the mechanisms driving inherent and acquired resistance to FGFR inhibitors. However, studies in models of FGFR3-fusion harbouring bladder cancer lines have so far been limited to a single cell line, RT-112, which harbours a FGFR3-TACC3 fusion as well as an ampli cation (26-28). Mechanisms of resistance described in this model include epithelialto-mesenchymal transition (26), activation of EGFR (27), ERBB2, and ERBB3 (26), and increased activation of AKT (28). However, whether these mechanisms extend to other FGFR3-fusion driven bladder cancer cell lines is unknown.
To address this, we undertook this preclinical study to investigate potential resistance mechanisms to FGFR inhibitors in two previously uncharacterised bladder cancer cell lines harbouring FGFR3-fusions, RT4 (FGFR3-TACC3 fusion) and SW780 (FGFR3-BAIP2L1 fusion). Through continuous culture in the presence of the FGFR inhibitor BGJ398, we derived lines highly resistant to BGJ398. These cell lines were also cross-resistant to erda tinib (balversa) and TAS-120 (futifatinib), an irreversible FGFR inhibitor that is currently undergoing phase II clinical testing (29). Pro ling of receptor tyrosine kinases revealed increased pERBB3 in both cell lines with acquired resistance to FGFR inhibition, and we demonstrated that combination treatment with the pan-ERBB inhibitor AZD8931 can re-sensitise these cell lines to FGFR inhibitors. We also demonstrate that pERK and pERBB3 are rapidly reactivated in FGFR3-driven cell lines following FGFR inhibitor treatment, which could also be overcome by ERBB receptor blockade. These ndings suggest combination treatment with an FGFR and pan-ERBB inhibitor from the outset may represent a more effective approach for treating FGFR3-fusion driven bladder cancers.

Methods
Cell lines, culture and reagents

Cell cycle and apoptosis assays
Changes in cell cycle kinetics and apoptosis were assessed following 24 hours of drug treatment using Propidium Iodide staining described previously (30), followed by FACS analysis using a BD FACS Canto II ow cytometer (BD Biosciences San Jose, CA). The percentage of apoptotic cells was determined by calculating the proportion of cells with a sub-diploid DNA content using FLOWJO software V10.0 (FlowJO LLC, Ashland, OR, USA).
Phospho-receptor tyrosine kinase (RTK) arrays RT4-RS and SW780-RS cells were cultured in fresh media without drug for 24 hours before collection to negate effects induced by acute drug exposure. Control and resistant lines were then lysed in Radio immunoprecipitation assay (RIPA) buffer (Sigma-Aldrich, St Louis, MO, USA) containing complete Protease Inhibitor Cocktail Tablets (Roche, Basel, Switzerland) and PhosSTOP (Roche). A total of 200 mg of lysate was then incubated with human Phospho-RTK Arrays (ARY001B, R&D Systems, Minneapolis, MN, USA) as per manufacturer's instructions, and blots were read using the ChemiDoc X Imaging System (Bio-Rad, Hercules, CA, USA).

Statistical analysis
Statistical analyses were performed using Prism v5 (GraphPad Software, La Jolla, CA, USA). Data shown is mean±SEM from 3 technical replicates from a representative experiment unless stated otherwise. Biological replicates were performed for the majority of experiments and are stated in the gure legends.
Groups were compared using parametric unpaired Student's t-test with Welch's correction. P-values ≤0.05 were considered to be statistically signi cant. The effect of drug combinations on cell growth in vitro was assessed using the BLISS synergy and antagonism model in the Combene t software

Results
Generation of bladder cancer cell lines with acquired resistance to FGFR inhibitors SW780 and RT4 cell lines with acquired resistance to the FGFR inhibitor BGJ398 were generated by continuous culture in 1 µM BGJ398 for 3 months, a dose which is approximately 2-fold higher than the clinically reported Cmax of BGJ398 (31). The cell lines were named SW780-RS and RT4-RS and were subsequently maintained at 1 µM BGJ398. Response of the resistant cell lines to BGJ398 was subsequently assessed by CellTitre-Glo assay, which con rmed that SW780-RS and RT4-RS cells were signi cantly more resistant to BGJ398 than matched parental control cell lines that had been cultured in parallel. Similar resistance was demonstrated in SW780-RD and RT4-RD cells lines, generated by a graduated increase in exposure to BGJ398 ( Figure 1A, E). To determine whether resistance to BGJ398 resulted in cross-resistance to other FGFR inhibitors, including erda tinib which was recently approved for FGFR2/3 driven bladder cancers, and the covalently binding irreversible FGFR inhibitor TAS-120 (futibatinib) (32) which is currently in early-phase clinical trials, sensitive and resistant clones were treated with increasing doses of both agents and growth inhibition assessed by CTG assay. As shown in Figure 1, SW780 and RT4 clones that were resistant to BGJ398 were also found to be cross-resistant to erda tinib ( Figure 1B, F) and TAS-120 ( Figure 1C, G). Comparatively, all clones responded similarly to the MEK inhibitor trametinib ( Figure 1D, H), indicating resistance was speci c to FGFR inhibitors.

Investigation Of The Mechanisms Of Acquired Drug Resistance
To investigate the mechanistic basis for acquired resistance to FGFR inhibition, the phosphorylation status of 49 receptor tyrosine kinases (RTKs) was compared between parental and SW780-RS and RT4-RS cells using phospho-RTK arrays. This screen identi ed an increase in pERBB3 and pAXL in both SW780-RS and RT4-RS cell lines ( Figure 2A) and increased pEGFR and pERBB2 in SW780-RS cells. However, only the increase in pERBB3 was con rmed by western blot, which was also observed in SW780 RD and RT4 RD cells ( Figure 2B).

Investigation of combination therapy targeting FGFR3 and ERBB3 in FGFR-resistant bladder cancer lines
Based on the consistent increase in pERBB3, we examined the effect of combining BGJ398 with the pan-ERBB family inhibitor AZD8931 (sapitnib) in FGFR inhibitor-resistant cell lines (33). Combination treatment of SW780-RS and RT4-RS cells with BGJ398 and AZD8391 synergistically inhibited cell growth in both SW780-RS and RT4-RS cells ( Figure 3A, D). Assessment of the effect of this combination on cell cycle kinetics revealed a signi cant reduction in the percentage of cells in S phase and a concomitant increase in the percentage of cells in G0/G1, even when 10-fold (SW780-RS) or 20-fold (RT4 RS) lower concentrations of BGJ398 were used compared to the 1 mM dose used to generate resistance ( Figure 3B, E). Comparatively, minimal induction of apoptosis was observed, indicating this combination predominantly induces G0/G1 cell cycle arrest ( Figure 3C, F).
Reactivation of pERBB3 and MAPK signalling is an early adaptive mechanism of FGFR inhibition While these analyses identi ed mechanisms of acquired resistance associated with long-term FGFR inhibition, it is now evident that several tumours, including bladder cancers, can also rapidly adapt to targeted therapies and reactivate signalling through various mechanisms (26, 34,35). We therefore assessed the effect of FGFR inhibition on MAPK signalling in SW780 and RT4 parental cells over 72 hours. Remarkably, while BGJ398 initially suppressed pERK levels at 4 hours, the magnitude of suppression gradually diminished, and pERK levels rebounded to close to basal levels by 72 hours ( Figure   4A, B). We next investigated whether this feedback was associated with changes in ERBB3 and other ERBB family members. While minimal induction of pERBB2 or pEGFR was observed over the 72-hour time course in either cell line, robust induction of pERBB3 was observed in both cell lines within 24 hours.
We therefore investigated whether the rapid reactivation of MAPK signalling in FGFR-fusion harbouring bladder cancer cells could be attenuated by combined treatment with an FGFR inhibitor and the pan-ERBB inhibitor, AZD8931, and whether this could enhance the growth inhibitory effect of FGFR inhibitors. Indeed, combined treatment of both RT4 and SW780 parental cells with BGJ398 and AZD8931 signi cantly attenuated the induction of pERBB3, and further suppressed pERK levels compared to the effect of either agent alone ( Figure 5A, E). Furthermore, combination treatment with BGJ398 and AZD8931 synergistically inhibited cell growth in both SW780 and RT4 cells ( Figure 5B, F). Notably, this effect was observed at a concentration of 0.1 and 0.05 µM BGJ398 in SW780 and RT4 cells respectively, which is 5-10-fold lower than the clinically achievable concentration of 0.5 µM (31). As observed in the long-term resistance setting, the combination induced a signi cant reduction in the percentage of cells in S phase and a concomitant increase in the percentage of cells in G0/G1 ( Figure 5C, F). Comparatively, minimal induction of apoptosis was observed, indicating this combination predominantly induces G0/G1 cell cycle arrest ( Figure 5D, H).
Finally, we assessed whether synergistic suppression of cell proliferation of these FGFR3-fusion harboring cell lines was also induced when the FGFR inhibitors erda tinib and TAS-120 were combined with AZD8931. Indeed, combining either erda tinib or TAS-120 with AZD-8931 induced synergistic suppression of proliferation of SW780 and RT4 parental cells, demonstrating this is a highly effective combination regimen that can enhance the activity of multiple FGFR inhibitors ( Figure 6A-D).
Discussion FGFR inhibitors such as BGJ398 and the clinically approved agent erda tinib induce objective responses in ~25-40% of patients with metastatic urothelial cancer harbouring FGFR3 alterations (24,25,36,37). However, the median progression-free survival in these patients is typically less than six months. These ndings point to the existence of two types of resistance -inherent resistance which precludes 60-70% of patients responding at all, and acquired resistance, where tumours which initially respond invariably develop resistance. These observations are similar to the extensive clinical experience with other targeted therapies where initial responses are often followed by tumour progression after 6-12 months (38-40). To better understand the mechanisms of inherent and acquired resistance to FGFR inhibition, we undertook in vitro modelling of these processes in two previously uncharacterised FGFR3-fusion driven cell lines, SW780 (FGFR3-BAIP2L1 fusion) and RT4 (FGFR3-TACC3 fusion).
Interestingly, a prior study reported that not all FGFR3 altered cell lines remain dependent on FGFR for growth in vitro, with some cell lines switching their dependence to ERBB family members (27). While the basis for this switch is unknown, it may re ect differences in the availability of ERBB family ligands in the serum, as the exogenous provision of the ERBB2/3 ligands NRG1 or NRG2 has been shown to attenuate the sensitivity of FGFR3-fusion harboring RT-112 cells to BGJ398 (26). Notably, while both of the FGFR3fusion harbouring cell lines investigated in the current study were sensitive to FGFR inhibition, we did note considerable variability in response in the SW780 cell line with different serum batches and serum concentrations (data not shown), indicating this is an important consideration in these studies.
In this study, we examined resistance mechanisms to FGFR inhibition in two contexts -acquired resistance following long term exposure, and rapidly following 4-72 hours treatment -in two previously uncharacterised models of bladder cancer cell lines harbouring FGFR3-fusions. While pro ling of changes in receptor tyrosine kinases in cell lines with acquired resistance identi ed increased expression of several candidates, only expression of pERBB3 was consistently increased in both cell line models and was independent of the mode of resistance induction (RS vs RD). Importantly, combination treatment with the pan-ERBB inhibitor AZD8931 profoundly re-sensitised these lines to FGFR inhibition, indicating a direct role for pERBB3 in conferring resistance. This nding is consistent with observations previously reported in RT-112 cells (26, 41), suggesting this may be a common resistance mechanism in these tumours.
An important nding of the current study is the capacity of FGFR3-fusion harbouring bladder cancer cells to rapidly reactivate MAPK signalling. We demonstrated that this is associated with increased pERBB3 levels, and that the increase in pERBB3 and reactivation of pERK could be attenuated by combined treatment with the pan-ERBB inhibitor, AZD8931. In parallel, combined FGFR/ERBB inhibition further inhibited cell growth of FGFR3-fusion driven bladder cancer lines. These ndings are consistent to those previously reported in RT-112 cells, where pERBB2 and pERBB3 were both rapidly elevated within 24 hours of BGJ398 treatment (26). Furthermore, Herrera-Abreu et al also demonstrated that RT-112 cells express high basal levels of EGFR, and that EGFR-signaling is rapidly induced following BGJ398 treatment, limiting its response to BGJ398 (27). While we did not observe a rapid increase in pERBB2 or pEGFR in SW780 and RT4 cells following BGJ398 treatment, both receptors are expressed in these cells, particularly EGFR which is expressed at high levels. As ERBB3 can activate signalling through heterodimerisation with ERBB2 and EGFR (42), it is possible that ERBB3 may heterodimerise with existing ERBB2 and EGFR to reactivate MAPK signalling in these cells. On the other hand, cross-talk between ERBB and FGFR receptors has also been described, with ERBB3 required for the maintenance of FGFR2 phosphorylation and proliferation in some FGFR2-ampli ed gastric cancer cells (43). A further possibility therefore is that ERBB3 may cooperate with FGFR3 to reactivate MAPK/ERK signalling. While additional studies are required to de ne the speci c mechanisms by which ERBB receptors contribute to signalling in FGFR3-fusion driven bladder cancers following FGFR inhibition, our ndings in the SW780 and RT4 cell lines adds to the previous ndings in RT-112 cells to demonstrate compensatory activation of ERBB3driven signalling as a consistent mechanism of rapid adaptive resistance to FGFR inhibitors in these tumours.
Critical to preventing the emergence of resistance is the more e cient targeting and elimination of FGFRdriven tumours from the onset of treatment. In this regard, the nding that compensatory activation of ERBB3-driven signalling occurs rapidly in response to FGFR inhibition in multiple models of FGFR3-fusion harbouring bladder cancer, and our demonstration that dual targeting of FGFR and ERBB family receptors synergistically inhibits growth, suggests that dual blockade of FGFR and ERBB-driven signalling may represent a more effective treatment strategy for these tumours. Additionally, ERBB3-targeting antibodies are now entering clinical trial for other indications (e.g seribantumab for NRG1-fusion harboring tumours) (44) (NCT04383210), which may provide further options for clinical investigation of this treatment concept.

Conclusions
In summary, we identify consistent increased activation of pERBB3 as a resistance mechanism to FGFR inhibition in FGFR3-fusion driven bladder cancer cell lines. Our ndings suggest that upfront combination treatment with FGFR and ERBB3 inhibitor warrants further investigation for FGFR3-fusion driven bladder cancers.

Declarations
Ethics approval and consent to participate: Not applicable Consent for publication: Not applicable Availability of data and materials: All data generated or analysed during this study are included in this published article [and its supplementary information les].