Characterization of T‐DM1‐resistant breast cancer cells

Abstract The development of targeted therapies has drastically improved the outcome of patients with different types of cancer. T‐DM1 (trastuzumab‐emtansine) is an antibody‐drug conjugate used for the treatment of HER2‐positive breast cancer combining the FDA approved mAb (monoclonal antibody) trastuzumab and the microtubule cytotoxic agent DM1 (emtansine). Despite clinical successes achieved by targeted therapies, a large number of patients develop resistance during treatment. To explore mechanisms of resistance to T‐DM1, the MDA‐MB‐361 HER2‐positive breast cancer cell line was exposed in vitro to T‐DM1 in the absence or presence of ciclosporin A. Previously reported mechanisms of resistance such as trastuzumab‐binding alterations, MDR1 upregulation, and SLC46A3 downregulation were not observed in these models. Despite a decrease in HER2 expression at the cell surface, both resistant cell lines remained sensitive to HER2 targeted therapies such as mAbs and tyrosine kinase inhibitors. In addition, sensitivity to DNA damaging agents and topoisomerase inhibitors were unchanged. Conversely resistance to anti‐tubulin agents increased. Resistant cells displayed a decreased content of polymerized tubulin and a decreased content of βIII tubulin although the downregulation of βIII tubulin by siRNA in the parental cell line did not modified the sensitivity to T‐DM1. Both cell lines resistant to T‐DM1 also presented giant aneuploid cells. Several SLC (solute carrier) transporters were found to be differentially expressed in the resistant cells in comparison to parental cells. These results suggest that some characteristics such as increased baseline aneuploidy and altered intracellular drug trafficking might be involved in resistance to T‐DM1.


| INTRODUC TI ON
Human epidermal growth factor receptor 2 (HER2) is amplified/overexpressed in approximately 20% of breast cancers and is associated with poor outcome and a high risk of recurrence. [1][2][3][4] Direct targeting of HER2 has dramatically improved the management of HER2positive breast cancer patients. 5,6 The monoclonal antibody (mAb) trastuzumab represented the first breakthrough in HER2-targeted therapies and was followed by pertuzumab, followed by the development of antibody-drug conjugates (ADCs). 7,8 ADCs combine the selectivity of mAbs with highly potent cytotoxic molecules, allowing to directly deliver the drug within the targeted cancer cell while reducing off-target toxicity. 9,10 Consequently, this approach increases the therapeutic window of the drug.
Trastuzumab emtansine (T-DM1, Kadcyla) is the first FDAapproved ADC for the treatment of HER2-positive metastatic breast cancer as second-line therapy, and is also active in earlier stages of the disease. [11][12][13] T-DM1 is comprised of the monoclonal antibody trastuzumab conjugated to DM1 (emtansine), a derivative of maytansine, via a noncleavable linker (SMCC). 14 T-DM1 selectively binds to HER2 and delivers a potent tubulin-binding agent within targeted cancer cells while maintaining Fc-related antitumor activities of trastuzumab. 9 Despite the clinical success of T-DM1, some patients initially responding develop resistance during treatment while others present primary resistance to this agent. 15 ADCs are relatively recent agents in the clinic and exact mechanisms of resistance to this family of agents still require in-depth studies. However, numerous resistance mechanisms to T-DM1 and others FDA-approved ADCs have been highlighted so far, affecting antibody binding, ADC degradation, or conjugate availability or toxicity. T-DM1 is prone to resistance mechanisms involving reduced trastuzumab binding, including HER2 downregulation, mutations, and masking, such as enzymatic cleavage of HER2 into p95HER2 and MUC4 masking, respectively. [16][17][18] Others possible mechanisms involve the overexpression of efflux transporters such as ATPbinding cassette (ABC) transporters. 15,[19][20][21] More recently it has been shown that reduced intracellular release of the conjugate from the lysosomal compartment, due to reduced lysosomal proteolytic activity or altered lysosomal efflux transport, might be involved in resistance. 22,23 The solute carrier (SLC) transporter SLC46A3 is a substrate for Lys-MCC-DM1, the active metabolite of T-DM1, and its downregulation is involved in T-DM1 resistance, by limiting its cytoplasmic release. 23 The upregulation of proteins involved in the regulation of the actin/tubulin cytoskeleton has been observed in resistance models to trastuzumab-maytansinoid antibody-drug conjugates. 24,25 In order to identify novel mechanisms of resistance to T-DM1, we selected and characterized cells resistant to T-DM1 using the MDA-MB-361 breast cancer cell line. To avoid the exclusive emergence of MDR-overexpressing variants, we performed selection in the presence or absence of ciclosporin, a potent MDR inhibitor.

| Cell culture
The human breast adenocarcinoma cell line MDA-MB-361 was cultured in DMEM medium supplemented with 10% fetal calf serum and 100 µg/mL streptomycin at 37°C and 5% CO2. Cells were counted using a Cellometer Auto T4 (Nexcelom Bioscience LLC).
Selection of TR and TCR cells was performed by exposure to increasing concentrations of T-DM1 for 6 months. Ciclosporin A (CsA, C3662; Sigma-Aldrich) at 1 µg/mL was also added at the same time as T-DM1 for the selection of the TCR cell line.

| Cytotoxicity assays
Cells were seeded in 96-well plates at a density of 8000 cells per well and incubated overnight. Increasing concentrations of chemotherapy agents were added to the media and 6 days later, viability was determined by the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) assay. 20 µL of a 5 mg/mL MTT solution was added to each well and plates were incubated at 37°C for 4 hours. The media/MTT mix were removed and 100 µL of 4% HCl 1N/isopropanol per well were added to dissolve the purple formazan crystals. The absorbance was measured at 570 nm with 690 nm as a reference readout using a Thermo MultiSkan EX microplate reader. The absorbance of drug-exposed and control cells was compared to determine the percentage of living cells.
IC50 values were calculated using CompuSyn and Graphpad Prism softwares.

| Real-time cell analysis
The xCELLigence RTCA DP instrument (ACEA Bioscience) monitors cell impedance in real time. Cells were inoculated in E-plate 16 at a density of 10 000 cells per well and incubated overnight before the addition of cytotoxic agents. The cell index was monitored for one week.

| Efflux assays
Cell suspension was prepared with 4 e 6 cells in 10 mL of DMEM media containing 0.5 µg/mL Rhodamine 123 (Santa Cruz, sc-208306) and incubated for 30 minutes at 37°C, 5% CO 2 . Cells were washed three times in cold DPBS on ice and some cells were taken for flow cytometry analysis ("uptake"). The remaining cells were incubated in DMEM media in the absence or presence of 3 µg/mL of CsA and incubated for 24 hours. Cells were suspended using trypsin and washed on ice before flow cytometry analysis. 26

| Annexin-X/propidium iodide apoptosis assay
Cells were seeded in 6 well plates at a density of 2 e 5 cells per well and incubated overnight. Then cells were exposed to T-DM1 for 72 hours. Cells were harvested, washed with cold DPBS containing 10% SVF and stained using Annexin-V-FLUOS Staining Kit (Roche) according to the manufacturer's protocol. Analysis was performed by flow cytometry. Annexin-V-positive cells exposed to T-DM1 were normalized to a control for each cell line.

| Cell cycle distribution analyses
Cells were seeded as described for the apoptosis assay, incubated and exposed to 100 nmol/L T-DM1 for 24 hours. Cells were harvested, washed with cold DPBS and incubated for 30 minutes at 4°C with propidium iodide (0.05 mg/mL) containing Nonidet-P40 (0.05%) and 4 µmol/L of trisodium citrate. Cells were filtered using Falcon tubes with cell-strainer cap (352 235) before flow cytometry analysis.

| Western blots
Protein extraction was performed using complete RIPA buffer (RIPA buffer, 1 mmol/L DTT, 1M NaF, 100 mmol/L sodium orthovanadate, phosphatase inhibitor buffer and protease inhibitors). After SDS PAGE separation, proteins were transferred onto a PVDF membrane by iBlot dry blotting system (Invitrogen). Membranes were incubated overnight at 4°C with primary antibodies and 1 hour at room temperature with secondary antibodies (IRDye Infrared Dyes from LI-COR Biosciences). Primary antibodies used were as follows: HER2 (GTX50425; Genetex), βIII-tubulin (clone TUJ1), βII-tubulin (clone 7B9), α-tubulin (T6199), β-tubulin (T4026), and β-actin (A5441) from Sigma-Aldrich. Membranes were scanned using Odyssey infrared imaging system (LI-COR Biosciences) and densitometric quantification was performed with Odyssey software. Expression levels of proteins were normalized against β-actin. Cells were ultracentrifuged (100 000 g for 1 hour at 20°C) to separate soluble tubulin (supernatant) and microtubules (pellet). The pellet was suspended in 100 µL of PEM 50DP buffer, incubated on ice for 30 minutes to depolymerize tubulin and ultracentrifuged at 50 000 g for 45 minutes at 4°C to recover the supernatant. The supernatant was incubated with 1 mmol/L GTP for polymerization at 35°C for 30 minutes and ultracentrifuged at 50 000 g for 45 minutes at 35°C. After centrifugation, the supernatant was discarded and the pellet containing was suspended in 50 µL of PEM 50DP buffer. Tubulin amount in both fractions was determined by Western blot.

| RT-qPCR
RNA extraction was performed with the QIAamp RNeasy Mini Kit (QIAGEN) and followed by reverse transcription. Primers were designed according to Roche sequences and the NCBI primer-BLAST designer (Table S1) and quantitative PCR analysis were performed with the LightCycler 480 Real-Time PCR system (Roche Life Science). Gene expression was normalized using ribosomal 28S as the house-keeping gene.

| Statistical analyses
Each experiment was performed at least three times and the results were presented in graphs as the mean ± SD. Graphs and statistical analyses were performed using GraphPad Prism software. Statistics on cell survival experiments such as AnnexinV/PI staining or MTT assay were performed by two-way ANOVA followed by a Bonferroni posttest. Statistics on gene expression by RT-qPCR and fold calculation in MTT assays were performed with the Student's t test.

| Materials
T-DM1 and S-methyl DM1 were kindly provided by Roche and ImmunoGen, respectively. Pertuzumab and cisplatin were purchased from Mylan. Trastuzumab was purchased from Virbac. Afatinib, vinorelbine, and lapatinib were purchased from Vidal. Fluorouracil and doxorubicin were purchased from Accord Healthcare. DM1 (emtansine) and colchicine were purchased from Abcam and Sigma, respectively. Paclitaxel and vincristine were purchased from Bristol Myers and Teva, respectively. Irinotecan was purchased from Hospira. PNU-159682 was kindly provided by Mablink Bioscience.

| In vitro generation of MDA-MB-361 models resistant to T-DM1
MDA-MB-361-resistant cells were selected in vitro by constant exposure to increasing concentrations of T-DM1. The initial concentration of T-DM1 was 20% of the IC50 measured after a 72hour exposure and was gradually increased. The final concentration of T-DM1 reached 0.4 nmol/L, which corresponds to two times the initial IC50. Cell line selection was performed in the absence or presence of ciclosporin, a modulator of MDR1, a member of the ABC transporter family, as this transporter has been reported to perform efflux of DM1 outside the cells. 27,28 Consequently, ciclosporin A (CsA) was used to inhibit MDR1 and avoid increased efflux activity.
Two cell lines resistant to T-DM1 were therefore selected in the ab-

| Sensitivity to anti-cancer agents
Regarding resistance to T-DM1 the IC50 determined by MTT assay was increased by fivefold in the TR cell line and by eightfold in the TCR cell line when compared to the parental cell line ( Figure 1A).
The IC50 calculated by xCELLigence was also increased in TR cells by 73-fold and TCR cells by 12-fold compared to S cells ( Figure 1B).
Apoptosis was analyzed by Annexin V staining after exposure to T-DM1 for 6 days and a decreased sensitivity to T-DM1-induced apoptosis in TR and TCR cells was observed, compared to S cells ( Figure 1C). Altogether, these results indicate that the selected TR and TCR cell lines are resistant to T-DM1.
To assess the sensitivity of parental and resistant cell lines to various anti-cancer agents, cytotoxicity assays were performed using targeted therapies and chemotherapy agents (Table 1). We did not observe altered sensitivity to HER2-targeting agents, including the two FDA-approved antibodies pertuzumab and trastuzumab nor against the tyrosine kinase inhibitors lapatinib while the TR line displayed low-level resistance to afatinib. Both cell lines resistant to T-DM1 remained sensitive to DNA damaging agents and topoisomerase I inhibitors, indicating that DNA-repairing machinery may not be involved in resistance to T-DM1. Both cells developed resistance to tubulin-binding agents such as paclitaxel, vincristine, vinorelbine, colchicine, DM1, and S-methyl DM1.

| HER2 expression is decreased in cells resistant to T-DM1
HER2 expression at the cell surface is required for T-DM1 activity. Therefore, its expression was investigated by RT-qPCR, western blot, and flow cytometry. A downregulation of HER2 was observed at the mRNA and the protein levels in both resistant cell lines compared to the parental cell line (Figure 2A-C). Surprisingly, a heterogeneous population expressing HER2 high and HER2 low in TR and TCR cell types was identified. This indicates that during the prolonged exposure to T-DM1 a subpopulation of cells expressing low levels of HER2 was selected, with presumably reduced sensitivity to the cytotoxic agent.

| Efflux activity is increased in resistant models
Overexpression of ABC transporters is a common mechanism of multidrug resistance. [29][30][31] To verify whether resistance to T-DM1 was performed by two-way ANOVA followed by Bonferroni posttests and differences are shown for TR (***: P < .001; **: P < .01; *: P < .05) and TCR (+) compared to S. (B) Parental and resistant cells were exposed to increasing concentrations of T-DM1 and the cell index was followed by xCELLigence. The slopes of the normalized cell index determined by the RTCA software were plotted. Statistical analysis was performed by Two-way ANOVA followed by Bonferroni posttests and differences are shown for each cell line between control and exposed conditions (*: P < .05; ***: P < .001). (C) Annexin-positive cells were studied by flow cytometry after a 6-day exposure to T-DM1. The percentage of Annexin-positive cells decreased in TR and TCR compared to parental cells. Statistical analysis was performed by two-way ANOVA followed by Bonferroni posttest (*: P < .05; ***: P < .001) was due to the enhanced expression of ABC transporters, a rhodamine 123 (Rho 123) accumulation assay was performed ( Figure 3A).  Note: The sensitivity after a 6 day exposure to the indicated anti-cancer agents was studied by MTT cytotoxic assays. Data shown as the IC50 values of a single experiment, representative of 3-4 independent experiments, and the relative resistance is the mean ratio of the IC50 of each resistant cell line over the parental cell line from 3-4 independent experiments (#: IC75). Statistics were calculated using the Student t test (*P < .05; **P < .01; ***P < .001).
Bold values indicate relative resistance numbers

| SLC gene expression is altered in resistant cell lines
Solute carrier (SLC) transporters are involved in the transport of a wide range of substrates across membranes. Particularly, SLC46A3 has been found to allow the cytoplasmic release of Lys-MCC-DM1 from lysosomes. RNAseq data have shown expression changes in 52 SLCs (data not shown) among which 17 are presumably located in endosomes and/or lysosomes (Table S1). Their expression was assessed by RT-qPCR in parental and resistant cell lines (Figure 4).
Most of the results are in line with RNAseq data (Table S1)

| Tubulin βIII expression and polymerized tubulin fraction were decreased in resistant models
The microtubule/tubulin complex is the major intracellular target Overall these results suggest that T-DM1-resistant variants display complex modifications of their microtubule/tubulin complex.
The contribution to resistance is likely to be multifactorial since alteration of a single parameter such as tubulin III beta content was not sufficient to reproduce the resistance phenotype.

| T-DM1-induced cell cycle arrest is maintained in resistant cell lines
Cell cycle distribution after a 24-hour exposure to T-DM1 was stud-

| T-DM1 resistant cells are giant and aneuploid
To confirm the differences in cell cycle distribution observed in resistant cells, cell line DNA profiles were investigated by propidium iodide staining and flow cytometry in confluent cells ( Figure 7A).

| D ISCUSS I ON
Clinical studies have demonstrated that T-DM1 is prone to lose its benefit in some patients despite continuous treatment. 37,38 We se-  In spite of a slightly increased efflux activity, we did not observe cross-resistance to other agents such as doxorubicin, suggesting that the increased efflux was not sufficient to protect cells against the toxicity of these molecules.
Both resistant cell lines remained sensitive to T-DM1-induced cell cycle arrest, while increased fractions of aneuploid cells were observed in the TR and TCR cell lines both at baseline and after exposure to T-DM1. Additionally, resistant cells were larger than parental cells. These data suggest that endoreduplication is more common in resistant cells than in parental cells. Aneuploidy is associated with chromosomal instability, an increase in gene transcription, heterogeneity and adaptability, thus potentially favoring subpopulation evolution and resistance to treatment. 40,41 Correlations between aneuploidy and resistance to anti-cancer agents have been described in previous studies suggesting that the aneuploid character of both TR and TCR cell lines compared to the sensitive one is potentially involved in the resistance to T-DM1. [42][43][44] The solute carrier (SLC) transporter family includes 418 members, divided into 52 families. The substrates for these transporters are highly diverse and some of the families are known to interact specifically with a type of substrate such as organic anions and cations (SLC22 or SLCO, respectively), nucleosides (SLC28A and SLC29A), oligopeptides (SLC15A), 45 and amino acids (SLC1, SLC3, SLC6, SLC7, SCL25, and SLC36). 46 Also, some SLC families have been reported to confer sensitivity and resistance to cytotoxic drugs (SL21, SLC22, SLC47, SLC31, SLC15, SLC7A, SLC38A2). 47 SLC transporters mainly allow their substrates to cross the cell membrane. However, as some of them are located in endosomes or lysosomes, they could be responsible for nonmembrane permeable drug transport across the endosomal or lysosomal membrane. Thus, a variable expression of these SLC may affect the ADC's active metabolite release and therefore reduce its intracellular concentration. SLC46A3 is a substrate for noncleavable ADCs metabolites such as Lys-MCC-DM1, the active metabolite of T-DM1 and it has been shown that a downregulation of its expression is a mechanism of resistance to T-DM1. 15,48 In our resistant variants, significant modifications of expression of four SLCs presumably located in the endosome and/or lysosome (SLC27A6, SLC29A3, SLC2A12, and SLC30A2) may affect lysosomal trafficking, degradation or release of Lys-MCC-DM1.
ZnT2 (encoded by SLC30A2) is a zinc transporter from the cytoplasm into lysosomes. 49 The upregulation of this transporter may induce zinc accumulation into the lysosome and trigger lysosomal dysfunction. 50 Other SLC transporters presumably located in lysosomes or endosomes were found to be differentially expressed in resistant cell lines. SLC9B2 is under-expressed in the TCR cell line and encodes for NHA2 protein that belongs to the sodium hydrogen antiporter family. 51 Aside from being involved in cell pH and sodium regulation, it has also been reported that a downregulation of NHA2 protein could affect endocytosis 52 and therefore could limit T-DM1 internalization. Further studies are required to determine whether these SLCs are involved in T-DM1 resistance mechanisms.
In conclusion the variants resistant to T-DM1 did not display enhanced drug efflux while a decreased target antigen expression is not likely to be the main mechanism for drug resistance. Our results suggest that resistant cells possess an altered phenotype, including size and ploidy. Numerous alterations in SLC transporters, including several reported to be contained in lysosomes, were observed, warranting additional investigations of these transporters in the determination of resistance to T-DM1.

ACK N OWLED G M ENTS
This work was supported in part by Roche France.

E TH I C A L S TATEM ENT
Ethical statement is not required.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.