Resistance to the Antibody–Drug Conjugate T-DM1 Is Based in a Reduction in Lysosomal Proteolytic Activity Free

Increased levels of the transmembrane tyrosine kinase HER2 are present in 20% of breast tumors and are associated with poor patient outcome (1). Those findings, together with preclinical studies that linked HER2 overexpression to a tumorogenic phenotype (2), led to the development of agents directed to HER2 (3, 4). Two types of agents have reached the oncology clinic (4, 5). On the one hand, humanized monoclonal antibodies, such as trastuzumab (6) or pertuzumab (7), that bind to the extracellular domain of HER2, preventing HER2 signaling and also triggering immune-mediated responses (8). On the other hand, the small molecule kinase inhibitor lapatinib acts by competitively displacing ATP from its binding pocket in the intracellular region of HER2, neutralizing the kinase activity of HER2 and ultimately provoking cell-cycle arrest and apoptosis (9).

In addition to the potential oncogenic role of HER2, its overexpression on the membrane of tumor cells opened the possibility of using it as a docking site to preferentially deliver cytotoxic agents to tumor cells (10). To fight HER2 tumors, such a composite molecule has been created using trastuzumab as the backbone to which the anti-microtubular maytansinoid drug emtansine (DM1) has been linked (11). The conjugate termed trastuzumab-emtansine (T-DM1) is stable in the bloodstream and, upon interaction with cell surface HER2, is internalized and brought to intracellular compartments, where it is degraded or recycled to the cell surface. The process of degradation of T-DM1 appears critical for its antitumoral activity. In the lysosomes, acidic proteases progressively fragment the antibody part of T-DM1 releasing the active payload product Lys-MCC-DM1, which is then transported to the cytosol where it can act on tubulin (12). The targeted cells are then arrested in mitosis, which leads to cell death by mitotic catastrophe (13, 14).

T-DM1 reached the market in 2013 for the therapy of patients with HER2⁺ metastatic breast cancer that progressed to the combination of trastuzumab and a taxane (15). Although the clinical results with T-DM1 are impressive, some patients do not respond to the drug or relapse (16, 17). In this study, we have explored the mechanism(s) of resistance to T-DM1. Genomic as well as functional studies demonstrated the presence of several mechanisms of resistance to the drug. One of them, uncovered by systematic analysis of the path followed by the drug, was linked to altered lysosomal function.

BT474 cell line was obtained from the ATCC (Lot number: 59538606, purchased in 2012) and cultured as recommended. Authenticity of the cells was regularly tested by short tandem repeat (STR). T-DM1 resistant clones were derived from parental cells by prolonged exposure to T-DM1 following an initial dose-response study of T-DM1 (0.1 nmol/L–100 nmol/L) over 5 days. Resistant BT474 T-DM1R cell lines were obtained by single-cell cloning and continuously cultured in the presence of 5 nmol/L T-DM1 for 3 months. Clones were routinely tested for resistance.

Cell proliferation was assessed by MTT metabolization, cell counting, or crystal violet as described (18, 19). Cell cycle analysis was performed as previously described (14). To determine the mitotic index, cells were seeded at low density, treated and stained with DAPI. Coverslips were observed in an Axiophot 2 Zeiss microscope with the 40× objective. For each condition, at least 1,500 cells were counted. Those in which chromosomes were condensed were considered mitotic cells. The number of apoptotic cells was determined according to the protocol accompanying the FITC Annexin V Apoptosis Detection Kit I (BD Biosciences). Cells were acquired using an Accuri C6 Flow Cytometer (BD Biosciences). Typically, 50,000 events were collected and analyzed using the C6 software. Knockdown of HER2 was performed by infection with lentiviral particles as described (14, 20).

The procedure for the microarray experiments was as previously published (21). The enrichment analysis was performed using Enrichr software (22, 23). Microarray data were obtained from three independent replicates, and are available through the GEO repository database (GSE100192).

Cells were treated with 10 nmol/L T-DM1 for 30 minutes at room temperature. Then, cells were washed and incubated with Cy3-conjugated anti-human antibody. T-DM1 mean fluorescence levels were measured by using a BD Accuri C6 flow cytometer and the C6 software.

For cell surface immunoprecipitation, cells were washed twice with Krebs-Ringer-HEPES buffer [KRH, (24)] and incubated with 10 nmol/L T-DM1 in the same buffer for 2 hours at 4°C. Monolayers were washed twice with PBS and lysed. Cell debris was removed by centrifugation and supernatants incubated for 30 minutes with protein A-Sepharose. Immunocomplexes were then washed and loaded into SDS-PAGE and detected with the anti-human-HRP secondary antibody.

The preparation of cell lysates and Western blotting procedures have been described (24–26). For the antibody arrays, 750 μg of BT474 parental and resistant clones cell lysates were hybridized to Human Phospho-Kinase Array Kit (R&D Systems; ref. 27).

T-DM1 was biotinylated with succinimidyl-2-(biotinamido)ethyl-1,3-dithiopropionate (EZ-Link NHS-SS-Biotin, Thermo Scientific) according to the manufacturer's instructions. Briefly, 2 mg T-DM1 was incubated with a 20-fold molar excess of a 10 mmol/L NHS-SS-Biotin solution in DMSO. The nonreacted NHS-SS-Biotin was quenched with 50 mmol/L of NH₄Cl.

Cells were pulsed with 10 nmol/L biotin-labeled T-DM1 for 30 minutes at 4°C, then washed to remove any unbound biotin-T-DM1 and chased for the indicated times at 37°C. Noninternalized biotin-labeled T-DM1 was removed by washing three times with ice-cold freshly prepared cleavage buffer (50 mmol/L glutathione, 75 mmol/L NaCl, 1 mmol/L MgCl₂, 0.1 mmol/L CaCl₂, 80 mmol/L NaOH, and 10% FBS. pH = 8.6). Cells were then washed twice with PBS and lysed in ice-cold lysis buffer (150 mmol/L NaCl, 1 mmol/L EDTA, 10% glycerol, 40 mmol/L HEPES pH 7.4, 1% Nonidet P-40, and phosphatase and protease inhibitors). Biotin-labeled T-DM1 was immunoprecipitated with Streptavidin-Sepharose High Performance (GE Healthcare), for at least 2 hours at 4°C. Immunocomplexes were recovered, washed, and resolved by SDS-PAGE and standard Western blotting.

T-DM1 detection is detailed in Supplementary Materials and Methods. Acidic organelle staining was followed with 100 nmol/L LysoTracker Red DND-99 (Molecular Probes), which was added 30 minutes before fixing. Phase contrast images were obtained using conventional photomicroscopy.

To evaluate lysosomal pH, T-DM1 was labeled with pHrodo Red Microscale Labeling Kit (Thermo Scientific), according to the manufacturer's instructions Fluorescence emission was monitored by regular epifluorescence microscopy, using a Nikon Eclipse Ti-S microscopy (Nikon). Cells were collected and acquired using an Accuri C6 Flow Cytometer to quantify FL2H-positive cells. Cathepsin B enzymatic activity in live cells was measured using Magic Red Cathepsin B Detection Kit (Immunochemistry Technologies), following the manufacturer's instructions.

Female BALB/c nu/nu mice (7 weeks old) were purchased from Charles River Laboratories and kept in pathogen free housing in the Animal Facility following legal guidelines. Experiments received approval from the National Research Council Bioethics Subcommittee. A total of 5 × 10⁶ cells resuspended in 50 μL of DMEM and 50 μL of Matrigel (BD Biosciences) were injected orthotopically into the caudal mammary gland of each mouse. When tumors reached a volume of 700 mm³, animals (n = 10 per cell line) were randomized to two groups with similar mean tumor volumes that were intravenously treated with vehicle alone or T-DM1 (3 mg/kg). Tumor diameters were measured as described (27). Mice were sacrificed on day 21 or when the tumor volume approached 2,000 mm³. Tumor samples were then collected and immediately frozen in liquid nitrogen. T-DM1 was analyzed by immunoprecipitation with protein A-Sepharose, and Western blotting with HRP-conjugated anti-human secondary antibody.

PDX118 model was established from a cutaneous metastasis biopsy obtained at Vall d'Hebron Institute of Oncology (VHIO). The breast tumor sample used in this study was from a surgical resection at Vall d'Hebron University Hospital and was obtained following the Declaration of Helsinki on ethical principles for medical research involving human subjects. Written informed consent for the performance of tumor molecular studies was obtained from the patient who provided tissue. NOD.CB17-Prkdc^scid/J (NOD/SCID) mice used to establish the patient-derived xenograft (PDX) were maintained and treated in accordance with institutional guidelines of Vall d'Hebron University Hospital Care and Use Committee.

The cell cultures derived from PDX118 were established and characterized at Vall d'Hebron Institute of Oncology (VHIO) as previously described (19). The continuous treatment of these cells with increasing concentrations of T-DM1 led to the appearance of two independent resistant cells called TD44 and TD55.

For microarray analyses, differentially expressed genes (DEG) were selected with a fold-change ≥ 2 and ANOVA P value ≤ 0.05. Comparisons of continuous variables between two groups were performed using a two-sided Student t test. Differences were considered to be statistically significant when P values were less than 0.05. Statistical data are presented as the mean ± SD. All data were analyzed using the statistical software SPSS 21.0 (SPSS Inc.).

Because of their longstanding use as model for the study of HER2 biology and targeting, we selected the BT474 cell line to generate T-DM1–resistant cells. BT474 cells were treated for a total of 3 months in the presence of different concentrations of T-DM1 (Fig. 1A). Four resistant clones (C#1, C#2, C#3, and C#6) were obtained, as indicated by (i) failure of T-DM1 to provoke cell rounding (Fig. 1B) and (ii) failure of T-DM1 to affect their proliferation (Fig. 1C). Moreover, cell-cycle profiling of the T-DM1–resistant clones showed that they were largely resistant to the effect of the drug (Fig. 1D). Quantitative analyses showed accumulation of BT474 cells in mitosis (Fig. 1E). In contrast, very few effect of the drug was observed in the clones. As expected, the mitotic arrest caused by T-DM1 provoked cell death in parental cells (Fig. 1F), probably as a consequence of mitotic catastrophe (13, 14). Such loss of viability caused by T-DM1 was marginal in T-DM1–resistant clones.

Microarray analyses were performed to explore potential transcriptomic differences among the distinct cell lines. Volcano plots of DEGs indicated that BT474 cells differed in 277, 320, 254, and 389 genes from clones C#1, C#2, C#3, and C#6, respectively (Fig. 2A). Principal component analyses (PCA) of variance among the array data demonstrated clustering of the replicates obtained from each cell line (Fig. 2B). Such clustering is indicative of homogeneity or relationship among the gene expression profiles of different samples. Substantial differences in the three dimensional (3D) representation of PCA were observed among wild-type BT474 cells and the other clones. Importantly, clones C#1 and C#2 fell in the same region of the 3D plot (Fig. 2B). Moreover, comparative gene expression profiling between those two clones showed that only 11 genes were differentially expressed (Fig. 2C). The same comparative analyses between the other clones were performed. Despite the low number of DEGs observed between clones C#1 and C#2, a much larger number of genes appeared differentially expressed when performing comparison among the rest of them (Fig. 2D). Therefore, clones C#1 and C#2 were probably constituted by identical populations of cells and, for that reason, we decided to subsequently use clone C#1 as representative of both.

Having established that clones C#1, C#3, and C#6 were different from the gene expression profiling point of view, we analyzed genes shared by them, but differentially expressed when compared with BT474 cells. We observed 55 DEGs shared by the three clones (Fig. 2E). Clones C#1 and C#3 shared 70 DEGs, while clones C#1 and C#6 shared 122, and clones C#3 and C#6 shared 113. Together, these results suggested that clones C#1, C#3, and C#6 were different, but shared some DEGs when compared with wild-type BT474 cells.

In vivo studies also showed differences among the resistant clones. In mice injected with BT474 cells, treatment with the drug decreased tumor volume, which was not observed in the tumors of mice injected with cells from the resistant clones (Fig. 3A). However, the growth kinetics of tumors in mice treated with T-DM1 varied among the different clones. Those derived from clone C#1 were the most resistant to the in vivo action of T-DM1. Tumors derived from C#3 cells initially responded to T-DM1, although no regression of the tumor masses was observed. At later time points, tumors from C#3 cells regained growth. In mice injected with C#6 cells, T-DM1 treatment initially arrested tumor growth, but as occurred with the tumors generated by the other clones, no regression of tumor masses occurred. The low growth rate of the tumors of C#6 in mice treated with T-DM1 allowed us to follow for a longer time the behavior of these tumors. Of note, extension of the treatment with T-DM1 for 9 weeks caused complete regression of the tumors in mice injected with wild-type BT474 cells (Fig. 3B). However, such treatment did not prevent progressive growth of the tumor masses created by injecting C#6 cells.

Tumors from mice injected with the resistant cells were obtained and T-DM1 levels analyzed by Western blotting. Curiously, tumors derived from mice injected with cells from C#1 and treated with T-DM1 accumulated the drug to a higher extent than those derived from C#3 and C#6 cells (Fig. 3C and D).

The mode of action of T-DM1 includes inhibition of microtubule polymerization during mitosis (16). T-DM1 affected the formation of normal mitotic spindles in BT474 cells, while the resistant clones were much less affected (Fig. 4A and B). As DM1 binds tubulin, altered or mutant tubulins might constitute a mechanism of resistance to T-DM1 (28). Tubulin β1 sequencing failed to show any mutations in BT474 or the resistant clones.

Cell proliferation experiments indicated that the sensitivity of the different clones to the free DM1 compound (mertansine) and other anti-microtubular agents used in the clinic, such as vinorelbine or docetaxel, was preserved (Fig. 4C). However, and in line with the different behavior of the distinct clones observed when analyzing other biological properties, differences in the anti-proliferative potency of vinorelbine or docetaxel were observed among the different clones. Cell-cycle analyses demonstrated that these agents provoked G₂–M arrest in the resistant clones as well as in wild-type BT474 cells (Fig. 4D). The above data suggested that tubulin functionality was unaltered in these clones.

One of the described mechanisms of resistance to anti-HER2 therapies is the loss of HER2, or the presence of truncated forms of the protein lacking the ectodomain (29, 30). As shown in Fig. 5A, no major differences among the levels of total HER2 present in the different clones and BT474 cells were observed. Moreover, immunofluorescence studies using trastuzumab to label HER2 showed similar patterns of HER2 staining in BT474 cells and the T-DM1–resistant clones (Fig. 5B). The immunofluorescence experiments also suggested that HER2 was present at the cell surface of the different cells studied at similar amounts. This was verified through quantitative analyses of cell surface HER2 levels using cell-surface immunoprecipitation (Fig. 5C) or flow cytometry (Fig. 5D). These quantitative experiments were performed using T-DM1, and, therefore, served to also demonstrate that binding of the drug to cell-surface HER2 was preserved in the resistant clones. No differences were found in HER2 levels after treatment of the cell lines with T-DM1 for prolonged incubation times (Fig. 5E).

Loss of HER2 dependency has also been proposed as another mechanism of resistance to anti-HER2 therapies (31). To explore whether such mechanism may account for the T-DM1 resistance observed in the different clones, we knocked down HER2 (Fig. 5F) and analyzed the effect on cell proliferation (Fig. 5G). These experiments demonstrated the strong dependency on HER2 of the parental BT474 cells as well as the resistant clones, indicating that they remain addicted to HER2. Together, the above data indicated that HER2 levels and dependency are preserved in the T-DM1–resistant clones.

Resistance to anti-HER2 therapies may also arise through compensatory upregulation of downstream signaling pathways, particularly the PI3K route (32, 33). The status of several signaling pathways was, therefore, analyzed using antibody arrays. The capture antibodies used in these arrays allow detection of activated forms of signaling proteins, which are used as readouts of the activity of several signaling pathways, including some activated by HER2 or PI3K activation (a schematic representation of these routes can be consulted online at the Kyoto Encyclopedia of Genes and Genomes-KEGG). Antibody arrays of signaling intermediates indicated that the patterns of phosphorylation of the different proteins analyzed in C#1 and C#3 cells were similar. Clone C#6 gave a pattern of phosphoproteins more similar to BT474 cells than to the C#1 and C#3 cells (Fig. 5H). These data indicate that no evident upregulation in major downstream signaling routes was present in the T-DM1–resistant clones.

Because the resistance to T-DM1 did not appear to rely on molecular alterations in HER2 or tubulin, the lack of action of T-DM1 in the resistant cells could be due to increased extrusion of the payload. In fact, maytansinoids have been reported to be substrates of drug transporters such as MDR1/PgP and high levels of these proteins may confer resistance to DM1 (12). Gene expression analyses of drug transporters known to act in efflux of chemotherapeutics showed no difference in their mRNA levels (Supplementary Fig. S1), including SLC46A3, a lysosomal membrane protein, which may participate in delivery of noncleavable antibody–drug conjugate (ADC) catabolites into the cytoplasm (34).

Gene expression profiling data were also used to explore cellular components, which could be altered in T-DM1–resistant clones. To that end, we used the Enrichr web-based tool, which is an enrichment analysis software that provides summaries of functions assigned to gene lists, including a cellular component analysis (22, 23). The gene ontology cellular component networks obtained for the gene lists corresponding to clones C#1, C#3, and C#6 showed that terms related to the endocytic pathway were enriched in all the resistant clones, having some of them in common, such as lysosome or vacuole (Fig. 6A, Supplementary Table S1). These studies pointed to endosomal trafficking or lysosomal function as subcellular processes that could be responsible for the emergence of T-DM1 resistance. To measure T-DM1 internalization, the drug was labeled using a cleavable biotin compound and the translocation of biotinylated T-DM1 to intracellular compartments was analyzed. These experiments showed that biotin/S-S/T-DM1 underwent internalization in BT474 and the resistant clones (Fig. 6B).

To gain further insights into the internalization of T-DM1 in BT474 cells and the T-DM1–resistant clones, immunofluorescence analyses were performed. Fixation of cells right after a 15 minutes pulse with T-DM1 showed that T-DM1 staining was restricted to the cell surface (Fig. 6C, 0 minute). After 1 day, cell-surface staining was still appreciable; however, punctate staining was also observed, suggesting intracellular accumulation of T-DM1. This accumulation was even more evident when using chloroquine, an agent that causes entrapment of endocytosed materials (35). In vivo time-lapse studies confirmed redistribution of T-DM1 from the cell surface to a punctate pattern in all the cell lines (Supplementary Movie S1). Extension of the incubation times showed differences in the patterns of T-DM1 staining. After 3 days, most of the drug had disappeared from the surface of BT474 cells and few intracellular spots remained (Fig. 6C). In contrast, in clones C#1 and C#3, a substantial amount of T-DM1 remained in punctate structures. In the case of clone C#6, some T-DM1 was still detected at the cell surface. Time-lapse studies showed progressive accumulation of the drug in such punctate structures along the time course of the experiment (Supplementary Movie S2).

Colocalization experiments with markers of intracellular organelles were carried out to identify the intracellular site of T-DM1 accumulation in the resistant cells. Substantial colocalization was detected between the acidic vesicle marker LysoTracker Red DND-99 and T-DM1 after 3 days of incubation (Fig. 6D). Quantitative analysis of colocalization showed LysoTracker/T-DM1 colocalization rates of 13.21%, 79.06%, 83.88%, and 55.07% in BT474, C#1, C#3, and C#6, respectively (Fig. 6D). Moreover, colocalization with the lysosomal marker LAMP-1 (Supplementary Fig. S2) was detected in C#1 and C#3 cells.

The intracellular accumulation of T-DM1 in clones C#1 and C#3 suggested that the mechanism of resistance to the drug could be linked to deficient processing of T-DM1, likely because of altered lysosomal function. Because an acidic pH is crucial for the optimal function of lysosomal proteolytic enzymes, we evaluated whether T-DM1 reached such an acidic environment within lysosomes. To this end, T-DM1 was labeled with pHrodo, a red fluorogenic dye that is almost non-fluorescent at neutral pH, but which fluoresces in acidic environments (36). Cells were incubated with 5 nmol/L pHrodo-T-DM1 for 24 hours, and fluorescent signals analyzed in vivo by microscopy or by FACS. In BT474 and clone C#6 cells, fluorescence accumulated intracellularly (Fig. 7A). In contrast, in cells from clones C#1 and C#3, the fluorescent signal was much weaker. Quantitative cytometric analyses of pHrodo-T-DM1 corroborated the difference in the fluorescent signal in cells from clones C#1 and C#3 with respect to wild-type BT474 cells (Fig. 7B). These experiments suggested that the pH in the compartments reached by pHrodo-T-DM1 was substantially higher in C#1 and C#3 when compared with wild-type cells.

The proteolytic activity of lysosomes was next explored using a cathepsin B activity assay as readout (37). Cathepsin B activity of C#6 was similar to that of wild-type BT474 cells (Fig. 7C). In contrast, significantly lower cathepsin B activities were observed in clones C#1 and C#3. Taken together, the above data indicated that lysosomal pH was higher in clones C#1 and C#3, and this likely resulted in impaired overall proteolytic activity, which prevented adequate cleavage of T-DM1.

As lysosomal dysfunction could affect autophagy, the latter process was evaluated by Western blotting of LC3 processing and p62 degradation. LC3-II is degraded during autophagy. However, lysosomal degradation of LC3-II can be avoided if fusion between autophagosomes and lysosomes is blocked, using compounds such as bafilomycin A1 (38). This drug is a V-ATPase inhibitor that causes an increase in lysosomal/vacuolar pH and, ultimately, inhibits autophagy (39). As shown in Supplementary Fig. S3A and S3B, bafilomycin A1 caused an increase in LC3-II levels in parental BT474 and the resistant clones. On the other hand, p62 is primarily degraded during autophagy (38). The inductor of autophagy rapamycin provoked a decrease in p62 in all the cell lines (Supplementary Fig. S3A and S3C). Together, these data suggest that the autophagy process was preserved in all the cell lines analyzed.

If lysosomal alkalization plays a role in T-DM1 resistance, it would be expected that artificial manipulation of lysosomal pH could render BT474 cells resistant to the drug. To address that bafilomycin A1 was used. Preincubation with bafilomycin A1 caused a decrease in pHrodo-T-DM1 fluorescence, indicative of lysosomal alkalization (Fig. 7D). Moreover, bafilomycin A1 diminished the anti-proliferative action of T-DM1 in BT474 cells (Fig. 7E).

Finally, we explored whether the resistance mechanism described in BT474 cells also occurred in patient-derived material. To that end, we used cell cultures established from a patient-derived HER2⁺ xenograft (PDX118; ref. 19). The initially T-DM1 sensitive cell cultures were maintained in the presence of increasing concentrations of T-DM1 for 45 days to obtain two independent cultures (TD44 and TD55, Supplementary Fig. S4A). These cells were resistant to the anti-proliferative action of T-DM1, as compared with parental cells (Supplementary Fig. S4B). Western blotting analyses confirmed the HER2⁺ phenotype of the parental and the TD44 and TD55 cells (Supplementary Fig. S4C). Although smaller amounts of total HER2 were observed in the resistant clones, preliminary data demonstrated that resistance to T-DM1 was not due to such decrease in HER2. Staining with pHrodo-T-DM1 showed a statistically significant decrease in TD44 and TD55 cells, as compared with parental cells (Supplementary Fig. S4D). Moreover, immunofluorescence studies showed statistically significant accumulation of T-DM1 in the lysosomes of TD55 cells (Supplementary Fig. S4E and S4F).

The development of drug resistance is a central clinical problem in oncology that, if prevented, could result in long lasting anti-tumoral responses. Because of this, efforts to uncover the mechanisms of resistance to anti-HER2 therapies are being made with the purpose of overcoming them.

We initiated the study herewith reported with the intention of identifying mechanisms of resistance to T-DM1, given the scarce knowledge on them and the favorable clinical responses obtained in patients treated with this drug (17). To that end, we modeled resistance using BT474 cells, and evaluated cellular processes that T-DM1 uses to carry out its effect.

Experiments aimed at exploring whether loss or truncation of HER2 were responsible for such resistance, in analogy to trastuzumab (29), failed to demonstrate an alteration in the levels of HER2 in the three T-DM1–resistant clones isolated. Furthermore, T-DM1 binding was preserved, in contrast to mechanisms of resistance to anti-HER therapies that are accompanied by loss of therapeutic antibody binding to the target (40). Moreover, knockdown experiments demonstrated oncogenic dependence on HER2 in the resistant clones, excluding the possibility that resistance to T-DM1 was caused by molecular alterations of HER2, which could cause loss of function of the protein. Finally, exploration of several signaling routes, such as PI3K, whose alteration has been linked to trastuzumab resistance (41), failed to define a clear signaling alteration in the resistant clones. So far, the data generated indicate that T-DM1 resistance in those clones is not due to a molecularly altered HER2 protein, or altered HER2-dependent signaling.

Bioinformatic analyses of gene functions pointed to the endosomal/lysosomal pathway as a route that could be involved in T-DM1 resistance. Deficient internalization of HER2 has been hypothesized as a possible mechanism of resistance to T-DM1 (42). Moreover, the already reported slow rate of internalization of HER2 (43, 44) could prime the raising of resistance to T-DM1. However, several experiments indicated that the first steps of the internalization of HER2 were preserved in the resistant clones at levels similar to those of wild-type BT474 cells. Cleavable biotinylation experiments showed that the rates of internalization of biotin-labeled T-DM1 were similar in BT474, C#1, C#3, and C#6 cells. Furthermore, in vivo follow-up of dye-labeled T-DM1 failed to show any defect in its early internalization steps. These results suggest that internalization was preserved in the resistant clones.

Video analyses in live cells together with more conventional immunofluorescence studies showed differences in the subcellular distribution of T-DM1, especially in long-term incubation experiments. These analyses showed that T-DM1 accumulated in intracellular compartments in C#1 and C#3 cells, but not in C#6 cells. Such accumulation of T-DM1 was also observed in vivo, particularly in tumors derived from C#1 cells. Studies to identify the intracellular compartment where T-DM1 was trapped in the resistant clones showed colocalization of T-DM1 with (i) the acidic compartment marker LysoTracker, (ii) antibodies recognizing LAMP-1, and (iii) a GFP-LAMP-1 fusion protein, all of them considered bona fide lysosomal markers. Assessment of lysosomal pH and cathepsin B protease activity indicated that both were altered in C#1 and C#3 cells with respect to wild-type or C#6 cells. The increase in lysosomal pH in C#1 and C#3 cells could restrict the intrinsic proteinase activities of lysosomal enzymes, impairing proteolytic cleavage of endocytosed T-DM1 (schematically illustrated in Fig. 7F). In fact, experiments in which lysosomal pH was raised in parental BT474 cells by the use of bafilomycin A1 showed that such compound decreased the anti-proliferative action of T-DM1. Importantly, analyses of lysosomal pH and anti-tumoral properties of T-DM1 in resistant cells obtained from a patient-derived HER2⁺ tumor passaged as a PDX confirmed that lysosomal derangement may lead to T-DM1 inefficacy. Not only this scenario approaches to the clinical setting, but also offers confirmation that such mechanism is not exclusive of BT474 cells made resistant to the drug.

Several important conclusions can be extracted from the studies herewith reported. On one hand, our studies suggest the existence of several mechanisms of resistance to T-DM1 in HER2⁺ breast cancer cells, one of then linked to deficient lysosomal activity. Such altered lysosomal function may impair processing of T-DM1, limiting its anti-tumoral action. Various experimental data support the existence of diverse mechanisms of T-DM1 resistance, and included (i) genomic analyses, which indicated differences in the DEGs among the T-DM1–resistant clones, (ii) animal studies, which revealed variety in the in vivo behavior of tumors in mice injected with cells of the resistant clones, and (iii) cell biological and biochemical data, which showed lysosomal accumulation of T-DM1 and impaired lysosomal function in some of the resistant clones. These interclonal differences demonstrate the existence of multiple mechanisms of resistance to T-DM1, as occurs in the case of other anti-HER2 therapies (45). This could somehow be expected, but is unfortunate from the clinical point of view, as it will imply the knowledge of all the resistance mechanisms to design active anti-tumoral strategies to fight T-DM1 resistance.

Finally, the description of such resistance mechanism may have several important general implications, beyond its identification as a mechanism of T-DM1 resistance. Given the expansion of ADC-based therapeutics, our findings may be relevant to understand resistances raised to other ADCs using the same internalization-degradation pathway for generation of active payloads (46). In addition, our findings suggest the possibility of manipulating the lysosomal pH to augment/restore the anti-tumoral properties of T-DM1 and other ADCs. In this respect, it is interesting to mention a recent report that used photoactivatable nanoparticles to manipulate intralysosomal pH (47). Strategies in that direction could improve ADC-based therapies.

No potential conflicts of interest were disclosed.

Conception and design: C. Ríos-Luci, S. García-Alonso, E. Díaz-Rodríguez, J. Arribas, A. Ocaña, A. Pandiella

Development of methodology: C. Ríos-Luci, S. García-Alonso, M. Nadal-Serrano, A. Ocaña, A. Pandiella

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Ríos-Luci, S. García-Alonso, E. Díaz-Rodríguez, M. Nadal-Serrano, A. Pandiella

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Ríos-Luci, S. García-Alonso, E. Díaz-Rodríguez, A. Pandiella

Writing, review, and/or revision of the manuscript: C. Ríos-Luci, S. García-Alonso, E. Díaz-Rodríguez, J. Arribas, A. Ocaña, A. Pandiella

Study supervision: C. Ríos-Luci, S. García-Alonso, A. Pandiella

This work was supported by the Ministry of Economy and Competitiveness of Spain (BFU2012-39151 and BFU2015-71371-R), the Instituto de Salud Carlos III through the Spanish Cancer Centers Network Program (RD12/0036/0003), and the Scientific Foundation of the Spanish Association Against Cancer (AECC) and the CRIS Cancer Foundation. S. García-Alonso is recipient of a predoctoral contract from the MINECO (BES-2013-065223). The Cancer Research Institute and the work carried out at our laboratory receive support from the European Community through the Regional Development Funding Program (FEDER). Work in J. Arribas lab was supported with funds from the Instituto de Salud Carlos III (PI16/00253) and the Breast Cancer Research Foundation (BCRF).

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.