Trastuzumab emtansine is active on HER-2 overexpressing NSCLC cell lines and overcomes gefitinib resistance

The total level of HER-2 protein was detected by immunoblotting on cell lysates in a panel of NSCLC cell lines (H1781, H3255, H322, H1299, H1975, Calu-6, H596, H460, A549, PC9, HCC827 and Calu-3). As shown in Figure 1A, HER-2 expression varied widely among the analyzed cell lines, ranging from barely detectable levels in Calu-6 to high levels in Calu-3. The latter was an expected finding, due to known amplification of HER-2 in Calu-3 cell line [21]. Considering that trastuzumab and T-DM1 have a common targeted receptor on the cell surface, we quantified HER-2 expression levels on the plasma membrane by flow cytometry. Indeed, the ability of the antibody to interact with its target is strictly related to the presence of the receptor on the cell surface. As reported in Figure 1B, Calu-3 and H3255 cells displayed the highest levels of HER-2 at the plasma membrane. The total level of HER-2 in H3255 was similar to that observed in other cell lines such as H460 and A549 (Figure 1A) indicating that the total level of proteins detected on cell lysate is not a good predictor of HER-2 level on plasma membrane.

Based on the previous analysis, the effect of T-DM1 and trastuzumab on cell viability was focused on NSCLC cell lines expressing different cell surface levels of HER-2: Calu-3 (very high), H3255 (high) and Calu-6 (low). T-DM1 showed strong anti-proliferative effect in HER-2 over-expressing Calu-3 cells (IC₅₀ = 0.40 ± 0.08 μg/ml, Figure 1C), whereas no effects were detected on Calu-6 cell line, with low levels of HER-2 on the plasma membrane. The inhibition observed at 10 μg/ml was related to a non-specific toxic effect of T-DM1, as previously reported in MCF-7 HER-2 negative breast cancer cell line [22]. Intermediate results were seen on H3255 cells. Trastuzumab, administered at the same dosages of T-DM1, did not show notable effect on cell growth in any of the cell lines tested. H1781 cell line, harbouring mutated HER-2 (G776insV_G/C), was also included in this study in order to investigate whether this mutation influenced the anti-proliferative effect of T-DM1. As shown in Figure 1C, T-DM1 at 1 μg/ml induced about 35% inhibition of cell viability. A similar inhibition profile was observed in T-DM1 (1 μg/ml) treated H322 cells, displaying comparable levels of HER-2 on cell surface (Figure 1B), suggesting that H1781 sensitivity to the drug is not affected by the mutated receptor. We then analyzed the percentage of inhibition of cell viability induced by 1 μg/ml T-DM1 on eight different cell lines, as a function of HER-2 level on plasma membrane and, as shown in Figure 1D, we confirmed that sensitivity to T-DM1 is strictly correlated to HER-2 expression on the cell surface.

To exclude the hypothesis that different effects of T-DM1 could be ascribed to dissimilar sensitivity to the microtubule polymerization inhibitor DM1, we treated two cell lines harbouring high and low levels of HER-2 on the plasma membrane, respectively Calu-3 and Calu-6, with increasing concentration of Vinorelbine, an anti-mitotic drug, which acts by a similar mechanism of action of the maytansinoid DM-1. Vinorelbine inhibited viability of Calu-3 and Calu-6 cells in a comparable manner (Figure 1E).

To determine the effect of T-DM1 on cell cycle, Calu-3 treated with 1 μg/ml T-DM1 or Trastuzumab for 24 h were analyzed by flow cytometry. As shown in Figure 2A, T-DM1 caused an increase in the proportion of cells in G2-M phase with a decrease in G1 and S phases, whereas no alterations on cell cycle distribution were detected in cells treated with trastuzumab. The arrest of Calu-3 in G2-M phase of the cell cycle, as a result of T-DM1 exposure, was also supported by increased levels of Cyclin B1 as measured by western blot analysis (Figure 2B) whereas pRb and Cyclin A were unchanged.

We then tested the effect of T-DM1 and trastuzumab on phosphorylation status of HER-2, AKT and p42-44 MAPK in Calu-3 cell line. Differently of trastuzumab, T-DM1 significantly inhibited the phosphorylation of AKT and p42-44 after 24 h of treatment (Figure 2C) with a decrease in HER-2 total level and phosphorylation at 48 h.T-DM1 (1 μg/ml) exerted a significant cytotoxic effect already after 24 h of treatment, with appearance of floating dead cells (Figure 3A). By contrast trastuzumab did not modify cell proliferation nor induced cell death up to 72 h of treatment.

As shown in Figure 3B, 48 h exposure of Calu-3 cells to T-DM1 at 0.1 and 1 μg/ml induced the activation of caspases-7 and −9 and the release of cytochrome-c into the cytoplasm (Figure 3C) indicating that the intrinsic pathway is involved in T-DM1-triggered apoptotic cell death. Vinorelbin was used as internal control. A lower activation of caspases and a weak release of cytochrome-c was also induced by trastuzumab treatment even if no significant cell death was observed (Figure 3A).

Since antibody-dependent cell-mediated cytotoxicity (ADCC) is one of the main mechanisms of action of specific mAbs directed to ErbB family members in vivo [23], we examined whether the capability to activate natural killer (NK)-mediated ADCC is preserved by T-DM1. As shown in Figure 3D, T-DM1-dependent cytotoxicity in the presence of IL-2 activated NK cells was similar to trastuzumab-dependent cytotoxicity in Calu-3 overexpressing HER-2. In the low HER-2 expressing H1299 cells, neither T-DM1 nor trastuzumab significantly induced mAb-dependent cytotoxicity.

As previously reported [14] and independently confirmed by our laboratory, the clone PC9/HER2c1 (a generous gift from Dr. William Pao), obtained by stably transfection of PC9 cells with HER-2 expression vector, is more resistant to gefitinib than parental cells. HER-2 expression on plasma membrane was 10 time higher in the clone compared to the parental cell line (data not shown).

Based on these results we tested the effect of T-DM1 on PC9/HER2c1 and in the parental PC9 cells. As shown in Figure 4A, HER-2 overexpression significantly enhanced the efficacy of T-DM1 with 40% inhibition of cell viability at 1 μg/ml in the PC9/HER2c1 clone. With respect to PC9 cells, the clone showed a marked increase in AKT, p70S6K and p42-44 activation. After 48 h of treatment with T-DM1 a reduction in AKT and p70S6K phosphorylation was observed (Figure 4B) suggesting that T-DM1 might improve gefitinib treatment. In Figure 4C the dose–response curves of gefitinib in the presence of a fixed concentration of T-DM1 (0.1 μg/ml) are shown. Comparing the experimental combination points with that expected by the Bliss criterion, an additive effect was observed. In fact, no significant differences between experimental and theoretical points were observed.

It has been reported that cell density can influence the expression of EGFR in breast cancer [24] and in pancreatic cancer cell lines [25] and that surface expression of HER-2 is regulated post-transcriptionally in mammary epithelial cells by the culture cell density [26]. We investigated the dependence of HER-2 membrane protein expression on cell density as well as the effect of T-DM1 on cells seeded at different densities. Confluent Calu-3 cells exhibited a significant decrease of HER-2 at cell surface level detected by immunohistochemistry (Figure 5A ii), as compared to cells seeded at low density (i). At low density, more than 80% of cells showed a strong surface expression of the receptor whereas in almost confluent Calu-3 cultures a significant downregulation of HER-2 was observed (nearly 35%). Consequently, as reported in Figure 5B, the inhibition of cell viability induced by T-DM1 at 1 μg/ml was markedly decreased with increasing cell density.

We performed an in vivo experiment, aimed to determine whether T-DM1 efficacy might be affected by tumor size and structural organization. Tumors were clearly visible in all mice inoculated with 4 × 10⁶ or 8 × 10⁶ Calu-3 with a mean volume of 161 ± 15 mm³ (Figure 5C) and 370 ± 50 mm³ (Figure 5D) respectively. Trastuzumab (15 mg/Kg intraperitoneal) or T-DM1 (15 mg/Kg intravenously) were given every six days. T-DM1 or trastuzumab treatments in animals carrying tumors of small size were able to strongly inhibit tumor growth compared to vehicle treated mice (Figure 5C). Treatment with T-DM1 not only inhibited tumor growth, but a reduction of tumor dimension was observed in five out of six mice.

On the other hand, when treatments were performed on larger tumors, only T-DM1 was able to significantly reduce tumor growth compared to control group, whereas no significant effect was seen with trastuzumab (Figure 5D). We did observe neither rapid tumor shrinkage nor long term complete response in T-DM1 treated animals. T-DM1 in vivo efficacy was confirmed by the reduction in weight of tumors excised at sacrifice, compared to control (Inset to Figure 5C and D).

To define at tissue level the response of small and large tumors to T-DM1, a morphometric analysis of neoplastic tissue composition was performed. To this purpose, the fractional volume occupied by PanCK^pos cells was assessed (Figure 6A,B). Quantitatively, compared to control in small tumors a 67% and 73% reduction of neoplastic epithelial cells within the nodules was observed with trastuzumab and T-DM1, respectively (A). In large tumors, the amount of neoplastic tissue was reduced by 16% and 37%, respectively, with trastuzumab and T-DM1 compared to control (B). Importantly, both drugs significantly decreased mitotic index in small and large tumors, however, the antiproliferative activity of T-DM1 was superior to that of trastuzumab (data not shown). To gain insights on the mechanism underlying these effects of T-DM1, we sought to determine whether the results obtained in vitro on the different sensitivity to T-DM1 according to cell density and HER-2 expression had an in vivo counterpart. To this hand, sections of small and large tumors were immunostained with HER-2 antibodies (Figure 6C,D). Again, differences in the surface expression of the HER-2 was striking, as neoplastic cells composing small tumors (C) showed higher extent and intensity of HER 2 immunolabelling than large tumors (D). Thus, our data clearly document that the anti-tumor potency of T-DM1 is strictly dependent on HER-2 expression, which in turn is intrinsically modulated among neoplastic cells and their structural organization. Characteristic and atypical mitotic figures and giant multinucleated cells were detected both morphologically on HER-2^pos cells (Figure 6E,F,G,H) and by the nuclear labelling of PH-H3 in PanCK expressing cells (Figure 6I,J,K) both in small (Figure 6E,F,G,H) and large tumors (not shown).

The human NSCLC cell lines used in this study were purchased from American Type Culture Collection (ATCC) (Manassas, VA, USA) and banked at early passage (P2). Furthermore, the cells we culture, are regularly verified on the basis of cell morphology and never cultured for more than 3 months. The PC9, HCC827, HCC827 GR5 and H1781 cell lines were kindly provided in 2013 by Dr P. Jänne (Dana-Farber Cancer Institute, Boston MA, USA). The PC9/HER2c1 was kindly provided in 2013 by Dr.William Pao (Vanderbilt-Ingram Cancer Center, Nashville, Tennessee). All cells were cultured as recommended and maintained at 37°C in a humidified atmosphere of 5% CO₂ and 95% air.

Trastuzumab and vinorelbine were kindly provided by inpatient pharmacy. T-DM1 was supplied from Genentech Inc. (South San Francisco, CA) through a Materials Transfer Agreement. For the in vitro experiments, stock solutions of drugs were prepared in distilled water, stored at 4°C and diluted in fresh medium for use, whereas for the in vivo experiments trastuzumab and T-DM1 were daily dissolved in sterile saline solution (NaCl 0.9%).

Cell protein extraction, solubilization, and analysis by 1-D PAGE were performed as previously described [41]. Antibodies against HER2; p-HER2 ^Tyr1221/1222; p70S6K; p-p70S6K ^{Thr421/Ser424}; Akt; p-Akt ^Ser473; p44/42 MAPK; p-p44/42 MAPK; caspase-7 and 9; cyclin A and B1; Rb; p-Rb were from Cell Signaling Technology (Beverly, MA). Antibody against cytochrome-c (7H8) was form Santa Cruz Biotechnology Inc. (Dallas, TX). Antibodies against actin was from Sigma–Aldrich (St Louis, MO). Antibody against GAPDH was from Ambion (Austin, TX). HRP-conjugated secondary antibodies were from Pierce (Rockford, IL) and chemiluminescence system (ImmobilionTM Western Cemiluminescent HRP Substrate), was from Millipore (Temecula, CA).

One million of NSCLC cell lines were incubated, for one hour at room temperature, with Isotype control Monoclonal Mouse IgG1/R-PE (Ancell IRP, Bayport, MN, USA) or PE mouse anti-Human HER-2 (BD Biosciences, San Josè CA) to determine HER-2 protein membrane levels as previously described [42]. After the incubation the analysis was performed using an EPICS-XL flow cytometer. Mean fluorescence intensity (MFI) values were converted in units of equivalent fluorochrome (MEF) using the FluoroSpheres 6-Peak Kit (Dako, CA, USA).

Cell viability was evaluated by tetrazolium dye [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium- bromide] (MTT) assay and by crystal violet staining as previously described [43]. Data are expressed as percent inhibition of cell proliferation versus control cells. Distribution of the cells in the cell cycle was determined by PI staining and flow cytometry analysis as described elsewhere [44].

Apoptosis was assessed by morphological study: stained (Hoechst 33342, propidium iodide) or unstained cells were observed using light-, phase-contrast- and fluorescence-microscopy. Western blot analysis was performed to evaluate cleavage products of caspase-7 and caspase-9. Cytosolic and mitochondrial fractions for cytochrome-c detection were generated using a digitonin-based subcellular fractionation technique as previously described [45].

Highly purified CD56+ natural killer (NK) cells were obtained by magnetic separation and ADCC was measured with the CytoTox 96 non-radioactive cytotoxicity assay (Promega, Madison, WI, USA) as previously described [42].

All experiments involving animals and their care were performed with the approval of the Local Ethical Committee of the University of Parma, in accordance with the institutional guidelines that are in compliance with national (DL116/92) and international (86/609/CEE) laws and policies.

Balb/c-Nude female mice (Charles River Laboratories, Calco, Italy) 6 weeks old, were housed in a protected unit for immunodeficient animals with 12-hour light/dark cycles and provided with sterilized food and water ad libitum [41, 42]. We performed an in vivo experiment in order to investigate whether the tumor dimension could influence drug efficacy. 200 μl of matrigel (BD Biosciences) and sterile PBS (1:1) containing 4 × 10⁶ or 8 × 10⁶ Calu-3 cells were subcutaneously injected on the right flank of each mouse. Ten days after cells injection, tumor volume reached an average size of 160 mm³ or 370 mm³ respectively, and animals were randomized into three different groups (n = 6): control, trastuzumab and T-DM1. Once every six days trastuzumab was intraperitoneally administered at a dosage of 15 mg/Kg, whereas T-DM1 was injected into the lateral tail vein at a dosage of 15 mg/Kg. Control mice received intravenous injection of sterile saline solution (NaCl 0.9%) according to the same schedule. Drug dosages were chosen accordingly to previous studies conducted on breast and gastric xenograft models [15, 17, 18, 22].

Tumor xenografts were measured three times per week using a digital caliper and tumor volume was determined using the formula: (length × width²)/2. At the end of the experiments, mice were euthanized by cervical dislocation and tumors weighted and collected for immunohistochemical and histological analysis.

Formalin fixed samples were embedded in paraffin. On each tumor serial sections of 5 μm thickness were stained with Haematoxylin and Eosin (H&E), Masson’s Trichrome and subjected to immunohistochemistry. A morphometric analysis was performed on the entire section in order to evaluate the volume of neoplastic tissue, connective tissue and vascular interstitium. To better define the fraction occupied by neoplastic cells, sections were stained with pancytokeratin antibody (PanCK monoclonal mouse, 1:500, o.n. 4°C, Dako) revealed through biotin-streptavidin-DAB system (Dako). The volume fraction of fibrosis and vascular interstitium was assessed on Masson’s Trichrome stained samples. To this end, the number of points overlying each tissue components was counted and expressed as percentage of the total number of points explored. All these morphometric measurements were obtained with the aid of a grid defining a tissue area of 0.23 mm² and containing 42 sampling points each covering an area of 0.0052 mm².

Combining the entire tumor volume with the above morphometric measurements, the total volume occupied by neoplastic cells, connective tissue and vascular interstitium was computed for each sample.

Moreover, immunohistochemical analysis of HER-2 was performed on each tumor of each experimental group. Sections were stained with HER-2 antibody (monoclonal mouse clone 4B5, Ventana, USA) and revealed through biotin-streptavidin-DAB system. This analysis was performed on the entire tumor by an optical microscope (Olympus, BX60- 100X magnification) to evaluate the area occupied by cell expressing HER-2 and their intensity. The latter was expressed as Integrated Optical density (IOD), as detected using a software for image analysis (Image Pro Plus, Media Cybernetics, USA).

In addition, the nuclear expression of the phosphorylated form of Histone H3 (pH-H3, rabbit poyclonal, 1:100, Millipore, USA) on Pancytokeratin^pos cells was detected by immunofluorescence to document mitotic figures.

For all tested epitopes negative controls were represented by immunostaining the sample with an irrelevant antibody or by exposing the sections only to the secondary antibody.

Statistical analyses were carried out using GraphPad Prism version 5.0 software (GraphPad Software Inc., San Diego, CA, USA). Results are expressed as mean values ± standard deviations (SD) for the indicated number of independent measurements. Differences between the mean values recorded for different experimental conditions were evaluated by Student’s t test, and P values are indicated where appropriate in the figures and in their legends. A P value <0.05 was considered as significant. Bliss interaction was calculated as previously described [46]. For in vivo studies comparison among groups was made using analysis of variance (two-way ANOVA repeated measures) followed by Bonferroni’s post-test.