Combining Epitope-Distinct Antibodies to HER2: Cooperative Inhibitory Effects on Invasive

Combining epitope-distinct antibodies to HER2: cooperative inhibitory effects on invasive growth

Anna Emde 1, Pradeep Chaluvally Raghavan 1, Benjamin Ribba 2, Zvi Kam 3 and Yosef Yarden 1*

Departments of 1Biological Regulation and 3Molecular Cell Biology, the Weizmann Institute of Science, Rehovot 76100, Israel, and 2INRIA project-team NUMED, ENS Lyon, Lyon 69007, France

Running title: Anti-HER2 mAb combinations

Character count: 5,621 (excluding abstract, references and figure legends)

*Corresponding author: Department of Biological Regulation, Candiotty Building (room 302), The Weizmann Institute of Science, 1 Hertzl Street, Rehovot 76100, Israel. Tel. 972-8-9343974, FAX: 972-8-9342488, e-mail: )

Abstract

Monoclonal antibodies (mAbs) to HER2 are currently used to treat breast cancer, but low clinical efficacy, along with primary and acquired resistance to therapy, commonly limit clinical applications. We previously reported that combinations of antibodies directed at non-overlapping epitopes of the same receptor increase anti-tumor effects, probably by accelerating receptor degradation. Here we extend these observations to 3-dimensional mammary cell models, and aim at comparative examination of the effects of single drugs with the effects of mAb combinations. Collectively, our in vitro assays, mathematical modelling and computational image analyses indicate that combining mAbs against different epitopes of HER2 better inhibits invasive growth. Importantly, while growth factors are able to reduce intraluminal apoptosis and induce an invasive phenotype of HER2-overexpressing spheroids, combinations of mAbs better than single mAbs reverse the growth factor-induced phenotype. Thus, our studies propose that mAb combinations negate the collaborative effect of HER2 and growth factors on invasive growth. Hence, combining mAbs offers a potential therapeutic strategy, able to enhance clinical efficacy of existing immunotherapeutics.

Introduction

The ErbB-family of receptor tyrosine kinases consists of four members, namely ErbB-1, ErbB2/HER2, ErbB-3, and ErbB-4. Upon activation by a variety of ligands, such as EGF, TGF-alpha and amphiregulin, the receptors form hetero- and homodimers able ofto signalling along multiple downstream pathways, such as the phosphoinositide-3-kinase (PI3K)/AKT pathway,and the mitogen activated protein kinase (MAPK) pathway, and such as the extracellular signal-regulated kinase (ERK) pathway(Yarden and Sliwkowski, 2001) NOT CLEAR IF THESE ARE EXAMPLES OF THREE DIFFERENT PATHWAYS?. Excessive ErbB signalling plays crucial roles in the initiation and progression of human epithelial malignancies. In the last decade, various treatments have been developed to target the ErbB signalling network (Baselga, 2006); nevertheless certain features of this receptor family such as redundancy and complexity of signalling, limit the clinical efficacy of targeted therapy(Citri and Yarden, 2006).

The HER2/neu proto-oncogene is amplified in 25-30% of human primary breast tumors, and its alteration is linked to disease behaviour (Slamon et al., 1987). Moreover, HER2 overexpression/amplification was observed with high frequencies also in lung, gastric and oral cancers (Hou et al., 1992; Schneider et al., 1989; Weiner et al., 1990; Xia et al., 1997; Yoshida et al., 1989). An effective drug targeting HER2 is the humanized mouse monoclonal antibody trastuzumab, also known as herceptin, which is currently in clinical use both in the adjuvant and metastatic settings of HER2-overexpressing breast cancers(Lemieux et al., 2009). Despite clinical efficacy in responding patients, the majority of trastuzumab-treated patients with metastatic breast cancer, who initially respond to trastuzumab therapy, develop resistance within one year; therefore mechanisms of action and processes underlying resistance are of high clinical interest (Nahta and Esteva, 2006). In need of circumventing resistance to trastuzumab, new therapeutic strategies have been developed. One emerging drug is the dual tyrosine kinase inhibitor lapatinib, targeting both EGFR and HER2 (Xia et al., 2002). Another important candidate is the monoclonal antibody pertuzumab, which inhibits heterodimerization of HER2 with EGFR and with HER3 (Agus et al., 2002).

Another promising strategy might be to combine mAbs against the same receptor. Our previous studies revealed that HER2 comprises at least seven prominent antigenic sites; one of the most immunogenic epitopes mediates heterodimerization with other members of the ErbB/HER family (Klapper et al., 1997). We later reported that pairs of mAbs specific to distinct epitopes of HER2, of which one is involved in heterodimerization, are highly effective in inhibiting growth in vivo and in vitro of HER2-overexpressing cells (Ben-Kasus et al., 2009). The strategy of combining two or more antibodies against distinct epitopes of the same receptor was previously demonstrated (Drebin et al., 1988; Kasprzyk et al., 1992), and its added benefit has been attributed to various factors, including enhanced receptor degradation (Friedman et al., 2005) and improved recruitment of immune effector cells (Spiridon et al., 2002). Yet, another study concluded that the cooperative effects of mAb combinations might be due to different, but complementary mechanisms of action, such as inhibition of HER2 dimerizationand prevention of formation of an active fragment of HER2, p95HER2 (Scheuer et al., 2009). It is notable that a recent phase II clinical trial, which combined trastuzumab and pertuzumab in patients with HER2-positive breast cancer whose disease had progressed during prior trastuzumab-based therapy, achieved a 24.2% objective response rate and 7.6% complete response (Baselga et al., 2009).

The present study aimed at the establishment of an in vitro model system able to quantitatively reflect cooperative effects of monoclonal anti-HER2 antibodies on cellular phenotypes. To this end we employed human breast cancer cell lines, as well as an engineered normal mammary cell line, MCF10A, overexpressing HER2. When tested under conditions permitting mammary cells to form duct-like spheroids in extracellular matrix, combinations of anti-HER2 mAbs negated an invasive phenotype promoted by growth factors. Computational image analyses attributed the inhibitory action of mAb combination to an ability to increase intraluminal cell death and to abrogate morphological alterations. These effects are discussed in light of the potential clinical applications of antibody combinations in the treatment of HER2-overexpressing breast cancer.

Results

The growth-inhibitory effects of anti-HER2monoclonal antibodies and their combination are reflected in vitro

Unlike single mAbs to HER2, specific combinations of mAbs were able to persistently eliminate N87 gastric tumours in mice (Ben-Kasus et al., 2009; Friedman et al., 2005). Internalization and degradation induced by combinations of mAbs targeting the same receptor can contribute to therapeutic efficacy, especially when one antibody of the pair prevents heterodimerization of HER2 (Ben-Kasus et al., 2009). INTRODUCTION??? Because the proposed mechanism involves no immune effector cells, it is conceivable that the cooperative action of antibody combinations would be reflected in vitro, especially when applied on 3D structures mimicking structural and functional features of epithelial organs. To this end, our initial analyses tested the effect of different anti-HER2 therapeutics in T47D cells, grown either in 2D or 3D configurations (Fig. 1). Cell viability assays performed on monolayers showed that the humanized antibody trastuzumabcaused a 17% reduction in T47D cell proliferation (Fig. 1A). In comparison, 14% and 18% reductions were observed with our murine mAbs N12 and 431, the latter is directed against the dimerization site of HER2. Strikingly, we observed a 38% reduction by the combination of mAbs N12 and 431 (Fig. 1A). Similarly enhanced growth inhibitory effects were observed when two other breast cancer cell lines were used, BT474 and MCF7 (Supplementary Figure S1). It is notable that the latter expresses relatively low levels of HER2, raising the possibility that the combination of mAbs sensitizes low expressors. Using the same assay, we found that lapatinib reduced viability of T47D cells in a dose dependent manner, reaching 42% at 250nM(Fig. 1B). Thus, a combination of two antibodies cooperatively inhibits growth of HER2-overexpressing mammary tumor cells, but a kinase inhibitor is even more potent.

I AM NOT SURE A PEDANTIC REFEREE WOULD NOT PICK ON THE USE OF COOPERATIVITY. IF EACH Ab GAVE 20% EFFECT AND TWO TOGETHER SHOW 40% EFFECT, THIS MAY INDICATE COOPERATIVITY NON-COOPERATIVITY OR EVEN ANTAGONISM DEPENDING ON THE WHOLE RESPONSE CURVE OF EACH DRUG, AND THE MODELING OF THE COMBINED ACTION MECHANISM (E.G. ADDITIVE OR INDEPENDENT).

OK: you discuss this nicely later. Maybe then point that you do to avoid Qs.

Our preliminary experiments indicated that T47D cells organize into spheroids when cultured in extracellular matrix and stimulated with NRG, but therapeutic anti-HER2 agents slightly reduced spheroid size (data not shown). Hence, in the next step we monitored antibody effects on the average size of T47D spheroids (n=80) grown in 3D cultures (Figs. 1C and 1D). T47D spheroids were grown without and with NRG, and from day 4 they were treated with lapatinib, trastuzumab, single mAbsor a combination. Notably, this analysis indicated that only the combination of mAbs induced a statistically significant (p=0.0008) effect on spheroid size. Neither NRG alone nor each single mAb elicited statistically significant alterations, although we noted positive and negative trends, respectively. Similar analyses of EGF-treated 3D structures formed by two additional mammary cell lines, SKBR3 and BT474, which highly and moderately overexpress HER2, respectively, confirmed that the effect of antibody combination is general (Supplementary Fig. S2). Once again,only the combination of two mAbs resulted in statistically significant shrinking of spheroids, reinforcing the notion that antibody-induced morphogenic effects may not be induced by each mAb alone, but they require cooperative actions of two mAbs.

Overexpression of HER2 enhances the growth inhibitory effects of an antibody combination in a mammary model cell line

MCF10A cells are spontaneously immortalized, normal breast epithelial cells of human origin. This cell system and 3D derivative structures have been extensively used to analyze effects of HER2 and EGFR. For example, activation of HER2, but not of EGFR, led to re-initiation of intraluminal proliferation within 3D spheroids of MCF10A cells (Muthuswamy et al., 2001). Using retroviral vectors, we previously established a derivative of MCF10A cells that co-expresses HER2 and the green fluorescence protein (GFP) (C.R. Pradeep et al., manuscript submitted for publication). Employing this experimental system, we found that HER2-overexpressing spheroids protrude invasive arms into the surrounding extracellular matrix upon EGF stimulation. Hence, we assumed that MCF10A and their derivatives, MCF10A-HER2 cells, will reflect the ability of antibody combinations to alter morphology and invasiveness of 3D mammary structures depending on HER2 expression levels.

Western blot analyses of whole extracts prepared from day eight spheroids of MCF10A-HER2 cells and the parental, MCF10A cells, reflected a difference in HER2 expression, as well as high basal HER2 phosphorylation (Fig. 2A). Quantification of HER2 levels (data not shown) indicated similar expression level to that presented by BT474 cells, but lower than HER2 expression by SKBR3 cells. To observe effects on phosphorylation of HER2, we treated the spheroids for 48 hours with either an analogue of lapatinib (GW2974; 100 nM) or with trastuzumab (10 µg/ml), both applied after a 30-minute long stimulation with EGF (20 ng/ml). GW2974 was able to erase diminish HER2 phosphorylation (at tyrosine 1248), independent of EGF. In addition, trastuzumab was able to partly block phosphorylation of HER2 in intact cells (Fig. 2A). In contrast with MCF10A-HER2 cells, the parental MCF10A cells exhibited low levels of HER2 expression and tyrosine phosphorylation, even in the presence of EGF.

Next, we examined the effect of lapatinib, as well as mAbs N12, 431 and their combination, in a cell viability assay, following 48 hours of treatment of monolayers with increasing drug concentrations (Fig. 2B). As observed when using mammary cancer cells (Fig. 1A), the combination of mAbs was able to reduce viability to a larger extent than each antibody alone. Notably, the inhibitory effect of the combination appeared at lower concentrations compared to each single mAb. In addition, it is important noting that the effects of both lapatinib and the antibody combination on inhibition of cell viability were more pronounced in MCF10A-HER2 cells than in the parental cells. In conclusion, the cooperative inhibitory effects of mAbs to HER2 observed on cancer cells (Fig. 1) may be extended to a model cellular system overexpressing HER2. It is nevertheless imperative noting that MCF10A cells, whose HER2 level is low, also displayed inhibition by the mAb combination, suggesting that the mixture may sensitize low expressors of HER2.

To answer the question whether the mAb combination displays synergistic or only additive effects on cell viability, we analyzed the results presented in Figures 1 and 2 by employing area under the curve (AUC) calculations (see Supplementary Materials and Methodsfor details). Essentially, we compared the values of the model parameters Em (lowest overall cell viability reached by the specific treatment) and Kd ((1+ Em)/2; lower Kd means higher drug affinity). The calculated AUC values are shown in Supplementary Figure S3. It is clear that adding a second mAb significantly increased AUC, especially in T47D and MCF10A-HER2 cells. Next, we calculated for each cell line the percentage of increase of inhibitory effect of the mAb combination, beyond additive effects of single mAbs (Fig. S3B). A synergistic effect of the mAb combination was observed in all 3 cell lines, with the strongest effect observed in T47D cells. In the last step, we calculated the equally effective concentrations of single mAbs and their combination in MCF10A-HER2 cells (Fig. S3C), and found that the mAb-combination elicits a strong reduction in equally effective doses. In aggregates, these calculations clearly attribute a synergistic inhibitory effect to the combination of two mAbs to HER2.

EGF induces an invasive phenotype in MCF10A-HER2 spheroids, but anti-HER2 targeted therapies are able to reverse this phenotype

As a prelude to testing morphogenic effects of mAb combinations, MCF10A-HER2 spheroids were grown with EGF (20 ng/ml), and from day 4 onward they were treated with a single mAb, N12 or 431, or with their combination. When analyzing average spheroid size, we observed that the mixture of two mAbs exerted a significantly stronger effect on the reduction of average spheroid size (Supplementary Fig. S4; p=10-5 for thecombinationvs. N12, p=0.002for the combination vs. mAb 431; n>50 for each analysis). This observation is reminiscent of the results we obtained with neuregulin-treated breast cancer cells (Fig. 1C), implying that spheroid size may serve as an indicator of therapeutic efficacy. The detailed microscopic analysis, which also included lapatinib and trastuzumab, is shown in Figure 3. In contrast to MCF10A spheroids, which displayed a lumen in the un-stimulated state, luminal filling occurred in untreated MCF10A-HER2 spheroids, confirming the ability of an overexpressed HER2 to increase luminal proliferation (Debnath et al., 2002). Importantly, EGF had no effect on the outer shape of MCF10A spheroids, but in MCF10A-HER2 spheroids, EGF was able to induce a multi-acinar invasive phenotype (Fig. 3A). All drug treatments shown in Figure 3B were able to inhibit ligand-induced formation of invasive structures in MCF10A-HER2 spheroids. Alongside with inhibition of outgrowths, upon long incubation with the drugs (and EGF), a lumen occurred in all treatment groups. In conclusion, luminal filling is propelled by HER2 overexpression, as previously shown using chimeric HER2 molecules (Muthuswamy et al., 2001). On the other hand, formation of invasive structures is driven by EGF, but both phenotypes can be reversed to variable extents by anti-HER2 agents, such as a kinase inhibitor and a mAb combination.

Computational image analyses indicate that stimulation with EGF significantly alters morphological parameters and reduces intraluminal death of MCF10A-HER2 spheroids

Because of the subtle effects of HER2 overexpression and EGF treatment on spheroid morphology, we developed applied a home-writtena quantitative image analysis software that usesto photomicrographs captured by confocal microscopy. Briefly, intraluminar cell death and of evasive extentson from spheroids surface were evaluated as follows: in a first step, the outer rimsArea of the spheroids were smoothed identified (segmented?) by a threshold (Fig. 4A).In the next step, weAnother higherreduced threshold level (set as a constant fraction of the intensity inside the spheroid) [PLEASE ALSO CORRECT LABEL “Threshold reduction” to “higher threshold” IN FIGURE 4. ALSO YOU SHOULD BE CONSISTENT IN NUMENCLATURE: CTH=SPHEROID AREA/CONVEX HULL AREA.

ALSO: USE SAME COLOR (RED) TO OUTLINE SPHEROID AREA, AND OUTLINE (BLUE) THE NECROTIC AREA] the threshold to visualize identify (or segment?) internal details (e.g. low fluorescence area due to cell death in the lumen). The fractions of non-apoptotic luminal areas were calculated (called Core Factor) from the total threshold reducedspheroidarea and smoothed pictureswere calculated (“Core Factor”). The Cell spheroid to Convex Huall area Ratio (“CTH”) (Fig. 4B) assessed the dispersion of the outer border of spheroids in terms of convexity. Another imaging parameter that evaluates the smoothness of spheroids surface, “Solidity”, measuresd the relation ratio between spheroid area and spheroid perimeter (Fig. 4B).In the first step, we quantified the effects of EGF using a home-made program (Priisma) and spheroid cross-sections. Control MCF10A spheroids, which were cultured without or with EGF (20 ng/ml) for 15 days, displayed no significant differences when the values of Cell to Hall Ratio (CTH) and Core Factor were compared (Figs. 5A and 5C). In contrast, MCF10A-HER2 spheroids displayed a significant effect of EGF on both the Core Factor (p=0.003; Fig. 5B) and CTH (p=0.004; Fig. 5D) on day 15. This difference is not yet detectable at day 10 (results not shown), indicating that the multi-acinar invasive phenotype develops over time following stimulation of HER2-overxpressing spheroids with EGF. We conclude that EGF exerts a significant effect on invasive growth and intraluminal proliferation, only when HER2 is overexpressed.

A combination of two mAbs better than single antibodies reverses EGF-induced changes of CTH and Core Factor in a model cellular system overexpressing HER2

To quantify the effects of EGF and anti-HER2 targeted treatments of MCF10A-HER2 spheroids, the 3D structures were treated with lapatinib, mAb N12, mAb 431, or a mAb combination, and cross-section photomicrographs were captured on day 15 (Fig. 6). The results we obtained clearly indicated that the combination of two mAbs was able to reverse the effect of EGF on the Core Factor significantly stronger than each antibody alone (combination versus mAb N12: p=0.0002; combination versus mAb 431: p=0.0001). In case of CTH (Cell to Hall Ratio), the mAb-combination was also superior compared to single antibody treatments in reversing the EGF-induced invasive phenotype (combination versus mAb N12: p=0.02; combination versus mAb 431: p=0.04). As expected, lapatinib (100nM) was able to significantly reverse the EGF-induced phenotype both for Core Factor (p=0.01) and Cell to Hall CTHRatio (p=0.03). In conclusion, the computational image analyses we performed on multiple 3D structures firmly establish the ability of an antibody combination to restore the circular shape (CTH and Solidity) and increase intraluminal cell death (Core Factor). Consistent with the sensitivity of the image analysis and parameters we derived, treatment with a single antibody, mAb 431, which inhibits heterodimerization of HER2, reached show statistically significant p values only forcewhen CTH ??? is this what you meant??? was quantified. I AM NOT SURE I UNDERSTAND THE STATISTICAL STATEMENTS??? ARE YOU QUOTING STUDENT T-TEST P FACTORS BETWEEN TWO DISTRIBUTIONS OF THE PARAMETERS??? Why not quote also the t-test or z-test values???