Oncogene (2014) 33, 966–976 & 2014 Macmillan Publishers Limited All rights reserved 0950-9232/14 www.nature.com/onc

ORIGINAL ARTICLE Identification of novel determinants of resistance to in ERBB2-amplified cancers

D Wetterskog1, K-K Shiu1, I Chong1,2, T Meijer1, A Mackay1, M Lambros1, D Cunningham2, JS Reis-Filho1, CJ Lord1 and A Ashworth1

The encoding the ERBB2, also known as HER2, is amplified and/or overexpressed in up to 15% of breast cancers. These tumours are characterised by an aggressive phenotype and poor clinical outcome. Although therapies targeted at ERBB2 have proven effective, many patients fail to respond to treatment or become resistant and the reasons for this are still largely unknown. Using a high-throughput functional screen we assessed whether found to be recurrently amplified and overexpressed in ERBB2 þ ve breast cancers mediate resistance to the ERBB2-targeted agent lapatinib. Lapatinib-resistant ERBB2-amplified breast cancer cell lines were screened, in the presence or absence of lapatinib, with an RNA interference library targeting 369 genes recurrently amplified and overexpressed in both ERBB2-amplified breast cancer tumours and cell lines. Small interfering RNAs targeting a number of genes caused sensitivity to lapatinib in this context. The mechanisms of resistance conferred by the identified genes were further investigated and in the case of NIBP (TRAPPC9), lapatinib resistance was found to be mediated through NF-kB signalling. Our results indicate that specific amplified and/ or overexpressed genes found in ERBB2- amplified breast cancer may mediate response to ERBB2-targeting agents.

Oncogene (2014) 33, 966–976; doi:10.1038/onc.2013.41; published online 11 March 2013 Keywords: HER2; ERBB2-positive breast cancer; lapatinib resistance; RNAi screen; NIBP; TRAPPC9

INTRODUCTION The mechanism of action of has also been shown to 21,22 Breast cancer is a heterogeneous disease characterised by a series involve both innate and adaptive immunity. Although both of molecular subtypes, which have distinct molecular features and treatments have provided substantial benefits for patients, clinical behaviour.1,2 Some of these molecular features, such as resistance to treatment remains a major problem. Only about half amplification and overexpression of genes, provide biomarkers of the patients with ERBB2-overexpressing metastatic breast cancer 4 and drug targets that can be utilised for treatment of the specific respond when trastuzumab is given in addition to subgroup.3 For example, expression of the oestrogen receptor in and even less (32–34%) respond with trastuzumab monothe- 4,23 tumours is used to direct the use of anti-oestrogen agents such as rapy. In addition, many of the patients who initially respond to 4,16,23 tamoxifen and aromatase inhibitors, whereas tumour overexpres- anti-ERBB2 therapy relapse within 1 year. sion of ERBB2 or ERBB2 gene amplification are both a drug target Several potential mechanisms of resistance to ERBB2-targeted as well as a prognostic and predictive biomarker for anti-HER2 treatments have been proposed, including activation of growth agents.4,5 factor receptors, alterations of ERBB2 signalling molecules as well 24 Although approximations of ERBB2 amplification and overexpres- as the expression of drug-resistant ERBB2 isoforms. From studies sion in breast cancer vary from study to study, conservative in trastuzumab-resistant breast cancer cell lines, elevated EGFR estimates suggest that somewhere between 15–20% of all breast signalling has been implicated as a potential resistance 25 cancers display this characteristic.6–10 Overexpression of ERBB2 mechanism to trastuzumab. Similarly, heregulin-mediated results in accumulation of ERBB2 receptors in the plasma membrane activation of ERBB3 has also been shown to provide resistance 26 of the cell. ERBB2 interacts with itself and with other members of to trastuzumab treatment in ERBB2-amplified cells. Over- the EGFR family of growth factor receptors (EGFR, ERBB3 and ERBB4) expression of the -like growth factor 1 receptor has also to form homo- or heterodimers; these dimerisation events activate been shown to provide resistance to ERBB2-targeted treatments, downstream molecular cascades that promotes cell proliferation, probably via activation of the ERBB2 insulin-like growth factor 1 27 survival as well as tumour metastasis.11–15 receptor common mediators AKT and p42/p44. Loss of PTEN 19 The current standard of care for early-stage and metastatic function, found in up to 25% of ERBB2-positive breast cancers, ERBB2 ve breast cancer is ERBB2-targeted treatment in the form and activating of PIK3CA, found in up to 30% of breast þ 28,29 of the monoclonal antibody trastuzumab in combination with cancer, appear to mediate resistance to ERBB2-targeted chemotherapy.4 In the case of metastatic ERBB2 þ ve breast cancer, treatment through their activation of the ERBB2 downstream 19,30,31 a small-molecule inhibitor of ERBB2, lapatinib, has also been PI3K/AKT/mTOR pathway. A different mechanism of approved for treatment.16 The mode of cell inhibition caused by resistance has also been identified with the discovery of the these agents is not fully understood but in part involves truncated form of ERBB2 (p95ERBB2), which is generated by downregulation of ERBB2 intracellular signalling cascades.17–20 shedding of the extra cellular domain to which trastuzumab

1The Breakthrough Breast Cancer Research Centre, The Institute of Cancer Research, London, UK and 2GI Unit, Royal Marsden Hospital, Sutton, UK. Correspondence: Professor JS Reis-Filho or Dr CJ Lord or Professor A Ashworth, The Breakthrough Breast Cancer Research Centre, The Institute of Cancer Research, Breakthrough, ICR, 237 Fulham Road, London SW3 6JB, UK. E-mail: [email protected] or reisfi[email protected] or [email protected] Received 14 September 2012; revised 8 January 2013; accepted 11 January 2013; published online 11 March 2013 RNAi screen for lapatinib resistance D Wetterskog et al 967 binds.32 This variant of ERBB2 retains its kinase activity and thus the effect on cell viability (in the absence of lapatinib) of siRNA can still activate downstream signalling while being resistant to targeting ERBB2. We found that the effect of ERBB2 siRNA on cell trastuzumab.33 It should be noted, however, that although some growth inhibition significantly correlated with the effect of cell studies suggested that expression of p95ERBB2 predicted growth by lapatinib (r2 ¼ 0.79, Po0.001, Figure 1c) suggesting that resistance to trastuzumab treatment,34,35 others failed to find the siRNA transfection procedure used could deliver biologically such an association and p95 appears to have no role in resistance relevant screening data. to lapatinib.36 Internal validation of lapatinib sensitisation effects identified by ERBB2-amplified breast cancers have complex genomic profiles the screen was provided by the demonstration that siRNAs and harbour a number of discrete amplified genomic regions in targeting RAC1 caused lapatinib sensitisation, and that siRNA addition to the ERBB2 amplicon. Of these, amplifications of regions targeting PTEN caused resistance to lapatinib as previously on 1q, 8q and 20q are the most common.37–39 reported30,44 (Figure 1d). For each primary screen we identified However, the biological significance of these additional amplifi- the siRNAs with significant effects on lapatinib response, which cation events is not yet clear, though they likely encompass targeted an amplified gene. Of the identified sensitising hits oncogenes and potential therapeutic targets.40,41 Given the (Table 2), 21 genes fulfilled these criteria. Full results of the screens presence of co-amplifications in ERBB2-amplified breast cancers are supplied as Supplementary Table 2. we hypothesised that these amplicons may contain genes whose overexpression results in resistance to anti-ERBB2-targeted Validation of primary screen hits therapy. To assess this hypothesis, we used a high-throughput To exclude hits resulting from off-target effects the four individual lapatinib small interfering RNA (siRNA) sensitivity screen (HTS) siRNA oligonucleotides constituting each SMARTpool were investi- using a siRNA library targeting recurrently amplified genes, which gated for their ability to reproduce the effect seen for the are overexpressed in ERBB2-amplified primary breast cancer SMARTpool in the primary screen. A hit was defined as likely to tumours and cell lines. We demonstrate that genes mapping to be on-target if two or more of the four siRNAs targeting the same recurrently amplified regions in ERBB2-amplified breast cancer gene resulted in sensitisation to lapatinib. Although different siRNAs provide resistance to ERBB2-targeted treatments. targeting the same gene had differing effects on the survival fraction in DMSO-treated cells (an effect possibly due to the different off-target profiles of different siRNAs), we were still able to RESULTS validate the following genes as causing lapatinib sensitivity: RAB34, Lapatinib sensitisation high-throughput siRNA Screen TP53INP1, RAC1, ATP6C1V1, C11ORF73, MLLT6, NIBP (TRAPPC9), NUFIP, PROCA1, RAB7L1, RAD21, SCRN2 and SPOP (Figure 2a). Full results of To identify suitable breast cancer cell line models for our study we the screens are found in Supplementary Table 3 and the survival determined the ERBB2-amplification status and lapatinib sensitiv- effects of the siRNAs with or without lapatinib in Supplementary ity of 11 breast cancer cell lines (Figure 1a, Supplementary Table 4. Lapatinib dose-response survival analysis also confirmed Figure 1 and described in Shiu et al.42 Of the cell lines confirmed the effects of a majority of the candidate sensitising genes, as seen to be ERBB2-amplified by array-based comparative genome by the shift of the lapatinib response curve to a less resistant profile hybridisation and in situ hybridisation, three (Figure 2b). The sensitising effect of the siRNAs was considered proved to be lapatinib sensitive, namely BT474, SKBR3 and validated if they caused at least a two-fold sensitisation to lapatinib ZR75.30, as previously described.18 Seven cell lines, however, were when compared with control-transfected cells (Table 3). We did, defined as lapatinib resistant (SF 40.2 mM, Table 1). The lapatinib- 50 however, note that none of the siRNA sensitization effects caused a resistant ERBB2-amplified cell lines, HCC202, VP229, HCC1569, level of lapatinib sensitivity observed in inherently sensitive MDAMB453, HCC1954, JIMT1 and MDAMB361, where transfection ERBB2 þ ve models; it is possible that the transient and incomplete conditions had been previously successfully optimised,43 were nature of siRNA-induced gene silencing precludes seeing a total taken forward for analysis with the HTS. sensitization to the level in inherently sensitive models. To enable the study of the significance of amplification events in ERBB2 þ ve breast cancer, a siRNA library had previously been constructed (as described in Shiu et al.42 and Supplementary Cross-validation screen in ERBB2-amplified oesophageal cancer Table 1). The siRNA library used in our HTS-targeted genes Approximately 20% of patients with oesophageal adenocarcinoma (n ¼ 369) recurrently amplified and overexpressed in ERBB2- (EAC) have ERBB2 þ ve tumours.45,46 To assess whether our amplified breast cancer. These genes had previously been findings in ERBB2-amplified lapatinib-resistant breast cancer identified using array-based comparative genome hybridisation were relevant for ERBB2-amplified lapatinib-resistant EAC, we and microarray analysis of 45 ERBB2-amplified invasive breast investigated the effect of silencing of our HTS hits in ERBB2- cancers and 14 ERBB2-amplified breast cancer cell lines. The amplified EAC cell lines. The ERBB2 amplification status of EAC cell criteria for inclusion of a gene as a target in the library was that it line models was determined by array-based comparative genome was recurrently (n42) amplified and overexpressed in both hybridisation. Three of the four cell lines analysed (OE33, the tumours and cell lines. This criterion was set to triage the number cisplatin-resistant OE33 clone (CROE33) and OE19) exhibited of amplified and overexpressed genes down to a gene-set more ERBB2 amplification (Supplementary Figure 3). Of these cell lines likely to be of importance to the biology of ERBB2-amplified breast only OE19 was lapatinib sensitive, CROE33 was intermediate cancer and to increase the chance of having cell line models, sensitive and OE33 and the non-ERBB2-amplified FLO-1 cell line which could be screened and perform functional analysis in. were resistant (Figure 3a). As expected, the lapatinib-sensitive cell The HTS involved reverse transfecting lapatinib-resistant cell line OE19 was also sensitive to ERBB2 gene silencing (Figure 3b). lines with siRNA in a 96-well plate format after which cells were Validation siRNA screens were performed in the lapatinib-resistant exposed to lapatinib for 5 days and the cell viability estimated cell lines OE33 and CR-OE33. In both cell lines silencing of (Figure 1b). The screen was designed to detect modestly TP53INP1 sensitised to lapatinib for three out of four siRNAs sensitising effects by using doses of lapatinib at or close to SF80. (Figure 3c). Full results of the validation screens can be found in Quality control analysis of the screens indicated a high transfec- Supplementary Table 3. Moreover, TP53INP1 silencing sensitised to tion efficiency in each cell line as demonstrated by the satisfactory lapatinib over a wide range of concentrations in CROE33 and OE33 dynamic range between the effect on cell viability of non- cells (Figure 3d). Although TP53INP1 was not amplified in either targeting siRNA when compared with siRNA targeting PLK1 OE33 or CROE33 both cell lines had the highest expression (Supplementary Figure 2). As an additional control, we assessed levels of TP53INP1 as compared with FLO and OE19 (Figure 3e).

& 2014 Macmillan Publishers Limited Oncogene (2014) 966 – 976 RNAi screen for lapatinib resistance D Wetterskog et al 968

BT474 1.0 SKBR3 ZR75.30 HCC202 Amplicon siRNA library VP229 0.5 HCC1569 HCC1954 JIMT1 MDAMB361 Surviving fraction 0.0 MDAMB453 0 -8 -7 -6 -5 MCF7 Divide into replicates and reverse transfect cells Lapatinib log[M]

2 r2=0.83 0 Control Lapatinib at ~SF80 -2 Expose for 5 days

Z-score -4

siERBB2 viability -6

-7.5 -7.0 -6.5 -6.0 -5.5 -5.0 Measure cell viability

Lapatinib SF50 log[M]

MDAMB361 JIMT1 HCC1954 HCC1569 MDAMB453 VP229 HCC202 6.0

4.5

siPTEN 3.0

siPTEN 1.5

0.0

-1.5 Drug effect Z-score

-3.0

-4.5 siRAC1

-6.0 siRNA 1-400 for each cell line Figure 1. ERBB2 amplicon siRNA library screen identifies 45 genes as potential modulators of lapatinib response. (a) Identification of lapatinib- resistant cell lines. Eleven breast cancer cell lines (10 ERBB2-amplified and the non-amplified MCF7) were plated in 96-well plates and treated with lapatinib ranging from 0–10 mM. Each treatment was done in triplicate. Twenty-four and 96 h after plating, cells were treated with lapatinib. Cell viability was assessed after 5 days of lapatinib exposure using CellTiter-Glo Luminescent Cell viability assay. Error bars represent s.d. (b) Overview of the high-throughput siRNA screen performed in lapatinib-resistant cell lines. Individual siRNA SMARTpools targeting 369 genes recurrently amplified and overexpressed in ERBB2-amplified breast cancer and 31 genes hypothesised to provide resistance to lapatinib were plated in 96-well plates. Each transfection plate contained 80 experimental siRNAs supplemented with 4 wells each of Mock, siControl1, siControl2 and siPLK1. Transfection mixes were prepared in the transfection plates and divided into six-replica plates whereupon cells were added to each plate. Every 48 h after transfection half of the replica plates were treated with lapatinib at 0.4 or 1 mM. Cell viability was assessed after 5 days of lapatinib exposure using CellTiter-Glo Luminescent Cell viability assay. (c) Comparison of lapatinib SF50 concentrations with viability response to ERBB2 silencing. (d) Analysis of the ERBB2 amplicon siRNA library screen. Scatter plot of averaged drug effect Z-scores from lapatinib sensitivity screens carried out in triplicate. The horizontal grey line indicates the drug effect Z-score value of À 1.5. Grey dots indicate siRNA SMARTpools targeting 369 recurrently amplified and overexpressed genes as well as 31 supplemented SMARTpools targeting genes hypothesised for having an effect on lapatinib response.

Functional analysis of validated hits enhancer of activated B cells) signalling, a known oncogenic We originally hypothesised that amplicons in ERBB2 þ ve breast process (see later) suggested the potential importance of this tumours may contain genes whose overexpression results in gene. To confirm the amplification status of these genes and to resistance to anti-ERBB2-targeted therapy. On this basis we determine the degree of correlation with , we selected two genes for further study, NIBP and TP53INP1, which compared NIBP and TP53INP1 copy number status and mRNA were amplified and overexpressed in cell lines in which siRNA expression in a cohort of 45 patients with ERBB2 þ ve tumours targeting caused a strong lapatinib sensitisation. Furthermore, the (Figure 4a). For both genes a significant association between casual relationship between NIBP (aka TRAPPC9 trafficking protein amplification and overexpression was observed. The ERBB2-ampli- particle complex 9) and NF-kB (nuclear factor kappa--chain- fied cell line HCC1569, where TP53INP1 and NIBP where amplified

Oncogene (2014) 966 – 976 & 2014 Macmillan Publishers Limited RNAi screen for lapatinib resistance D Wetterskog et al 969 and silencing sensitised to lapatinib, showed the highest TP53INP1 Table 1. Lapatinib SF concentrations 50 and high NIBP protein expression (Figures 4b and c). To identify signalling pathways/networks involved in the Cell line Lapatinib SF (nM) 50 lapatinib resistance conferred by the NIBP and TP53INP1, BT474 61 þ / À 7.5 functional protein association networks involving these HCC202 560 þ / À 49 were investigated using STRING 9.0 (http://string-db.org/). Based HCC1569 2440 þ / À 390 on this analysis we identified HIPK2 and IKBKB as interacting HCC1954 3160 þ / À 1280 partners of TP53INP1 and NIBP, respectively (Figure 4d). As JIMT1 5170 þ / À 240 these two genes represented potential druggable kinases, which MDAMB361 410000 could potentially be used in combination treatments with MDAMB453 2060 þ / À 730 SKBR3 79 þ / À 19 lapatinib we silenced them and assessed their ability to modulate VP229 1360 þ / À 410 lapatinib response. Silencing of IKBKB was shown to sensitise to ZR75.30 122 þ / À 54 lapatinib in HCC1569 where silencing of NIBP (TRAPPC9) sensitised MCF7 410000 to lapatinib, but not in the control cell lines JIMT1 and VP229 (Figure 4e). No sensitisation to lapatinib was observed The table gives a list of cell lines used in the study and their lapatinib upon silencing of HIPK2. SF50 values.

Table 2. Drug effect Z-scores of primary screen hits

Gene Symbol MDAMB361 JIMT1 HCC1954 HCC1569 MDAMB453 VP229 HCC202

ADAM9 0.77 0.36 À 1.56 0.80 0.14 0.97 0.11 AP1GBP1 1.83 3.84 À 0.91 À 2.17 0.16 À 2.67 À 0.33 ATP6V1C1 0.27 0.85 À 1.83 À 2.32 À 0.75 À 4.55 0.51 C11ORF73 À 1.61 3.07 0.34 0.05 0.09 À 0.33 À 2.17 CTSC 1.35 À 0.15 À 1.73 À 1.02 À 0.77 À 1.12 À 0.46 DDX42 À 0.58 1.34 À 1.30 À 2.14 À 0.36 À 2.59 À 0.90 E(Y)2 À 0.55 À 0.85 À 0.18 À 0.18 À 2.21 À 4.35 À 0.98 EIF3S3 0.30 À 0.50 À 0.25 À 1.78 1.45 À 3.87 À 1.03 FBXO43 0.05 0.44 À 0.51 À 1.51 0.69 À 0.27 0.38 FLJ20291 À 0.02 0.04 0.99 À 1.25 À 0.06 À 1.57 À 0.57 FLJ22578 2.26 À 2.49 0.34 À 0.49 0.85 À 2.00 À 0.56 GALGT2 0.13 À 1.25 À 0.48 À 1.65 0.13 À 2.59 0.12 HOXB9 À 1.88 0.40 0.35 0.22 À 0.64 0.25 0.85 IRX4 0.67 À 2.87 À 0.83 3.02 0.83 À 0.49 1.44 MGC23280 À 2.77 0.25 À 0.48 À 0.74 0.25 1.94 À 0.02 MLLT6 À 1.26 À 1.39 À 0.81 À 1.47 À 1.74 À 1.95 À 1.91 MRPL13 0.87 À 0.62 1.04 À 1.25 À 0.30 À 3.36 0.68 MRPL36 À 0.69 À 1.44 À 2.18 À 1.52 2.00 À 0.31 1.53 MYOHD1 À 1.84 À 0.92 0.79 À 3.60 0.23 0.40 0.10 MYST2 À 0.11 0.10 À 0.74 À 3.13 0.89 À 3.56 À 1.63 NIBP À 0.53 0.40 À 0.80 À 2.21 À 3.15 À 0.23 À 0.03 NUFIP 0.56 0.80 À 1.92 0.25 0.46 À 1.47 À 0.93 ORMDL3 0.00 À 0.64 À 0.54 À 2.36 À 3.10 À 0.55 À 1.35 OTUD6B À 0.47 À 0.72 À 0.86 À 1.85 À 0.54 À 0.29 À 0.56 PDGFRB 0.12 0.75 0.81 À 0.74 À 0.55 À 1.97 À 0.88 PDK2 À 0.48 À 1.35 À 1.02 À 1.17 À 0.71 À 0.30 À 1.59 PFDN4 À 3.52 0.99 À 0.85 0.20 0.20 1.06 0.70 PPARBP 0.02 À 0.05 1.85 0.11 0.80 À 1.95 0.29 PROCA1 1.01 À 0.89 À 0.44 À 1.82 0.06 0.18 À 1.67 RAB34 À 1.34 0.01 À 0.22 À 2.56 À 3.09 1.24 À 1.84 RAB7L1 À 0.15 À 0.21 À 1.11 À 0.74 À 4.85 1.35 À 0.21 RAC1 À 0.26 À 4.02 À 2.65 À 0.93 0.22 À 1.01 0.23 RAD21 0.47 À 3.65 2.82 À 0.21 0.92 À 2.50 1.44 RECQL5 À 1.49 0.33 À 1.00 À 0.08 À 3.77 0.48 À 1.31 SCRN2 À 0.13 À 0.16 À 1.41 À 1.06 À 4.09 0.61 À 1.79 SPOP 0.17 À 0.67 0.02 À 0.58 À 0.84 À 1.21 À 1.79 STAU2 À 0.06 À 0.21 À 1.34 0.59 À 3.64 À 0.41 À 1.00 THRA À 0.42 À 0.31 À 0.39 0.17 À 5.29 À 0.85 0.23 TP53INP1 0.15 À 0.92 À 1.33 À 2.25 À 4.56 À 0.89 À 0.97 UBE2W À 0.18 0.85 À 0.59 0.05 À 6.97 0.27 0.20 USP32 0.77 0.59 0.28 À 0.74 À 0.87 0.00 À 2.19 WSB1 0.32 0.21 À 0.85 À 3.41 À 0.55 À 0.42 À 1.40 ZNF217 À 1.23 À 1.71 2.40 À 0.17 0.77 À 0.66 1.59 ZNF572 0.09 À 3.64 2.52 0.57 0.48 0.05 0.93 ZNF704 0.72 À 1.55 0.61 À 3.04 À 1.09 0.53 À 1.17 The table shows the results of the lapatinib-sensitisation amplicon siRNA library screen and list of the 45 sensitising siRNA SMARTpools from the siRNA screen displayed with siRNA SMARTpool target gene name and drug effect Z-score for each cell line. Drug effect Z-scores in bold indicates scores below À 1.5.

& 2014 Macmillan Publishers Limited Oncogene (2014) 966 – 976 RNAi screen for lapatinib resistance D Wetterskog et al 970

1.6 HCC1569 1.6 VP229 1.4 1.4 1.2 1.2 1.0 1.0 0.8 0.8 0.6 0.6 Surviving fraction Surviving fraction 0.4 0.4 0.2 0.2 0.0 0.0 siCTRL siCTRL NIBP Pool NIBP-dup1 NIBP-dup2 NIBP-dup3 NIBP-dup4 SPOP Pool SPOP-dup1 SPOP-dup2 SPOP-dup3 SPOP-dup4 MLLT6 Pool RAB34 Pool RAD21 Pool MLLT6-dup1 MLLT6-dup2 MLLT6-dup3 MLLT6-dup4 RAB34-dup1 RAB34-dup2 RAB34-dup3 RAB34-dup4 RAD21-dup1 RAD21-dup2 RAD21-dup3 RAD21-dup4 PROCA1 Pool C11orf73 Pool C11orf73 Pool PROCA1-dup1 PROCA1-dup2 PROCA1-dup3 PROCA1-dup4 C11orf73-dup1 C11orf73-dup2 C11orf73-dup3 C11orf73-dup4 C11orf73-dup1 C11orf73-dup2 C11orf73-dup3 C11orf73-dup4 TP53INP1 Pool TP53INP1-dup1 TP53INP1-dup2 TP53INP1-dup3 TP53INP1-dup4

HCC1954 JIMT1 1.6 HCC202 1.6 1.6 DMSO 1.4 Lapatinib 1.4 1.4 1.2 1.2 1.2 1.0 1.0 1.0 0.8 0.8 0.8 0.6 0.6 0.6 Surviving fraction

0.4 Surviving fraction 0.4

Surviving fraction 0.4 0.2 0.2 0.2 0.0 0.0 0.0 siCTRL siCTRL siCTRL THRA Pool RAC1 Pool NUFIP Pool THRA-dup1 THRA-dup2 THRA-dup3 THRA-dup4 MLLT6 Pool RAC1-dup1 RAC1-dup2 RAC1-dup3 RAC1-dup4 NUFIP-dup1 NUFIP-dup2 NUFIP-dup3 NUFIP-dup4 RAD21 Pool MLLT6-dup1 MLLT6-dup2 MLLT6-dup3 MLLT6-dup4 SCRN2 Pool RAD21 Pool RAD21-dup1 RAD21-dup2 RAD21-dup3 RAD21-dup4 SCRN2-dup1 SCRN2-dup2 SCRN2-dup3 SCRN2-dup4 RAD21-dup1 RAD21-dup2 RAD21-dup3 RAD21-dup4 RAB7L1 Pool RAB7L1-dup1 RAB7L1-dup2 RAB7L1-dup3 RAB7L1-dup4 ATP6Vc1 Pool ATP6Vc1-dup1 ATP6Vc1-dup2 ATP6Vc1-dup3 ATP6Vc1-dup4

1.6 MDAMB361 1.6 MDAMB453 1.4 1.4 1.2 1.2 1.0 1.0 0.8 0.8 0.6 0.6 Surviving fraction Surviving Surviving fraction Surviving 0.4 0.4 0.2 0.2 0.0 0.0 siCTRL siCTRL NIBP Pool NIBP-dup1 NIBP-dup2 NIBP-dup3 NIBP-dup4 MLLT6 Pool DDX42 Pool MLLT6-dup1 MLLT6-dup2 MLLT6-dup3 MLLT6-dup4 DDX42-dup1 DDX42-dup2 DDX42-dup3 DDX42-dup4 UBE2W Pool RAB7L1 Pool UBE2W-dup1 UBE2W-dup2 UBE2W-dup3 UBE2W-dup4 RAB7L1-dup1 RAB7L1-dup2 RAB7L1-dup3 RAB7L1-dup4 PROCA1 Pool ORMLD3 Pool ORMLD3dup 2 PROCA1-dup1 PROCA1-dup2 PROCA1-dup3 PROCA1-dup4 ORMLD3-dup1 ORMLD3-dup3 ORMLD3-dup4 TP53INP1 Pool TP53INP1-dup1 TP53INP1-dup2 TP53INP1-dup3 TP53INP1-dup4

HCC1569 VP229

1.0 1.0

siCTRL 0.5 siC11ORF73 0.5 siMLLT6 siCTRL siNIBP siRAD21 Surviving fraction siPROCA1 Surviving fraction siRAB34 siC11ORF73 siSPOP 0.0 siTP53INP1 0.0 -8 -7 -6 -5 -8 -7 -6 -5 Lapatinib log[M] Lapatinib log[M]

JIMT1 HCC202 HCC1954 1.0 siCTRL 1.0 1.0 siNUFIP siATP6C1V1 siRAB7L1 siSCRN2 siTHRA 0.5 0.5 0.5 Surviving fraction Surviving

siCT RL Surviving fraction siCTRL fraction Surviving siRA D2 1 siRAD21 siM LLT 6 0.0 0.0 0.0 -8 -7 -6 -5 -8 -7 -6 -5 -7 -6 -5 Lapatinib log[M] Lapatinib log[M] Lapatinib log[M] MDAMB361 MDAMB453

1.0 1.0

0.5 0.5 siCT RL siPR OC A1

siCTRL fraction Surviving siNIBP Surviving fraction siDDX42 siRAB7L1 siML LT6 siTP53INP1 0.0 siUBE2W 0.0 -7 -6 -5 -7 -6 -5 Lapatinib log[M] Lapatinib log[M] Figure 2. Validation screen confirms 16 genes as potential modulators of lapatinib response. (a) Validation of hits from the lapatinib HTS. Lapatinib sensitivity assay was repeated with the four individual siRNA oligos originally included in each SMARTpool. The differences in survival fractions as compared with siCON survival fractions between lapatinib treatment and DMSO drug effect are shown. Error bars represent the s.e.m. (b) Identification of changes in lapatinib response upon silencing of the eight novel validated hits from the screen. Validated siRNA oligos were pooled and reverse transfected into the cell line they were a hit in. Twenty-four hours after plating and every 48 h after that, cells were treated with lapatinib ranging from 0–10 mM. Each treatment was done in triplicate. Cell viability was assessed after 7–10 days of lapatinib exposure using CellTiter-Glo Luminescent Cell viability assay. Error bars represent s.d.

Oncogene (2014) 966 – 976 & 2014 Macmillan Publishers Limited RNAi screen for lapatinib resistance D Wetterskog et al 971 targeting IKBKB. To test if the IKBKB inhibitor IMD-0354 could Table 3. Changes in lapatinib SF values upon gene silencing 50 sensitise to lapatinib, as IKBKB silencing was able to, we performed Cell line Treatment Fold sensitivity Lapatinib combination treatments of lapatinib and IMD-0354. For the lapatinib-resistant cell lines HCC1569 and HCC1954, a synergistic change SF50 (nM) effect could be seen on both NF-kB reporter activity as well as cell JIMT1 siCTRL 1 4369 inhibition (Figures 5d and e and Supplementary Figure 4). Despite siRAD21 1.9 2303 MDAMB453 cells being more sensitive to single agent IMD-0354, a HCC202 siCTRL 1.0 436 clear but more modest effect on lapatinib sensitivity was also siRAD21 0.5 937 observed (Supplementary Figure4). siMLLT6 1.9 235 HCC1569 siCTRL 1.0 1963 siC11ORF73 1.5 1294 DISCUSSION siMLLT6 2.2 887 siNIBP 2.2 912 De novo and acquired resistance to ERBB2-targeted therapies siPROCA1 2.6 751 remains a significant problem, which reduces the clinical benefit in siRAB34 3.1 635 patients. To address this issue, considerable efforts have been siTP53INP1 2.3 849 made to identify the molecular changes that could lead to VP229 siCTRL 1.0 1122 resistance. Although genome-wide studies, together with the siRAD21 2.3 493 generation of cell lines resistant to ERBB2-targeting agents, have siC11ORF73 1.3 877 led to insights into the immediate signalling network in which siSPOP 0.9 1224 ERBB2 acts and identified potential resistance-causing alterations, MDAMB453 siCTRL 1.0 5127 this has of yet not affected clinical practice.31,34,49–51 Here, we siPROCA1 4.3 1181 siTP53INP1 1.2 4355 show that by focusing on recurrently amplified and overexpressed siNIBP 1.3 4002 genes, found in primary ERBB2 þ ve tumours and cell lines, we siMLLT6 11.4 450 could identify genes that result in resistance to anti-ERRB2 agents MDAMB361 siCTRL 1.0 2462 without an obvious connection to previously described resistance siDDX42 0.6 3891 mechanisms. Our confirmation of genes involved in ERBB2- siORMLD3 1.2 2097 targeting agent resistance such as PTEN and RAC1, using this siRAB7L1 1.2 2099 approach, demonstrates the potential of this method. siUBE2W 3.4 721 This is the first study to identify NIBP as being important for HCC1954 siCTRL 1.0 768 lapatinib response in a subgroup of ERBB2-amplified breast cancer siNUFIP 2.4 320 siATP6C1V1 3.8 202 models defined by NIBP amplification and/or overexpression. siRAB7L1 0.7 1037 Importantly, NIBP amplification can be found in up to 11.1% of siSCRN2 0.9 832 ERBB2 þ ve breast cancers and, in these patients, amplification siTHRA 0.6 1350 correlates with NIBP mRNA overexpression (Po0.05). Thus far, NIBP has neither been implicated in cancer nor in therapeutic resistance. The table shows the results of the lapatinib drug response changes upon Instead, the best described roles of NIBP are in protein trafficking52 silencing of validated genes and list of the different lapatinib-resistant cell lines, 47 fold change in lapatinib SF s and the actual lapatinib SF s. and NF-kB signalling. Although there is only one report describing 50 50 this alternative role of NIBP, the data presented here support these previous observations. Hu et al 47 found that NIBP interacted with and activated the IKBKB kinase and activated NF-kB signalling. Although NIBP and the involvement of the NF-kB pathway in lapatinib ERBB2haslongbeenknowntobeabletoactivateNF-kB signalling,53 resistance the mechanisms leading to this activation has yet to be fully 54 We hypothesised that the effect of NIBP depletion on lapatinib elucidated. Merkhofer et al. showed that ERBB2 activates NF-kB sensitivity could be explained by its ability to modulate NF-kB through the canonical NF-kB signalling protein IKKa and not via AKT/ signalling. To study the role of NIBP and the NF-kB pathway in PI3K. In addition, activation of RelA has been shown to be linked to 55 lapatinib resistance, we first generated NF-kB-mediated transcrip- the development of resistance in lapatinib-sensitive cell lines. Taken tion reporter cell lines. The cell lines HCC1569 and HCC1954 were together, these observations could suggest a synthetic gene transduced with a lentivirus containing a NF-kB reporter construct interaction scenario where ERBB2 and NIBP activate NF-kB where tandem repeats of the NF-kB transcriptional response signalling via IKKa and IKBKB, respectively. Inhibition of either element were regulating luciferase expression. After 2 weeks of ERBB2 or NIBP on their own does not significantly affect viability but puromycin selection, the cell lines were tested for the ability of the when simultaneously inhibited causes cell death. NF-kB reporter to respond to the NF-kB signalling activator In conclusion, we described how a focused functional genomics tumour necrosis factor (TNFa). All reporter containing cell lines approach on genes recurrently amplified and overexpressed in showed increased reporter activity upon TNFa treatment con- ERBB2 breast cancer may be used to identify new mechanisms of firming the successful generation of the reporter cell lines resistance to ERBB2-targeted therapies. We demonstrated that by (Figure 5a). As the role of NIBP in NF-kB signalling has only been inhibiting NIBP, its interacting partner IKBKB or the NF-kB pathway partially validated,47 we tested the ability of NIBP silencing to using the IKBKB inhibitor IMD-0354, we sensitised resistant cells to reduce NF-kB reporter activity in the cell lines and compared normally sub-lethal doses of lapatinib. Targeting the NF-kB these effects with the effect of silencing of the well-validated pathway in ERBB2 þ ve cancers dependent on this pathway for NF-kB signalling component RelA48 (Figure 5b). For both cell lines, resistance could offer a therapeutic strategy for improving the siRNA targeting RelA markedly decreased NF-kB reporter activity. response to treatment with ERBB2-targeting agents. NIBP silencing was also able to reduce NF-kB reporter activity. As no small molecule NIBP inhibitors exist that could be used in MATERIALS AND METHODS combination with lapatinib, we assessed the ability of a panel of Cell lines inhibitors to other NF-kB pathway components to modify NF-kB Ten commercially available breast cancer cell lines known to harbour reporter activity (Figure 5c). The strongest effect on NF-kB reporter ERBB2 gene amplification and 1 non-ERBB2-amplified control were activity after 2 days of drug treatment of the NF-kB reporter cell selected. They were purchased from their primary sources and were lines HCC1569 and HCC1954 was seen for IMD-0354, a drug grown as per their exact recommendations. These included BT474,

& 2014 Macmillan Publishers Limited Oncogene (2014) 966 – 976 RNAi screen for lapatinib resistance D Wetterskog et al 972 2 r2=0.39 1.0 0

-2 0.5 OE19 -4 FLO Drug effect Z-score

Surviving fraction CR OE33 -6 0.0 OE33 0 -8 -7 -6 -5 -7.5 -7.0 -6.5 -6.0 -5.5 -5.0 Lapatinib log[M] Lapatinib SF log[M] 50

DMSO SF OE33 DMSO SF CROE33 LAP SF LAP SF 1.0 1.0

0.5 0.5 Surviving fraction Surviving fraction 0.0 0.0

siCON2 siCON2

TP53INP1TP53INP1-dup1 Pool TP53INP1-dup2TP53INP1-dup3TP53INP1-dup4 TP53INP1TP53INP1-dup1 Pool TP53INP1-dup2TP53INP1-dup3TP53INP1-dup4

OE33 CROE33 siCTRL 1.0 1.0 siTP53INP1

0.5 0.5

Surviving fraction siCTRL Surviving fraction siTP53INP1 0.0 0.0 -7.0 -6.5 -6.0 -5.5 -5.0 0 -7 -6 -5 Lapatinib log (M) Lapatinib log (M)

OE19 OE33 CROE33FLO TP53INP1

Actin

Figure 3. Characterisation of ERBB2-amplified oesophageal cell lines. (a) Identification of lapatinib-resistant ERBB2-amplified oesophageal cell lines. Four oesophageal cell lines (3 ERBB2-amplified and the non-amplified FLO) were plated in 96-well plates and treated with lapatinib ranging from 0–10 mM. Each treatment was done in triplicate. Twenty-four and 96 h after plating, cells were treated with lapatinib. Cell viability was assessed after 5 days of lapatinib exposure using CellTiter-Glo Luminescent Cell viability assay. Error bars represent s.d. (b) Comparison of lapatinib SF50 concentrations with viability response to ERBB2 silencing. (c) Assessment of lapatinib sensitising effect of hits from the lapatinib HTS in ERBB2-amplified oesophageal cancer. A lapatinib sensitivity assay was performed in ERBB2-amplified oesophageal cancer using the individual siRNA oligos constituting the HTS hit SMARTpool. The differences in survival fractions as compared with siCON survival fractions between lapatinib treatment and DMSO are shown. Error bars represent the s.e.m. Asterisk indicates a validated siRNA oligo. (d) Identification of changes in lapatinib response upon silencing of TP53INP1. Validated siRNA oligos were pooled and reverse transfected into the cell lines they were a hit in. Twenty-four hours after plating and every 48 h after that, cells were treated with lapatinib ranging from 0–10 mM. Each treatment was done in triplicate. Cell viability was assessed after 7–10 days of lapatinib exposure using CellTiter-Glo Luminescent Cell viability assay. Error bars represent s.d. (e) TP53INP1 expression in oesophageal cancer cell lines. Forty-eight hours after seeding cells, protein lysates were prepared and western blots performed for TP53INP1 and actin as a loading control.

HCC202, HCC1569, HCC1954, MCF7, MDA-MB-361, MDA-MB-453, SKBR3, Drug sensitivity assays ZR-75.30 (ATCC) and JIMT-1 (DSMZ). In addition, the oesophageal cell lines; Cells were seeded into 96-well plates. Twenty-four and 96 h after seeding, FLO, OE19 and OE33 and the breast cell line VP229 were obtained from drugs were added with a final concentration ranging from 0–10 mM for European Collection of Cell Cultures (ECACC, Teddington, UK). lapatinib. After 5 days of drug exposure, cell viability was assessed with CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, WI, USA) Drugs as per the manufacturer’s instructions. To determine SF50 values drug Lapatinib was obtained from Selleck Chemicals (Munich, Germany). BAY11- concentrations were logged, survival fractions calculated by dividing drug 7085, Betulinic acid, IMD-0354, Parthenolide and Tanshinone, was obtained luminescence value with the luminescence value in DMSO wells and from Tocris bioscience (Bristol, UK). plotted against each other using GraphPAD Prism version 5.01 (GraphPad

Oncogene (2014) 966 – 976 & 2014 Macmillan Publishers Limited RNAi screen for lapatinib resistance D Wetterskog et al 973 3 TP53INP1 3 NI BP

2 2

1 1 Relative expression Relative expression 0 0

-1 -1

tumour id tumour id BT474 HCC202 HCC1569 HCC1954 JIMT1 MDAMB361 MDAMB453 SKBR3 VP229 siCTRL siTP53INP1_1 siTP53INP1_2 siTP53INP1_3 siTP53INP1_4

TP53INP1 TP53INP1

Β-tubulin Β-tubulin siCTRL siNIBP_1 siNIBP_2 siNIBP_3 siNIBP_4 BT474 HCC202 HCC1569 HCC1954 JIMT1 MDAMB361 MDAMB453 SKBR3 VP229 NIBP NIBP

Β-tubulin Β-tubulin

HCC1569 VP229 JIMT1

1.0 1.0 1.0

siCTRL 0.5 siCTRL 0.5 TP53INP1 0.5 siCTRL TP53INP1 NIBP TP53INP1 NIBP NIBP Surviving fraction Surviving fraction HIPK2 Surviving fraction HIPK2 HIPK2 IKBKB IKBKB 0.0 IKBKB 0.0 0.0 0 -8 -7 -6 -5 0 -8 -7 -6 -5 0 -8 -7 -6 -5 Lapatinib log[M] Lapatinib log[M] Lapatinib log[M] Figure 4. Characterisation of NIBP and TP53INP1 amplification, overexpression and associated resistance mechanisms. (a) Correlation between amplification and overexpression of TP53INP1 and NIBP in an ERBB2 þ ve patient cohort. Bars in red indicate tumours with gene amplification. (b) Assessment of expression and silencing efficiency of TP53INP1 in the ERBB2-amplified cell lines. Red text indicates TP53INP1-amplified cell lines. (c) Assessment of expression and silencing efficiency of NIBP in the ERBB2-amplified cell lines. Red text indicates NIBP-amplified cell lines. (d) TP53INP1 and NIBP functional protein association networks. (e) Effect of silencing of hit associated genes on lapatinib response. SMARTpool siRNA oligos against hit associated genes were reverse transfected into JIMT1, HCC1569 and VP229. Twenty-four hours after plating and every 48 h after that, cells were treated with lapatinib ranging from 0–10 mM. Each treatment was done in triplicate. Cell viability was assessed after 7–10 days of lapatinib exposure using CellTiter-Glo Luminescent Cell viability assay. Error bars represent s.d.

Software, La Jolla, CA, USA). A non-linear curve fit was performed and SF50 to affect resistance to lapatinib was obtained in 5, 96-well plates concentrations extrapolated from the resulting curve. from Dharmacon (Waltham, MA, USA). Each plate was further supple- mented with 4 wells each of Mock, the negative controls siCON1 and RNAi libraries siCON2 and the positive control siPLK1. Each well in the library contained The amplicon RNA interference (RNAi) library targeting 369 recurrently a SMARTpool of four distinct siRNA species targeting different sequences amplified and overexpressed genes as well as 31 genes hypothesised of the target transcript. The validation library, targeting 45 genes,

& 2014 Macmillan Publishers Limited Oncogene (2014) 966 – 976 RNAi screen for lapatinib resistance D Wetterskog et al 974 a b 16.0 DMSO 2.5 siCTRL TNFa siNIBP siRelA p=0.008 p<0.0001 2.0 4.0 p<0.0001 p<0.0001

1.5 p=0.03 p=0.002 1.0 1.0

0.3 0.5 Fold NFkB reporter activity Fold NFkB reporter activity

0 0 HCC1569 HCC1954 HCC1569 HCC1954

c 3.5 DMSO 3.0 Lapatinib Parthenolide 2.5 Tanshinone IMD-0354 2.0 BAY11-7085

1.5 Betulinic acid TNFa 1.0 Lap+Part Fold NFkB reporter activity Lap+Tans 0.5 Lap+IMD-0354

0.0 HCC1569 HCC1954

d HCC1569 HCC1954 0 uM IMD 0uM IMD 1.0 0.15uM IMD 1.0 0.15uM IMD 0.31uM IMD 0.31uM IMD 0.63uM IMD 0.63uM IMD 1.25uM IMD 0.5 1.25uM IMD 0.5 NFkB reporter activity 0.0 NFkB reporter activity 0.0 0-7.0 -6.5 -6.0 -5.5 -5.0 0-7.0 -6.5 -6.0 -5.5 -5.0 Lapatinib log [M] Lapatinib log [M]

e HCC1569 HCC1954 0um IMD 0um IMD 1.0 0.1uM IMD 1.0 0.1uM IMD 0.4uM IMD 0.4uM IMD 0.5uM IMD 0.5uM IMD 0.6uM IMD 0.6uM IMD 0.5 0.5 0.7uM IMD 0.7uM IMD 0.8uM IMD 0.8uM IMD Survival Fraction Survival Fraction 0.9uM IMD 0.9uM IMD 0.0 1.0uM IMD 0.0 1.0uM IMD 0 -8 -7 -6 -5 1.1uM IMD 0 -8 -7 -6 -5 1.1uM IMD Lapatinib log[M] Lapatinib log[M] Figure 5. Characterisation of the involvement of the NF-kB pathway in lapatinib resistance. (a) Confirmation of inducible NF-kB reporter activity in NF-kB reporter stably infected ERBB2-amplified cell lines. The ERBB2-amplified cell lines HCC1569 and HCC1954 transduced with an NF-kB reporter were plated in 96-well plates and treated with 0 or 10 ng/ml TNFa. Each treatment was done in quadruplicate. NF-kB reporter activity was assessed 48 h after treatment with TNFa using the Dual-glo assay. Error bars represent s.e.m. (b) Verification of the role of NIBP in NF-kB signalling. siRNAs against NIBP and RelA were reverse transfected in triplicate into the NF-kB reporter cell lines HCC1569 and HCC1954 in 96-well plates. NF-kB reporter activity was assessed 48 h after transfection the Dual-glo assay. Error bars represent s.e.m. (c) Characterisation of the effect of lapatinib and NF-kB pathway targeting drugs on NF-kB signalling. The ERBB2-amplified cell lines HCC1569 and HCC1954 transduced with an NF-kB reporter were plated in 96-well plates and treated with 0 or 1 mM of drug. Each treatment was done in quadruplicate. NF-kB reporter activity was assessed 48 h after treatment with TNFa using the Dual-glo assay. Error bars represent s.e.m. (d) Identification of synergistic effects of lapatinib and the NF-kB inhibitor IMD-0354 on NF-kB signalling. The ERBB2-amplified cell lines HCC1569 and HCC1954 transduced with an NF-kB reporter were plated in 96-well plates. Twenty-four and 96 h after plating cells were treated with 0 to 10 mM lapatinib and 0–1 mM IMD-0354. Each treatment was done in triplicate. NF-kB reporter activity was assessed 5 days after plating using the Dual-glo assay. Error bars represent s.d. (e) Identification of synergistic effects of lapatinib and the NF-kB inhibitor IMD-0354 on cell survival. The ERBB2-amplified cell lines HCC1569 and HCC1954 were plated in 96-well plates. 24 and 96 h after plating cells were treated with 0–10 mM lapatinib and 0–1 mM IMD-0354. Each treatment was done in triplicate. NF-kB reporter activity was assessed 5 days after plating using the CellTiter-Glo Luminescent assay. Error bars represent s.d.

contained the individual siRNA species constituting the SMARTpool HTS screen method divided into one well each and was obtained in 2, 96-well plates from ERBB2-amplified cell lines were reverse transfected with siRNA (final Dharmacon. concentration 50 nM) using either RNAiMAX or Dharmafect 3 as per the

Oncogene (2014) 966 – 976 & 2014 Macmillan Publishers Limited RNAi screen for lapatinib resistance D Wetterskog et al 975 manufacturers’ instructions in a 96-well format. Master transfection plates 7 Tse CH, Hwang HC, Goldstein LC, Kandalaft PL, Wiley JC, Kussick SJ et al. Deter- were divided into 6 replica plates whereupon cells were added. Forty-eight mining true HER2 gene status in breast cancers with polysomy by using alter- hours after transfection, replica plates were either treated or not treated native chromosome 17 reference genes: implications for anti-HER2 targeted with lapatinib (0.4 mM for HCC202 and VP229, 1 mM for HCC1569, HCC1954, therapy. J Clin Oncol 2011; 29: 4168–4174. JIMT1, MDAMB361 and MDAMB453). Media in replica plates where 8 Lal P, Salazar PA, Hudis CA, Ladanyi M, Chen B. HER-2 testing in breast cancer replenished every 48 h and cell viability was assessed after 5 days of drug using immunohistochemical analysis and fluorescence in situ hybridization: a exposure using CellTiter-Glo Luminescent Cell Viability Assay (Promega) as single-institution experience of 2279 cases and comparison of dual-color and per manufacturer’s instructions to determine the drug effect of individual single-color scoring. Am J Clin Pathol 2004; 121: 631–636. SMARTpool siRNAs (that is, genes that when silenced only had an effect on 9 Owens MA, Horten BC, Da Silva MM. HER2 amplification ratios by fluorescence viability with lapatinib treatment), we first normalised cell viability data in situ hybridization and correlation with immunohistochemistry in a cohort of from each well to the median of all effects in that plate for no treatment 6556 breast cancer tissues. Clin Breast Cancer 2004; 5: 63–69. (DMSO) and lapatinib treatment, respectively. The drug effect value was 10 Saini KS, Azim Jr. HA, Metzger-Filho O, Loi S, Sotiriou C, de Azambuja E et al. generated by subtracting the centred lapatinib value from the centred Beyond trastuzumab: new treatment options for HER2-positive breast cancer. DMSO value for each SMARTpool. For each SMARTpool the Z score/median Breast 2011; 20(Suppl 3): S20–S27. absolute deviation method was used to identify hits. For the Z-score the 11 Graus-Porta D, Beerli RR, Daly JM, Hynes NE. ErbB-2, the preferred hetero- s.d. of the screen was estimated from the median absolute deviation of all dimerization partner of all ErbB receptors, is a mediator of lateral signaling. EMBO 400 SMARTpools adjusted by a factor of 1.4826 for equivalence with an J 1997; 16: 1647–1655. asymptotically normal distribution. 12 Tzahar E, Waterman H, Chen X, Levkowitz G, Karunagaran D, Lavi S et al. A hierarchical network of interreceptor interactions determines signal transduc- Validation of HTS screen tion by Neu differentiation factor/ and . Mol Cell Biol 1996; 16: 5276–5287. Four distinct siRNA species targeting each gene were used to validate 13 Graus-Porta D, Beerli RR, Hynes NE. Single-chain antibody-mediated intracellular hits from the high-throughput screen. For each siRNA, survival retention of ErbB-2 impairs Neu differentiation factor and epidermal growth fractions compared with siCTRL were calculated. Next, DMSO siRNA factor signaling. Mol Cell Biol 1995; 15: 1182–1191. survival fractions were compared with corresponding lapatinib siRNA 14 Spencer KS, Graus-Porta D, Leng J, Hynes NE, Klemke RL. ErbB2 is necessary for survival fractions. siRNAs with survival fraction differences 40.15, where induction of carcinoma cell invasion by ErbB family receptor tyrosine kinases. survival fraction of siRNA with DMSO was over 0.8 and where survival J Cell Biol 2000; 148: 385–397. fraction of siRNA with lapatinib was under 0.8 were considered validated. 15 Beerli RR, Graus-Porta D, Woods-Cook K, Chen X, Yarden Y, Hynes NE. Neu dif- ferentiation factor activation of ErbB-3 and ErbB-4 is cell specific and displays a Western blotting differential requirement for ErbB-2. Mol Cell Biol 1995; 15: 6496–6505. Cell lysates were separated using 4–12% SDS–PAGE Bis-Tris gel, transferred 16 Geyer CE, Forster J, Lindquist D, Chan S, Romieu CG, Pienkowski T et al. Lapatinib to nitrocellulose membrane, blocked in 5% milk and blotted with plus capecitabine for HER2-positive advanced breast cancer. N Engl J Med 2006; antibodies. Immunoblotting was performed according to the manufac- 355: 2733–2743. turer’s instructions using enhanced chemiluminescence. Antibodies 17 Xia W, Mullin RJ, Keith BR, Liu LH, Ma H, Rusnak DW et al. Anti-tumor targeting the following epitopes were used: NIBP (Abcam, Cambridge, activity of GW572016: a dual tyrosine kinase inhibitor blocks EGF activation of UK), and TP53INP1 (GTX112066S; Source Bioscience, Nottingham, UK). EGFR/erbB2 and downstream Erk1/2 and AKT pathways. Oncogene 2002; 21: b-tubulin (T4026; Sigma, St Louis, MO, USA). 6255–6263. 18 Konecny GE, Pegram MD, Venkatesan N, Finn R, Yang G, Rahmeh M et al. Activity of the dual kinase inhibitor lapatinib (GW572016) against HER-2-over- Statistical analysis expressing and trastuzumab-treated breast cancer cells. Cancer Res 2006; 66: Statistical analysis was performed with SPSS package version 15 (Chicago, 1630–1639. IL, USA) or GraphPad Prism version 5.01 (GraphPad Software). 19 Nagata Y, Lan KH, Zhou X, Tan M, Esteva FJ, Sahin AA et al. PTEN activation contributes to tumor inhibition by trastuzumab, and loss of PTEN predicts trastuzumab resistance in patients. Cancer Cell 2004; 6: 117–127. CONFLICT OF INTEREST 20 Yakes FM, Chinratanalab W, Ritter CA, King W, Seelig S, Arteaga CL. Herceptin- The authors declare no conflict of interest. induced inhibition of phosphatidylinositol-3 kinase and Akt Is required for anti- body-mediated effects on p27, cyclin D1, and antitumor action. Cancer Res 2002; 62: 4132–4141. ACKNOWLEDGEMENTS 21 Clynes RA, Towers TL, Presta LG, Ravetch JV. Inhibitory Fc receptors modulate in vivo cytoxicity against tumor targets. Nat Med 2000; 6: 443–446. We thank Breakthrough Breast Cancer and AACR SU2C for funding this work. We 22 Park S, Jiang Z, Mortenson ED, Deng L, Radkevich-Brown O, Yang X et al. The acknowledge NHS funding to the NIHR RMH Biomedical research Centre. We thank Dr therapeutic effect of anti-HER2/neu antibody depends on both innate and David Watkins from the Royal Marsden hospital for the kind donation of the CR-OE33 adaptive immunity. Cancer Cell 2010; 18: 160–170. cell line. 23 Vogel CL, Cobleigh MA, Tripathy D, Gutheil JC, Harris LN, Fehrenbacher L et al. Efficacy and safety of trastuzumab as a single agent in first-line treatment of HER2-overexpressing metastatic breast cancer. J Clin Oncol 2002; 20: REFERENCES 719–726. 1 Reis-Filho JS, Pusztai L. Gene expression profiling in breast cancer: classification, 24 Spector NL, Blackwell KL. Understanding the mechanisms behind trastuzumab prognostication, and prediction. Lancet 2011; 378: 1812–1823. therapy for human epidermal 2-positive breast cancer. 2 Curtis C, Shah SP, Chin SF, Turashvili G, Rueda OM, Dunning MJ et al. The genomic J Clin Oncol 2009; 27: 5838–5847. and transcriptomic architecture of 2000 breast tumours reveals novel subgroups. 25 Ritter CA, Perez-Torres M, Rinehart C, Guix M, Dugger T, Engelman JA et al. Human Nature 2012; 486: 346–352. breast cancer cells selected for resistance to trastuzumab in vivo overexpress 3 Alvarez RH, Valero V, Hortobagyi GN. Emerging targeted therapies for breast epidermal growth factor receptor and ErbB ligands and remain dependent on the cancer. J Clin Oncol 2010; 28: 3366–3379. ErbB receptor network. Clin Cancer Res 2007; 13: 4909–4919. 4 Slamon DJ, Leyland-Jones B, Shak S, Fuchs H, Paton V, Bajamonde A et al. Use of 26 Junttila TT, Akita RW, Parsons K, Fields C, Lewis Phillips GD, Friedman LS et al. chemotherapy plus a monoclonal antibody against HER2 for metastatic breast Ligand-independent HER2/HER3/PI3K complex is disrupted by trastuzumab cancer that overexpresses HER2. N Engl J Med 2001; 344: 783–792. and is effectively inhibited by the PI3K inhibitor GDC-0941. Cancer Cell 2009; 15: 5 Slamon DJ, Clark GM, Wong SG, Levin WJ, Ullrich A, McGuire WL. Human breast 429–440. cancer: correlation of relapse and survival with amplification of the HER-2/neu 27 Lu Y, Zi X, Zhao Y, Mascarenhas D, Pollak M. Insulin-like growth factor-I receptor oncogene. Science 1987; 235: 177–182. signaling and resistance to trastuzumab (Herceptin). J Natl Cancer Inst 2001; 93: 6 Baehner FL, Achacoso N, Maddala T, Shak S, Quesenberry Jr. CP, Goldstein LC et al. 1852–1857. Human epidermal growth factor receptor 2 assessment in a case-control study: 28 Saal LH, Holm K, Maurer M, Memeo L, Su T, Wang X et al. PIK3CA mutations comparison of fluorescence in situ hybridization and quantitative reverse tran- correlate with hormone receptors, node metastasis, and ERBB2, and are scription polymerase chain reaction performed by central laboratories. J Clin mutually exclusive with PTEN loss in human breast carcinoma. Cancer Res 2005; Oncol 2010; 28: 4300–4306. 65: 2554–2559.

& 2014 Macmillan Publishers Limited Oncogene (2014) 966 – 976 RNAi screen for lapatinib resistance D Wetterskog et al 976 29 Stemke-Hale K, Gonzalez-Angulo AM, Lluch A, Neve RM, Kuo WL, Davies M et al. 42 Shiu KK, Wetterskog D, Mackay A, Natrajan R, Lambros M, Sims D et al. Integrative An integrative genomic and proteomic analysis of PIK3CA, PTEN, and AKT molecular and functional profiling of ERBB2-amplified breast cancers identifies mutations in breast cancer. Cancer Res 2008; 68: 6084–6091. new genetic dependencies. Oncogene 2013; 33: 619–631 30 Berns K, Horlings HM, Hennessy BT, Madiredjo M, Hijmans EM, Beelen K et al. 43 Brough R, Frankum JR, Sims D, Mackay A, Mendes-Pereira AM, Bajrami I et al. A functional genetic approach identifies the PI3K pathway as a major Functional viability profiles of breast cancer. Cancer Discov 2011; 1: 260–273. determinant of trastuzumab resistance in breast cancer. Cancer Cell 2007; 12: 44 Dokmanovic M, Hirsch DS, Shen Y, Wu WJ. Rac1 contributes to trastuzumab 395–402. resistance of breast cancer cells: Rac1 as a potential therapeutic target for the 31 Eichhorn PJ, Gili M, Scaltriti M, Serra V, Guzman M, Nijkamp W et al. Phosphati- treatment of trastuzumab-resistant breast cancer. Mol Cancer Ther 2009; 8: 1557–1569. dylinositol 3-kinase hyperactivation results in lapatinib resistance that is reversed 45 Tanner M, Hollmen M, Junttila TT, Kapanen AI, Tommola S, Soini Y et al. by the mTOR/phosphatidylinositol 3-kinase inhibitor NVP-BEZ235. Cancer Res Amplification of HER-2 in gastric carcinoma: association with Topoisomerase 2008; 68: 9221–9230. IIalpha gene amplification, intestinal type, poor prognosis and sensitivity to 32 Codony-Servat J, Albanell J, Lopez-Talavera JC, Arribas J, Baselga J. Cleavage of trastuzumab. Ann Oncol 2005; 16: 273–278. the HER2 ectodomain is a pervanadate-activable process that is inhibited by the 46 Bang YJ, Van Cutsem E, Feyereislova A, Chung HC, Shen L, Sawaki A et al. Tras- tissue inhibitor of metalloproteases-1 in breast cancer cells. Cancer Res 1999; 59: tuzumab in combination with chemotherapy versus chemotherapy alone for 1196–1201. treatment of HER2-positive advanced gastric or gastro-oesophageal junction 33 Christianson TA, Doherty JK, Lin YJ, Ramsey EE, Holmes R, Keenan EJ et al. NH2- cancer (ToGA): a phase 3, open-label, randomised controlled trial. Lancet 2010; terminally truncated HER-2/neu protein: relationship with shedding of the 376: 687–697. extracellular domain and with prognostic factors in breast cancer. Cancer Res 47 Hu WH, Pendergast JS, Mo XM, Brambilla R, Bracchi-Ricard V, Li F et al. NIBP, a 1998; 58: 5123–5129. novel NIK and IKK(beta)-binding protein that enhances NF-(kappa)B activation. 34 Scaltriti M, Rojo F, Ocana A, Anido J, Guzman M, Cortes J et al. Expression of J Biol Chem 2005; 280: 29233–29241. p95HER2, a truncated form of the HER2 receptor, and response to anti-HER2 48 Campbell KJ, Perkins ND. Reprogramming RelA. Cell Cycle. 2004; 3: 869–872. therapies in breast cancer. J Natl Cancer Inst 2007; 99: 628–638. 49 Harris LN, You F, Schnitt SJ, Witkiewicz A, Lu X, Sgroi D et al. Predictors of resis- 35 Sperinde J, Jin X, Banerjee J, Penuel E, Saha A, Diedrich G et al. Quantitation of tance to preoperative trastuzumab and vinorelbine for HER2-positive early breast p95HER2 in paraffin sections by using a p95-specific antibody and correlation cancer. Clin Cancer Res 2007; 13: 1198–1207. with outcome in a cohort of trastuzumab-treated breast cancer patients. Clin 50 Dave B, Migliaccio I, Gutierrez MC, Wu MF, Chamness GC, Wong H et al. Loss of Cancer Res 2010; 16: 4226–4235. phosphatase and tensin homolog or phosphoinositol-3 kinase activation and 36 Scaltriti M, Chandarlapaty S, Prudkin L, Aura C, Jimenez J, Angelini PD et al. Clinical response to trastuzumab or lapatinib in human epidermal growth factor receptor benefit of lapatinib-based therapy in patients with human epidermal growth 2-overexpressing locally advanced breast cancers. J Clin Oncol 2010; 29: 166–173. factor receptor 2-positive breast tumors coexpressing the truncated p95HER2 51 Scaltriti M, Eichhorn PJ, Cortes J, Prudkin L, Aura C, Jimenez J et al. Cyclin E receptor. Clin Cancer Res 2010; 16: 2688–2695. amplification/overexpression is a mechanism of trastuzumab resistance in 37 Chin K, DeVries S, Fridlyand J, Spellman PT, Roydasgupta R, Kuo WL et al. Genomic HER2 þ breast cancer patients. Proc Natl Acad Sci USA 2011; 108: 3761–3766. and transcriptional aberrations linked to breast cancer pathophysiologies. Cancer 52 Sacher M, Barrowman J, Wang W, Horecka J, Zhang Y, Pypaert M et al. TRAPP I Cell 2006; 10: 529–541. implicated in the specificity of tethering in ER-to-Golgi transport. Mol Cell 2001; 38 Marchio C, Natrajan R, Shiu KK, Lambros MB, Rodriguez-Pinilla SM, Tan DS et al. 7: 433–442. The genomic profile of HER2-amplified breast cancers: the influence of ER status. 53 Galang CK, Garcia-Ramirez J, Solski PA, Westwick JK, Der CJ, Neznanov NN et al. J Pathol 2008; 216: 399–407. Oncogenic Neu/ErbB-2 increases ets, AP-1, and NF-kappaB-dependent gene 39 Staaf J, Jonsson G, Ringner M, Vallon-Christersson J, Grabau D, Arason A et al. expression, and inhibiting ets activation blocks Neu-mediated cellular transfor- High-resolution genomic and expression analyses of copy number alterations in mation. J Biol Chem 1996; 271: 7992–7998. HER2-amplified breast cancer. Breast Cancer Res 2010; 12: R25. 54 Merkhofer EC, Cogswell P, Baldwin AS. Her2 activates NF-kappaB and induces 40 Shiu KK, Natrajan R, Geyer FC, Ashworth A, Reis-Filho JS. DNA amplifications invasion through the canonical pathway involving IKKalpha. Oncogene 2010; 29: in breast cancer: genotypic-phenotypic correlations. Future Oncol 2010; 6: 1238–1248. 967–984. 55 Xia W, Bacus S, Husain I, Liu L, Zhao S, Liu Z et al. Resistance to ErbB2 tyrosine 41 Santarius T, Shipley J, Brewer D, Stratton MR, Cooper CS. A census of amplified kinase inhibitors in breast cancer is mediated by calcium-dependent activation of and overexpressed human cancer genes. Nat Rev Cancer 2010; 10: 59–64. RelA. Mol Cancer Ther 2010; 9: 292–299.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene (2014) 966 – 976 & 2014 Macmillan Publishers Limited