Inhibition of Casein Kinase 1 Alpha Prevents Acquired Drug Resistance to Erlotinib in EGFR-Mutant Non-Small Cell Lung Cancer

Home , Casein kinase 1, CSNK1E

Author Manuscript Published OnlineFirst on October 21, 2015; DOI: 10.1158/0008-5472.CAN-15-1113 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Inhibition of casein kinase 1 alpha prevents acquired drug resistance to erlotinib in EGFR-mutant non-small cell lung cancer

Alexandra B Lantermann*, Dongshu Chen, Kaitlin McCutcheon, Greg Hoffman, Elizabeth Frias, David Ruddy, Daniel Rakiec, Joshua Korn, Gregory McAllister, Frank Stegmeier, Matthew J Meyer and Sreenath V Sharma*

Oncology Drug Discovery, Novartis Institutes for Biomedical Research; 250 Massachusetts Avenue, Cambridge MA 02139

Running title: CK1 inhibition prevents erlotinib resistance

Keywords: RNA interference screen, acquired resistance, EGFR tyrosine kinase inhibitors, casein kinase 1 alpha (CSNK1A1, CK1), EGFR-mutant non-small cell lung cancer (NSCLC)

Financial support: All studies were funded by the Novartis Institutes for BioMedical Research. A. Lantermann is a recipient of the postdoctoral fellowship from Novartis Institutes for BioMedical Research.

*Corresponding authors: Sreenath Sharma: Email: [email protected], Telephone: 617-971-6616 Alexandra Lantermann: Email: [email protected], Telephone: 617-871-5656, Fax: 617-726-7808

Disclosure of potential conflict of interests: All the listed authors are employees of Novartis Institutes for BioMedical Research.

Author contributions: A.B.L., D.C., F.S., M.J.M. and S.V.S. designed research. A.B.L., D.C., K.M., G.H., E.F., D.R. and D.R. performed research. A.B.L., D.C., K.M., D.R., D.R., J.K., G.M., F.S., M.J.M. and S.V.S. analyzed and interpreted data. A.B.L., F.S. and S.V.S. wrote and reviewed the manuscript.

Notes about the manuscript: Word count: 4963. Total number of figures: 5.

Abstract

Patients with lung tumors harboring activating mutations in the EGF receptor (EGFR) show good initial treatment responses to the EGFR tyrosine kinase inhibitors (TKIs) erlotinib or gefitinib. However, acquired resistance invariably develops. Applying a focused shRNA screening approach to identify genes whose knockdown can prevent and/or overcome acquired resistance to erlotinib in several EGFR- mutant non-small cell lung cancer (NSCLC) cell lines, we identified casein kinase 1 alpha (CSNK1A1,

CK1). We found that CK1suppressioninhibits the NF-B pro-survival signaling pathway. Furthermore, down-regulation of NF-B signaling by approaches independent of CK1 knockdown can also attenuate acquired erlotinib resistance, supporting a role for activated NF-B signaling in conferring acquired drug resistance. Importantly, CK1 suppression prevented erlotinib resistance in an HCC827 xenograft model in vivo. Our findings suggest that patients with EGFR-mutant NSCLC might benefit from a combination of

EGFR TKIs and CK1 inhibition to prevent acquired drug resistance and prolong disease-free survival.

Precis

In this study we show that in EGFR-mutant NSCLC, acquired resistance to the EGFR TKI erlotinib can be prevented by co-inhibition of the serine/threonine kinase CK1. The ability to prevent acquired resistance raises the possibility that inhibition of CK1may be able to block the wide variety of mechanisms that engender resistance to EGFR inhibitors in the clinic, before they can occur, rather than dealing with them individually after they develop.

Introduction

Lung cancer is the leading cause of cancer-related deaths worldwide. Approximately 85% of lung carcinomas are non-small cell lung carcinoma (NSCLC), with adenocarcinoma representing the most frequently occurring histologic subtype (1). 10-40% of patients with adenocarcinoma harbor activating mutations in the receptor tyrosine kinase (RTK) gene EGFR, and frequently show good initial responses to treatment with the EGFR tyrosine kinase inhibitors (TKIs) erlotinib and gefitinib (1, 2). However, most patients eventually acquire resistance, resulting in a modest overall survival benefit compared to standard chemotherapeutics (1).

Modeling acquired resistance in cancer cell line models and examining relapsed patient tumor samples has identified both genetic and epigenetic mechanisms of resistance to EGFR targeted TKIs (3, 4). Most frequently, acquired resistance is achieved through reactivation of signaling pathways that were originally suppressed by RTK inhibition (3, 4). This can result from secondary mutations in the drug target, such as the T790M mutation in EGFR, observed in 50% of EGFR-mutant lung cancers with acquired resistance (5-7). Alternatively, key downstream signaling pathways can be reactivated through mechanisms independent of the original target, for example through mutational activation of components of the PI3K/AKT and MEK/ERK pathways, or through activation of bypass RTKs, such as MET

(8-10), ErbB2 (11), IGF1R (12-14), FGFR1 (15) or AXL (16, 17). Furthermore, activation of alternative pathways, such as upregulation of the pro-survival NF-B signaling pathway, as well as a switch from epithelial to mesenchymal cell morphology (EMT) have been connected to both intrinsic and acquired resistance to EGFR TKIs (16, 18, 19).

Cancer cell lines used to model acquired resistance show initial sensitivity to the drug and undergo apoptosis, similar to the initial treatment response observed in cancer patients. However, as in patients, the treatment response in cell lines is rarely complete, resulting in a subpopulation of initially quiescent

drug-tolerant persisters (DTPs), that eventually start proliferating in the presence of the drug and give rise to a drug-resistant cell population (3, 20, 21). In such initially drug-sensitive cancer cell lines, genetic loss-of-function screens have been successfully applied to identify genes whose suppression can confer resistance to the drug (22-25). Loss-of-function screens to identify genes whose suppression enhance drug treatment and thereby potentially prevent drug tolerance or drug resistance are more challenging in these cell line models. This is due to the fact that only a small subpopulation of DTPs survives drug treatment, thus making it more difficult to achieve a high representation of individual shRNAs and a robust signal-to-noise ratio in the screen.

Using loss-of-function genetic screens, we identified that suppression of casein kinase 1 alpha

(CSNK1A1, CK1) can prevent and/or overcome erlotinib resistance in several erlotinib-sensitive, EGFR- mutant, NSCLC cell lines. Suppression of CK1 inhibits the NF-B signaling pathway. Moreover, direct down-regulation of NF-B signaling, by approaches independent of CK1 suppression, also attenuates the development of acquired resistance to erlotinib, supporting a role for activated NF-B signaling in conferring erlotinib resistance. Importantly, suppression of CK1 can also prevent resistance to erlotinib in an HCC827 xenograft model in vivo, making CK1 a potential novel target to attenuate resistance to erlotinib in EGFR-mutant NSCLC.

Materials and Methods

Cell lines and culture conditions

HCC827, HCC4006, HCC2935, H3255 and HEK293T cells were obtained and cultured as recommended by the American Type Culture Collection (ATCC). PC9 cells were kindly provided by Prof. Nishio (Kinki

University of Medicine, Osaka, Japan) and cultured in RPMI1640 supplemented with 10% FBS (Hyclone).

All cell lines were authenticated by single-nucleotide polymorphism fingerprinting and used for experiments within 20 passages.

Generation of lentiviruses

Viral constructs were packaged in HEK293T cells plated at 80-90% confluency in a 10 cm-culture dish by co-transfection with 2.4 g of shRNA-encoding plasmid, 0.6 g of VSV-G envelope plasmid and 2.4 g of

D8.9 (gag/pol/rev) packaging plasmid, using TransIT-293 reagent (Mirus) for transfection. Growth media was exchanged the following day and lentivirus-containing supernatant was harvested 48 hours later.

Target cell lines were transduced at MOI < 0.5.

RNAi shRNA sequences were designed to include EcoRI and AgeI restriction sites to allow subsequent cloning into the pLKO-Tet-On inducible vector system in which the H1-TetO promoter had been replaced with

U6-TetO. The CTCGAG stem loop was used for shNTC. For shCK1#1, shCK1#2, shCK1#3 and shCK1#4, the Cellecta stem loop sequence GTTAATATTCATAGC was used for the top oligonucleotide and GCTATGAATATTAAC for the bottom oligonucleotide. The sequences of the oligonucleotides used are as follows from 5’ to 3’:

shCK1#1-top:

CCGGCATCTATTTGGCGATTAACATGTTAATATTCATAGCATGTTGATCGCCAAATAGATGTTTTT; shCK1#1- bottom: AATTAAAAACATCTATTTGGCGATCAACATGCTATGAATATTAACATGTTAATCGCCAAATAGATG; shCK1#2-top:

CCGGCTGCCTGTTTAATTGTGTTAGGTTAATATTCATAGCCTAGCACAATTAAGCAGGCAGTTTTT; shCK1#2- bottom: AATTAAAAACTGCCTGCTTAATTGTGCTAGGCTATGAATATTAACCTAACACAATTAAACAGGCAG; shCK1#3-top:

CCGGCTGTACGAGAGTAAGCTTTATGTTAATATTCATAGCATAGAGCTTGCTCTCGTACAGTTTTT; shCK1#3- bottom: AATTAAAAACTGTACGAGAGCAAGCTCTATGCTATGAATATTAACATAAAGCTTACTCTCGTACAG; shCK1#4-top:

CCGGGACTTTGCCTGTTTAATTGTGGTTAATATTCATAGCCACAATTAAGCAGGCAGAGTCTTTTT; shCK1#4- bottom: AATTAAAAAGACTCTGCCTGCTTAATTGTGGCTATGAATATTAACCACAATTAAACAGGCAAAGTC; shNTC-top: CCGGGGATAATGGTGATTGAGATGGCTCGAGCCACTCAATCACCATTATCCTTTTT; shNTC-bottom:

AATTAAAAAGGATAATGGTGATTGAGATGGCTCGAGCCACTCAATCACCATTATCC. shCK1#3 caused a stronger antiproliferative effect in HCC4006 cells than in other cell lines, potentially due to cell line specific off-target effect(s). Therefore, shCK1#4 was used in HCC4006 cells. For viral transduction, 200,000 cells were plated per 6-well dish, and virus-containing medium with 8 g/ml polybrene (Millipore) was added the following day. 24 h later, fresh medium containing 2 g/ml puromycin was added, and cells were cultured for one week to eliminate non-transduced cells. shRNAs were induced with 100 ng/ml doxycycline (Clontech).

Ectopic expression of mCK1- and IB-superrepressor alleles

The kinase-dead mCK1(D136N) was generated from the murine CK1-cDNA construct (Life

Technologies, clone ID: IOM18458), using the Q5-mutagenesis kit (New England Biolabs) with primers 5’-

TATACACAGAAACATTAAACCAG-3’ and 5’-AAATTCTTTGTATGCACATATTC-3’. Additional silent mutations were added to further increase the resistance of murine CK1 to CK1-shRNA #1 and #3 using the primers 5’-AATATAACCAATGGCGAGGAAGTG-3’ and 5’-TATGGCTAGATAAATGTCCCCGAAAGAG-3’ as well as 5’-TAAACTGTACAAGATTCTTCAAGGTGGGGTTG-3’ and 5’-CTTTCGTACAGCAACTGGGGATGCCTGGC-3’, respectively. The stop codon was deleted with primers 5’-TACCCAGCTTTCTTGTAC-3’ and 5’-

GAAACCTGTGGGGGTTTG-3’, and constructs were cloned into pLenti6.3-V5-CMV plasmid (Life

Technologies) to generate V5-tagged fusion constructs. Plasmids were packaged into lentiviral particles, and after infection cells were selected with 25 g/ml blasticidin for one week to eliminate non- transduced cells. IB-superrepressor was cloned into a tetracycline-inducible pLKO-TREX plasmid.

Lentivirus-transduced cells were selected with 2 g/ml puromycin. Expression of the IB- superrepressor was induced with 100 ng/ml doxycycline (Clontech). shRNA screen

A custom shRNA library containing 6,500 shRNAs against ~350 potentially cancer-relevant genes (~17 shRNAs per gene) described previously (26) was screened. Cells were transduced at an MOI of 0.5.

Selection with 2 g/ml puromycin was started 24 h after transduction and continued for 72 h. Cells were trypsinized and the percentage of positively transduced cells was >90% for all cell lines screened as tested by RFP expression. 15 million cells, representing the baseline of shRNA representation, were harvested. The remaining cells were plated for DMSO or 2 M erlotinib treatment, respectively. DMSO- treated cells were split every 3 days and at each time point 20 million cells were re-plated. Erlotinib- treated cells were retreated with fresh media and compound every 3 days. After 10 and 24 days of treatment, respectively, at least 15 million cells were harvested per condition, ensuring a representation of >2,000 cells per shRNA. Purification of genomic DNA, PCR for library production, and next generation

sequencing were performed as described previously (26). Screens for all cell lines and conditions were performed in duplicates.

shRNA screen data analysis

Sequencing counts from each sample were normalized to 15 million reads. The fold change of shRNA representation in erlotinib-treated versus DMSO-treated condition was calculated. In addition, the fold change of shRNA representation in DMSO-treated versus baseline condition was calculated to estimate the effects of shRNAs on cell proliferation. A robust z-score was calculated using the median and MAD for the calculated fold changes across the entire shRNA library. For gene based hit calling, two statistical measures were used, redundant siRNA activity (RSA) (27) and Q1-z-score (26, 28). The RSA-score is a statistical score (p-value) representing the probability of a gene hit based on the collective activities of multiple shRNAs per gene. It is a measure of how significantly the rank order of shRNAs against a given gene differs from the population of other shRNAs in the library. In this approach genes with multiple moderately active shRNAs score higher than genes with few but highly active shRNAs (27). The Q1-z- score represents the z-score of the shRNA at the first quartile (Q1), i.e. the z-score of the 4th best (in a library that contains an average of 16-17 shRNAs per gene) performing shRNA per gene. By using both

RSA-score and Q1-z-score for hit calling, the most robust hits with both consistent and high shRNA activity are identified.

Cell viability and proliferation analysis

For short-term cell viability assays, 2,000-4,000 cells were seeded in triplicate in 96-wells one day before compound addition, and incubated for 5 days with various concentrations of compounds. Cell viability was determined by CellTiter-Glo Luminescence Assay (Promega). Luminescence signal was recorded on

an EnVision plate reader (PerkinElmer) and the inhibition of viability relative to DMSO-treated cells was calculated. For long-term colony formation assays, 100,000-200,000 cells were plated per 6-well the day before treatment start. Cells were retreated with fresh media with or without compound every 3 days until the appropriate confluency as estimated by the control condition was reached. Each condition was done in triplicate. Plates were stained with 0.2% Crystal violet/10% formaline, and crystal violet signal was quantified on Odyssey CLx (LI-COR) at 700 nm. To measure proliferation over time, cells were plated at 10,000 cells per well in a 24-well plate in triplicate. 24 h later, the indicated treatment was started and confluency measurements were taken every 12 hours using an Incucyte Kinetic Imaging System

(Essen BioScience).

Luciferase assay

Cell lines stably expressing an NF-B- luciferase reporter plasmid (pLenti6-NF-B-Luc2P) or a WNT- responsive Super-Topflash (STF) luciferase reporter plasmid (pLenti6-STF) were generated by lentiviral infection. 4,000 cells per 96-well were plated and the luciferase signal was measured after 4 days using

Luciferase Assay System (Promega) and normalized to cell viability assessed by CellTiter-Glo

Luminescence Assay (Promega). For transient NF-B-luciferase reporter assays, cells were co- transfected with firefly luciferase construct driven by NF-B binding sequences (pTranslucent-NFBI-

Luc) and Renilla luciferase plasmid (pRL-SV40-Renilla) at a ratio of 25 to 1. Luciferase signal was measured using Dual Luciferase Reporter Assay System (Promega) 48 h after transfection, and firefly luciferase signal was normalized to Renilla luciferase signal. All conditions were done in triplicate.

Caspase 3/7-apoptosis assay

10,000 cells were plated per 96-well in triplicate and treated with 2 M erlotinib or DMSO the following day. Caspase activity was measured 24 h post-treatment using the Caspase Glo 3/7 assay (Promega), and normalized to cell viability assessed by CellTiter-Glo Luminescence Assay (Promega).

Quantitative real-time PCR

Total RNA was extracted using RNeasy Plus Mini Kit (Qiagen). cDNA was generated using iScript cDNA

Synthesis Kit (BioRad). The qRT-PCR reactions were performed using TaqMan Universal PCR Master Mix

(Life Technologies) and were run on 7900HT Fast Real-Time PCR System (Applied Biosystems). TaqMan probes for Serpine1, TNF and -actin were purchased from Life Technologies. All experiments were performed in triplicate and normalized to -actin transcript levels using comparative CT method.

Microarray analysis

HCC827 and PC9 cells were treated for 8 days or 30 days with DMSO or 2 M erlotinib. Total RNA was isolated using the RNeasy Mini Kit (Qiagen). Gene expression profiling was performed using Affymetrix

U133Plus2 Arrays and data were analyzed as in (29). Differential analysis was performed by computing fold changes of erlotinib-treatment to DMSO. The data have been deposited at the National Center for

Biotechnology information (NCBi Gene Expression Omnibus with the accession number GSE67051).

GeneGo pathway enrichment analysis was performed to identify differentially regulated pathways between treated and untreated condition.

Immunoblotting

Cell pellets were lysed in RIPA buffer and immunoblot analysis was performed using standard methods.

The following antibodies were used: CK1 (Santa Cruz Biotechnology, sc-6477), V5 (Invitrogen, 46-0705),

GAPDH (Millipore, MAB374), pERK (T202, Y204) (CST, 4377S), ERK (CST, 4695S), pAKT (S473) (CST,

4058L), AKT (CST, 4685S), pEGFR (Y1086) (Abcam, ab32086), EGFR (CST, 4267S), cleaved PARP (Asp214)

(CST, 5625S).

Xenograft studies

Mice were handled in accordance with the Novartis Institutes for BioMedical Research (NIBR) Animal

Care and Use Committee protocols and regulations. 10 million cells in 50% matrigel (BD Biosciences) were implanted subcutaneously into female athymic nude mice (20-25 grams; 7-8 weeks old). The cells were free of mycoplasma and viral contamination (IMPACT VIII PCR Profile, IDEXX BioResearch).

Treatment with compound (formulation 40% Captisol, 0.1M tartaric acid) and doxycycline- supplemented food (PharmaServ (diet #5SL5: PharmaServ 5053 with 400 PPM doxycycline)) started 10-

14 days after implant when average tumor volume reached approximately 350 mm3. Animals were administered once daily with vehicle or erlotinib (LC Laboratories; 1, 2.5, 5 or 10 mg/kg, oral) for the duration of the study. Tumor volume was measured using calipers and calculated as (length x width x width)/2. At the end of the study and one hour after the last dose, tumor tissues were excised and snap frozen in liquid nitrogen for biomarker analysis.

Statistics

Except noted otherwise all average results are presented as mean ± STDV. P-values were calculated using a two-tailed t-test. * P<0.05 by t-test; ** P< 0.01 by t-test; *** P<0.001; n.s. = not significant by t- test.

Results shRNA screening identifies casein kinase 1 alpha as a synergizer with erlotinib in EGFR-mutant NSCLC cells

To identify novel genes involved in acquired resistance to the EGFR TKI erlotinib, we performed loss-of- function genetic screens in several EGFR-mutant non-small cell lung cancer (NSCLC) cell lines in the presence or absence of the drug. We screened three EGFR-mutant NSCLC cell lines, HCC827, HCC4006 and PC9 cells, all of which are highly sensitive to erlotinib (Fig. 1A left, Supplementary Fig. S1A and S1B), but acquire resistance upon long-term treatment with the drug. We purposely chose highly drug sensitive cell lines in which treatment with the clinically relevant dose of 2 M erlotinib allows only the survival of a small subpopulation of drug tolerant persisters (DTPs), since an enhancer screen in such cell models has the potential to identify genes whose knockdown can prevent the occurrence of DTPs and thereby prevent drug resistance in the long-term. In addition, in order to identify genes whose knockdown can overcome drug resistance, we screened an erlotinib-resistant PC9 cell line generated by continuous treatment and then maintained with a clinically relevant dose of erlotinib (2 M) (Fig. 1A right, Supplementary Fig. S1A and S1B). Importantly, as described previously, erlotinib-resistant PC9 cells did not show any additional resistance conferring mutations in the EGFR gene or in other genes when analyzed by RNA sequencing (data not shown, (20)). We then screened a focused, deep coverage

(17 shRNAs per gene) shRNA library of 6,500 shRNAs against ~350 potentially cancer-relevant genes

(Supplementary Table S1) (26). The deep coverage shRNA library of 17 shRNAs per gene enables high confidence hit calling at the gene level rather than analysis of individual shRNAs in the data set. HCC827,

HCC4006, PC9 cells and erlotinib-resistant PC9 cells were infected with shRNA library containing lentiviral particles, and library-infected cells were cultured for 10 or 24 days in the absence or presence of 2 M erlotinib (Fig. 1A), corresponding to approximately 50-fold of the IC50 in the cell lines used

(Supplementary Fig. S1A and S1B). To avoid the follow-up of cell line specific as opposed to more generalizable phenotypes, we focused on shRNA drop-out hits that were shared across several NSCLC cell lines. We prioritized those hits that came up in the 10 d treatment condition since long-term culture can increase the noise of shRNA data.

Using these criteria for hit selection, casein kinase 1 alpha (CSNK1A1, CK1) was identified as a novel drop-out hit that decreased cell survival in the presence of erlotinib in parental HCC827, HCC4006 and

PC9 cells, as well as in erlotinib-resistant PC9 cells (Fig. 1B, Supplementary Fig. S3). In the presence of

DMSO, CK1 shRNAs dropped out only in PC9 cells (Supplementary Fig. S2), explaining the lower scores of CK1 in PC9 in the comparison of erlotinib versus DMSO treatment (Fig. 1B). Other isoforms of the

CK1 family represented in the shRNA library, namely CSNK1D (CK1) and CSNK1E (CK1), did not exhibit significant drop-out in the presence of erlotinib (Fig. 1B, Supplementary Fig. S3), suggesting a specific role of the alpha isoform of CK1 (CK1 in resistance to erlotinib.

Suppression of CK1 increases the maximum dose response (Amax) and enhances the apoptotic response to erlotinib

To further confirm the role of CK1 in cell survival in the presence of erlotinib, we generated stable cell lines expressing three independent doxycycline-inducible shRNAs against CK1 or a non-targeting control (NTC-) shRNA. Cells were initially treated with or without doxycycline for 72 hours to induce knockdown of CK1. Subsequently, cells were plated in the presence or absence of doxycycline and treated with different concentrations of erlotinib for 5 days. Knockdown of CK1 increased the maximum dose response to erlotinib (Amax) in HCC827 and HCC4006 cells and sensitized erlotinib- resistant PC9 cells to the drug in this short-term assay (Fig. 2A). No separation of the dose response

curves was detectable upon knockdown of CK1 in the presence of doxorubicin or cisplatin, suggesting that CK1-knockdown specifically increases the response to erlotinib (Supplementary Fig. S4).

Since knockdown of CK1 specifically increased the Amax rather than shift the IC50 in combination with erlotinib, the effect of CK1-knockdown on the apoptotic response to erlotinib was examined. Indeed, knockdown of CK1 enhanced apoptosis in the presence of erlotinib in HCC827, HCC4006 and erlotinib- resistant PC9 cells, as shown by increased PARP cleavage (Fig. 2B), and increased caspase 3/7-activity

(Fig. 2C). Collectively, these findings indicate that knockdown of CK1 increases the maximum dose response, at least in part, by increasing the extent of the apoptotic response to erlotinib in EGFR-mutant

NSCLC cells.

Knockdown of CK1 attenuates resistance to erlotinib in EGFR-mutant NSCLC cells

To determine the effects of CK1-knockdown on the development of resistance to erlotinib in long-term colony formation assays, cell lines generated to express doxycycline-inducible CK1- or NTC-shRNAs were cultured in erlotinib in the presence or absence of doxycycline until colonies formed (generally 21-

35 days). Strikingly, in this long-term assay, CK1-knockdown attenuated the development of resistance to erlotinib in HCC827, HCC4006 and PC9 cells. The effects of CK1-knockdown on cell proliferation in the absence of erlotinib were less pronounced, with no significant effect in HCC827 cells, but moderate and significant effects in HCC4006 and PC9 cells (Fig. 3A, Supplementary Fig. S5A). Interestingly, knockdown of CK1 was also effective when induced in DTPs, i.e. after initial erlotinib treatment for 8 days, showing that CK1 -knockdown can also attenuate the outgrowth of DTPs to resistant clones

(Supplementary Fig. S6A right), in addition to preventing the emergence of DTPs (Fig. 3A, Supplementary

Fig. S6A left). Furthermore, knockdown of CK1 inhibited the proliferation of resistant PC9 cells, and this

effect was more pronounced in the presence than in the absence of erlotinib (Supplementary Fig. S6B).

In addition, the CK1 inhibitor D4476 (30) in combination with erlotinib attenuated resistance to erlotinib in HCC827, HCC4006 and PC9 cells albeit with modest effects on cell proliferation when D4476 was used as a single agent (Supplementary Fig. S7). We extended the finding that CK1-knockdown attenuates acquired resistance to erlotinib to two additional EGFR-mutant and erlotinib-sensitive NSCLC cell lines, HCC2935 and H3255, that were not part of the shRNA screen (Fig. 3B, Supplementary Fig. S1A and S1B, Supplementary Fig. S5B). Taken together, these data suggest that inhibition of CK1 can attenuate resistance to erlotinib across a panel of EGFR-mutant NSCLC cell lines and that the synergy between CK1 inhibition and erlotinib represents a general mechanism.

We next tested whether the effect of CK1-knockdown on preventing resistance to erlotinib could be reversed by ectopic expression of CK1. To this end, we generated cell lines that stably express V5- tagged shRNA-resistant murine wildtype or kinase-dead mCK1 cDNA (31) and doxycycline-inducible

CK1- or NTC-shRNAs. As shown in Fig. 3D, shRNA-resistant mCK1-V5 was refractory to knockdown by the human specific CK1-shRNAs, and wildtype and kinase-dead mCK1-V5 were expressed at similar levels. Importantly, expression of wildtype mCK1-V5, but not kinase-dead mCK1(D136N)-V5, partially rescued the attenuation of resistance to erlotinib in the presence of human CK1-knockdown (Fig. 3C).

These results support the importance of CK1 and its kinase-activity in acquired resistance to erlotinib in EGFR mutant NSCLC.

Knockdown of CK1 does not affect the MEK/ERK or PI3K/AKT survival pathways, neither does it inhibit Wnt signaling

Resistance to erlotinib often results from reactivation of key downstream signaling pathways, such as the PI3K/AKT and MEK/ERK pathways. This is particularly true in the case of erlotinib-resistant PC9 cells which downregulate pEGFR but maintain robust activation of pERK and pAKT even in the presence of erlotinib (Supplementary Fig. S8, bottom right). To rule out a direct effect of CK1 on ERK/AKT signaling we investigated pERK and pAKT levels upon CK1-knockdown in the erlotinib resistant PC9 cells. pERK and pAKT levels were not significantly affected by CK1-knockdown either in the presence or absence of erlotinib, suggesting an ERK/AKT-pathway-independent mechanism for the synergy between CK1- knockdown and erlotinib treatment (Supplementary Fig. S8, bottom right). Consistent with these findings, CK1-knockdown also did not affect pERK, pAKT or pEGFR levels in HCC827, HCC4006 and PC9 cells under acute treatment in the presence or absence of erlotinib (Supplementary Fig. S8).

Previous studies have shown that CK1 by virtue of being part of the beta-catenin destruction complex, plays a role in WNT-pathway regulation, with CK1-knockdown activating WNT signaling in the gut (32-

34). However, CK1-knockdown did not activate WNT-signaling in HCC827 and PC9 cells as measured by a WNT-responsive Super-Topflash (STF) luciferase reporter assay (Supplementary Fig. S9), thereby excluding WNT-pathway deregulation as the mechanism for preventing resistance to erlotinib in EGFR- mutant NSCLC cells. Consistently, GeneGo pathway enrichment analysis of erlotinib- versus DMSO- treated HCC827 and PC9 cells did not suggest altered WNT-signaling upon erlotinib treatment

(Supplementary Fig. S10A and S10B).

Knockdown of CK1 decreases NF-B-signaling and inhibition of NF-B-signaling attenuates resistance to erlotinib

CK1 has been implicated in several cellular signaling pathways (35). To better understand the effect of erlotinib on EGFR-mutant NSCLC cells, we generated microarray data from NSCLC cells treated for 8 days

(DTPs) or 30 days (resistant cells) with 2 M erlotinib. GeneGO pathway enrichment analysis of erlotinib- versus DMSO-treated cells revealed a prominent upregulation of NF-B-signaling in DTPs and in erlotinib-resistant cells (Supplementary Fig. S10A and S10B). Based on these findings we hypothesized that upregulation of NF-B-signaling contributes to tolerance/resistance to erlotinib, and that knockdown of CK1 attenuates resistance to erlotinib through downregulation of NF-B-signaling. To test this hypothesis, we generated HCC827 and PC9 cell lines stably expressing both an NF-B-luciferase reporter and doxycycline-inducible CK1- or NTC-shRNAs. Knockdown of CK1 significantly decreased the NF-B-luciferase reporter signal (Fig. 4A). Consistent with these findings, CK1-knockdown reduced the expression of the NF-B target genes TNFand Serpine1 in the presence of erlotinib

(Supplementary Fig. S10C).

Based on these results, we next investigated whether inhibition of NF-B-signaling by independent approaches would prevent resistance to erlotinib. To this end, we generated cell lines expressing a doxycycline-inducible dominant-negative version of IB, the IB-superrepressor. Two point mutations at Serine 32 and Serine 36 to Alanine in the IB-superrepressor prohibit phosphorylation and thus degradation of IB, thereby inhibiting NF-B-signaling (36). Indeed, induction of the IB- superrepressor by doxycycline decreased NF-B-signaling, as measured by NF-B-luciferase reporter assay (Fig. 4B). Importantly, expression of the IB-superrepressor attenuated resistance to erlotinib in

HCC827 and PC9 cells, whereas it only modestly affected cell proliferation in the absence of erlotinib

(Fig. 4B, Supplementary Fig. S11A and S11B). These findings recapitulated the observations made for

CK1-knockdown, which also downregulated NF-B-signaling and attenuated resistance to erlotinib.

Underscoring the importance of the NF-B-signaling pathway in acquired resistance, combination treatment with erlotinib and AFN700, an inhibitor of IB-kinase (IKK) (37, 38), the primary kinase promoting IB-phosphorylation and NF-B-activation, attenuated resistance to erlotinib in HCC827,

HCC4006 and PC9 cells at concentrations where AFN700 alone did not affect cell proliferation (Fig. 4C,

Supplementary Fig. S11C and S11D). Furthermore, treatment with AFN700 inhibited proliferation of resistant PC9 cells, and this effect was more pronounced in the presence than in the absence of erlotinib

(Supplementary Fig. S11E), suggesting that resistant PC9 cells depend on activated NF-B-pathway for survival and that erlotinib modulates this dependency.

In summary, our data show that the development of drug tolerance to erlotinib is associated with activated NF-B-signaling in EGFR-mutant NSCLC cell lines, and that down-regulation of NF-B-signaling by CK1-knockdown, as well as by other approaches can attenuate resistance to erlotinib.

Knockdown of CK1 attenuates resistance to erlotinib in an HCC827 xenograft model

Since knockdown of CK1 attenuated acquired resistance in vitro, we next investigated whether CK1- knockdown could affect resistance to erlotinib in vivo. In an initial experiment we determined the minimum dose of erlotinib that was sufficient to achieve maximum HCC827 tumor regression in vivo to be 5 mg/kg (Fig. 5A). Treatment of mice with 5 mg/kg erlotinib reduced pEGFR levels, demonstrating the effectiveness of 5 mg/kg erlotinib on target modulation in vivo (Fig. 5B). In addition, two CK1-shRNAs, but not the NTC-shRNA, induced robust knockdown of CK1 in vivo (Fig. 5B). Mice harboring HCC827- derived tumors expressing doxycycline-inducible CK1- or NTC-shRNAs were kept on a diet supplemented with or without doxycycline and dosed daily with vehicle or 5 mg/kg erlotinib. As shown in Fig. 5C, CK1-knockdown alone did not affect tumor growth in the vehicle treatment group.

Treatment of mice with 5 mg/kg erlotinib strongly inhibited tumor growth (relative to vehicle treatment), with resistance to erlotinib starting to emerge after approximately 70 days of erlotinib treatment. Importantly, expression of two independent CK1-shRNAs but not of the NTC-shRNA prevented the emergence of resistance to erlotinib in the HCC827 xenograft model (Fig. 5C).

These in vivo data further support a role for CK1 as potential target to prevent resistance to erlotinib in

EGFR-mutant NSCLC.

Discussion

We used a loss of function screening approach to investigate mechanisms of acquired resistance to erlotinib across a panel of EGFR-mutant NSCLC cells. CK1 was identified as a novel hit whose knockdown could both prevent and overcome acquired resistance to erlotinib. Suppression of CK1 downregulated the pro-survival NF-B-signaling pathway and NF-B-signaling was upregulated in erlotinib-treated drug-tolerant NSCLC cells and in resistant PC9 cells. In addition, inhibition of NF-B- signaling by approaches independent of CK1 suppression similarly attenuated resistance to erlotinib, supporting a causal link between increased NF-B-signaling and erlotinib resistance. Importantly, suppression of CK1 in vivo also prevented resistance to erlotinib in an HCC827 xenograft model.

Together, our data suggest CK1 inhibition as a novel combination treatment strategy with EGFR TKIs to attenuate acquired resistance in patients with lung cancer harboring activating EGFR mutations.

Functional shRNA screens have been successfully used in the past to identify novel players in drug resistance. Several factors determine the quality of shRNA screen data and the ability for high confidence hit calling: First, the use of deep coverage shRNA libraries, in our case 17 shRNAs per gene, reduces the likelihood of false-positive hit calling (39). Second, the recovery of sufficiently high cell

numbers at the end of the screen is necessary to ensure a high representation of each shRNA in the library and to reduce noise in the screen (39). And third, screening a panel of cell lines of the same lineage helps to avoid the follow-up of cell line specific as opposed to more generalizable phenotypes. In drug-sensitive cell lines, drug treatment often kills the bulk population (20), hampering the recovery of sufficient shRNAs per gene. Therefore, most shRNA screens to date have been set up to identify genes whose knockdown confers acquired resistance, for which gene hits can also be identified at lower shRNA representation (22-25). Alternatively, shRNA screens with the aim to identify drop-out hits have mostly been applied in cell line models of intrinsic resistance that lack drug sensitivity (18, 40, 41). In contrast, we successfully performed drop-out shRNA screens in a panel of drug-sensitive EGFR-mutant NSCLC cells of acquired resistance in the presence of clinically relevant erlotinib concentrations. One challenge of this approach is that it requires handling of initially large cell numbers to account for the treatment- induced cell loss throughout the screen, and is therefore more applicable for screening of focused as opposed to genome-wide shRNA libraries. Nonetheless, this screening approach is highly valuable since it can inform new combination treatment strategies. In our study it identified CK1 as a potential novel target to prevent resistance to erlotinib in EGFR-mutant NSCLC.

CK1 has previously been shown to play a role in NF-B-signaling pathway in DLBCL (diffuse large B-cell lymphoma) (31). Suppression of CK1 was selectively lethal for activated B-cell-like DLBCL cells that rely on constitutive NF-B-activation for proliferation and survival, but not for NF-B-pathway independent germinal center B-cell-like DLBCL cells (31, 42). We demonstrate that suppression of CK1 can downregulate NF-B-pathway activity in EGFR-mutant NSCLC cells and attenuate resistance to erlotinib.

Furthermore, we show that DTPs and resistant PC9 cells show NF-B-activation and that downregulation of NF-B-signaling, through expression of the IB-superrepressor or through treatment with an IKK- inhibitor, can attenuate resistance to erlotinib. Together, these findings suggest a causal role for activated NF-B-signaling in resistance to erlotinib and provide evidence that inhibition of NF-B-

signaling can attenuate resistance to erlotinib in EGFR-mutant NSCLC cells. Accordingly, it has been described previously that the extent of NF-B-activity may determine the response to EGFR TKI treatment in EGFR-mutant lung cancer patients, with low IB expression, an indicator of high NF-B- activation, being predictive of worse progression-free survival and decreased overall survival of patients with EGFR-mutant lung cancer treated with erlotinib (18). Furthermore, adaptive NF-B-activation has recently been shown to enable survival of NSCLC cells upon initial EGFR-targeted therapy, thereby allowing tumor cell persistence manifesting as an incomplete tumor response that may ultimately promote acquired resistance (19).

The present finding that CK1-knockdown is able to prevent as well as overcome acquired resistance to erlotinib opens up some interesting possibilities. Previous studies have shown that a variety of resistance mechanisms operate to drive acquired resistance to erlotinib and that these can frequently co-occur in patients, suggesting that completely overcoming all resistance mechanisms in a patient may be extremely difficult (4). Thus, preventing acquired resistance has a better chance of being curative. In this regard, the preventative potential of targeting CK1 in EGFR-mutant NSCLC may be well worth exploring further.

References

1. Pao W, Chmielecki J. Rational, biologically based treatment of EGFR-mutant non-small-cell lung cancer. Nature reviews Cancer. 2010;10:760-74. 2. Sharma SV, Bell DW, Settleman J, Haber DA. Epidermal growth factor receptor mutations in lung cancer. Nature reviews Cancer. 2007;7:169-81. 3. Garraway LA, Janne PA. Circumventing cancer drug resistance in the era of personalized medicine. Cancer discovery. 2012;2:214-26. 4. Niederst MJ, Engelman JA. Bypass mechanisms of resistance to receptor tyrosine kinase inhibition in lung cancer. Science signaling. 2013;6:re6. 5. Kobayashi S, Boggon TJ, Dayaram T, Janne PA, Kocher O, Meyerson M, et al. EGFR mutation and resistance of non-small-cell lung cancer to gefitinib. The New England journal of medicine. 2005;352:786-92. 6. Pao W, Miller VA, Politi KA, Riely GJ, Somwar R, Zakowski MF, et al. Acquired resistance of lung adenocarcinomas to gefitinib or erlotinib is associated with a second mutation in the EGFR kinase domain. PLoS medicine. 2005;2:e73. 7. Kwak EL, Sordella R, Bell DW, Godin-Heymann N, Okimoto RA, Brannigan BW, et al. Irreversible inhibitors of the EGF receptor may circumvent acquired resistance to gefitinib. Proceedings of the National Academy of Sciences of the United States of America. 2005;102:7665-70. 8. Bean J, Brennan C, Shih JY, Riely G, Viale A, Wang L, et al. MET amplification occurs with or without T790M mutations in EGFR mutant lung tumors with acquired resistance to gefitinib or erlotinib. Proceedings of the National Academy of Sciences of the United States of America. 2007;104:20932-7. 9. Engelman JA, Zejnullahu K, Mitsudomi T, Song Y, Hyland C, Park JO, et al. MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling. Science. 2007;316:1039-43. 10. Turke AB, Zejnullahu K, Wu YL, Song Y, Dias-Santagata D, Lifshits E, et al. Preexistence and clonal selection of MET amplification in EGFR mutant NSCLC. Cancer cell. 2010;17:77-88. 11. Takezawa K, Pirazzoli V, Arcila ME, Nebhan CA, Song X, de Stanchina E, et al. HER2 amplification: a potential mechanism of acquired resistance to EGFR inhibition in EGFR-mutant lung cancers that lack the second-site EGFRT790M mutation. Cancer discovery. 2012;2:922-33. 12. Cortot AB, Repellin CE, Shimamura T, Capelletti M, Zejnullahu K, Ercan D, et al. Resistance to irreversible EGF receptor tyrosine kinase inhibitors through a multistep mechanism involving the IGF1R pathway. Cancer research. 2013;73:834-43. 13. Guix M, Faber AC, Wang SE, Olivares MG, Song Y, Qu S, et al. Acquired resistance to EGFR tyrosine kinase inhibitors in cancer cells is mediated by loss of IGF-binding proteins. The Journal of clinical investigation. 2008;118:2609-19. 14. Fidler MJ, Basu S, Buckingham L, Walters K, McCormack S, Batus M, et al. Utility of insulin-like growth factor receptor-1 expression in gefitinib-treated patients with non-small cell lung cancer. Anticancer research. 2012;32:1705-10. 15. Terai H, Soejima K, Yasuda H, Nakayama S, Hamamoto J, Arai D, et al. Activation of the FGF2- FGFR1 autocrine pathway: a novel mechanism of acquired resistance to gefitinib in NSCLC. Molecular cancer research : MCR. 2013;11:759-67. 16. Byers LA, Diao L, Wang J, Saintigny P, Girard L, Peyton M, et al. An epithelial-mesenchymal transition gene signature predicts resistance to EGFR and PI3K inhibitors and identifies Axl as a therapeutic target for overcoming EGFR inhibitor resistance. Clinical cancer research : an official journal of the American Association for Cancer Research. 2013;19:279-90. 17. Zhang Z, Lee JC, Lin L, Olivas V, Au V, LaFramboise T, et al. Activation of the AXL kinase causes resistance to EGFR-targeted therapy in lung cancer. Nature genetics. 2012;44:852-60.

18. Bivona TG, Hieronymus H, Parker J, Chang K, Taron M, Rosell R, et al. FAS and NF-kappaB signalling modulate dependence of lung cancers on mutant EGFR. Nature. 2011;471:523-6. 19. Blakely CM, Pazarentzos E, Olivas V, Asthana S, Yan JJ, Tan I, et al. NF-kappaB-Activating Complex Engaged in Response to EGFR Oncogene Inhibition Drives Tumor Cell Survival and Residual Disease in Lung Cancer. Cell reports. 2015;11:98-110. 20. Sharma SV, Lee DY, Li B, Quinlan MP, Takahashi F, Maheswaran S, et al. A chromatin-mediated reversible drug-tolerant state in cancer cell subpopulations. Cell. 2010;141:69-80. 21. Sosa MS, Bragado P, Aguirre-Ghiso JA. Mechanisms of disseminated cancer cell dormancy: an awakening field. Nature reviews Cancer. 2014;14:611-22. 22. Huang S, Holzel M, Knijnenburg T, Schlicker A, Roepman P, McDermott U, et al. MED12 controls the response to multiple cancer drugs through regulation of TGF-beta receptor signaling. Cell. 2012;151:937-50. 23. Oosterkamp HM, Hijmans EM, Brummelkamp TR, Canisius S, Wessels LF, Zwart W, et al. USP9X downregulation renders breast cancer cells resistant to tamoxifen. Cancer research. 2014;74:3810-20. 24. Berns K, Horlings HM, Hennessy BT, Madiredjo M, Hijmans EM, Beelen K, et al. A functional genetic approach identifies the PI3K pathway as a major determinant of trastuzumab resistance in breast cancer. Cancer cell. 2007;12:395-402. 25. Iorns E, Turner NC, Elliott R, Syed N, Garrone O, Gasco M, et al. Identification of CDK10 as an important determinant of resistance to endocrine therapy for breast cancer. Cancer cell. 2008;13:91- 104. 26. Hoffman GR, Rahal R, Buxton F, Xiang K, McAllister G, Frias E, et al. Functional epigenetics approach identifies BRM/SMARCA2 as a critical synthetic lethal target in BRG1-deficient cancers. Proceedings of the National Academy of Sciences of the United States of America. 2014;111:3128-33. 27. Konig R, Chiang CY, Tu BP, Yan SF, DeJesus PD, Romero A, et al. A probability-based approach for the analysis of large-scale RNAi screens. Nature methods. 2007;4:847-9. 28. Birmingham A, Selfors LM, Forster T, Wrobel D, Kennedy CJ, Shanks E, et al. Statistical methods for analysis of high-throughput RNA interference screens. Nature methods. 2009;6:569-75. 29. Hubbell E, Liu WM, Mei R. Robust estimators for expression analysis. Bioinformatics. 2002;18:1585-92. 30. Rena G, Bain J, Elliott M, Cohen P. D4476, a cell-permeant inhibitor of CK1, suppresses the site- specific phosphorylation and nuclear exclusion of FOXO1a. EMBO reports. 2004;5:60-5. 31. Bidere N, Ngo VN, Lee J, Collins C, Zheng L, Wan F, et al. Casein kinase 1alpha governs antigen- receptor-induced NF-kappaB activation and human lymphoma cell survival. Nature. 2009;458:92-6. 32. Elyada E, Pribluda A, Goldstein RE, Morgenstern Y, Brachya G, Cojocaru G, et al. CKIalpha ablation highlights a critical role for p53 in invasiveness control. Nature. 2011;470:409-13. 33. Amit S, Hatzubai A, Birman Y, Andersen JS, Ben-Shushan E, Mann M, et al. Axin-mediated CKI phosphorylation of beta-catenin at Ser 45: a molecular switch for the Wnt pathway. Genes & development. 2002;16:1066-76. 34. Liu C, Li Y, Semenov M, Han C, Baeg GH, Tan Y, et al. Control of beta-catenin phosphorylation/degradation by a dual-kinase mechanism. Cell. 2002;108:837-47. 35. Knippschild U, Kruger M, Richter J, Xu P, Garcia-Reyes B, Peifer C, et al. The CK1 Family: Contribution to Cellular Stress Response and Its Role in Carcinogenesis. Frontiers in oncology. 2014;4:96. 36. Boehm JS, Zhao JJ, Yao J, Kim SY, Firestein R, Dunn IF, et al. Integrative genomic approaches identify IKBKE as a breast cancer oncogene. Cell. 2007;129:1065-79. 37. Naylor TL, Tang H, Ratsch BA, Enns A, Loo A, Chen L, et al. Protein kinase C inhibitor sotrastaurin selectively inhibits the growth of CD79 mutant diffuse large B-cell lymphomas. Cancer research. 2011;71:2643-53.

38. Rahal R, Frick M, Romero R, Korn JM, Kridel R, Chan FC, et al. Pharmacological and genomic profiling identifies NF-kappaB-targeted treatment strategies for mantle cell lymphoma. Nature medicine. 2014;20:87-92. 39. Bassik MC, Kampmann M, Lebbink RJ, Wang S, Hein MY, Poser I, et al. A systematic mammalian genetic interaction map reveals pathways underlying ricin susceptibility. Cell. 2013;152:909-22. 40. Sudo M, Mori S, Madan V, Yang H, Leong G, Koeffler HP. Short-hairpin RNA library: identification of therapeutic partners for gefitinib-resistant non-small cell lung cancer. Oncotarget. 2015;6:814-24. 41. Prahallad A, Sun C, Huang S, Di Nicolantonio F, Salazar R, Zecchin D, et al. Unresponsiveness of colon cancer to BRAF(V600E) inhibition through feedback activation of EGFR. Nature. 2012;483:100-3. 42. Davis RE, Brown KD, Siebenlist U, Staudt LM. Constitutive nuclear factor kappaB activity is required for survival of activated B cell-like diffuse large B cell lymphoma cells. The Journal of experimental medicine. 2001;194:1861-74. 43. Fellmann C, Lowe SW. Stable RNA interference rules for silencing. Nature cell biology. 2014;16:10-8.

Figure legends

Figure 1. shRNA screening identifies casein kinase 1 alpha (CK1, CSNK1A1) as a drop-out hit in EGFR- mutant NSCLC cells treated with the EGFR TKI erlotinib.

(A) Schematic outline of the drop-out shRNA screens for genes whose inhibition prevents or overcomes resistance to erlotinib in EGFR-mutant NSCLC cells. Cells were infected with lentiviral particles containing a shRNA library comprised of 6,500 shRNAs against ~350 genes, and then treated with DMSO or 2 M erlotinib for 10 days or 24 days. shRNA representation from both populations was determined by next- generation sequencing. Strategy 1 (left) outlines screening of erlotinib-sensitive EGFR-mutant NSCLC cells, strategy 2 (right) outlines screening of erlotinib-resistant PC9 cells. (B) Representation of the abundance of shRNAs at gene level in erlotinib- versus DMSO-treated conditions. For high confidence hit calling at the gene level, RSA-score and Q1-z-score were calculated for the 17 shRNAs for each gene.

RSA-score < -1.5 (p-value: 0.05) and Q1-z-score < -1 were used as parameters for hit selection (boxed area) and hits shared across several NSCLC cell lines were prioritized. Red dots highlight CSNK1A1, blue dots CSNK1D and green dots CSNK1E.

Figure 2. Suppression of CK1 increases the maximum dose response (Amax) and enhances the apoptotic response to erlotinib.

(A) Knockdown of CK1 increases the maximum dose response (Amax) to erlotinib. Lentiviral infection was used to introduce three independent doxycycline inducible CK1- or NTC-shRNAs into HCC827,

HCC4006 or erlotinib-resistant PC9 cells. Cells pretreated for 72 h +/- doxycycline were plated in triplicate, and treatment with various doses of erlotinib was started the following day. Cell viability was assessed by Cell Titer Glo after 5 days, and inhibition of viability relative to DMSO-treated cells was calculated. Data represent one of three independent experiments. * highlight the significance of the difference between the maximum dose response of plus versus minus doxycyclin condition at the

highest concentration of erlotinib (5M). (B) Knockdown of CK1 enhances the apoptotic response to erlotinib as measured by cleaved PARP. HCC827 and HCC4006 cells expressing doxycycline-inducible

CK1- or NTC-shRNAs were treated for 72 h +/- doxycycline followed by +/- 2 M erlotinib treatment for

24 h. Erlotinib resistant PC9 cells cultured in the presence of 2 M erlotinib were treated +/- doxycycline and +/- 2 M erlotinib for 96 h. Immunoblot analysis of total cell lysates was performed. (C) Knockdown of CK1 enhances the apoptotic response to erlotinib as measured by caspase3/7-activity. Treatment of cells was performed as in (B). Caspase-activity was normalized to cell viability measured by Cell Titer Glo.

Data are represented as fold change to - dox/+ erlotinib-condition +/-STDV. Data represent one of three independent experiments. The subtle apoptotic signal caused by the NTC-shRNA may result from cell- line specific small seed homology off-target effects or from general cell perturbations due to the presence of exogenous vectors and/or RNA species (43).

Figure 3. Suppression of CK1 attenuates acquired resistance to erlotinib in EGFR-mutant NSCLC cells.

(A) HCC827, HCC4006 and PC9 cells expressing doxycycline-inducible CK1- or NTC-shRNAs were treated

+/- doxycycline and +/- 2 M erlotinib. Cells were fixed, stained and photographed at the indicated time points. Each condition was done in triplicate and a representative image is shown. Data represent one of three independent experiments. The level of knockdown of CK1 protein induced by each shRNA was analyzed by immunoblotting. (B) As (A) but for two additional EGFR-mutant NSCLC cell lines, HCC2935 and H3255. H3255 cells were treated with 75 nM erlotinib. (C) Ectopic expression of V5-tagged shRNA- resistant wildtype mCK1 cDNA, but not of kinase-dead mCK1(D136N) cDNA, rescues CK1- knockdown cells from erlotinib-induced cell death. HCC827 cells expressing doxycycline-inducible CK1- or NTC-shRNAs and wildtype or kinase-dead mCK1-cDNA constructs were generated by lentiviral infection. Cells were treated +/- doxycycline and +/- 2 M erlotinib. Cells were fixed, stained and photographed at the indicated time points. Each condition was done in triplicate and a representative

image is shown. Levels of ectopically expressed mCK1 protein were analyzed by immunoblotting.

Human hCK1 was detected by anti-CK1 antibody, murine mCK1-V5 was detected by anti-V5 antibody.

Figure 4. Down-regulation of NF-B-signaling by CK1-knockdown attenuates acquired resistance to erlotinib in NSCLC cells.

(A) Knockdown of CK1 inhibits NF-B-signaling as measured by NF-B luciferase reporter assay.

HCC827 and PC9 cells expressing doxycycline-inducible CK1- or NTC-shRNAs were transduced with NF-

B luciferase reporter plasmid. Cells were treated for 72 h +/- doxycycline followed by +/- 2 M erlotinib treatment for 24 h. Luciferase signal was normalized to cell viability assessed by Cell Titer Glo. Data are represented as fold change to - dox-condition +/-STDV. Data represent one of three independent experiments. (B) Inhibition of NF-B signaling by expression of IB-superrepressor attenuates acquired resistance to erlotinib. HCC827 and PC9 cells expressing doxycycline-inducible IB-superrepressor were cultured +/- doxycycline and +/- 2 M erlotinib. Cells were fixed, stained and photographed at the indicated time points. Each condition was done in triplicate and a representative image is shown. Data represent one of two independent experiments. Inhibition of NF-B-signaling upon induction of IB- superrepressor was assessed by NF-B luciferase reporter. Cells treated for 48 h +/- doxycycline were co-transfected with NF-B-firefly luciferase plasmid and renilla luciferase plasmid. 48 h after transfection, the signal of firefly and renilla luciferase was measured using Luciferase Dual Glo assay.

Firefly luciferase signal was normalized to renilla luciferase signal. Data are represented as fold change to - dox-condition +/- STDV. Data represent one of two independent experiments. (C) Co-treatment of

HCC827, HCC4006 and PC9 cells with the IKK-inhibitor AFN700 and 2 M erlotinib attenuates acquired resistance. Cells were cultured in the presence of the indicated concentrations of AFN700 and +/- 2 M

erlotinib. At the indicated time points cells were fixed, stained and photographed. Data represent one of two independent experiments.

Figure 5. Knockdown of CK1 attenuates acquired resistance to erlotinib in HCC827 xenograft model.

(A) Determination of minimum erlotinib dose to achieve maximum tumor regression. Mice bearing

HCC827 xenografts expressing doxycycline-inducible NTC-shRNAs were dosed with vehicle or 1 mg/kg,

2.5 mg/kg, 5 mg/kg or 10 mg/kg erlotinib, respectively, for the indicated time frame. Data are represented as mean tumor volume (mm3) +/- SEM. Each group included 6 mice. (B) Inducible CK1- shRNAs downregulate CK1 protein, and erlotinib modulates pEGFR in the HCC827 xenograft model.

Immunoblot analysis of tumor pharmacodynamic markers from three animals per treatment group.

Mice bearing HCC827 xenografts expressing doxycycline-inducible CK1- or NTC-shRNAs were kept on a diet supplemented +/- doxycycline and dosed once daily with 5 mg/kg erlotinib or vehicle for 7 days. (C)

Suppression of CK1 attenuates acquired resistance to erlotinib in HCC827 xenograft model. Mice bearing HCC827 xenografts expressing doxycycline-inducible CK1- or NTC-shRNAs were kept on a diet supplemented +/- doxycycline and dosed once daily with 5 mg/kg erlotinib or vehicle for the indicated time frame. Data are represented as mean tumor volume (mm3) +/- SEM. Vehicle treatment groups included 6 mice, erlotinib treatment groups included 10 mice.

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2015 American Association for Cancer Research. Author Manuscript Published OnlineFirst on October 21, 2015; DOI: 10.1158/0008-5472.CAN-15-1113 Figure 1 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

A Screening Strategy 1 Screening Strategy 2

HCC827 / HCC4006 / PC9 PC9 Resistant PC9

Infect with shRNA library Infect with shRNA library

DMSO Erlotinib DMSO Erlotinib Allow selection Allow selection

10d/24d 17d

Determine shRNA representation by next generation sequencing and identify shRNAs that are depleted from the erlotinib-treated arm (relative to DMSO).

B HCC827 HCC4006 PC9 Resistant PC9

10d 17d RSA-score RSA-score RSA-score RSA-score

Q1-z-score Q1-z-score Q1-z-score Q1-z-score

CSNK1A1 24d CSNK1D CSNK1E RSA-score RSA-score RSA-score RSA-score

Q1-Z-score Q1-z-score Q1-z-score

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2015 American Association for Cancer Research. Author Manuscript Published OnlineFirst on October 21, 2015; DOI: 10.1158/0008-5472.CAN-15-1113 Figure 2 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

A sh #1 sh #2 sh #3 sh NTC 120 120 120 120 + Dox - Dox 80 80 80 80 %Survival %Survival %Survival %Survival %Survival %Survival 40 40 40 40 HCC827 DAmax DAmax DAmax ] ] ] DAmax *** *** ] * 0 *** 0 0 0 0.0001 0.1 0.0001 0.1 0.0001 0.1 0.0001 0.1 Erlotinib (mM) Erlotinib (mM) Erlotinib (mM) Erlotinib (mM)

sh #1 sh #2 sh #4 sh NTC 120 120 120 120

80 80 80 80

40 %Survival %Survival %Survival 40 %Survival 40 %Survival 40 HCC4006 DAmax DAmax DAmax DAmax ] ] ] 0 *** 0 *** 0 ] *** 0 * 0.0001 0.1 0.0001 0.1 0.0001 0.1 0.0001 0.1 Erlotinib (mM) Erlotinib (mM) Erlotinib (mM) Erlotinib (mM)

sh #1 sh #2 sh #3 sh NTC 120 120 120 120

80 80 80 80

40 40 40

%Survival %Survival %Survival 40 %Survival %Survival

Resistant PC9 Resistant 0 0 0 0 0.1 10 0.1 10 0.1 10 0.1 10 Erlotinib (mM) Erlotinib (mM) Erlotinib (mM) Erlotinib (mM)

B HCC827 HCC4006 Resistant PC9

#1 #3 NTC #1 #3 NTC #1 #4 NTC #1 #4 NTC #1 #3 NTC #1 #3 NTC + + + + + + ------+ + + + + + ------+ + + + + + ------:Erl + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - :Dox CK1a cl. PARP GAPDH

C HCC827 HCC4006 Resistant PC9

8 12 4 *** *** 6 *** *** 9 *** 3 *** + Dox activity activity activity -

- - ** 4 6 2 - Dox *** n.s. n.s. n.s. 2 ** n.s. ** ** 3 ** ** ** 1 0 0 0 Caspase 3/7 Caspase 3/7 Caspase 3/7 M M Erl M Erl M Erl M Erl M Erl M Erl M Erl M Erl M Erl DMSO DMSO DMSO DMSO DMSO DMSO DMSO DMSO DMSO m m m m m m m m m DMSO DMSO DMSO DMSO DMSO DMSO DMSO DMSO DMSO 2uM ERL 2uM ERL 2uM ERL 2 2 2 2 2 2 2 2 2 2uM ERL 2uM ERL 2uM ERL 2uM ERL 2uM ERL 2uM ERL shsh #1 #1 shsh #3 #3 shsh NTC shsh #1 #1 shsh #4 #4 shsh NTC shsh #1 #1 shsh #3 #3 shsh NTC

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2015 American Association for Cancer Research. Author Manuscript Published OnlineFirst on October 21, 2015; DOI: 10.1158/0008-5472.CAN-15-1113 Figure 3 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

sh #1 sh #2 sh #3 sh NTC A - + - + - + :Dox + - #1 #2 #3 NTC + - + - + - + - :Dox 7d CK1a DMSO DMSO

GAPDH HCC827 HCC827 21d Erl Erl sh #1 sh #2 sh #4 sh NTC + - + - + - + - :Dox #1 #2 #4 NTC + - + - + - + - :Dox 7d CK1a DMSO DMSO GAPDH

HCC4006 HCC4006 21d Erl Erl sh #1 sh #2 sh #3 sh NTC + - + - + - + - :Dox #1 #2 #3 NTC + - + - + - + - :Dox 7d CK1a DMSO DMSO PC9 PC9 33d GAPDH Erl Erl

B sh #1 sh #2 sh NTC + - + - + - :Dox #1 #2 NTC + - + - + - :Dox 7d CK1a DMSO DMSO GAPDH 30d HCC2935 HCC2935 Erl Erl sh #1 sh #2 sh NTC + - + - + - :Dox #1 #2 NTC + - + - + - :Dox 7d CK1a DMSO DMSO

H3255 H3255 GAPDH 14d Erl Erl

sh #1 sh #3 sh NTC C + - + - + - :Dox D + empty + mCK1a- + mCK1a vector V5 (D136N)-V5 + empty vector #1 #3 NTC #1 #3 NTC #1 #3 NTC Erl Erl + - + - + - + - + - + - + - + - + - :Dox + mCK1a- hCK1a V5 Erl Erl HCC827 HCC827 mCK1a-V5 + mCK1a (D136N)-V5 GAPDH Erl Erl

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2015 American Association for Cancer Research. Author Manuscript Published OnlineFirst on October 21, 2015; DOI: 10.1158/0008-5472.CAN-15-1113 Figure 4 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. A HCC827 PC9 * ** ** 1.5 *** 1.5 *** *** *** n.s. *** *** *** n.s. *** *** *** * *** *** *** * 1.0 1.0 B-activity B-activity B-activity

0.5 0.5 Rel. NF- k Rel. NF-k Rel. 0.0 0.0 sh #1 sh #2 sh #3 sh sh #1 sh #2 sh #3 sh sh #1 sh #2 sh #3 sh sh #1 sh #2 sh #3 sh NTC NTC + Dox NTC NTC DMSO Erlotinib - Dox DMSO Erlotinib B HCC827 PC9 IkBa-SR pLKO-TREX IkBa-SR pLKO-TREX + - + - :Dox + - + - :Dox

7d 7d DMSO DMSO

23d 28d Erl Erl

1.5 1.5 *** n.s. ** n.s. 1.0 1.0 + Dox B-activity B-activity B-activity - Dox 0.5 0.5

Rel. NF- k Rel. 0.0 NF-k Rel. 0.0 IkBa-SR pLKO-TREX IkBa-SR pLKO-TREX C HCC827 HCC4006 PC9 AFN700 (mM) AFN700 (mM) AFN700 (mM) 5 2.5 1 0 Erl 5 2.5 1 0 Erl 5 2.5 1 0 Erl

7d 7d 7d DMSO DMSO DMSO

21d 21d 35d Erlotinib Erlotinib Erlotinib Erlotinib Erlotinib Erlotinib

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2015 American Association for Cancer Research. Author Manuscript Published OnlineFirst on October 21, 2015; DOI: 10.1158/0008-5472.CAN-15-1113 Figure 5 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Vehicle, po, qd Erlotinib 1 mg/kg, po, qd Erlotinib 2.5 mg/kg, po, qd Erlotinib 5 mg/kg, po, qd Erlotinib 10 mg/kg, po, qd

B sh #1 sh #3 sh NTC

+ + + + + + ------+ + + + + + ------+ + + + + + ------:Erl + + + - - - + + + - - - + + + - - - + + + - - - + + + - - - + + + - - - :Dox CK1a

pEGFR

EGFR

GAPDH

C sh #1 sh #3 sh NTC Vehicle, po, qd / Dox diet Vehicle, po, qd Erlotinib 5 mg/kg, po, qd / Dox diet Erlotinib 5 mg/kg, po, qd

Inhibition of casein kinase 1 alpha prevents acquired drug resistance to erlotinib in EGFR-mutant non-small cell lung cancer

Alexandra B Lantermann, Dongshu Chen, Kaitlin J McCutcheon, et al.

Cancer Res Published OnlineFirst October 21, 2015.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-15-1113

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2015/10/17/0008-5472.CAN-15-1113.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet been Manuscript edited.

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2015/10/21/0008-5472.CAN-15-1113. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.