Quantitative Proteomic Verification of Membrane Proteins as Potential

Jul 7, 2015 - An important strategy for defining therapeutically relevant targets in these situations is to ascertain which genes are amplified at the...
1 downloads 4 Views 6MB Size
Article pubs.acs.org/jpr

Quantitative Proteomic Verification of Membrane Proteins as Potential Therapeutic Targets Located in the 11q13 Amplicon in Cancers Heather Hoover, Jun Li, Jason Marchese, Christopher Rothwell, Jason Borawoski, Douglas A. Jeffery,† L. Alex Gaither,* and Nancy Finkel* Novartis Institutes for Biomedical Research, Cambridge, Massachusetts 02139, United States S Supporting Information *

ABSTRACT: Tumor types can be defined cytologically by their regions of chromosomal amplification, which often results in the high expression of both mRNA and proteins of certain genes contained within the amplicon. An important strategy for defining therapeutically relevant targets in these situations is to ascertain which genes are amplified at the protein level and, concomitantly, are key drivers for tumor growth or maintenance. Furthermore, so-called passenger genes that are amplified with driver genes and a manifest on the cell surface can be attractive targets for an antibody−drug conjugate approach (ADC). We employed a tandem mass spectrometry proteomics approach using tumor cell lines to identify the cell surface proteins whose expression correlates with the 11q13 amplicon. The 11q13 amplicon is one of the most frequently amplified chromosomal regions in human cancer, being present in 45% of head and neck and oral squamous cell carcinoma (OSCC) and 13−21% of breast and liver carcinomas. Using a panel of tumor cell lines with defined 11q13 genomic amplification, we identified the membrane proteins that are differentially expressed in an 11q13 amplified cell line panel using membrane-enriched proteomic profiling. We found that DSG3, CD109, and CD14 were differentially overexpressed in head and neck and breast tumor cells with 11q13 amplification. The level of protein expression of each gene was confirmed by Western blot and FACS analysis. Because proteins with high cell surface expression on selected tumor cells could be potential antibody drug conjugate targets, we tested DSG3 and CD109 in antibody piggyback assays and validated that DSG3 and CD109 expression was sufficient to induce antibody internalization and cell killing in 11q13-amplified cell lines. Our results suggest that proteomic profiling using genetically stratified tumors can identify candidate antibody drug conjugate targets. Data are available via ProteomeXchange with the identifier PXD002486. KEYWORDS: antibody-drug conjugate, mass spectrometry, head and neck carcinoma, tandem mass tag



INTRODUCTION A patient stratification to enrich the optimal patient population for a cancer treatment is becoming an essential feature in the early phase of drug development.1,2 For example, melanoma treatment uses the Raf V600E mutation to identify patients that will respond to mutant selective compounds, enriching treatment in patients that will likely respond to the drug. More than 80% of patients showed partial response to the treatment, while the treatment of nonstratified melanoma patients using the BRAF inhibitor sorafenib had a more variable response.2 In addition, secondary mutations and resistant mutations can guide the choice of the combination of drugs in a patient-centric fashion.3 Defining the subpopulations of patients has also been shown to be effective in treating colon cancers with EGFR inhibitors in the presence of KRAS mutations.2 Another effective therapeutic paradigm is targeting lineage-specific oncogenic targets. HER2-amplified breast and gastric cancer treatment with trastuzumab is effective in breast tissue where Her2 is overexpressed but not in ovarian or endometrial cancers where expression is low.3 In contrast, some © XXXX American Chemical Society

drug targets are effective across multiple lineages including BRCA1/2 mutations using olaparib in breast, ovarian, and prostate cancers.3 Thus, depending on the context of the drug target, either a lineage-selective or mutation-specific lesion with highly expressed cell surface proteins in a defined genetic context can provide new therapeutic targets in cancer. Antibody drug conjugate therapy leverages the high expression of cell surface proteins that get internalized upon antibody binding selectively on tumor cells. The key requirements for a good antibody drug conjugate are high cell surface expression and internalization upon antibody conjugate binding. There are also rare examples such as Her2, where the protein is both highly expressed on the cell surface and has a defined link to tumor growth. If these features of Her2 can be Special Issue: The Chromosome-Centric Human Proteome Project 2015 Received: June 2, 2015

A

DOI: 10.1021/acs.jproteome.5b00508 J. Proteome Res. XXXX, XXX, XXX−XXX

Article

Journal of Proteome Research

415, Cama-1, and HCC1954), and four liver lines (LI-7, JHH-7, SNU-387, and SNU-878) were grown at 37 °C 5% CO2 in either RPMI or DMEM supplemented with 10% fetal bovine serum until 80% confluent. SCC25 cells were also supplemented with 10 mL hydrocortisone (catalog no. H6909, Sigma) per recommendation by ATCC.

linked to a secondary cancer-dependent genetic lesion such as the 11q13 amplicon, the patient stratification and treatment landscape can be expanded. To discover the cell surface proteins associated with known tumor drivers, we employed a membrane-enriched proteomics approach to identify highly expressed cell surface proteins that correlated with the 11q13 amplicon in a panel of tumor cells. The 11q13 amplicon is one of the most frequently amplified chromosomal regions in human cancer. The 11q13 region is amplified in 45% of head and neck and oral squamous cell carcinoma (OSCC) and 13−21% of breast and liver carcinomas.4 The identification of a protein that is highly and selectively expressed on the surface of 11q13-amplified tumors in combination with antibody−drug conjugates could provide a targeted therapeutic approach that may benefit cancer patients. The current standard treatment for head and neck cancer is surgery or radiation or a combination of the two, typically resulting in severe deformity.5 An in-depth profiling of the chromosome 11 is currently underway as part of the Chromosome-Centric Human Proteome Project (C-HPP).6,7 We expect that the strategy of focusing on the 11q13-amplified region can be applied more broadly, which is in accordance with the aim of the C-HPP.6,7 A natural follow-up to this study would be to integrate the RNAseq data of these cell lines with the proteomics data to identify new splice variants, as has been shown in breast cancer.8 Transmembrane targets from these various lineages could be identified that are putative antibody−drug conjugate targets. The 11q13 amplicon contains the known oncogene Cyclin D1 and tumor-promoting genes such as FGF19, FGF23, and ANO1.9−11 We used 11q13 genomic amplification as the criteria to pick cell lines to profile for the high cell surface expression of genes whose protein expression was up-regulated when compared to the results from nonamplified cell lines. We validated that the target proteins were highly expressed in amplified, lineage-matched cell lines using Western blots and FACS analysis. The top three differentially expressed proteins in head and neck and breast cancer were DSG3, CD109, and CD14. These were further characterized as potential antibody− drug conjugate targets. Selective antibodies that bind to each of the proteins were characterized using FACS in the 11q13amplified and nonamplified cell lines. The primary antibodies were added to cell lines together with secondary antibodies that were conjugated to the toxin saporin and added to cells in culture (the piggyback assay). DSG3 and CD109 antibody− antibody saporin complexes could induce cell death in culture in highly expressing cell lines. This killing was dose-, temperature-, and time-dependent, suggesting that antibody internalization was required. Our results suggest that DSG3 and CD109 could be putative ADC therapeutic targets in 11q13amplified cancers, where their expression correlates with this tumor-promoting genetic lesion. Furthermore, our results demonstrate the utility of using transmembrane protein profiling in genetically defined cell line panels to identify highly expressed cell surface proteins that could serve as ADC drug targets.



Copy Number Analysis

Genomic DNA was isolated using Qiagen’s DNeasy Kit according to the manufacturer’s instructions. RT-PCR analysis was performed as described using a probe for ANO1: Hs04399219_cn, CCND1: Hs69455941_cn, and RNaseP (4401631, Invitrogen, Carlsbad, CA). The copy number was calculated using Copy Caller v1.0 freeware (Applied Biosystems), normalized to RNaseP, and expressed relative to normal human tissue (human placenta, female D3035, Sigma). Enrichment and Preparation of Membrane Proteins

Membrane proteins were enriched using the Thermo Pierce cell surface isolation kit (catalog no. 89881, Thermo Scientific, San Jose, CA). The kit uses a cell impermeable, cleavable Sulfo−NHS−SS−Biotin to label primary amines on the cell surface proteins of living cells. The samples are collected and centrifuged at 2000 rpm at 4 °C for 5 min. The cells were lysed using probe sonication. The protein concentration was determined using Pierce BCA protein assay reagent (catalog no. 23227, Pierce). The labeled proteins were isolated using Neutravidin Agarose beads (catalog no. 20361, Pierce). Each cell line was prepared in biological triplicate, and 400 μg of protein of each sample was reduced with 20 mM dithiothreitol (DTT, Thermo Scientific, Rockford, IL) at 37 °C for 1 h and then alkylated using 50 mM iodoacetamide (IAM, Sigma) in 50 mM triethylammonium bicarbonate (TEAB, Sigma) at room temperature for 1 h in the dark. The beads were washed twice with 80 μL of 50 mM TEAB. The samples were then digested overnight at 37 °C with sequencing grade (1:50) trypsin (catalog no. V5111, Promega, Madison, WI). The samples were acidified with formic acid (FA) and dried down using a SpeedVac. The samples were resuspended in 30 μl 500 mM TEAB. TMT (catalog no. 90063, Thermo Scientific, San Jose, CA) was brought to room temperature and resuspended in 70 μL acetonitrile (ACN). The TMT reagent was transferred into the respective sample vial. The vials were kept at room temperature for 1 h, and the reaction was stopped with 10% hydroxylamine. Samples from all 6 channels were combined and cleaned using the Oasis HLB elution plate (30 uM; catalog no. 186001828BA, Waters, Milford, MA). The samples were dried in a SpeedVac and reconstituted in 20 μL of 20 mM ammonium formate (pH 10) and 2% acetonitrile. All samples were then fractionated using a high-pH fraction method as previously described.12 The peptides were fractionated into 12 fractions using a Dionex HPLC (Thermo Scientific) and 2.1 mm × 50 mm Xterra column (catalog no. 186000408, Waters). The samples were lyophilized and stored at −20 °C until mass spectrometer analysis. Mass Spectrometry Analysis

The tryptic peptides were reconstituted in 10 μL of FA (2% ACN and 0.2% FA). The sample was first loaded at 5 μL/min onto a μ-Precolumn 300 μm i.d. × 5 mm C18 PepMap100, 5 μm, 100 Å trap column (catalog no. 160454, Thermo Scientific). Digested samples (2 μL) were analyzed using nanoflow liquid chromatography coupled to a data-dependent mass spectrometer (LC−MS/MS) using the Eksigent nano-LC

MATERIALS AND METHODS

Cell Culture

A total of four head and neck cell lines (A253, Detroit562, FaDu, and SCC25), four OSCC lines (TE-11, TE-9, KYSE-10, and TE-15), five breast lines (SK-BR-3, HCC1143, MDA-MBB

DOI: 10.1021/acs.jproteome.5b00508 J. Proteome Res. XXXX, XXX, XXX−XXX

Article

Journal of Proteome Research

Supporting Information). Within each TMT reporter channel, ion intensities were summed by the peptide sequence with isotopic correction factors applied per manufacturer guidelines. Peptide fold-changes were calculated across lineages and within lineages for the 11q13-amplified and nonamplified cell lines. The peptide fold-changes were renormalized using the median fold-change of all quantified peptides. The protein fold-changes were derived from the median peptide fold-changes. Significance was determined using ANOVA statistical testing and p values were calculated. Data were visualized using Spotfire DXP, substituting gene names for IPI accession numbers.

(Applied Biosystems/MDS Sciex, Foster City, CA) coupled to an LTQ Orbitrap Velos mass spectrometer (Thermo Scientific, San Jose, CA). A 75 μm i.d. Picotip emitter with a 15 μm diameter tip (catalog no. PF360-75-15-N-5, New Objective, Woburn, MA) was hand-packed using Magic C18 100 Å 3 μm resin to a length of 13 cm. Tryptic peptides were eluted over a 68 min gradient at a flow rate of 400 nL/min using a water/ acetonitrile (ACN) gradient (mobile phase A: 100% water and 0.2% FA; mobile phase B: 100% ACN and 0.2% FA). The gradient was ramped from a minimum of 2% B over 68 min to 40% B and then ramped to 95% B over 8 min, held for 2 min at 95% B, ramped to 5% B in 2 min, and then ramped to 2% for 5 min. The Velos system was operated in the standard-scan mode with positive ionization. The electrospray voltage was 2.75 kV, and the ion transfer tube was 275 °C. Full MS spectra were acquired in the Orbitrap mass analyzer over the 350−1800 m/z range with mass resolution at 30 000 (at 400 m/z); the target value was 1.00 × 1006. The 30 most intense peaks with charge states greater than or equal to 2 were fragmented in the HCD collision cell with a normalized collision energy of 40%. The tandem mass spectra were acquired in the orbitrap mass analyzer with a mass resolution of 7500 with a target value of 5.00 × 1006. The ion selection threshold was 500 counts, and the maximum allowed ion accumulation times were 500 ms for full scans and 250 ms for HCD. Dynamic exclusion was enabled with a repeat count of 1, a repeat duration of 15 s, an exclusion list of 500, and an exclusion duration of 15 s. All samples were analyzed in biological triplicate and subjected to duplicate LC− MS/MS analysis.

Western Blot Analysis

Cell lysates were analyzed by SDS-PAGE and Western blotting using standard protocols and the following primary antibodies: DSG3 clone 3G133 (catalog no. sc-59776, Santa Cruz Biotechnology Inc., Dallas, Texas), CD109 clone C-9 (catalog no. sc-271085, Santa Cruz Biotechnology), CD14 clone 134620 (catalog no. MAB3832, R & D Systems, Minneapolis, MN) used in nonreducing conditions, GAPDH clone GA1R (catalog no. MA5-15738, Thermo Scientific), and β-actin (catalog no. ab8227, Abcam, Cambridge, UK). Secondary antibodies used were LI-COR IR dye 680LT (catalog no. 827−11080, LI-COR BioSciences, Lincoln, NE) and LI-COR IR dye 800CW (catalog no. 827−08365, LI-COR). Blots were analyzed using an Odyssey scanner (LI-COR). Flow Cytometry Analysis

The primary antibodies were DSG3 clone 3G133 (catalog no. sc-59776, Santa Cruz Biotechnology), CD109 clone 496920 (catalog no. MAB4385, R & D Systems), CD14 clone 134620 (catalog no. MAB3832, R & D Systems). The secondary antibody was Molecular Probes R-phycoerythrin F(ab′)2 fragment of goat anti-rabbit (H + L) (catalog no. A10542, Life Technologies, Woburn, MA) and Molecular Probes Rphycoerythrin F(ab′)2 fragment of goat anti-mouse (H + L) (catalog no. A10543, Life Technologies). Cells were washed in phosphate-buffered saline (PBS) and then harvested with versene (catalog no. 15040066, Invitrogen). Detached cells were resuspended with PBS 0.5% BSA. 1.6 × 105 cells per well were then incubated with 4 ng of the primary antibody or the respective isotype control. The cells were incubated for 1 h in the dark and then washed with PBS 0.5% BSA. Then, 0.5 μL of the PE secondary antibody was added to each well and incubated for 1 h at room temperature in the dark. The cells were washed with PBS 0.5% BSA and analyzed on a LSRFortessa cell analyzer (BD Biosciences, San Jose, CA). The data was analyzed using FloJo software version 7.6 (Tree Star, Inc., Ashland, OR).

Data Processing and Protein Identification

Peptide and protein identifications were obtained using the IPI_human v3.78 database with common contaminants and the reverse database appended (174 261 sequences; 70 553 677 residues.) The IPI algorithm clusters proteins on the basis of sequence similarity, yet it also includes isoforms for which there is sufficient experimental evidence. Raw data from the LTQ Orbitrap Velos were processed with Mascot (vs 2.0.0) using the default parameters. The data were searched using trypsin as the enzyme and allowing for up to 2 missed cleavages. The search criteria included peptide mass tolerance (±15 ppm), fragment mass tolerance (±0.05 Da), fixed modifications of carbamidomethyl (C), and variable modifications of oxidation (M), phospho (ST), and phospho (Y). Sulfo−NHS−S−S−biotin−IOA (K), Sulfo−NHS−S−S− biotin−IOA (N-term), TMT6plex (K), and TMT6plex (Nterm) were used. Mass values are monoisotopic and the protein mass is unrestricted. Mascot results for the sample fractions were aggregated and submitted to the PeptideProphet and ProteinProphet algorithms for peptide and protein validation, respectively (ISB/SPC Trans-Proteomic Pipeline TPP v4.3 JETSTREAM rev. 1, build 200909091257 (MinGW)). The protein results were then filtered using a false discovery rate of less than 1%. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium13 via the PRIDE partner repository with the dataset identifiers PXD002486 and 10.6019/PXD002486.

Piggyback Assay

The primary antibodies used were DSG3 clone 3G133 (catalog no. sc-59776, Santa Cruz Biotechnology), CD109 clone 496920 (catalog no. MAB4385, R & D Systems), CD14 clone 134620 (catalog no. MAB3832, R & D Systems). The isotype control used was isotype IgG (catalog no. MAB002, Becton Dickinson, Franklin Lakes, NJ). Cells were seeded at the following densities for the 96 h cytotoxicity assay: FaDu 1000 cells/well, Detriot562 5000 cells/well, Cama-1 at 8000 cells/well, SCC25 1500 cells/well, and HCC1143 at 2200 cells/well. The plates were incubated for 24 h at 37 °C in 5% CO2 in the respective media. A MabZAP secondary (catalog no. IT-04, Advanced Targeting Systems, San Diego, CA) was added at a 4-fold excess over

Protein Quantitation and Statistical Analysis

Nondegenerate peptides were included in the quantitative analysis by default, whereas degenerate peptides were used only in cases where they were shared between a high-scoring protein and another with a probability score of zero (annotated in the C

DOI: 10.1021/acs.jproteome.5b00508 J. Proteome Res. XXXX, XXX, XXX−XXX

Article

Journal of Proteome Research

Figure 1. Membrane proteins overexpressed with 11q13 amplfication. (A) A total of four cell lines from the head and neck (A253, FaDu, SCC25, and Detroit562), OSCC (TE-11, TE-9, KYSE-410, and TE-15), breast (HCC1143, MDA-MB-415, HCC1954, and Cama-1) and liver (Li-7, JHH-7. SMU-387, and SNU-878) were screened by proteomic mass spectrometry. Amplification was based on a copy number >6 of CCND1 (gray) and ANO1 (black) genes. (B) The protein fold-changes across all of the lineages from amplified to nonamplified were derived as the median peptide fold-changes. The p values were calculated using a two-way t-test. DSG3, CD109, and CD14 were up-regulated, with the largest fold-change across lineages and within breast and head and neck, and were prioritized because they are membrane proteins.

an increase in mRNA levels but also in protein levels.6 We were trying to define cell lines that most closely represented the human tumor pathology. It is possible that additional distinct amplicon cores exist within HNSCC and could represent multiple driver lesions.13 A cell surface enrichment protocol, in combination with LC−MS/MS, was used to identify proteins that are highly expressed in 11q13-amplified tumor lines within and across lineages. A total of four lineages were screened: breast, liver, head and neck, and a specific subset of head and neck (oral squamous cell carcinoma). The cell lines were grouped in sets of six and screened using the TMT6 quantitation kit. A channel combining all 16 cell lines was run in each of the TMT6 plex experiments to bridge and combine the data. A total of four TMT six-plex experiments were used, mixing one cell line from each lineage, in addition to the pool and a control cell line (SK-BR-3). The data was normalized to the bridging pooled channel and then combined. The different hypotheses were tested using a two-tailed t-test, and the significance was ranked using the calculated p value. The top hits were considered to be proteins that were found to be differentially expressed across lineages in 11q13-positive versus 11q13-negative cells. The proteins that showed a foldchange greater than 2 within lineages and were annotated to be transmembrane proteins were picked for further analysis (Figure 1). A total of 2916 unique proteins were identified with a false discovery rate of 1%. A total of 66 proteins were up-regulated with a fold-change greater than log2 = 0.5, and 65 proteins were down-regulated with a fold-change greater than log2 = 0.5 when comparing all of the 11q13-amplified lines regardless of lineage (66 proteins total), and the list was further narrowed to proteins annotated as transmembrane proteins. There were three proteins found to be up-regulated in all three lineages and were chosen for further follow-up. CD14 had the most significant fold-change in the breast lineage comparison, with an average breast lineage fold-change (log2 = 3.97), a head and

the primary antibody and allowed to sit for 30 min at room temperature to form the complex. A total of nine 5× serial dilutions in PBS were diluted from the starting concentration of primary antibody at 20 nM and secondary antibody at 80 nM. The nine 5× serial dilutions of saporin (catalog no. PR-01, Advanced Targeting Systems) started at 1000 nM. Mouse IgG was used as the isotype control (MAB002 from R & D systems). Cell titer glow was added at 96 h post-cell-plating, and the signal was read from CTG on a PerkinElmer brand EnVision 2101 multilabel reader using Wallac Envision Manager (version 1.12).



RESULTS

Proteomic Mass Spectrometry of Cell-Surface-Enriched Protein for Identification of Plasma Membrane Proteins in 11q13-Amplified Cell Lines

Using DNA amplification analysis, we identified a series of 11q13-amplified cell lines to identify the overexpressed plasma membrane proteins using a membrane-enriched proteomics approach. The broader amplicon was defined from 0.45 to 132.32 Mb on the chromosome arm of 11q13, and the “peak” of the amplicon was defined as 60.02 to 75.92 Mb.6 Within this peak were 16 genes: SHANK2, CTTN, PPFIA1, FADD, ANO1, FGF3, FGF4, FGF19, ORAOV1, CCND1, MYEOV, TPCN2, MRGPRF, IGHMBP2, MRPL21, and CPT1A. We measured the copy number of both ANO1 and CCND1, which lie within the core of the amplicon peak. If both ANO1 and CCND1 were found to have >8 copies in the cell line, we considered the cell line to contain the 11q13 amplicon. It is possible that if we measured additional genes in the amplicon, we would define additional cell lines as amplicon-positive. Based on the TCGA data, ANO1 and CCND1 captured the majority of amplicon-positive disease; thus, therapeutic strategies to target this subset of the disease were considered clinically relevant. ANO1 and CCND1 also represented a subset of disease wherein the amplification resulted in not only D

DOI: 10.1021/acs.jproteome.5b00508 J. Proteome Res. XXXX, XXX, XXX−XXX

Article

Journal of Proteome Research

Figure 2. Higher DSG3 expression in 11q13-amplified head and neck and breast lineages. (A) Proteomics data were normalized to HCC1954 to demonstrate the fold-change between the 11q13-amplifed (black) and nonamplified (gray) lineages. The 11q13 amplification status was based on an ANO1 copy number expression greater than 6. (B) Western blots of DSG3 in whole-cell lysate correlates with the mass spectrometry results. (C) FACS analysis of DSG3 surface expression in Detroit562 and SCC25 also correlates to the mass spectrometry data.

Validation of Targets by Western Blot Analysis and Flow Cytometry

neck average fold-change of log2 = 2.43, and an overall average fold-change of log2 = 0.62. CD109 in the breast lineage comparison had an average fold-change of log2 = 1.86, with a head and neck average fold-change of log2 = 1.28 and an overall average fold-change of log2 = 0.96. DSG3 in the breast lineage comparison had an average fold-change of log2 = 2.18, with a head and neck average fold-change of log2 = 1.21 and an overall average fold-change of log2 = 1.22. The liver and OSCC within the linage data generated a number of additional protein hits, and that data is included in the Supporting Information. Although we did not pursue the validation of proteins identified in the liver and OSCC experiments, there are several proteomics studies in liver that could be integrated with our findings (see references and 15). The complete data set is available in Tables 1 and 2 in the Supporting Information.

To confirm that the protein expression results observed in the mass spectrometry experiment, we performed Western blot and FACS analysis on DSG3, CD109, and CD14 in a panel of cell lines. Western blots were performed on whole-cell lysate (Figures 2−4). The Western blot expression was consistent with the mass spectrometry values, and the levels of cell surface protein were confirmed by FACS analysis (Figures 2, 3, and 4C). Functional Validation of Antibody Drug Conjugates by Piggyback Assay

The antibodies that were validated using Western blot and FACS for DGS3 and CD109 we used to determine if antibodyinduced internalization of the proteins was sufficient to induce cell killing in highly expressed cell lines. For the piggyback E

DOI: 10.1021/acs.jproteome.5b00508 J. Proteome Res. XXXX, XXX, XXX−XXX

Article

Journal of Proteome Research

Figure 3. Higher CD109 expression in 11q13-amplified head and neck and breast lineages. (A) Proteomics data were normalized to HCC1954 to demonstrate the fold-change between the 11q13-amplifed (black) and nonamplified (gray) lineages. The 11q13 amplification status was based on an ANO1 copy number expression greater than 6. (B) The Western blot of CD109 correlates with the mass spectrometry results. The FACS analysis of CD109 surface expression in Detroit562 and FaDu also correlates to the mass spectrometry data.

killed the FaDu and SCC25 cell lines with an IC50 of 1.3 nM and 2.4 nM (Figure 5 D,E). There was no efficacy of the piggyback assays in the breast cell line MDS-MD-415 for both DSG3 and CD109. The slow growth rate of the MDA-MD415s was most likely the reason the piggyback assay was not effective over 96 h of treatment. The other 11q13-amplified breast cell line, HCC1143, was not very sensitive to the saporin toxin alone, so another toxin payload would need to be used to be efficacious in these cells (data not shown). Control treatments for piggy back assays are shown in Figure 1 in the Supporting Information.

assay, the primary Ab recognizes the differentially expressed protein epitope and is bound by a secondary antibody conjugated to the saporin payload. If antigen binding to the cell surface is internalized and trafficked to the lysosome, the saporin toxin is released in a pH-dependent cleavage event, releasing it into the cytoplasm and resulting in cell death. The number of molecules of payload taken up by the cell should be proportional to the number of cell surface receptors on the membrane. The piggyback assay for DSG3 was shown to selectively kill the 11q13-amplified cell lines with an IC50 of 0.011 nM in the SCC25 cells and 0.02 nM in FaDu cells (Figure 5 A,B). The piggyback assay for CD109 selectively F

DOI: 10.1021/acs.jproteome.5b00508 J. Proteome Res. XXXX, XXX, XXX−XXX

Article

Journal of Proteome Research

Figure 4. Higher CD14 expression in 11q13-amplified head and neck and breast lineages. (A) Proteomics data were normalized to HCC1954 to demonstrate fold-changes between the 11q13-amplifed (black) and nonamplified (gray) lineages. The 11q13 amplification status was based on an ANO1 copy number expression greater than 6. (B) The Western blot of CD14 expression correlates with the mass spectrometry results. The FACS analysis of CD14 surface expression in Cama-1 and MDA-MB-415 also correlates to the mass spectrometry data.



DISCUSSION Transmembrane discovery proteomics is a quantitative method used to measure cell surface proteins from a variety of tissue sources. We employed this technology to look at the differentially expressed proteins across a panel of tumor cell lines that were genetically stratified by the 11q13 amplicon. Our goal was to identify plasma membrane proteins that were selectively highly expressed in 11q13-amplified cell lines. In this way, the proteins could be tested as putative ADC therapeutic targets in 11q13-amplified HNSCC. We were using cell models that genetically matched the amplification and expression found in primary HNSCC tumor data so that the cell lines would better represent the epidemiology of the disease. It is possible that distinct amplicons exist within HNSCC and that within each amplicon core could exist multiple driver mutations. A more comprehensive look at the additional amplicon cores

across multiple cell lines will be an important follow up study to further define relevant drivers depending on the amplicon core that defines the subtype.13 C-HPP is an in-depth investigation established to gain a better picture of the complete human proteome, including the missing proteins, variants, mutations, and polymorphisms.6,7 The C-HPP project is also studying the association of every chromosome in the disease setting.6,7 In this paper, we focused in on a specific region of chromosome 11 and its relation to cancer as well as the potential to develop targets as antibody− drug conjugates. We performed a focused study on targets within the amplicon that could be potential antibody therapeutic targets in cancer. DSG3 has been shown to be highly expressed in head and neck cancer, but no link to the 11q13 amplicon has been observed.5,16 Although DSG3 has been shown to be involved in G

DOI: 10.1021/acs.jproteome.5b00508 J. Proteome Res. XXXX, XXX, XXX−XXX

Article

Journal of Proteome Research

Figure 5. Selective killing of DSG3 and CD109 high-expressing cell lines. In a piggyback assay, the DSG3 Mab-ZAP shows the killing of the highexpressing DSG3 head and neck cell lines FaDu and SCC25 at IC50 of 0.02 nm and 0.011, respectively (A and B). No effect was noted on the DSG3-low line Detroit562 (C). The CD109 piggyback assay was also effective in CD109 high-expressing head and neck cell lines FaDu and SCC25 at IC50 of 1.3 and 2.4 nm, respectively (E and F). No effect was noted on the CD109-low line Detroit562 (F). The saporin, DSG3, and CD109 curves were at baseline except in Detroit562. The Detroit562 cell line was more susceptible to cell death using the DSG3 Ab alone than FaDu or SCC25.

possible that the antibodies generated from autoimmune patients could be used as possible ADC molecules, although the antibodies will potentially cross-react with DSG1 in addition to DSG3. Recent studies into treatment for pemphigus vulgaris has revealed that antibodies to DSG3 alone do not result in blistering in neonatal mice.17 Further studies have revealed that the pathogenicity of the DSG3 antibodies were dependent on the binding epitopes of the DSG3 antibodies.18 CD109 is a GPI-anchored glycoprotein that is found on the surface of platelets, endothelial cells, and activated T cells. CD109 binds to TGFB with high affinity and inhibits TGFB and SMAD signaling. CD109 has been shown to be highly expressed in squamous cell lung and esophageal carcinomas20 and epithelioid sarcoma cell lines,17 and its expression was correlated to poor prognosis in 80 clinical samples.17 CD109 is found to be expressed in normal tissues (Broad microarray data) including platelets, endothelial cells, and activated T-

lung tumor cell migration and correlates with late-stage head and neck cancer, little is known about its functional role.5,16 A recent article has suggested that DSG3 regulates activator protein 1 and protein kinase C-dependent Ezrin activation.16 It has also been suggested that DSG3 is a functional target in head and neck cancers, driving tumor growth by an unknown mechanism.5 The knockdown of DSG3 using RNAi inhibited the tumor growth in lung cancer cell lines, so it is possible that DSG3 plays a functional role in head and neck cancers, but this hypothesis requires further investigation. DSG3 is highly expressed in the skin and is a member of the desmosome family. The desmosomes are structural cells that are important to maintaining cell−cell adhesion and tissue integrity. There is an autoimmune disease associated with decreased DSG3 expression, pemphigus vulgaris.17−19 It is possible that an ADC would mimic this autoimmune disease, presenting with blisters and sores on the skin and mucous membranes. It is H

DOI: 10.1021/acs.jproteome.5b00508 J. Proteome Res. XXXX, XXX, XXX−XXX

Article

Journal of Proteome Research cells.20,21 Although there would be some concern with developing CD109 as an ADC target, on the basis of its high differential expression in tumors, it warrants further preclinical investigation. CD14 was originally identified as a protein expressed on the surface of monocytes and macrophages.22 It is now known the CD14 is a receptor in innate immunity for a variety of ligands and is expressed on a number of cells beyond myeloid cells.23 CD14 is found both on the surface of cells as a GPI-anchored glycoprotein and in a soluble form.24 From our results, we determine that CD14 could be an interesting antibody therapeutic target, but the concern for toxicity with its potentially broad expression in the immune system would require careful follow-up studies to confirm it viability as an antibody−drug conjugate target.

proteomic scale coverage of the genome and to begin profiling cell lines, primary tumors, and normal tissues to build a next generation data set of tumor-expressed antigens.



ASSOCIATED CONTENT

S Supporting Information *

Supplementary Figure 1: control treatments for the piggy-back assays. Supplemental Table 1: 11q13 proteomic profiling peptides detected and sequence coverage. Supplemental Table 2: 11q13 proteomic profiling quantitative results and statistics. Supplemental Table 3: TMT6 experiment groupings and replicates. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/ acs.jproteome.5b00508.





CONCLUSIONS Our results serve as a paradigm to leverage quantitative transmembrane proteomics as a tool for the unbiased identification of cell surface proteins highly expressed on primary tumor samples and cell lines. Quantitative proteomics has several advantages over other technologies for identifying highly expressed tumor antigens. Proteomics profiling does not require the upfront validation of an antibody as required for IHC, FACS, protein arrays, and Western blot analysis. Profiling at the protein level bypasses the challenges associated with microarray or deep sequencing techniques because alterations such as the nucleic acid level do not always translate into changes in protein levels.18 Because proteomic profiling is quantitative, we have some degree of confidence that the highly expressing peptides represent a true over-representation of protein on the cell surface. One of the challenges of the proteomic approach is that not all peptides can be easily measured using Mass Spec so there will be peptides missed in our analysis. In addition, quantitation with TMT labeling has been shown to underestimate the relative protein foldchanges.25,26 There are several methods being employed to decrease this ratio compression, but we think Western blots are a robust reflection of the true fold-change of protein levels in the cell.25,27 These results highlight the need to perform a follow-up Western blot analysis of the MS data. Solubility has always been a challenge in membrane proteomics, but it is further complicated by proteins wrapped in the membrane consisting of large hydrophobic regions that do not generate easy-to-detect tryptic peptides.19 Proteins wrapped in the membrane with small intracellular or extracellular domains will be more difficult to detect. This limitation leads to overexpressed proteins being lost from the study. In these cases, we will refer to microarray data to overlap with our proteomic data to look for target genes that may be overexpressed at the mRNA level that we missed at the protein level. In conclusion, we think using quantitative proteomics to study differentially expressed cell surface antigens in primary tumors and tumor cell lines will be a valuable tool for identifying novel ADC therapeutic targets. It will be important to also profile normal tissues using this approach to enrich our data set for targets that not only have intratumor selective expression but also intertumor expression as a comparison to the results from a panel of normal tissues. We think that advancing this pipeline will augment current IHC, FACS, protein arrays, and Western blot analysis to map cell surface antigens without the need to validate an antibody early in target discovery. We are currently exploring ways to obtain the

AUTHOR INFORMATION

Corresponding Authors

*N.F. Tel: (617) 871-5705. E-mail: nancy.fi[email protected]. *L.A.G. Tel: (671)-871-7209. E-mail: alex.gaither@novartis. com. Present Address † FDA/CDRH/OIR/DIHD, 10903 New Hampshire Avenue Building 66, Room 5574 Silver Spring, MD 20993-0002, United States

Author Contributions

All authors have given approval to the final version of the manuscript. L.A.G. and N.F. contributed equally. Notes

The authors declare no competing financial interest.



ABBREVIATIONS ACN, acetonitrile; ADC, antibody−drug conjugate; MS, mass spectrometry; OSCC, oral squamous cell carcinoma; TMT, tandem mass tag



REFERENCES

(1) Dienstmann, R.; Jordi, R.; Tabernero, Josep. Biomarker-driven patient selection for early clinical trials. Curr. Opin. Oncol. 2013, 25, 305−312. (2) Moschos, S. A. Genomic biomarkers for patient slection and stratification: the cancer paradigm. Bioanalysis 2012, 4, 2499−2511. (3) Dienstmann, R.; Jordi, R.; Tabernero, J. Genomic Medicine Frotier in Human Solid Tumors: Prospects and Challenges. J. Clin. Oncol. 2013, 31, 1874−1884. (4) Huang, X.; Gollin, S. M.; Raja, S.; Godfrey, T. E. High-resolution mapping of the 11q13 amplicon and identification of a gene, TAOS1, that is amplified and overexpressed in oral cancer cells. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 11369−11374. (5) Chen, Y.-J.; Chang, J. T.; Lee, L.; Wang, H.-M.; Liao, C.-T.; Chiu, C.-C.; Chen, P.-J.; Cheng, A.-J. DSG3 is overexpressed in head neck cancer and is a potential molecular target for inhibition of oncogenesis. Oncogene 2007, 26, 467−476. (6) Chaiyarit, S.; Singhto, N.; Chen, Y.-J.; Cheng, C.-Y.; Chiangjong, W.; Kanlaya, R.; et al. Chromosome-centric Human Proteome Project (C-HPP): Chromosome 12. J. Proteome Res. 2014, 13, 3160−3165. (7) Paik, Y.-K.; Omenn, G. S.; Thongboonkerd, V.; Marko-Varga, G.; Hancock, W. S. Genome-wide Proteomics, Chromosome-centric Human Proteome Project (C-HPP), Part II. J. Proteome Res. 2014, 13, 1−4. (8) Menon, R.; Im, H.; Zhang, E.; Wu, S.-L.; Chen, R.; Snyder, M.; et al. Distinct Splice Variants and Pathway Enrichment in the Cell-Line Models of Aggressive Human Breast Cancer Subtypes. J. Proteome Res. 2014, 13, 212−227.

I

DOI: 10.1021/acs.jproteome.5b00508 J. Proteome Res. XXXX, XXX, XXX−XXX

Article

Journal of Proteome Research (9) Britschgi, A.; Bill, A.; Brinkhaus, H.; Rothwell, C.; Clay, I.; Duss, S.; et al. Calcium-activated chloride channel ANO1 promotes breast cancer progression by activating EGFR and CAMK signaling. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, E1026−E1034. (10) Shi, Z. Z.; Shang, L.; Jiang, Y.; Hao, J.; Zhang, Y.; Zhang, T.; Lin, D.; Liu, S.; Wang, B.; Gong, T.; Zhan, Q.; Wang, M. Consistent and differential genetic abberations between esophageal dysplasia and squamous cell carcinoma detected by array comparative genomic hybridization. Clin. Cancer Res. 2013, 19, 5867−5878. (11) Carles, A.; Millon, R.; Cromer, A.; et al. Head and neck squamouse cell carcinoma transcriptome analysis by comprehensive validate differential display. Oncogene 2006, 25, 1821−1831. (12) Gilar, M.; Olivova, P.; Daly, A. E.; Gebler, J. C. Twodimensional separation of peptides using RP-RP-HPLC system with different pH in first and second separation dimensions. J. Sep. Sci. 2005, 28, 1694−1703. (13) Wilkerson, P. M.; Reis-Filho, J. S. the 11q13-q14 amplicon: Clinicopathological correlations and potential drivers. Genes, Chromosomes Cancer 2013, 52, 333−355. (14) Wang, Q.; Wen, B.; Wang, T.; Xu, Z.; Yin, X.; Xu, S.; et al. Omics Evidence: Single Nucleotide Variants Transmissions on Chromosome 20 in Liver Cancer Cell Lines. J. Proteome Res. 2014, 13, 200−211. (15) Ponomarenko, E. A.; Kopylov, A. T.; Lisitsa, A. V.; Radko, S. P.; Kiseleva, Y. Y.; Kurbatov, L. K.; et al. Chromosome 18 Transcriptoproteome of Liver Tissue and HepG2 Cells and Targeted Proteome Mapping in Depleted Plasma: Update 2013. J. Proteome Res. 2014, 13, 183−190. (16) Brown, L.; Waseem, A.; Cruz, I. N.; Szary, J.; Gunic, E.; Mannan, T.; Unadkat, M.; Yang, M.; Valderrama, F.; O’Toole, E. A.; Wan, H. Desmoglein 3 promotes cancer cell migration and invasion by regulating activator protein 1 and protein C-dependent-Ezrin activation. Oncogene 2014, 33, 1−12. (17) Payne, A.; Ishii, K.; Kacir, S.; Lin, C.; Li, H.; Hanakawa, Y.; Tsunoda, K.; Amagai, M.; Stanley, J.; Siegel, D. Genetic and functional characterization of human pemphigus vulgaris monoclonal autoantibodies isolated by phage display. J. Clin. Invest. 2005, 115, 888−899. (18) Amagai, M.; Karpati, S.; Prussick, R.; et al. Autoantibodies against the amino-terminal cadherin-like binding domain of pemphigus vulgaris antigen are pathogenic. J. Clin. Invest. 1992, 90, 919−926. (19) Koga, H.; Tsuruta, D.; Ohyama, B.; Ishii, N.; Hamada, T.; Ohata, C.; Furumura, M.; Hashimoto, T. Desmoglein 3, its pathogenecity and a possibility for therapeutic target in pemphigus vulgaris. Expert Opin. Ther. Targets 2013, 17, 293−306. (20) Murray, L.; Bruno, E.; Uchida, N.; Hoffman, R.; Nayar, R.; Yeo, E.; Schuh, A.; Southerland, D. CD109 is expressed on a subpopulation of CD34+ cells enriched in hematopoietiic stem and progenitor cells. Exp. Hematol. 1999, 27, 1282−1294. (21) Sutherland, D.; Yeo, E.; Ryan, A.; Mills, G.; Bailey, D.; Baker, M. Identification of a cell-surface antigen associated with activated T lymphoblasts and activated platelets. Blood 1991, 77, 84−93. (22) Griffin, J. D.; Ritz, J.; Nadler, L. M.; Schlossman, S. F. Expression of myeloid differentiation antigens on normal and malignant myeloid cells. J. Clin. Invest. 1981, 68, 932−941. (23) Jersmann, H. P. A. Time to abandon dogma: CD14 is expressed by non-myeloid lineage cells. Immunol. Cell Biol. 2005, 83, 462−467. (24) Goyert, S.; Ferrero, E.; Rettig, W.; Yenamandra, A.; Obata, F.; Le Beau, M. The CD14 monocyte differentiation antigen maps to a region encoding growth factors and receptors. Science 1988, 239, 497− 500. (25) Savitski, M. M.; Mathieson, T.; Zinn, N.; Sweetman, G.; Doce, C.; Becher, I.; et al. Measuring and Managing Ratio Compression for Accurate iTRAQ/TMT Quantification. J. Proteome Res. 2013, 12, 3586−3598. (26) Ow, S. Y.; Salim, M.; Noirel, J.; Evans, C.; Wright, P. C. Minimising iTRAQ ratio compression through understanding LC-MS elution dependence and high-resolution HILIC fractionation. Proteomics 2011, 11, 2341−2346.

(27) Ting, L.; Rad, R.; Gygi, S. P.; Haas, W. MS3 eliminates ratio distortion in isobaric multiplexed quantitative proteomics. Nat. Methods 2011, 8, 937−940.

J

DOI: 10.1021/acs.jproteome.5b00508 J. Proteome Res. XXXX, XXX, XXX−XXX