Dysregulation of splicing proteins in head and neck squamous cell carcinoma

ABSRTRACT

Signaling plays an important role in regulating all cellular pathways. Altered signaling is one of the hallmarks of cancers. Phosphoproteomics enables interrogation of kinase mediated signaling pathways in biological systems. In cancers, this approach can be utilized to identify aberrantly activated pathways that potentially drive proliferation and tumorigenesis. To identify signaling alterations in head and neck squamous cell carcinoma (HNSCC), we carried out proteomic and phosphoproteomic analysis of HNSCC cell lines using a combination of tandem mass tag (TMT) labeling approach and titanium dioxide-based enrichment. We identified 4,920 phosphosites corresponding to 2,212 proteins in six HNSCC cell lines compared to a normal oral cell line. Our data indicated significant enrichment of proteins associated with splicing. We observed hyperphosphorylation of SRSF protein kinase 2 (SRPK2) and its downstream substrates in HNSCC cell lines. SRPK2 is a splicing kinase, known to phosphorylate serine/arginine (SR) rich domain proteins and regulate splicing process in eukaryotes. Although genome-wide studies have reported the contribution of alternative splicing events of several genes in the progression of cancer, the involvement of splicing kinases in HNSCC is not known. In this study, we studied the role of SRPK2 in HNSCC. Inhibition of SRPK2 resulted in significant decrease in colony forming and invasive ability in a panel of HNSCC cell lines. Our results indicate that phosphorylation of SRPK2 plays a crucial role in the regulation of splicing process in HNSCC and that splicing kinases can be developed as a new class of therapeutic target in HNSCC.

KEYWORDS:

• GeneSpring
• in vitro labeling
• mRNA processing
• phosphorylation
• spliceosome
• SR proteins

Abbreviations

HNSCC	=	Head and neck squamous cell carcinoma
TiO2	=	Titanium dioxide
TMT	=	Tandem mass tag

Introduction

Head and neck squamous cell carcinoma (HNSCC) is a heterogeneous group of cancers with high mortality.¹ Although targeted therapy for EGFR has been used in the treatment of HNSCC; the development of resistance is common.^2,3 Studies have shown that aberrant activation of kinase signaling pathways contribute to abnormality in cellular processes which results in tumorigenesis.^4,5 Oncogenic activation of PI3K, Notch and hedgehog signaling pathways have been reported in HNSCC and they are being tested in clinical studies as potential targets.^6-8

Mass spectrometry-based proteome/phosphoproteome profiling has become a reliable method for studying the signaling pathways/events that are dysregulated in any given system. In this study, we carried out LC-MS/MS analysis of TMT-labeled, titanium dioxide (TiO₂) enriched phosphopeptides to study the signaling pathways that are altered in HNSCC cells compared to a normal oral cell line OKF6/TERT1. Mass spectrometry-based analysis led to the identification of 4,920 phosphosites in 2,212 proteins. Among the phosphorylated proteins, we observed hyperphosphorylation (fold change ≥ 1.5 in at least four HNSCC cell lines compared to normal cell line) of various serine/threonine kinases including cyclin dependent kinase 1 (CDK1), epidermal growth factor receptor (EGFR), protein kinase D2 (PRKD2), MAP/microtubule affinity-regulating kinase 2 (MARK2), maternal embryonic leucine zipper kinase (MELK), protein kinase N2 (PKN2), ATR serine/threonine kinase (ATR), SRSF protein kinase 2 (SRPK2) and ribosomal protein S6 kinase (RPS6KB2). Bioinformatics analysis of the data revealed significant enrichment of proteins involved in splicing.

Gene expression in eukaryotes is mainly regulated by alternative splicing, where the pre-mRNA is modified to mature mRNA through post-transcriptional modifications. Genome-wide sequencing studies have revealed extensive changes in splicing patterns in more than 90% of human protein coding genes.^9,10 The complex of proteins that catalyze splicing mechanism known as ‘spliceosome’ is a macromolecular complex containing five small nuclear ribonuclear proteins and ˜200 accessory proteins including various kinases and signaling molecules.^11,12 Aberrant activation or expression of splice factors and kinases has been shown to influence cellular transformation and other oncogenic processes.^13,14 Among the proteins involved in splicing, we identified hyperphosphorylation of SRPK2, a splicing kinase known to affect subcellular localization and activity of splice factors.¹⁵ We also identified several substrates of SRPK2 to be aberrantly expressed or phosphorylated in HNSCC cells. Though studies have shown the significance of splicing proteins in multiple diseases including cancer, the key role of SRPK2 in HNSCC is not known. We studied the role of SRPK2 and its potential as a therapeutic target in head and neck squamous cell carcinoma.

Results

Quantitative proteomic profile of HNSCC

To study the dysregulated proteome of HNSCC cells, we employed TMT labeling approach. The experimental workflow employed for the study is depicted in Fig. 1. LC-MS/MS analysis of 12 bRPLC fractions was carried out using LTQ-Orbitrap Velos mass spectrometer. MS/MS data was searched against human protein database (RefSeq 65) using SEQUEST and Mascot search algorithms through Proteome Discoverer platform suite (version 1.4.1.14, Thermo Scientific). We identified a total of 3,255 proteins. Among these, 538 proteins were found to be upregulated (fold change ≥1.5) and 590 proteins were found to be downregulated (fold change ≤ 0.67) in at least four HNSCC cell lines used in this study compared to the normal cell line. A partial list of molecules differentially expressed in HNSCC cell lines compared to OKF6/TERT1 cell line is provided in Table 1. Among the differentially expressed proteins, proteins such as BAG family molecular chaperone regulator 2 (BAG2), actin-binding protein anillin (ANLN), Caprin 1(CAPRIN1), Lymphokine-activated killer T-cell-originated protein kinase (PBK) and WD repeat and HMG-box DNA-binding protein 1 (WDHD1) are some of the molecules which are overexpressed in all the HNSCC cell lines studied. The complete list of identified proteins and their corresponding peptides is provided in Tables S1 and S2, respectively.

Figure 1. Schematic workflow used to study the proteome and phosphoproteome in HNSCC.

Table 1. A partial list of molecules overexpressed in HNSCC cell lines.

		Relative protein abundance (Fold-change as compared to the normal OKF6/TERT1 cells)
Gene Symbol	Protein	JHU-O28	JHU-O11	FaDu	CAL 27	JHU-O22	JHU-O29
ANLN	Actin-binding protein anillin	3.4	2.3	1.5	3.5	2.3	1.6
KPNA2	Importin subunit α-1	5.9	3.5	6.7	3.3	5.2	5.6
BAG2	BAG family molecular chaperone regulator 2	3.7	3	2.7	4.8	3.6	4.1
WDHD1	WD repeat and HMG-box DNA-binding protein 1	2.6	1.9	2.4	1.7	2.1	2.5
SF3A2	Splicing factor 3A subunit 2	1.3	3	2.1	1.9	2.3	2.4
SRSF6	Serine/arginine-rich splicing factor 6	1.4	1.9	3.8	2.1	2.5	2.5
CAPRIN1	Caprin 1	2.1	2.2	1.9	2.1	1.9	1.8
PBK	Lymphokine-activated killer T-cell-originated protein kinase	5.2	2	3.2	2.4	3	2.7

Table 2. A partial list of molecules hyperphosphorylated in HNSCC cell lines.

				Relative phosphopeptide abundance (Fold-change as compared to the normal OKF6/TERT1 cells)
Gene Symbol	Protein	Site	Phosphopeptide sequence	JHU-O28	JHU-O11	FaDu	CAL 27	JHU-O22	JHU-O29
MPZL1	Myelin protein zero-like protein 1	S210	DYTGCSTSESLsPVK	2.1	2.7	2.3	1.9	1.6	2
PRPF4B	Serine/threonine-protein kinase PRP4 homolog	S366, S368	sRsPLLNDRR	2.9	2.6	2.8	2.2	2.8	2.4
SRSF9	Serine/arginine-rich splicing factor 9	S204	GRDsPYQSR	2.7	3	9	2	3.4	2.9
SRPK2	SRSF protein kinase 2	S494, S497	TVsASsTGDLPK	2.4	2.5	2.5	1.6	1.6	1.5
MELK	Maternal embryonic leucine zipper kinase	S311	sVELDLNQAHMEETPK	1.6	2.7	1.9	3	2.3	1.7
IGF2BP3	Insulin-like growth factor 2 mRNA-binding protein 3	S184	QGsPGSVSK	3.4	2.1	2.1	4.5	2.6	1.9
SRSF6	Serine/arginine-rich splicing factor 6	S314, S316	sVsPPPK	1.7	1.8	4.2	1.5	2.2	1.6
RRP36	rRNA processing protein 36	S73	QLVAGNsPK	8.8	2.4	4.3	5.8	3.5	4.4

Identification of dysregulated phosphoproteins in HNSCC cell lines

Since we observed widespread dysregulation of protein expression in HNSCC cell lines compared to the normal oral cell line, we sought to study the altered signaling events in these cells using phosphoproteomic approach. Protein phosphorylation plays a crucial role in intracellular signaling. Identification of post-translational modifications in cancer cells will not only help in elucidating the signaling mechanisms but also help in the identification of potential novel therapeutic candidates. We performed TMT-based quantitative phosphoproteomic analysis to identify dysregulated signaling in HNSCC cell lines compared to a normal oral cell line OKF6/TERT1 (Fig. 1). The enriched (pSer/Thr) peptides were analyzed on LTQ-Orbitrap Velos mass spectrometer. With a false discovery rate (FDR) cutoff of 1% and PhosphoRS probability cutoff of 75%, we identified a total of 4,920 phosphosites from 2,212 proteins (Table S3). The peptides with a fold change ±1.5 were considered as hyperphosphorylated or hypophosphorylated. Based on these, we identified hyperphosphorylation of 1,299 sites and hypophosphorylation of 639 sites in at least four HNSCC cell lines. These corresponded to 878 hyperphosphorylated proteins and 402 hypophosphorylated proteins respectively. Among the proteins identified, 23 kinases were found to be hyperphosphorylated in all HNSCC cell lines. The distribution of phosphosites identified in all cells is shown in Fig. 2A. Among the phosphosites identified, 88% of phosphorylation sites accounted for serine phosphorylation, 11% of phosphorylation sites for threonine residues and 1% of phosphorylation sites for tyrosine residues. A heat map representing the phosphorylation changes of a subset of kinases identified in HNSCC cell lines compared to OKF6/TERT1 is depicted in Fig. 2B. To identify consensus motifs enriched in the datasets, we further carried out motif analysis using motif-x algorithm. From the hyperphosphorylated phosphopeptides, we identified 4 distinct phosphorylation motifs including pSPxK, pSPxR, pSPxRK and pSPxxxK. These are proline directed motifs where serine is followed by proline at +1 position (Fig. 2C).

Figure 2. Summary statistics of the analysis. (A) Distribution of phosphoserine, phosphothreonine and phosphotyrosine sites (B) Heat map of the phosphorylation changes in a subset of kinases identified in HNSCC cell lines compared to OKF6/TERT1. (C) Motifs that were found to be enriched in the hyperphosphorylated kinase sites in the data set.

We performed bioinformatics analysis of differentially phosphorylated proteins and categorized them based on the cellular localization and biological function (Fig. S1). The classifications were based on annotations in HPRD.^16,17 Our analysis revealed that 21% of the proteins were involved in regulation of nucleic acid metabolism; 18% in signal transduction, 15% in cell communication and 11% in cellular growth and maintenance. Classification of the proteins based on subcellular localization revealed 40% of the hyperphosphorylated proteins localized to the nucleus and 25% localized to the cytoplasm. Among the other predominant protein groups, 7% of the proteins were localized to plasma membrane.

Pathway analysis revealed the dysregulation of proteins associated with splicing in HNSCC

To gain insights into the altered signaling pathways in HNSCC, we used GeneSpring GX v12.6.1 (Agilent Technologies) for analysis. The list of overexpressed proteins identified from proteomic data (538) and hyperphosphorylated proteins from phosphoproteomic (878) data were included for pathway analysis. The analysis was carried out using Pathway Architect module of GeneSpring wherein both data sets were mapped on to pathways. Among these, mRNA processing pathway was found to be significantly enriched in HNSCC cells compared to the normal cell line. Out of 127 proteins depicted in the pathway diagram (Fig. 3), 32 proteins from our phosphoproteomic data and 16 proteins from proteomic data were found to map on to this pathway. These include splicing kinase such as SRSF protein kinase 2 (SRPK2), Serine/threonine-protein kinase PRP4 homolog (PRPF4B) and various other splicing factors including, serine/arginine-rich splicing factor 2 (SRSF2), serine/arginine-rich splicing factor 9 (SRSF9), members of heterogeneous ribonuclear protein family (hnRNPs), RNA binding motif protein 17 (RBM17), splicing factor 3b (SF3B1) and DEAH (Asp-Glu-Ala-His) box polypeptide 16 (DHX16) (Fig. 3). MS/MS spectra for representative peptides from SRSF2 (T29) and SRPK2 (S494 and S497) are shown in Figs. 4A–B. In agreement with the mass spectrometry data, we observed hyperphosphorylation of SRPK2 (Ser 494) in HNSCC cells using western blot analysis (Fig. 4C). Further we analyzed the expression of SRPK2 in a panel of HNSCC cell lines. Western blot analysis of SRPK2 revealed an increase in the expression of the protein in HNSCC cells compared to the non-neoplastic oral cells (Fig. 4C).

Figure 3. Pathway analysis revealed dysregulation of proteins associated with splicing machinery in HNSCC. Gene spring version 12.6.1 was used for pathway analysis. Molecules that mapped to the mRNA processing pathway from the phosphoproteomic data are highlighted in green, molecules that mapped from the proteomic data are highlighted in orange and molecules that are common in both data sets are highlighted in yellow. The fold changes are represented using heat strips.

Figure 4. Representative MS and MS/MS spectra of phosphopeptides identified in HNSCC cells (A) serine/arginine-rich splicing factor 2 (SRSF2) (B) SRSF protein kinase 2 (SRPK2) (C) Western blot analysis shows overexpression and hyperphosphorylation of SRPK2 in a panel of HNSCC cell lines compared to OKF6/TERT1. (D) Western blot analysis shows a decrease in SRPK2 expression in HNSCC cell lines transfected with SRPK2 siRNA. β-actin was used as a loading control.

Invasive and colony forming properties of HNSCC cells

To understand if SRPK2 had any association with features of transformed cells such as loss of contact inhibition and increased cell migration, we examined the colony formation and invasive ability of the HNSCC cells. Except JHU-O11 and JHU-O28, all HNSCC cells were capable of forming well defined colonies. The colonies of JHU-O11 and JHU-O28 were less clearly defined than those of other HNSCC cells, indicating cell scattering as evidenced by the invasion data. Our data also revealed that all HNSCC cells except JHU-O29 have a high invasive ability. Overall, our data indicate that three of the HNSCC cell lines (FaDu, CAL 27 and JHU-O22) had both high colony forming and invasive ability, whereas the other cells exhibited only a high colony forming ability (JHU-O29) or only a high invasive ability (JHU-O11 and JHU-O28) (Fig. S2).

Inhibition of SRPK2 reduced the colony forming ability of HNSCC cells

Silencing of SRPK2 using its specific siRNA was confirmed by Western blot analysis (Fig. 4D). siRNA mediated silencing of SRPK2 resulted in a decrease in the colony forming ability of HNSCC cells (Fig. 5A–B) compared to control siRNA transfected cells. Using an alternative approach, we studied the colony forming ability of HNSCC cells by inhibiting the activity of SRPK2 using SRPIN340, a small molecule inhibitor of SRPK2.¹⁸ In agreement with the results observed using siRNA-mediated silencing, inhibition of SRPK2 by SRPIN340 led to a significant decrease in the colony formation ability of HNSCC cell lines (Fig. 5C–D).

Figure 5. Inhibition of SRPK2 reduces the colony forming ability of HNSCC cells. Colony formation assay following (A) siRNA mediated knockdown of SRPK2 or control siRNA in a panel of HNSCC cell lines, as indicated. Colonies formed were visualized after staining with methylene blue. (B) A graphical representation of the colony forming ability of the HNSCC cells upon SRPK2 silencing *p < 0.05. (C) Colony forming ability of the HNSCC cells upon inhibition of SRPK2 using SRPIN340 or control (DMSO), in the indicated panel of HNSCC cells (D) A graphical representation of the colony forming ability of HNSCC cells upon SRPK2 inhibition *p < 0.05.

Inhibition of SRPK2 decreases the invasive ability of HNSCC cells

Our data showed that SRPK2 plays a significant role in the proliferative ability of HNSCC cells. siRNA mediated silencing of SRPK2 led to a significant reduction in the invasive property of all HNSCC cells (Fig. 6A–B). In agreement with the siRNA results, inhibition of SRPK2 with SRPIN340 resulted in significant decrease in the invasive property of all HNSCC cells (Fig. 6C–D). Taken together, our results indicate that SRPK2 plays an important role in regulating the oncogenic potential of HNSCC cells.

Figure 6. Inhibition of SRPK2 reduces the invasive ability of HNSCC cells. Invasion assays were carried out using in a transwell system using Matrigel-coated filters and the number of cells that migrated to the lower chamber was counted. Cells that migrated are visualized following methylene blue staining in a panel of HNSCC cell lines as indicated. (A) HNSCC cell lines were transfected with either control (Scrambled) or SRPK2 siRNA and invaded cells were photographed (B) A graphical representation of the invasive ability of the HNSCC cells upon SRPK2 silencing *p < 0.05. (C) HNSCC cells were treated with a small molecule inhibitor of SRPK2 (SRPIN340) or vehicle control (DMSO) and invaded cells were photographed. (D) A graphical representation of the invasive ability of the HNSCC cells upon SRPK2 inhibition *p < 0.05.

Discussion

Activation of multiple signaling pathways can be directly assessed by studying the global phosphorylation events. This is because a large majority of activated kinases can be identified due to an increase in their phosphorylation status and the corresponding pathways can be determined by their substrates that are phosphorylated. In recent years, kinases have become one of the most intensively studied groups of proteins as drug targets. In this study, we used mass spectrometry-based quantitative proteomic and phosphoproteomic analysis to identify altered signaling pathways in HNSCC. Our analysis revealed significant enrichment of molecules involved in mRNA processing / splicing. Among the splicing proteins, serine arginine (SR) proteins are a group of RNA binding proteins, which possess a unique serine/arginine (SR) domain. These proteins are important for constitutive as well as alternative pre-mRNA splicing. The splicing function of SR proteins is modulated by kinases which results in altered localization of these proteins in the cell.

Recent studies have shown an association between altered splicing machinery and cancers.^19,20 Dysregulation of splicing has been reported as an important factor in leukemia.^21-23 Alterations in spliceosome associated genes/proteins can significantly impact alternative splicing events like exon skipping or intron retention as reported in breast cancer, lung adenocarcinoma and chronic lymphocytic leukemia.^22,24,25 We identified dysregulation of multiple proteins encoded by genes involved in pre-mRNA splicing. Overexpression (≥ 1.5 fold) of splicing regulators including SRSF5, SRSF6, SF3A2, SF3A3 and RBM17 was observed in HNSCC cells compared to a normal oral cell line. Additionally, we observed significant dysregulation in the phosphorylation pattern of splicing regulators across HNSCC cell lines. Our phosphoproteome data indicates hyperphosphorylation (≥ 1.5 fold) of proteins including SRSF2, SRSF6, SRSF9, SRSF11, PRPF4B and SRPK2 in the HNSCC cell lines. We observed a 2-fold increase in phosphorylation of serine/arginine rich protein kinase 2 (SRPK2) in the panel of HNSCC cell lines. This is also evident from our Western blot data, where the expression of pSRPK2 was below detectable levels in the normal oral keratinocytes.

Though multiple studies have looked into the landscape of alternate splice forms of various proteins as biomarkers, there are limited studies on the role of splicing kinases in cancer progression.^26-28 Several studies have shown the involvement of various members of splicing kinases, especially SRPKs in multiple cellular processes including proliferation and angiogenesis.^29,30 Overexpression of splicing factors (SRSF1, SRSF2) has been reported in lung, colon and breast cancers.^13,31,32 SR protein specific kinases (SRPK1, 2 and 3), which is a highly conserved group of kinases, phosphorylates the SR domain and regulates the splicing activity of SR proteins.^14,33 Among the members of SRPKs, the role of SRPK1 in cancer has been established. Overexpression of SRPK1 has been reported in breast, colon and pancreatic cancers.³¹ siRNA mediated silencing of SRPK1 in ovarian cancer cell lines resulted in reduced cell proliferation and increased sensitivity to cisplatin.³⁴ Inhibition of SRPK1 using siRNA or pharmacological inhibitor suppressed VEGF expression and decreased melanoma growth in vivo. Studies by Zhou et al., have shown that SRPKs constitute a major branch of Akt signaling pathway in EGF mediated alternate splicing.³⁵

SRPK2 is a serine arginine rich protein specific kinase that phosphorylates SR family of splicing factors at serine residue located in serine arginine rich regions.²⁹ It has been shown that phosphorylation of acinus by SRPK2 results in the expression of cyclin A1 and plays a key role in promoting leukemia cell proliferation.²⁹ SRPK2 phosphorylates SRSF2 in nuclear speckles and mediate the pre-mRNA splicing events.³³ Karni et al., have reported SRSF2 to be upregulated in multiple human tumors and its overexpression leads to alternative splicing of the tumor suppressor proteins bridging integrator 1(BIN1), mitogen-activated protein kinase-interacting serine/threonine kinase 2 (MNK2) and ribosomal S6 kinase (S6K1).³⁶ The alternative splice variant of BIN1 lacks tumor-suppressor activity whereas the MNK2 and S6K1 isoforms have oncogenic properties.³⁶ SRSF6, another target of SRPK2, is one of the most ubiquitous splicing factors that play an important role in constitutive splicing as well as in the selection of alternative splice sites.³⁷ Filippov et al., has shown that SRSF6 depletion leads to an increased resistance to DNA damage in p53 dependent way in U2OS cells.³⁷ Thus altered expression of SRSF6 upon DNA damage has been known to affect changes in the splicing activity/ pattern.³⁷ SRSF9, another target of SRPK2, has been reported to be overexpressed in different cancers including ovarian, prostate and bladder cancer.^38-40 Additionally, SRSF11 another regulator of alternative splicing⁴⁰ was hyperphosphorylated (>2 fold) in our data.⁴¹ Studies have reported SRPK1 and SRPK2 to be overexpressed in non-small cell lung cancer and their overexpression correlated with hyperphosphorylation of SRSF1 and SRSF2.¹³ In concurrence with these studies, our data reveals both hyperphosphorylation and overexpression of SRPK2 in HNSCC cell lines. Taken together, these studies highlight the crucial role of SRPKs and their substrates in the regulation of splicing in cancer. Our in vitro results demonstrate that inhibition of SRPK2 leads to reduction in growth, invasion and migration of HNSCC cells. Overall, our study indicates a widespread dysregulation in the genes/proteins involved in splicing machinery in the HNSCC cells. We hypothesize that the aberrant regulation of SRPK2 and other genes involved in splicing may lead to altered splicing in HNSCC cells, which needs to be systematically evaluated further. Overall, our study provides a scaffold for future studies to investigate the regulation of splicing and development of SR protein-specific kinases as a new class of therapeutic targets in HNSCC.

Materials and methods

Cell culture

FaDu and CAL 27 were purchased from American Type Culture Collection (ATCC, Manassas, VA). OKF6/TERT1, a normal oral keratinocyte cell line,⁴¹ was a gift from Dr. James Rheinwald (Brigham and Women's Hospital in Boston, MA).⁴² HNSCC cell lines, JHU-O11, JHU-O22, JHU-O28, JHU-O29 and FaDu were cultured in RPMI-1640 with 10% FBS. CAL 27 was cultured in DMEM with 10% FBS. All cell lines were grown in a humidified incubator with 5% CO₂ at 37°C as described previously.⁴³ The cell lines used for the study were authenticated by short tandem repeat analysis.

Cell lysis, protein digestion and TMT labeling

Cells were cultured till 70% confluency and serum starved for 12 h. After serum starvation, cells were washed three times with 1X phosphate buffer saline (PBS), harvested and lysed in lysis buffer (2% SDS, 5 mM sodium fluoride, 1 mM β-glycerophosphate, 1 mM sodium orthovanadate in 50 mM Triethyl ammonium bicarbonate (TEABC)). The cell lysates were sonicated and centrifuged at 12,000 rpm for 10 minutes. The protein concentration was estimated using bicinchoninic acid assay (BCA).⁴⁴ Equal amount of protein from each cell line was reduced and alkylated as described previously.⁴³ We employed filter aided sample preparation (FASP) protocol as described earlier to remove SDS.⁴⁵ Protein digestion was carried using TPCK treated trypsin (1:20) at 37°C for 12–16 h. Peptides from each cell line were labeled using TMT labeling method as per manufacturer's instruction (Catalog # 90110, Thermo Fisher Scientific). Peptides derived from OKF6/TERT1 were labeled with TMT tag 126, JHU-O28 with TMT tag 127N, JHU-O11 with TMT tag 127C, FaDu with TMT tag 128N, CAL 27 with TMT tag 128C, JHU-O22 with TMT tag 129N and JHU-O29 with TMT tag 130C.

Basic reversed-phase liquid chromatography (bRPLC) and phosphopeptides enrichment using TiO₂

The labeled peptides were subjected to bRPLC fractionation generating 96 fractions, as described previously.⁴⁶ The fractions were concatenated into 12 fractions. From these pooled fractions 100 µg equivalent peptides were taken for total proteome analysis. All fractions were lyophilized and subjected to TiO₂-based enrichment.^47,48 Briefly, the phosphopeptide enriched- TiO₂ beads were washed two times with 80% ACN in 3% TFA, eluted using 4% ammonia solution and neutralized with 3% TFA. Eluted peptides were vacuum dried and desalted using C₁₈ Stage Tips.

LC-MS/MS analysis

LC-MS/MS analysis of total proteome and enriched fractions of phosphopeptides was carried using an LTQ-Orbitrap Velos mass spectrometer interfaced with Proxeon Easy nanoLC II system (Thermo Scientific).⁴⁶ The settings used for phosphopeptide analysis were: a) MS1 scans were acquired from m/z: 350–1800 with resolution of 60,000 at 400 m/z, b) Ten most intense precursor ions were selected for MS2 analysis. Precursor ions were fragmented using HCD fragmentation at a resolution of 30,000 at m/z 400 (isolation width: 1.90 m/z; normalized collision energy: 43%; activation time: 0.1 ms.

Mass spectrometry data analysis

The mass spectrometry derived data was searched against Human RefSeq protein database (Version 65, containing 36211 protein entries with common contaminants added) using SEQUEST and MASCOT search algorithms through Proteome Discoverer platform (version 1.4.1.14, Thermo Scientific). For both the algorithms search parameters included maximum of two missed cleavages, carbamidomethylation at cysteine, TMT 6-plex (+229.163) modification at N-terminus of peptide and lysine were set as fixed modification while oxidation of methionine and phosphorylation at serine, threonine and tyrosine as variable modifications. For MS data, monoisotopic peptide mass tolerance was set to 20 ppm and MS/MS tolerance to 0.1 Da. PhosphoRS (version 3.1) was used to calculate the confident localization of phosphosite within the identified peptide (Phospho RS score ≥75).⁴⁹ False discovery rate of 1% was set at the PSM level. The complete mass spectrometry data generated from this study has been deposited to the ProteomeXchange Consortium (http://www.proteomexchange.org) via PRIDE partner repository with the dataset identifier PXD001862.

Bioinformatics analysis

Biological process and cellular localization information were obtained from Human Protein Reference Database (HPRD).^16,50 Pathway analysis was done using GeneSpring GX software (version 12.6.1). Using Pathway Architect module of GeneSpring GX, the proteins that were found to be overexpressed and hyperphosphorylated in at least four cell lines with a fold change of 1.5 was included for pathway analysis. For predicting consensus motifs based on the phosphosites identified in the study, we used motif-x algorithm.⁵¹ The parameters used include (i) the flanking sequences around the phosphosites with a window size of 15 aminoacids long were extracted from RefSeq 65 database and used for extracting consensus motif. For phosphorylation sites that were localized at the region of the N- or C-termini, the surrounding sequence could not be extended and were excluded, (ii) the significance threshold was set to p < 0.01, (iii) the minimum occurrence of motifs was set to 10.

Colony formation assays

Colony formation assays were carried out as described previously.⁵² Briefly, the cells were seeded at a density of 3×10³, transfected with either SRPK2 siRNA or a scrambled siRNA as a control. The cells were allowed to grow for 10–12 days and the colonies were fixed with methanol, stained with 4% methylene blue and counted. Similarly the colony forming ability of the HNSCC cells were studied in presence of SRPK1/2 inhibitor (SRPIN340 - 10 µM, Catalog # sc-394310, Santa Cruz Biotechnology). The number of colonies formed was counted for ten randomly selected viewing fields and representative images were photographed at 2.5x magnification. All experiments were done in triplicate.

Cell invasion assays

Invasion assay was carried out in a transwell system (BD Biosciences) using Matrigel-coated filters as described.⁵² Briefly, the invasive ability of the cells was assayed using polyethylene terephthalate (PET) membrane (8-μm pore size) (Catalog # 353097, BD Biosciences). The upper compartment of the culture insert was coated with Matrigel (Catalog # 354234, BD Biosciences). 2×10⁴ cells were seeded into the transwell chambers in presence of serum-free medium. Complete media was added to the lower compartment and the cells were incubated at 37°C in 5% CO₂ incubator for 48 h. After incubation, the non-migratory cells from the upper chamber were removed with a cotton-tip applicator. Cells that migrated to the lower side of membrane were fixed and stained using 4% methylene blue. The number of cells that penetrated through the matrigel and membrane was counted for ten randomly selected viewing fields at 10x magnification. All the experiments were done in triplicate.

Western blot analysis

Cells were cultured to 80% confluency and proteins were harvested in RIPA lysis buffer (10 mM Tris pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% Triton-X-100, 0.1% SDS containing protease and phosphatase inhibitor cocktails) and sonicated. Mouse anti-SRPK2 antibody was obtained from BD transduction laboratories (Catalog # 611118). Phospho-SRPK2 (S494) antibody was obtained from Millipore (Catalog # 07–1817). Western blot analysis was carried out as described previously.⁵³

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank the Department of Biotechnology (DBT), Government of India for research support to the Institute of Bioinformatics (IOB), Bangalore. We thank the “Infosys Foundation” for research support to IOB. AC was supported by Department of Science and Technology (DST) grants (SERC/LS-439/2011 and SR/SO/HS/0208/2013). IOB is supported by DBT Program Support on Neuroproteomics and infrastructure for proteomic data analysis (BT/01/COE/08/05). This work was supported by NCI's Clinical Proteomic Tumor Analysis Consortium initiative (U24CA160036) and FAMRI-funded 072017_YCSA. AR, GS and SC are recipients of Senior Research Fellowship from Council of Scientific and Industrial Research (CSIR), Government of India. RR is a recipient of research associateship from DBT. Harsha Gowda is a Wellcome Trust/DBT India Alliance Early Career Fellow. We thank Strand Life Sciences, Bangalore and Agilent Technologies for providing access to GeneSpring. We acknowledge Dr. Manoj Kumar Kashyap for critically reviewing the manuscript. We thank Dr. S. K. Shankar and Dr. Anita Mahadevan of National Institute of Mental Health and Neurological Sciences, for providing access to the microscopy imaging facility.