Inactivation of SLIT2-ROBO1/2 Pathway in Premalignant Lesions of Uterine Cervix: Clinical and Prognostic Significances

The SLIT2-ROBO1/2 pathways control diverse biological processes, including growth regulation. To understand the role of SLIT2 and ROBO1/2 in cervical carcinogenesis, firstly their RNA expression profiles were screened in 21 primary uterine cervical carcinoma (CACX) samples and two CACX cell lines. Highly reduced expressions of these genes were evident. Concomitant alterations [deletion/methylation] of the genes were then analyzed in 23 cervical intraepithelial neoplasia (CIN) and 110 CACX samples. In CIN, SLIT2 was deleted in 22% samples compared to 9% for ROBO1 and none for ROBO2, whereas comparable methylation was observed for both SLIT2 (30%) and ROBO1 (22%) followed by ROBO2 (9%). In CACX, alteration of the genes were in the following order: Deletion: ROBO1 (48%) > SLIT2 (35%) > ROBO2 (33%), Methylation: SLIT2 (34%) > ROBO1 (29%) > ROBO2 (26%). Overall alterations of SLIT2 and/or ROBO1 (44%) and SLIT2 and/or ROBO2 (39%) were high in CIN followed by significant increase in stage I/II tumors, suggesting deregulation of these interactions in premalignant lesions and early invasive tumors. Immunohistochemical analysis of SLIT2 and ROBO1/2 in CACX also showed reduced expression concordant with molecular alterations. Alteration of all these genes predicted poor patient outcome. Multiparous (≥5) women with altered SLIT2 and ROBO1 along with advanced tumor stage (III/IV) and early sexual debut (<19 years) had worst prognosis. Our data suggests the importance of abrogation of SLIT2-ROBO1 and SLIT2-ROBO2 interactions in the initiation and progression of CACX and also for early diagnosis and prognosis of the disease.


Introduction
On worldwide basis, uterine cervical carcinoma (CACX) is the second common gynecological cancer, mostly afflicting the developing nations like India [1][2][3]. Among Indian women CACX ranks second, next only to breast cancer [4]. High-risk Human Papilloma Virus (HPV) is an important etiological agent associated with CACX [5]. It is transmitted sexually and is consequently ubiquitous among sexually active women. Therefore determinants of sexual activity namely; parity, age at sexual debut etc significantly predicts the risk to CACX [6]. Existing evidences indicated an intricate interplay between HPV infection and genetic alterations that critically contributes towards cervical carcinogenesis [7]. Different chromosomal abnormalities namely; deletion, amplification, rearrangement etc involving several chromosomal regions were reportedly associated with the development of CACX [7][8][9][10][11]. Our previous study reported frequent deletion (35-38%) and methylation (52%) of SLIT2, located at chr. 4p15.31, in CACX of Indian patients [12]. Whereas, predominant hypermethylation (64%) and infrequent deletion (9%) of the same was reported in CACX of Western patients [13]. In normal non-neuronal cells, SLIT2 is a secreted tumor suppressor protein that through binding of its cognate receptors ROBO1/2 and via inhibition of WNT, SDF1 and HGF signaling regulates multiple cellular processes namely; cell cycle, apoptosis, cell-cell adhesion, cellular motility and invasion etc [14]. Interaction between SLIT2 and ROBO1/2 causes activation of srGAP molecules and consequent inactivation of CDC42 (by hydrolysis of bound GTP), leading to cell cycle arrest at G1-S transition [14]. Similarly, SLIT2-ROBO1/2 interaction relieves the inhibition on DCC receptors by ligand Netrin-1, leading to activation of apoptosis via caspases-3 and 9 [14]. In the event of abrogation of SLIT2 and ROBO1/2 interactions the cellular surveillance posited in the form of G1-S checkpoint and apoptosis gets aberrant, giving rise to mutated forms that has the property of conferring greater growth advantage to the cells. This ultimately results in uncontrolled cellular proliferation and promotion of tumorigenesis.
Alike SLIT2, frequent methylation (46%) and nominal deletion (10%) frequencies of ROBO1 was reported in CACX of Western patients [13]. To the best of our knowledge, alterations of ROBO1 have not been studied in CACX of Indian patients. In addition alterations of ROBO2 localized 1.3 Mb telomeric to ROBO1 at chr 3p12.3 region, have not been studied in details in CACX, though its alterations have already been reported in head & neck squamous cell carcinoma [HNSCC] [15].
Intron2 of ROBO1 harbors two non-coding RNAs [ncRNAs; BC017743 and BC043430], suggesting their probable role in modulating the expression of ROBO1. The expression pattern of these ncRNAs in CACX has not yet been studied but their differential expression has been detected in various carcinomas like; lung, breast, oral etc [15,16]. Therefore, to understand the importance of SLIT2-ROBO1/2 pathway in the development of CACX, it is pertinent to analyze the alterations of these genes in the same set of samples. Thus our study has been focused on the following aspects: (i) quantitative mRNA expression analysis, (ii) alteration (deletion/methylation) analysis and (iii) correlation of molecular alterations of the genes with clinico-pathological parameters (stage, nodes at pathology, HPV status, parity etc) and patient outcome. Our data revealed reduced expression and frequent alterations (37-55%) of SLIT2 and ROBO1/2 in primary cervical lesions. In addition, 84% and 80.5% samples showed co-alterations of SLIT2-ROBO1 and SLIT2-ROBO2 pairs respectively, indicating the importance of these ligand-receptor interactions in cervical carcinogenesis. Moreover Cox multivariate analysis revealed alterations of SLIT2 and ROBO1, in combination with advanced tumor stage (III/IV), multiparity ($5) and early sexual debut (,19 years) as determinants of worse prognosis.

Ethics Statement
The Institutional Ethical board of Chittaranjan National Cancer Institute, Kolkata approved the usage of Human specimens in this study. The above board approved usage of these human clinical samples specifically in this study, pertaining to the involvement of SLIT2-ROBO1/2 signaling in CACX. The tumor specimens were collected from the hospital section of Chittaranjan National Cancer Institute, Kolkata, after obtaining written, informed consent of the concerned patients, in stipulated format, approved by the above mentioned Institutional Ethical board of Chittaranjan National Cancer Institute, Kolkata, India.

Clinical Specimens and Cell lines
This study included a total of 133 primary cervical lesions and corresponding peripheral blood lymphocytes (PBL) collected from the hospital section of Chittaranjan National Cancer Institute, Kolkata, after institutional consent. The tumors were graded or staged according to FIGO classification. Among these samples, 23 were premalignant/CIN lesions (10 low grade CINI and 13 high grade CINII/III), 56 stage I/II tumors and 54 stageIII/IV tumors ( Table 1). The normal cervical tissues (n = 8) were collected from patients with clinically normal cervix, but underwent hystectomy due to other gynecological reasons. These served as controls for RNA/protein study. The tissues (normal/tumors) were either frozen to 280uC or taken in TRIzol reagent (Invitrogen, USA) for RNA isolation or fixed in formalin and paraffin embedded for immuno-histochemistry (IHC). Clinical/follow-up data of the patients were collected from the hospital records. Demographic details were obtained by personally questioning the patients. The CACX cell lines: SiHa and HeLa were purchased from National Centre for Cell Sciences, Pune, India and were grown according to supplier's instructions.

RNA Expression Analysis
Expression of ROBO1/2, SLIT2 RNAs and the ncRNAs were analyzed by Q-RT PCR in normal cervical tissue (n = 8), CACX samples (#T1-#T21) and cell lines: SiHa and HeLa, using primers given in Table S1. RNA isolation and cDNA preparation were done as outlined by Mitra et al (2010) [17] and Information S1. The real time quantitation of RNA expression was done by Power SYBR-green PCR assay (Applied Biosystems, USA) using ddCt method [18,19]. The b2-microglobulin was used as the internal control. Details of this method have been included in Information S1.

5-aza-dC Treatment of SiHa and HeLa
HeLa and SiHa cell lines were grown in presence and absence of 5-Aza-29-deoxycytidine (5-aza-dC) for 3-5 days at 5 mM, 10 mM and 20 mM concentrations. The cells were harvested followed by RNA isolation, cDNA preparation and real time quantitations using Power SYBR-green PCR assay according to protocols described above [18,19].

Microdissection and DNA Extraction
The clinical specimens comprising mostly of biopsy/surgical materials, were frozen, sectioned in cryomicrotome (5 mm), stained with hematoxylin-eosin and then the contaminant normal cells were removed by microdissection [20] under dissecting microscope (Leica MZ 16, Germany). Microdissected samples containing .60% dysplastic epithelium/tumor cells were taken for DNA isolation according to the standard procedure [20,21]. Details in Information S1.

Promoter Methylation Analysis
Methylation sensitive restriction analysis (MSRA) was performed to screen promoter methylation status of SLIT2 and ROBO1/2 in a cohort of 23 CIN and 110 CACX samples using primers listed in Table S1 [22]. Among these samples, 17 CIN and 41 CACX samples have already been analyzed for SLIT2 methylation by a previous study of our laboratory [12]. The methylation sensitive HpaII and its methylation insensitive isochizomer MspI have been used in the analysis. The b-3A adaptin gene (K1) was used as digestion control and RARb2 (K2) served as the control for DNA integrity [22].
Methylation analysis was validated in 20 (2 CIN and 18 CACX) randomly selected samples by methylation-specific-PCR (MSP) after bisulphite modification of DNA using primers listed in Table  S1B. The genomic DNA (5 mg) was subjected to bisulphite modification followed by PCR amplification of the modified DNA using primers for non-methylation (U) or methylation (M) specific alleles [23]. Details of MSRA/MSP protocols have been included in Information S1.

Deletion Analysis
Deletion of the ROBO1/2 and SLIT2 loci were done in 23 CIN and 110 CACX samples, using microsatellite and exonic markers [12,15]. Among these, 17 CIN and 41 CACX samples were previously analyzed only (not ROBO1/2) for SLIT2 deletion [12]. In total, 6 microsatellite and 2 exonic markers [(Ensembl release 49; Genome Database); Details in Table S1] were chosen. Among them D3S3507, D3S1274 and D4S1546 were informative (high polymorphic) and the rest were non-informative. The details of the deletion analysis have been included in Information S1. Scoring of loss of heterozygosity [LOH]/deletion and microsatellite size alteration [MA] for informative/non informative markers were done according to standard protocols [12,20,24] Details in Information S1.

Immunohistochemical and Immunocytochemical Analysis
Protein expression of ROBO1/2 and SLIT2 was studied by immunohistochemistry (IHC) in normal cervical tissue (n = 8) and primary CACX (n = 15) using primary antibodies [Goat polyclonal IgG sc-16611, sc-16615 and sc-16619 for ROBO1, ROBO2 and SLIT2 respectively] and HRP-conjugated rabbit anti-goat secondary antibody (sc-2768) from Santa Cruz Biotechnology, CA, USA, following standard protocols (Information S1). The scoring of staining frequency and intensity was done according to Perrone et al [2006] [25]. The expression of these proteins in SiHa and HeLa cells were studied by immunocytochemistry. The cells were cultured on coverslips till subconfluency, fixed with methanol, blocked with BSA, incubated with respective primary antibodies (as above) followed by fluorescein isothiocyanate (FITC)-conjugated secondary antibody (sc-2777), washed and briefly incubated with DAPI (sc-3598) and subsequently mounted on slides, viewed and photographed by fluorescence microscope (Leica DM 4000B; Germany) [17].

Detection of HPV-16 and HPV-18
Presence of HPV in the cervical lesions were detected by PCR using primers (MY09 and MY11) from the consensus L1 region followed by typing of HPV 16/18, frequent oncogenic variants, in the L1 positive samples [26].

Statistical Analysis
The x 2 analysis determined the association of genetic profile of tumors (alteration of ROBO1/2, SLIT2) with different clinicopathological parameters. Survival curves were obtained according to Kaplan-Meier method. Cox proportional hazards regression model predicted the significant determinants (genetic/epidemiological) of patient's survival. Overall survival (OS) was measured from the date of surgery to the date of most recent follow-up or death (upto 5 years). Consistent follow-up records were available for 86 CACX patients with mean and median follow-up of 14610 months and 10 months respectively. Only these patients were included in the Cox and Kaplan-Meier survival analyses. Probability value (P-value) #0.05 was considered statistically significant. SPSS was used to perform all the statistical analyses (SPSS Inc. Chicago, IL, USA).
Similarly the expressions of the two ncRNAs were also reduced in both primary CACX and cell lines in the following order: BC043430 (5.863.14) . BC017743 (362.9) ( Figure 1A,B). The ncRNA BC017743 showed 2 folds reduction in SiHa and BC043430 showed 26 folds reduction in HeLa, though ROBO1 was not reduced in any of the cell lines. A comparison of expression levels (ddCt) of the ncRNAs versus ROBO1, showed frequent, relatively more reduced expression of ROBO1 (data not shown), in the primary tumors. These facts probably suggest that the ncRNAs and ROBO1 were transcribed from separate promoters.
About, 15.5% (17/110) and 13% (14/110) of the CACX samples showed methylation exclusively for ROBO1 and ROBO2 respectively, indicating that none of the events were epiphenomena (Table S2B). However, significant association was also found between the methylation statuses of ROBO1 and ROBO2 in both CIN and CACX (Table S3A), suggesting their possible cooperativity during cervical carcinogenesis.
In case of SLIT2 and ROBO1 methylation was frequent (20-30%) in premalignant CIN lesions and remained comparable thereafter ( Figure 2D). However, the methylation frequency of ROBO2 increased gradually from CIN to subsequent stages of tumorigenesis, indicating that methylation was a late event for ROBO2 inactivation ( Figure 2D). Thus, promoter methylation of SLIT2 and ROBO1/2 might account for the lowered expression of these genes during cervical carcinogenesis.

Validation of Promoter Methylation of the Genes by 5-Aza Deoxycytidine
To confirm if downregulation of the SLIT2, ROBO1/2 and ncRNAs were due to methylation, demethylation experiment was performed by treatment of SiHa and HeLa cells with 5-Aza Deoxycytidine  (Figure 1D-F). The comparatively high expression of ROBO2 in both cell lines and SLIT2 in SiHa were in concordance with their methylation statuses. Differential increase in expression of ROBO1 and ncRNAs in both the cell lines, re-affirms transcription of these genes from different promoters. Thus, our data validates promoter methylation as one of the inactivating mechanisms of SLIT2 and ROBO1/2 in CACX development.
In CACX, deletion of ROBO2 showed significant association with ROBO1 and SLIT2 (Table S3B), indicating their correlation in this carcinogenesis. The deletion of ROBO2 is not an epiphenomenon of deletion of ROBO1 as 11% (12/110) of the CACX samples showed deletion only of ROBO2 (Table S2B). Microsatellite size alterations (MA) of ROBO1 and SLIT2 were infrequent in cervical tumors (Table S2A-D), unlike HNSCC (15). No homozygous deletion but other types of biallelic alterations [LOH+MA, MAII] was evident for SLIT2 and ROBO1 loci (Table  S2A-D). The deletion frequency of SLIT2 increased gradually during tumor progression ( Figure 2D). However, in case of ROBO1 there was significant increase in deletion frequency from CIN to stage I/II tumors and for ROBO2, frequency of deletion was high in stage I/II tumors and remained comparable in the subsequent stage ( Figure 2D). These facts indicate that deletion was a late event during inactivation of the receptor genes and contributed towards inactivation of both SLIT2 and ROBO1/2 in CACX.
Significant association between deletion and methylation of SLIT2 and ROBO1 in CIN, ROBO1 and ROBO2 in CACX and marginal significance of SLIT2 in CACX were in concordance with Knudson's modified two-hit hypothesis for candidate TSGs (Table S4).
The alterations of SLIT2 were frequent (35%) in CIN lesions and increased gradually during subsequent stages of tumorigenesis ( Figure 2E). However, for both ROBO1 and ROBO2, alteration frequencies increased significantly from CIN to stage I/II tumors and remained comparable thereafter ( Figure 2E), indicating that alterations of the receptors were late event during cervical carcinogenesis.
Overall alterations of ROBO1 and ROBO2 showed significant association in CIN and CACX, whereas SLIT2 and ROBO1 in CACX, suggesting their functional cooperativity in regulating the signaling pathways (Table S3C). Frequent alterations of SLIT2 and/or ROBO1 were seen in CIN (44%; 10/23) followed by significant increase in stage I/II (93%; 52/56) and stage III/IV tumors (93%; 50/54) ( Figure 2E). Similar was the trend for SLIT2 and/or ROBO2 alterations ( Figure 2E). This suggests that deregulation of these ligand-receptor interactions might have important role in the development of both premalignant cervical lesions and early invasiveness of the disease.

Analysis of SLIT2 and ROBO1/2 Protein Expression
The association of alterations (deletion/methylation) and RNA expressions of SLIT2 and ROBO1/2 with their protein expressions was studied in primary CACX and the cell lines: SiHa and HeLa. In IHC analysis, expression of ROBO1/2 and SLIT2 were observed in the membrane and cytoplasm of the basal and parabasal layers of normal cervical epithelium and comparatively low expression in the spinous layer ( Figure 3AI-III). The primary tumors without alterations (deletion/methylation) of these genes showed expression of these proteins similar to the basal/parabasal layers. However, significant association was seen between reduced/no expression of the proteins with their molecular alterations ( Figure 3BI-III, Table 2). SLIT2 and ROBO2 expressions were reduced in SiHa and HeLa cells. Expressions of ROBO1 and ROBO2 were observed with some discrete foci on the membrane of SiHa cells ( Figure 3CI-III). Similar localizations of these proteins were seen in HeLa, without any focus formation ( Figure S1). This might be due to some modifications of these receptors in SiHa, leading to their aggregation on the membrane and subsequent dysfunction of the downstream pathway(s). Cytoplasmic expression of SLIT2 was observed in SiHa cells, in concordance with the reports of Singh et al [2007] (12), in HeLa. In primary tumors, reduced/no expression of SLIT2 and/or ROBO1 was found in 93% (14/15) samples and of SLIT2 and/or ROBO2 in 67% (10/15) ( Table 2), indicating the deregulation of these ligand-receptor(s) interactions as critical events in cervical carcinogenesis.
Clinico-pathological Association of SLIT2 and ROBO1/2 The alterations of SLIT2 and ROBO1/2 were correlated with the clinico-pathological/survival parameters in order to uncover their prognostic significance (if any) for early diagnosis and prediction of patient outcome. HPV, an important causative agent of CACX, was detected in 86.5% (115/133) of the cervical lesions. Of these, 88% (101/115) were HPV16 positive and 12% (14/115) were HPV18 positive. HPV infection was not significantly associated with tumor stage, grade, nodal status, parity, age at sexual debut (Table 1) and SLIT2-ROBO1/2 alterations (data not shown). Similarly, chi-square analysis revealed significant association of SLIT2-ROBO1/2 alterations with tumor stage, however no such associations were observed with the other clinico-pathological parameters (grade, nodal status, parity etc; data not shown).
The Kaplan-Meier (K-M) survival analysis revealed significantly reduced overall survival (OS) of CACX patients with alterations of SLIT2, ROBO1 and ROBO2 ( Figure S2A-C), indicating their prognostic significances. Interestingly, alterations of SLIT2 and/or ROBO1 and SLIT2 and/or ROBO2 predicted poor OS of the patients, suggesting abrogation of these ligand-receptor interactions were directly equated to poor prognosis ( Figure 4A,B). The Cox multivariate analysis indicated that alterations of SLIT2 and ROBO1 along with advanced tumor stage (III/IV), multiparity ($5) and early sexual debut (,19 years) were determinants of poor prognosis for CACX patients ( Figure 4C), thereby enabling efficient classification of the high-risk patients.

Discussion
The aim of this study was to investigate the association of ligand-receptor genes SLIT2 and ROBO1/2 in cervical carcinogenesis. To this end, at first, expression (RNA) profiles of these genes were analyzed in primary CACX samples and CACX cell lines. Highly reduced expressions of these genes were revealed. Then alterations (deletion/methylation) of the genes were analyzed to understand the mechanism of reduced expression. Alterations of the genes were then correlated with their expression (RNA and protein) profile. In addition molecular alterations of the genes were correlated with different clinicopathological parameters. Reduced RNA and protein expression of SLIT2 and ROBO1/2 were observed in primary CACX, in concordance with the previous reports [12,13]. However, none prior to us reported the expression pattern of SLIT2, ROBO1 and ROBO2 concomitantly in primary CACX. Alike us reduced expression of these genes were also reported in a plethora of other malignancies including HNSCC, gliomas and carcinomas of liver, lung, breast, kidney etc [15,[27][28][29][30][31][32][33]. However, overexpression of these genes were also reported in prostate and breast cancers [34,35]. In addition, Angeloni et al., (2006) [16] suggested that expression of ROBO1 could be regulated by two ncRNAs encoded by the intron2 of ROBO1. However, our data suggested that there was no concordance between the expression profiles of ROBO1 and the ncRNAs, as observed in HNSCC [15].
Overall alterations of atleast one member of the SLIT2-ROBO1/ 2 cascade were observed in 43% CIN lesions and 95% CACX, indicating the importance of deregulation of this pathway for cervical carcinogenesis. Alterations of SLIT2 was observed in 35% CIN lesions with the frequency increasing gradually in subsequent stages, indicating this as an early event, in concordance with the reports of Narayan et al., (2006) [13]. In addition deletion or methylation mediated deregulation of SLIT2 has already been reported in various malignancies including gliomas, neuroblasto-ma, Wilm's tumor, leukemia and carcinomas of lung, breast, kidney, colon etc though, none reported the stage wise correlation pattern [15,[28][29][30]33,[36][37][38][39][40]. The alteration frequencies of ROBO1 and ROBO2 increased significantly from CIN to stage I/ II tumors, indicating that these receptors were inactivated during the development of early invasive cervical tumors. Alike SLIT2, deletion or methylation mediated inactivation of ROBO1 have also been reported in multiple malignancies including lung, breast, HNSCC etc [15,27,41]. However, alterations of ROBO2 have been reported only in HNSCC, as yet [15].
In premalignant CIN lesions, alterations of SLIT2 and/or ROBO1 and SLIT2 and/or ROBO2 ligand-receptor pairs were observed in 44% and 39% cases respectively, which increased significantly in the subsequent stage I/II tumors and remained comparable thereafter. This fact corroborates that abrogation of the ligand-receptor interaction was necessary for initiation and progression of CACX. As yet none has comprehensively reported the importance of loss of these ligand-receptor interactions in cervical carcinogenesis of Indian patients.
A close scrutiny of the methylation and deletion frequencies of SLIT2 and ROBO1 during progression of CACX, indicated that methylation was the more frequent alteration that occurred early during carcinogenic progression. Although, in case of ROBO2, promoter methylation was the late event nonetheless, it was predominant over deletion. Predominant methylation and in-frequent deletion of SLIT2 (64%, 9%) and ROBO1 (46%, 10%) were also reported in CACX of Western patients [13]. The deletion analysis was done in a cohort of 30 tumors compared to 133 in our analysis, thus accounting for the difference in results between the two studies [42]. Similarly the methylation frequency of SLIT2 in our study was 34% and differed widely from that reported by Narayan et al., (2006) [13], possibly due to age, ethnicity and lifestyle variation of the patients included in both the studies [42][43][44].
The interaction between SLIT2 and ROBO1 triggers a number of downstream pathways, controlling diverse cellular processes namely; cell proliferation, apoptosis, motility etc [14]. The most prominent effectors includes srGAP, CDC42, b-catenin, Netrin-1, PI3K, CXCR4 etc [14,45]. Detailed study of each of these pathways is warranted for identification of diagnostic or therapeutic targets of CACX, from among the effector molecules of the SLIT2-ROBO1 signaling pathway.
In this study molecular alterations of the individual genes, SLIT2 and/or ROBO1 and SLIT2 and/or ROBO2 pairs showed differential association with the disease progression and predicted poor patient outcome. Moreover Cox multivariate analysis revealed that alterations of SLIT2 and ROBO1 coupled with advanced tumor stage (III/IV), multiparity ($5) and early sexual debut (,19 years) were predictors of poor prognosis for CACX patients.
Thus we can conclude that abrogation of SLIT2-ROBO1 and SLIT2-ROBO2 signaling pathways are necessary for the initiation and progression of CACX. Since these ligand-receptor genes possess prognostic implications, they could be utilized as early diagnostic or prognostic markers for CACX.