SOX4, SOX11 and PAX6 mRNA expression was ...

RESEARCH ARTICLE For reprint orders, please contact: [email protected]

SOX4, SOX11 and PAX6 mRNA expression was identified as a (prognostic) marker for the aggressiveness of neuroendocrine tumors of the lung by using next-generation expression analysis (NanoString) Robert Fred Henry Walter‡,1,2, Fabian Dominik Mairinger‡,2, Robert Werner2, Saskia Ting2, Claudia Vollbrecht3, Dirk Theegarten2, Daniel Christian Christoph4, Konstantinos Zarogoulidis5, Kurt Werner Schmid2, Paul Zarogoulidis*,5 & Jeremias Wohlschlaeger2

ABSTRACT Background: Neuroendocrine tumors of the lung (NELC) account for 25% of all lung cancer cases and transcription factors may drive dedifferentiation of these tumors. This study was conducted to identify supportive diagnostic and prognostic biomarkers. Materials & methods: A total of 16 TC, 13 AC, 16 large cell neuroendocrine carcinomas and 15 small cell lung cancer were investigated for the mRNA expression of 11 transcription factors and related genes (MYB, MYBBP1A, OCT4, PAX6, PCDHB, RBP1, SDCBP, SOX2, SOX4, SOX11, TEAD2). Results: SOX4 (p = 0.0002), SOX11 (p < 0.0001) and PAX6 (p = 0.0002) were significant for tumor type. Elevated PAX6 and SOX11 expression correlated with poor outcome in large cell neuroendocrine carcinomas and small cell lung cancer (p < 0.0001 and p = 0.0232, respectively) based on survival data of 34 patients (57%). Conclusion: Aggressiveness of NELC correlated with increasing expression of transcription factors. SOX11 seems to be a highly valuable diagnostic and prognostic marker for aggressive NELC. Neuroendocrine tumors of the lung (NELC) include four heterogenic subtypes; well-differentiated typical carcinoids (TC), intermediate atypical carcinoids (AC) and poorly differentiated large cell neuroendocrine carcinomas (LCNEC) and small cell lung cancer (SCLC) [1–3] . All four entities together may account for 20–25% of all lung cancer cases [2] . The 5-year survival of neuroendocrine lung tumor subgroups differs significantly with TC having the best survival rates (>87%) [4,5] , followed by AC with more than 60% [2,6] , LCNEC with 15–57% [2,7] and SCLC with less than 5% [2,8] showing that the aggressiveness of NET depends on the clinicopathological subtype. Embryonic and neuronal development is driven by transcription factors and deregulations of these transcription factors are associated with congenital disorders, dysplasia and cancer development [9–16] .

KEYWORDS

• carcinoids • large cell lung cancer • lung cancer • mRNA • NanoString nCounter • neuroendocrine • small

cell lung cancer

Ruhrlandklinik, West German Lung Center, University Hospital Essen, University of Duisburg-Essen, Essen, Germany Institute of Pathology, University Hospital Essen, University of Duisburg-Essen, Essen, Germany 3 Institute of Pathology, University Hospital Cologne, Cologne, Germany 4 Department of Medical Oncology, West German Cancer Center, University Hospital Essen, University of Duisburg-Essen, Essen, Germany 5 Pulmonary Department, Oncology Unit, ‘G. Papanikolaou’ General Hospital, Aristotle University of Thessaloniki, Thessaloniki, Greece *Author for correspondence: Tel.: +30 697 727 1974; Fax: +30 231 099 2433; [email protected] ‡ Authors contributed equally 1 2

10.2217/FON.15.18 © 2015 Future Medicine Ltd

Future Oncol. (2015) 11(7), 1027–1036

part of

ISSN 1479-6694

1027

Research Article Walter, Mairinger, Werner et al. SOX transcription factors share a HMG domain (high-mobility group domain [DNAbinding domain]) and seem to be found in animals only [10,15,17] . For SOX transcription factors little is known about the interactome with other proteins, but recent data indicate that SOX members associate with a large number of other proteins to mediate activation of gene expression [16] . During embryogenesis [18] , SOX transcription factors show specific expression depending on the developmental status and in a tissue-specific context [16,17] . The SOX family comprises eight subgroups, A–H and member of one subgroup show high similarity in their structure, whereas members of different subgroups show lower similarity [10] . SOX4 and SOX11 belong to the SOX-C group that share high homology in their DNA-binding domain and C-terminus [10,19] . SOX4 and SOX11 seem to function as oncogenes in lung cancer and especially in SCLC high expression of both transcription factors was reported [19] . Knock down of SOX4 in lung cell lines induced apoptosis and reduced cell proliferation [19] , whereas a knockdown of SOX11 results in increased proliferation in a mantle cell lymphoma cell line [20] . That indicates that SOX transcription factors may play a tissue specific role as either oncogenes or tumor suppressors. PAX transcription factors belong to the paired box gene family and regulate developmental processes [9,21–22] . PAX6 is a member of this family and loss of function studies show that it is involved in the development of the eye, pancreatic Langerhans islet cells and neuroendocrine cells [21–23] . Additionally, PAX6 function as either oncogene or tumor suppressor seems to be tissue specific and is discussed controversially [9,21,24] . However, in non-SCLC cell lines inhibition of PAX6 hampered proliferation indicating that PAX6 expression in lung drives tumor maintenance [24] . Coutinho et al. reported 200 predicted target genes of PAX6 that are highly conserved in different species showing that the PAX6 interactome is highly complex and that merely a little is known about PAX6 function in developmental and tumorigenic processes [23] . After analysis of recent literature regarding cancerogenesis and association with transcription factors, this study was conducted to investigate the mRNA expression of eleven transcription factors and related genes (MYB, MYBBP1A, OCT4, PAX6, PCDHB, RBP1, SDCBP, SOX2, SOX4, SOX11, TEAD2) for

1028

Future Oncol. (2015) 11(7)

their impact on aggressiveness of neuroendocrine tumors of the lung. Searching the PubMed section of the National Center for Biotechnology Information [25] website for the Medical Subject Headings (MeSH) for the investigated genes and lung cancer (e.g., ‘SOX11’ and ‘lung’ resulting in 23 publications, not all focusing on lung cancer and SOX11) returns a tiny amount of related publications. Therefore, we saw a pressing need to investigate transcription factors and related genes in neuroendocrine tumors of the lung. For this approach, the nCounter technology was used. The method allows enzyme-free, digital detection of mRNAs [26,27] . Each investigated nucleic acid is targeted by a capture and a reporter probe that have a complementary region of approximately 50 nucleotides (nt) to the transcript of interest [26] . Each capture probe contains a biotin and each reporter probe contains a barcode of six fluorophores that is unique for each targeted sequence [26] . For mRNA analysis, total RNA and a mixture of target-specific capture and reporter probes are allowed to hybridize before excess probes are removed by affinity purification [26] . Afterwards, the RNA-probe hybrids are washed across a streptavidin-coated surface to capture the biotin-labeled probes [26] . Then, an electric field is applied to orient all hybrids in one direction. The fluorescent barcodes are imaged and the software analyses the amount of detected transcripts [26] . In this way, up to 800 different targets can be detected simultaneously [26,28] . Reis et al. showed that the nCounter technique is able to analyze mRNA from formalinfixed, paraffin-embedded (FFPE) tissue with results similar to fresh-frozen tissue [29] . The aim of this study was to identify supportive diagnostic and prognostic biomarkers in neuroendocrine lung tumors on the mRNA level derived from FFPE tissue. Material & methods Representative specimens of each tumor entity (16 typical and 13 atypical carcinoids, 16 large cell neuroendocrine lung cancer and 15 SCLC) were used for mRNA expression analysis. Patients that received chemotherapy before resection of tumor tissue specimens were excluded. The initial diagnosis was reevaluated by two experienced pathologists (J Wohlschlaeger and T Hager). Additional inclusion criteria were sufficient tumor material and minimal contamination by benign and stromal cells of the FFPE tissue. Specimens were collected from the tumor bank at the

future science group

SOX4, SOX11 & PAX6 are (prognostic) biomarkers in neuroendocrine lung tumors Department of Pathology and Neuropathology at the University Hospital Essen, Germany from 2005 till 2012. TNM-staging was based on the WHO Classification Of Tumours guidelines (2004) [30] and clinical data were obtained from the patients’ records. The study was conducted retrospectively for the identification of biomarkers to identify deregulation in neuroendocrine tumors regarding apoptosis and cell cycle. The study design was approved by the ethical committee of the University Hospital Essen (ID: 13-5382-BO). The investigation conforms to the principles outlined in the declaration of Helsinki. ●●RNA extraction & RNA integrity

assessment

Three to five paraffin sections with a thickness of 4 μm per sample were deparaffinized with xylene prior to RNA extraction using the RNeasy FFPE kit (Qiagen, Hilden, Germany) according to the manufacturer’s recommendations. RNA concentration was measured using a Nanodrop 1000 instrument (Thermo Fisher Scientific, MA, USA). RNA integrity was assessed using an Agilent 2100 bioanalyzer (Agilent Technologies, CA, USA) at the NanoString Core Facility at the University of Heidelberg (Germany). Smear analysis was performed using the Agilent 2100 expert software to determine the proportion of RNA ≥300 nt within a given sample. ●●nCounter CodeSet design & expression

quantification

Various relevant genes of different tumor-associated signaling pathways and neuroendocrine differentiation were included in the CodeSet. The CodeSet was designed to contain a total of 91 genes with different signature genes for each subgroup. The investigated genes and corresponding pathways encompassed the following: SOX signaling (MYB, MYBBP1A, OCT4, PAX6, PCDHB, RBP1, SDCBP, SOX2, SOX4, SOX11, TEAD2); MET pathway (GAB1, GRB2, MET, MST1R, PAX5); mTOR signaling (MTOR, RHOA, RICTOR, RPTOR); angiogenesis (CRHR2, FIGF, FLT4, HIF1A, KDR, MMP3); apoptosis (ASCL1, BAX, BCL2, CASP8, CASP-10, FAS, MDM2, TP53, PNN); neuroendocrine differentiation (CHGA, GABBR2, NCAM1, NTS, RTN1, SEMA3B, SYP); folate metabolism (ATIC, DHFR, FOLR1, FPGS, GART, GGT1, SLC19A1, TYMS); DNA-repair (ERCC1, MLH1, MSH2, MSH6, XRCC1); cell cycle regulation (CCND1, CCNE1, CDK1,


Research Article

CDK2, CDK4, CDK6, CDKN1A, CDKN1B, CDKN2A, MIB-1); tumor environment (LDHA, LDHB, LDHC, MAN2A1, MAN2B1, MAN2C1, TKTL1); growth factors (GF), GF receptors and downstream signaling (IGF-1, IGF-2, MET, EGFR, FGFR1, AKT1, ALK, PTEN ) as well as CAT, CYP1A1, FN1, NES, NKX21, SOD1, STK11, TWSG1, UCHL1. Three reference genes (ACTB, GAPDH and HPRT1) were also included in the CodeSet for biological normalization. The probes were designed to bind to the CDS (coding sequence) of the investigated genes to avoid a 5′ and/or 3′ bias and to compensate for degradation artefacts of the mRNA molecule tails. An overview of all genes is given in Table 1. Probe sets for each gene in the CodeSet were designed and synthesized at NanoString Technologies (WA, USA). Total RNA (100 ng) including miRNA from FFPE material was analyzed. ●●nCounter data processing & statistical

analysis

Raw mRNA counts for each gene were subjected to a technical normalization considering the counts obtained for positive control probe sets. After the technical normalization, a biological normalization using the three reference genes included in the CodeSet was performed. All statistical analyses were performed with the R statistical programming environment (v2.15.2). For dichotomous factors such as gender and expression level the Wilcoxon Mann–Whitney rank-sum test was applied. The Kruskal–Wallis test was used to correlate tumor type and gene expression. The same test was also used to perform subgroup analysis between low- and high-grade tumors, atypical and typical carcinoids and SCLC versus LCNEC. Correlations between gene expression and TNMstages were analyzed by Spearman’s rank correlation test. Significant differences in OS between groups were tested by using the COXPH-model (Wald-test, likelihood-ratio test and Score (logrank) test). Kaplan-Meier curves were created to visualize associations between gene expression and overall survival (OS). The level of statistical significance was defined as p < 0.05. Results Sixty tumor specimens were investigated including 16 typical (27%) and 13 atypical carcinoids (22%), 16 large cell neuroendocrine lung cancer (27%) and 15 SCLC (25%) cases. A total of 27 female (45%) and 25 male patients (42%)

www.futuremedicine.com

1029

Research Article Walter, Mairinger, Werner et al. Table 1. The investigated mRNA targets and associated cell processes are summarized. Cell process

Investigated targets

SOX signaling MET pathway mTOR signaling Angiogenesis Apoptosis Neuroendocrine differentiation Folate metabolism DNA repair Cell cycle regulation Tumor environment Growth factors and signaling Additional genes Reference genes

MYB, MYBBP1A, OCT4, PAX6, PCDHB, RBP1, SDCBP, SOX2, SOX4, SOX11, TEAD2 GAB1, GRB2, MET, MST1R, PAX5 MTOR, RHOA, RICTOR, RPTOR CRHR2, FIGF, FLT4, HIF1A, KDR, MMP3 ASCL1, BAX, BCL2, CASP8, CASP10, FAS, MDM2, TP53, PNN CHGA, GABBR2, NCAM1, NTS, RTN1, SEMA3B, SYP ATIC, DHFR, FOLR1, FPGS, GART, GGT1, SLC19A1, TYMS ERCC1, MLH1, MSH2, MSH6, XRCC1 CCND1, CCNE1, CDK1, CDK2, CDK4, CDK6, CDKN1A, CDKN1B, CDKN2A, MIB1 LDHA, LDHB, LDHC, MAN2A1, MAN2B1, MAN2C1, TKTL1 IGF1, IGF2, EGFR, FGFR1, AKT1, ALK, PTEN CAT, CYP1A1, FN1, NES, NKX21, SOD1, STK11, TWSG1, UCHL1 ACTB, GAPDH, HPRT1

were investigated. For eight patients the gender remained inconclusive. Regarding tumor type SOX4 (p = 0.0002), SOX11 (p < 0.0001) and PAX6 (p = 0.0002) showed the highest significances of the investigated mRNAs. Increasing SOX4 expression correlated directly with increasing malignancy of the tumor (Figure 1) . Elevated SOX11 and PAX6 expression was mainly found in SCLC (Figure 1) . Interestingly, one AC showed PAX6 expression of 2646 mRNA counts and SOX11 expression of 449 mRNA counts and this patient showed rapid progression of the disease and succumbed A 40,000

to the disease within one year. In this particular patient, a limited disease of a SCLC was diagnosed in 2002 and pneumonectomy of the left lung was performed. In 2006 this patient was diagnosed with an AC that was confirmed by three independent pathologists. Nevertheless, these biomarkers identified this AC to be related to the SCLC-group and the clinical behavior substantiates that finding. Additionally, SOX11 was highly expressed in three LCNEC and also one of these showed high PAX6 expression. Unfortunately, for these patients no follow-up data were available.

B

C p < 0.0001

p = 0.0002

2500

20,000

10,000

0

PAX6 mRNA counts

30,000

SOX11 mRNA counts

SOX4 mRNA counts

2500 2000 1500 1000 500

2 AC 3 LCNEC 4 SCLC Tumor subtype

2000 1500 1000 500 0

0 1 TC

p = 0.0002

1 TC


1 TC


Figure 1. Boxplots for tumor subtype in correlation with mRNA counts that were measured using the nCounter technology (NanoString). On the y-axis mRNA counts of the target gene are shown. The counts are directly proportional to the transcript abundance. On the x-axis the four subtypes of neuroendocrine lung tumors (TCs and ACs, LCNEC and SCLC) are shown. (A) SOX4 expression increased with increasing malignancy of the tumor. (B) SOX11 and (C) PAX6 expression was mainly found in SCLC. AC: Atypical carcinoid; LCNEC: Large cell neuroendocrine carcinoma; SCLC: Small cell lung cancer; TC: Typical carcinoid.

1030

Future Oncol. (2015) 11(7)


SOX4, SOX11 & PAX6 are (prognostic) biomarkers in neuroendocrine lung tumors

B

1.0

Absent PAX6 expression

0.8 0.6 0.4

Detectable PAX6 expression

0.2 0.0

Score (logrank) test; p < 0.0001 0

10

20

30

40

50

Time (months)

SOX11-dependent overall survival

PAX6-dependent overall survival

A

Research Article

1.0

Absent SOX11 expression

0.8 0.6 0.4 0.2

Detectable SOX11 expression

0.0

Score (logrank) test; p = 0.0232 0

10

20

30

40

50

Time (months)

Figure 2. Kaplan–Meier plot for overall survival in correlation to mRNA expression are shown. On the x-axis the survival time in months is shown. The y-axis shows survival rates in percentage. Elevated expression of (A) PAX6 and (B) SOX11 correlated with decreased survival rates. Absent mRNA expression correlated with prolonged overall survival.

The mean age at date of diagnosis was 58.58 years (median age: 58.99 years; 95% CI: 50.79–66.92 months). Survival data were available for 34 patients with eight reported deaths at the time of data collection. The Score (log-rank) test revealed that elevated PAX6 expression (p