- Email: [email protected]
Gene expression profiling of primary canine insulinomas and their metastases
The Veterinary Journal 197 (2013) 192–197
Contents lists available at SciVerse ScienceDirect
The Veterinary Journal journal homepage: www.elsevier.com/locate/tvjl
Gene expression profiling of primary canine insulinomas and their metastases Floryne O. Buishand a,⇑, Jolle Kirpensteijn a, Alexandra A. Jaarsma a, Ernst-Jan M. Speel b, Marja Kik c, Jan A. Mol a a b c
Department of Clinical Sciences of Companion Animals, Faculty of Veterinary Medicine, Utrecht University, Yalelaan 108, 3584 CM Utrecht, The Netherlands Department of Pathology, GROW-School for Oncology and Developmental Biology, Maastricht University Medical Centre, PO Box 5800, 6202 AZ Maastricht, The Netherlands Department of Pathobiology, Division Pathology, Faculty of Veterinary Medicine, Utrecht University, Yalelaan 1, 3584 CL Utrecht, The Netherlands
a r t i c l e
i n f o
Article history: Accepted 20 January 2013
Keywords: Insulinomas Canine Gene expression Microarrays Amphicrine Lipase
a b s t r a c t The gene expression profile of 10 primary canine insulinomas was compared with that of their accompanying metastases using microarray analysis and quantitative real time-PCR. Analysis of microarray data revealed 84 genes that were differentially expressed between primary insulinomas and their metastases, along with 243 genes differentially expressed between a low-metastatic and a high-metastatic subset of primary insulinomas. The genes differently expressed between primary insulinomas and their metastases clustered together in nine signalling pathways. Comparing the low-metastatic to the high-metastatic subset of primary insulinomas, 26 pathways appeared to be significantly influenced. The acinar enzymes pancreatic lipase (PNLIP) and chymotrypsinogen B1 (CTRB1) were amongst the most down-regulated genes in the malignant group of primary insulinomas and in metastases. Immunofluorescence demonstrated co-localisation of insulin and PNLIP in tumour cells. Different subsets of canine insulinomas can be identified on the basis of their gene expression profile. Canine insulinomas appear to contain amphicrine cells, which exhibit both endocrine and exocrine cell features. Ó 2013 Elsevier Ltd. All rights reserved.
Introduction Insulinomas, which cause clinical signs associated with hypoglycaemia, are the most common malignant pancreatic endocrine tumours in the dog. Although dogs treated by surgery survive longer than dogs treated with medication alone, the prognosis after surgery for insulinomas is guarded. Signs of hypoglycaemia often recur after surgery, because it may only be possible to achieve partial resection of tumour tissue or due to the presence of functional (micro)metastases (Tobin et al., 1999; Buishand et al., 2012). Since there are significant side effects in dogs with insulinomas following treatment with diazoxide, glucocorticoids, streptozocin or octreotide, new therapeutic strategies are warranted to improve prognosis in dogs with insulinomas (Steiner and Bruyette, 1996; Moore et al., 2002). In a previous study, we used quantitative real time-PCR (qPCR) to compare the expression of 16 candidate genes between primary canine insulinomas and their metastases (Buishand et al., 2012). Metastases of canine insulinomas had significantly increased expression of growth hormone (GH) and insulin-like growth factor-I (IGF-I). In addition to GH and IGF-I, other genes may be differen⇑ Corresponding author. Tel.: +31 30 2537563. E-mail address: [email protected] (F.O. Buishand). 1090-0233/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tvjl.2013.01.021
tially regulated between primary canine insulinomas and their metastases. The aim of the present study was to identify altered expression of genes and deregulation of gene signalling pathways in primary canine insulinomas and their corresponding metastases using a canine specific cDNA microarray representing 20,160 genes. Materials and methods Patients and tissue specimens Twenty-one primary canine insulinomas and 10 accompanying metastases (seven lymph node and three liver metastases) were resected at the Faculty of Veterinary Medicine, Utrecht University, from 1993 to 2009 (Table 1). Portions of tissue (5–15 mm diameter) for RNA extraction were removed from the central portion of the tumours, immediately frozen in liquid nitrogen and stored at 70 °C. Tissue fragments were fixed in 10% neutral-buffered formalin for 24–36 h for histopathology. After routine processing and paraffin embedding, sections were cut at 5 lm and stained with haematoxylin and eosin. RNA isolation and cRNA synthesis Total RNA was isolated using the RNeasy Mini Kit (Qiagen). To prevent contamination of the samples with genomic DNA, an on-column DNase treatment was performed. RNA concentrations were quantified by spectrophotometry (NanoDrop ND-1000, Isogen Life Sciences). Total RNA integrity was assessed using the Agilent BioAnalyzer 2100. Only RNA samples with an RNA integrity number (RIN) > 9.0 were selected for cRNA amplification. Two micrograms of total RNA from each sample were amplified by in vitro transcription using T7 RNA polymerase (Rao et al.,
193
F.O. Buishand et al. / The Veterinary Journal 197 (2013) 192–197 Table 1 Clinicopathological characteristics of insulinoma samples used in this study. Dog
Breeda
Sex
Glucose (mmol/L)
Specimen type
Diameter of tumour mass (cm)
TNM stageb
Microarray
ICH insulin (positive cells)c
ICH PNLIP (positive cells)c
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 9 11 13 14 15 16 17 19 20 21
Crossbreed Bouvier des Flandres Bearded collie Beagle West Highland white terrier German short haired pointer Boxer West Highland white terrier Bearded collie1 Irish setter Labrador retriever2 Belgian shepherd dog Crossbreed3 Rough collie4 German shepherd dog5 Anatolian shepherd dog6 Boxer7 Jack Russell terrier Rough collie8 Maltese9 Kooiker dog10 Bearded collie1 Labrador retriever2 Crossbreed3 Rough collie4 German shepherd dog5 Anatolian shepherd dog6 Boxer7 Rough collie8 Maltese9 Kooiker dog10
M Fx Mx Fx F M Fx M Fx Mx Mx Mx Fx Mx F Mx Fx Fx F M F Fx Mx Fx Mx F Mx Fx F M F
3.0 2.2 1.7 1.1 2.5 2.9 3.1 3.2 2.7 2.8 1.9 2.2 3.2 2.2 2.5 2.0 2.6 1.7 1.9 2.3 3.1 2.7 1.9 3.2 2.2 2.5 2.0 2.6 1.9 2.3 3.1
T T T T T T T T T T T T T T T T T T T T T N N N M M N N M N N
1.0 0.6 2.5 1.5 1.2 2.7 3.5d 1.5 1.5 4.0 4.0 1.5 0.8 5.0c 5.0 9.0d 2.5 3.5 4.0 1.0 1.5 1.0 4.0e 2.0 6.0f 5.0 2.0 4.0e 3.5 1.0 1.0
I III IV I I II II I III II III III III IV IV IV III II IV III IV – – – – – – – – – –
No No No No No No No No Yes No Yes No Yes Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
5 4 1 5 4 4 4 5 5 5 4 3 4 2 3 1 3 3 5 5 5 – 5 4 3 2 4 3 5 5 4
2 1 1 2 1 1 0 0 0 1 0 2 1 0 1 0 0 1 0 0 1 – 0 1 0 1 0 1 0 0 0
M, male; F, female, Mx, neutered male; Fx, neutered female; IHC, immunohistochemistry; PNILP, pancreatic lipase; T, primary tumour; N, lymph node metastasis; M, liver metastasis. a The same number indicates a metastasis (N or M) matched to a primary tumour from the same animal. b Staging was performed according to Buishand et al. (2010). c Grading proportion of IHC positive cells: 0, 0%; 1, <10%; 2, <24%; 3, 25–49%; 4, 50–74%; 5, 75–100%; –, not determined. d Two primary insulinomas present. e Multiple lymph node metastases present. f Multiple metastases present, diffusely spread through the liver.
2008). Three micrograms of amplified mRNA were coupled to Cy3 and Cy5 fluorophores (GE Healthcare). A pool of 60 lg amplified cRNA was formed, containing 3 lg of each amplified cRNA sample. Half of the cRNA samples from both the primary insulinoma group and the metastasis group were labelled with Cy5 and the other half from each group were labelled with Cy3. Equal quantities of pooled cRNA were also labelled with Cy5 and Cy3. ChromaSpin-30 columns (ClontechTakara Bio) were used to purify labelled cRNA and dye incorporation efficiency was measured by spectrophotometry (NanoDrop). Equal quantities of labelled cRNA samples with specific activity of 2–5% dye-labelled nucleosides were used for microarray hybridisation.
amplicons. A reaction solution was prepared for each gene using SYBR Green Supermix (Bio-Rad). The housekeeping genes HNRPH and RPL8 were used as reference genes for normalisation of target gene expression. qPCR was performed using the Bio-Rad My-iQ Single Color Real-Time PCR Detection System (Bio-Rad), with three-step reactions (denaturation, annealing and elongation) for RPL8 and two-step reactions (denaturation and annealing) for other primer sets (Brinkhof et al., 2006). All reactions were performed in duplicate and their efficiencies were assessed by a dilution series of pooled cDNA samples tested in each run. Efficiency cut-offs were 95–105%. Omission of the reverse transcriptase reaction showed no significant contamination with genomic DNA.
Microarray hybridisation
Immunohistochemistry
Gene expression profiling was performed using UltraGAPS slides (Corning) spotted with a canine specific collection of 20,160 non-redundant clones of 30 untranslated region cDNA fragments (Rao et al., 2008). Microarray hybridisation, high-stringency washing and scanning with an Agilent G2565AA DNA Microarray Scanner were performed according to Rao et al. (2008). The microarray specifications and derived data are accessible through National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO)1 accession numbers GPL5117 and GSE32934.
Paraffin-embedded blocks were available for immunohistochemistry for all 21 primary insulinomas and 9/10 metastases included in this study. Sections of normal canine pancreatic tissue were used as positive controls, while omission of primary antibodies served as negative controls. Rabbit polyclonal antibodies directed against canine PNLIP (a generous gift from Professor J.M. Steiner, College of Veterinary Medicine, Texas A&M University, Texas, USA) were used to detect expression of PNLIP protein. After deparaffinisation, sections were treated with 0.35% H2O2 in Tris buffered saline (TBS) for 15 min to block endogenous peroxidase activity. Slides were rinsed in TBS containing Tween-20 (TBST) and blocked for 30 min with 10% normal goat serum. Sections were then incubated for 30 min at room temperature with the primary antibody diluted 1:500 in TBST. The slides were rinsed with TBST and then incubated for 30 min with Envision peroxidase-conjugated anti-rabbit IgG (Dako). The slides were then rinsed with TBS and peroxidase activity was visualised following incubation with 3,3-diaminobenzidine. Sections of tissue were counterstained with haematoxylin, dehydrated in an ascending series of ethanol concentrations, incubated in xylene and mounted with Vectamount-permanent mounting medium (Vector Laboratories) under a coverslip. To detect expression of insulin protein, immunohistochemical staining was performed as described above, except that polyclonal guinea pig anti-human insulin (AR029-5R; Dako) diluted 1:1000 in phosphate buffered saline containing Tween-
Quantitative real-time PCR mRNA expression of four genes was evaluated by qPCR for a-1-antitrypsin (SERPINA1), pancreatic a-amylase precursor (PA), pancreatic lipase (PNLIP) and chymotrypsinogen B1 (CTRB1). cDNA was synthesised from 750 ng total RNA in 30 lL reaction volumes using the iScript cDNA Synthesis Kit (Bio-Rad). Primers were designed using OligoExplorer software version 1.2 (Gene Link) and PrimerSelect software version 5.05 (DNAStar) (Table 2). Sequencing confirmed specificity of the 1
See: http://www.ncbi.nlm.nih.gov/geo/.
194
F.O. Buishand et al. / The Veterinary Journal 197 (2013) 192–197
Table 2 Nucleotide sequences of primers used for quantitative real-time PCR. Symbol
Gene GenBank accession number
Primer sequences (50 –30 )
Ta (°C)
CTRB1
Chymotrypsinogen B1 XM_856586 Pancreatic a-amylase precursor XM_851071 Pancreatic lipase XM_535023 Serpin peptidase inhibitor, clade A NM_001080109 Ribosomal protein L8 XM_532360 Heterogeneous nuclear ribonucleoprotein H XM_538576
F: CCGGGGAGTTCGACCAGGGCT R: GGCAGGCACACGGCGGACAC F: GGTTCAGATTTCTCCACCC R: ACTCACAGCATTCCCACAC F: TGTGTGGACTGGAAGAGTGGC R: ACAAACTGGGCATCGCTGG F: CAACGCCACCGCCTTCTTCATC R: CCCATTTTGCTCAGGACGCTTTTC F: CCATGAATCCTGTGGAGC R: GTAGAGGGTTTGCCGATG F: CTCACTATGATCCACCACG R: TAGCCTCCATAACCTCCAC
77.5
PA PNLIP SERPINA1 RPL8 HNRPH
64.0 62.0 74.0 55.0 61.2
F, forward primer; R, reverse primer; Ta, annealing temperature. 20 (PBST) was used as the primary antibody. All slides were scored by determining the proportion of positive insulinoma cells: 0, 0%; 1, <10%; 2, <24%; 3, 25–49%; 4, 50–74%; 5, 75–100%.
Immunofluorescence Sections of insulinomas that demonstrated both insulin and PNLIP positivity on immunohistochemical staining, as well as normal pancreatic tissue sections, were evaluated for potential co-localisation by immunofluorescence using the insulin immunohistochemical staining protocol. Rabbit polyclonal anti-PNLIP (1:500) and mouse monoclonal anti-insulin (AB6995) (Abcam) together diluted 1:1000 in PBST were used as primary antibodies. Alexa Fluor-568 goat anti-mouse antibody (A11029) and Alexa Fluor-488 goat anti-rabbit antibody (A11036) (Invitrogen) were used as secondary antibodies in a mixture at a dilution of 1:70. Nuclei were counterstained with 40 ,6-diamidino-2-phenylindole dihydrochloride (DAPI) (D1306; Invitrogen). Sections were mounted with FluoroSave mounting medium under a coverslip (Calbiochem). Spectral imaging was performed as described by Van der Loos (2008) to exclude autofluorescence and to demonstrate potential co-localisation of insulin and PNLIP in insulinoma sections.
Statistical analysis Imagene 5.0 software (Biodiscovery) was used for microarray image analysis. Defective spots were flagged and data were normalised based on Lowess print-tip normalisation (Yang et al., 2002) and subjected to logarithmic (base 2) transformation. Significance analysis for microarrays (SAM) software was used for the identification of differences in gene expression at a false discovery rate (FDR) of 10%, with minimum fold changes of 2.0 (Tusher et al., 2001). Differentially expressed genes were identified by hierarchical clustering analysis and visualised as a two-dimensional dendrogram using Genespring version 7 (Agilent). The Enrichment Analysis algorithm of MetaCore Analytical Suite (GeneGo) was used to analyse gene regulatory networks among the differentially expressed genes identified by SAM. qPCR data were analysed with iQ5 software (Bio-Rad) and REST-384 version 1 (Pfaffl et al., 2002) was used to compare the Ct values by the pair-wise fixed reallocation test, with 2000 randomizations to correct for multiple testing. P values <0.05 were considered to be significant.
Differences in gene expression between two subsets of primary insulinomas Unpaired SAM analysis was used to compare the five primary insulinomas from the low-metastatic group with the five primary insulinomas from the high-metastatic group. SAM generated a list of 158 annotated down-regulated genes (see Appendix A: Supplementary Table 2) and 85 annotated up-regulated genes in the highmetastatic group compared to the low-metastatic group (see Appendix A: Supplementary Table 3). Validation of selected genes with quantitative real-time-PCR Changes in mRNA levels of 3/4 genes as assessed by qPCR were concordant with those observed by microarray analysis. There was significant down-regulation of PNLIP by 14.5 fold (P = 0.002), CTRB1 by 11.3 fold (P = 0.002) and PA by 10.0 fold (P = 0.002) in the highmetastatic group compared to the low-metastatic group. There was no significant difference in SERPINA1 mRNA expression between the high-metastatic group (+1.8 fold) and the lowmetastatic group (P = 0.44). Pathway analysis Pathway analysis of the down-regulated genes in metastases vs. primary insulinomas (data set 1) and the down-regulated and up-regulated genes in the high-metastatic group vs. the lowmetastatic group (data sets 2 and 3, respectively) was used to interrogate the microarray data for biological relevance. Using MetaCore, nine significantly changed pathways were detected in data set 1, 11 in data set 2 and 15 in data set 3 (see Appendix A: Supplementary Tables 4, 5 and 6).
Results
Immunohistochemistry
Genes differentially expressed in primary insulinomas and their metastases
All insulinoma sections were positive by immunohistochemistry for insulin (Table 1; Fig. 2A). Positive immunohistochemical staining for PNLIP was evident in 12/21 (57%) primary insulinomas and 3/9 (33%) metastases (Fig. 2B and C). Co-localisation of insulin and PNLIP was demonstrated in <10% insulinoma cells (Fig. 3).
Unpaired SAM analysis generated a list of 84 annotated genes that were significantly down-regulated in the group of metastases compared to the group of primary insulinomas (see Appendix A: Supplementary Table 1). A gene tree dendrogram revealed two distinct clusters; five primary insulinomas clustered together (lowmetastatic group), while the remaining five primary insulinomas (high-metastatic group) clustered together with metastases (Fig. 1). Four insulinomas from the low-metastatic group had metastasised only to the regional lymph nodes, whereas four insulinomas from the high-metastatic group had also metastasised to the liver (Table 1).
Discussion Gene expression profiles of primary insulinomas were compared with their metastases to identify genes associated with pancreatic b-cell tumorigenesis. Eighty-four annotated genes exhibited P2.0-fold down-regulation in metastases compared to primary insulinomas. Although insulinomas are tumours of endocrine cells, the 11 most down-regulated genes (fold changes 7.9 to 16.9) are
F.O. Buishand et al. / The Veterinary Journal 197 (2013) 192–197
195
Fig. 1. Gene tree dendrogram generated using significance analysis for microarrays (SAM) showing clustering of five primary insulinomas vs. 10 metastases and five additional primary insulinomas. T, primary tumour; M, metastasis. Sample numbers correspond to case numbers listed in Table 1.
Fig. 2. Immunohistochemistry of a primary insulinoma (sample 13 from Table 1). (A) Primary tumour: insulin. (B) Primary tumour: lipase. (C) Lymph node metastasis: lipase. Bar = 50 lm.
those normally expressed by exocrine pancreatic acinar cells rather than b-cells (Terada et al., 1997). Since care was taken to collect samples from the central portion of the insulinomas and since immunohistochemical staining of both primary insulinomas and their metastases demonstrated PNLIP expression, contamination of our insulinoma samples with exocrine pancreatic tissue is considered to be an unlikely explanation for detection of PNLIP expression in insulinomas. Moreover, co-localisation of insulin and PNLIP in insulinoma cells was demonstrated by immunofluorescence in insulinomas that did not show morphological features of acinar cell differentiation on histological examination. Therefore, it is concluded that acinar genes are expressed by b-cells in insulinomas and that primary neoplastic b-cells express acinar genes at a higher level than metastatic insulinomas. The presence of neoplastic cells with mixed endocrine–exocrine features has been reported in a small number of cases of human pancreatic endocrine tumours (Regitnig et al., 2001; Vortmeyer et al., 2004). Immunohistochemically, these so-called amphicrine cells are positive by immunostaining for endocrine markers, as well as pancreatic enzymes, and both neurosecretory and zymogen granules can be detected ultrastructurally (Klimstra et al., 1994; Kamisawa et al., 2002). Although, electron microscopy was not performed in the present study, co-localisation of insulin and PNLIP reactivity was demonstrated by immunofluorescence.
Therefore, it is concluded that canine insulinomas contain small subpopulations of amphicrine cells. Since acinar genes are downregulated in insulinoma metastases, it is further concluded that these amphicrine cells do not metastasise. Co-staining with insulin, synaptophysin and amylase, but not PNLIP, has been demonstrated in the rat amphicrine pancreatic cell line AR42J (Rosewicz et al., 1992; Huang et al., 2002). Five insulinomas (high-metastatic group) had a gene expression profile similar to that of their metastases, whereas the remaining five insulinomas (low-metastatic group) had a different gene expression profile. The high-metastatic group of insulinomas had a more malignant biological behaviour than the low-metastatic group: four tumours from the low-metastatic group had only metastasised to regional lymph nodes, while four tumours from the high-metastatic group had metastasised to both regional lymph nodes and the liver. Tumour-node-metastasis (TNM) stage is a prognostic marker in canine insulinomas; the survival time of dogs with distant metastases is significantly shorter than dogs in which insulinomas are restricted to the pancreas and regional lymph nodes (Buishand et al., 2010). Insulinomas from the high-metastatic group that metastasised to both lymph nodes and liver had similar gene expression profiles to lymph node and liver metastases. In contrast, insulinomas from the low-metastatic group that had only metastasised to the
196
F.O. Buishand et al. / The Veterinary Journal 197 (2013) 192–197
Fig. 3. Spectral imaging analysis of normal pancreas compared with a primary insulinoma (sample 13 from Table 1). (A) Fluorescence-like and pseudo-coloured composite of normal pancreas (DAPI in blue, insulin in green and pancreatic lipase in red). (B) Spectral imaging of (A) demonstrates the absence of co-localisation of insulin and pancreatic lipase. (C) Fluorescence-like and pseudo-coloured composite of a primary insulinoma. (D) Spectral imaging of (C) demonstrates co-localisation of insulin and pancreatic lipase in <10% of neoplastic cells in yellow. Bar = 50 lm.
lymph nodes had a different gene expression profile to their lymph node metastases. Neoplasms are composed of heterogeneous cell populations, from which only a small subset of cells has metastatic capacity (Fidler, 2002). On the basis of gene expression profiles generated in the present study, we hypothesise that insulinoma cells with metastatic capability have a similar gene expression profile, since lymph node and liver metastases cluster together. Primary insulinomas that metastasise only to the lymph nodes may contain a smaller subpopulation of these metastasising cells than primary insulinomas that also metastasise to the liver. Eighty-five genes were up-regulated P2.0 fold and 158 genes were down-regulated P2.0 in the high-metastatic group vs. the low-metastatic group of canine insulinomas. The 11 genes with the highest degree of down-regulation in the high-metastatic group (fold changes 99.5 to 33.1) were all acinar cell-related genes. This suggests that the high-metastatic group contains fewer amphicrine cells than the low-metastatic group. The existence of cells with mixed endocrine–acinar features supports the concept that endocrine and exocrine pancreatic cells may arise from a common progenitor stem cell (Ku, 2008; Puri and Hebrok, 2010; Pan and Wright, 2011). De Sá et al. (2007) speculated that insulinomas with a higher degree of differentiation have a higher level of expression of acinar cell markers than less differentiated insulinomas. On the basis of our microarray results, we suggest that insulinoma cells expressing acinar markers are less likely to metastasise. Whether this reflects a difference in differentiation potential between amphicrine cells and insulinoma cells without acinar gene expression remains to be elucidated. The acinar gene CTRB1 was down-regulated in the highmetastatic group of canine insulinomas in comparison with the low-metastatic group. In human insulinomas, CTRB1 was among the highest up-regulated genes in six well-differentiated endocrine tumours of benign behaviour (WDET-BB) compared to one well-
differentiated endocrine carcinoma (WDEC) and three metastases of endocrine carcinomas (MEC) (De Sá et al., 2007). Although De Sá et al. (2007) observed up-regulation of SERPINA1 in malignant insulinomas in humans, in our study there was no significant difference in SERPINA1 mRNA expression between the high-metastatic group and the low-metastatic group by qPCR. Down-regulation of the BRCA1 and ATM/ATR DNA damage response pathways was identified in the high-metastatic group of primary canine insulinomas and in metastases. Loss of control of cell cycle checkpoints and the DNA repair system are important in the pathogenesis of malignant cancers (Poehlmann and Roessner, 2010). Allelic loss of the tumour suppressor gene BRCA1 has been detected in human insulinomas (Hrasc´an et al., 2008). Since BRCA1/ BARD1 normally regulates ubiquitination during the DNA damage response, down-regulation of the BRCA1 pathway in the high-metastatic group of canine insulinomas and in metastases suggests that this pathway may have a role in tumour progression. Activation of the ATM/ATR pathway after DNA damage normally prevents cells from entering S phase of the cell cycle. Abundant data in humans and mice have demonstrated that the G1/S checkpoint in the cell cycle is critical in regulating pancreatic b-cell replication (Fiaschi-Taesch et al., 2009). Down-regulation of the ATM/ATR pathway in the high-metastatic group of insulinomas and in metastases suggests that down-regulation of this pathway in insulinomas contributes to ineffective cell cycle arrest and inadequate apoptotic responses after DNA damage. Aberrant expression of p53 protein, which is the central downstream mediator of this pathway, has been implicated in insulinoma tumorigenesis (Pavelic´ et al., 1995, 1996). Conclusions Differences in gene expression profiles were identified by microarray analysis in low-metastatic and high-metastatic subsets
F.O. Buishand et al. / The Veterinary Journal 197 (2013) 192–197
of primary canine insulinomas and their metastases. In particular, genes for the acinar enzymes pancreatic lipase (PNLIP) and chymotrypsinogen B1 (CTRB1) were down-regulated substantially in a more malignant subset of primary insulinomas and in metastases. Co-localisation of insulin and PNLIP was identified by immunofluorescence in insulinoma cells, supporting the hypothesis that canine insulinomas contain amphicrine cells. Conflict of interest statement None of the authors of this paper has a financial or personal relationship with other people or organisations that could inappropriately influence or bias the content of the paper. Acknowledgements The authors gratefully acknowledge the technical assistance of A. Slob, E.P. Timmermans-Sprang and M.E. van Wolferen. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.tvjl.2013.01.021. References Brinkhof, B., Spee, B., Rothuizen, J., Penning, L.C., 2006. Development and evaluation of canine reference genes for accurate quantification of gene expression. Analytical Biochemistry 356, 36–43. Buishand, F.O., Kik, M., Kirpensteijn, J., 2010. Evaluation of clinico-pathological criteria and the Ki67 index as prognostic indicators in canine insulinoma. The Veterinary Journal 185, 62–67. Buishand, F.O., van Erp, M.G., Groenveld, H.A., Mol, J.A., Kik, M., Robben, J.H., Kooistra, H.S., Kirpensteijn, J., 2012. Expression of insulin-like growth factor-1 by canine insulinomas and their metastases. The Veterinary Journal 191, 334– 340. De Sá, S.V., Corrêa-Giannella, M.C., Machado, M.C., Krogh, K., De Almeida, M.Q., Pereira, M.A.A., Siqeira, S.A.C., Patzina, R.A., Ibuki, F.S., Sogayar, M.C., Machado, M.C., Giannella-Neto, D., 2007. Serpin peptidase inhibitor clade A member 1 as a potential marker for malignancy in insulinomas. Clinical Cancer Research 13, 5322–5330. Fiaschi-Taesch, N., Bigatel, T.A., Sicari, B., Takane, K.K., Salim, F., Velazquez-Garcia, S., Harb, G., Selk, K., Cozar-Castellano, I., Stewart, A.F., 2009. Survey of the human pancreatic b-cell G1/S proteome reveals a potential therapeutic role for cdk-6 and cyclin D1 in enhancing human b-cell replication and function in vivo. Diabetes 58, 882–893. Fidler, I.J., 2002. Critical determinants of metastasis. Seminars in Cancer Biology 12, 89–96. Hrasc´an, R., Pec´ina-Slaus, N., Martic´, T.N., Colic´, J.F., Gall-Troselj, K., Pavelic´, K., Karapandza, N., 2008. Analysis of selected genes in neuroendocrine tumours: Insulinomas and phaeochromocytomas. Journal of Neuroendocrinology 20, 1015–1022. Huang, X., Sheu, L., Kang, Y., Eto, Y., Kojima, I., Gaisano, H.Y., 2002. Effects of selective endocrine or exocrine induction of AR42J on SNARE and Munc 18 protein expression. Pancreas 25, e56–e63.
197
Kamisawa, T., Tu, Y., Egawa, N., Ishiwata, J.I., Tsuruta, K., Okamoto, A., Hayashi, Y., Koike, M., Yamaguchi, T., 2002. Ductal and acinar differentiation in pancreatic endocrine tumors. Digestive Diseases and Sciences 47, 2254–2261. Klimstra, D.S., Rosai, J., Heffess, C.S., 1994. Mixed acinar–endocrine carcinomas of the pancreas. American Journal of Surgical Pathology 18, 765–778. Ku, H.T., 2008. Pancreatic progenitor cells – Recent studies. Endocrinology 149, 4312–4316. Moore, A.S., Nelson, R.W., Henry, C.J., Rassnick, K.M., Kristal, O., Ogilvie, G.K., Kintzer, P., 2002. Streptozocin for treatment of pancreatic islet cell tumors in dogs: 17 cases (1989–1999). Journal of the American Veterinary Medical Association 221, 811–818. Pan, F.C., Wright, C., 2011. Pancreas organogenesis: From bud to plexus to gland. Developmental Dynamics 240, 530–565. Pavelic´, K., Hrasc´an, R., Kapitanovic´, S., Karapandza, N., Vranes, Z., Belicza, M., Kruslin, B., Cabrijan, T., 1995. Multiple genetic alterations in malignant metastatic insulinomas. Journal of Pathology 177, 395–400. Pavelic´, K., Hrascan, R., Kapitanovic, S., Vranes, Z., Cabrijan, T., Spaventi, S., Korsic, M., Krizanac, S., Li, Y.Q., Stambrook, P., Gluckman, J.L., Pavelic, Z.P., 1996. Molecular genetics of malignant insulinoma. Anticancer Research 16, 1707– 1717. Pfaffl, M.W., Horgan, G.W., Dempfle, L., 2002. Relative expression software tool (REST) for 420 group-wise comparison and statistical analysis of relative expression results in real-time 421 PCR. Nucleic Acids Research 30, e36. Poehlmann, A., Roessner, A., 2010. Importance of DNA damage checkpoints in the pathogenesis of human cancers. Pathology – Research and Practice 206, 591– 601. Puri, S., Hebrok, M., 2010. Cellular plasticity within the pancreas – Lessons learned from development. Developmental Cell 18, 342–356. Rao, N.A., van Wolferen, M.E., van den Ham, R., van Leenen, D., Groot Koerkamp, M.J., Holstege, F.C., Mol, J.A., 2008. CDNA microarray profiles of canine mammary tumour cell lines reveal deregulated pathways pertaining to their phenotype. Animal Genetics 39, 333–345. Regitnig, P., Spuller, E., Denk, H., 2001. Insulinoma of the pancreas with insular– ductular differentiation in its liver metastasis – Indication of a common stemcell origin of the exocrine and endocrine components. Virchows Archives 438, 624–628. Rosewicz, S., Vogt, D., Harth, N., Grund, C., Franke, W.W., Ruppert, S., Schweitzer, E., Riecken, E.O., Wiedenmann, B., 1992. An amphicrine pancreatic cell line: AR42J cells combine exocrine and neuroendocrine properties. European Journal of Cell Biology 59, 80–91. Steiner, J.M., Bruyette, D.S., 1996. Canine insulinoma. Compendium on Continuing Education for the Practising Veterinarian 18, 13–23. Terada, T., Kitamura, Y., Ashida, K., Matsunaga, Y., Kato, M., Harada, K., Morita, T., Ohta, T., Nakanuma, Y., 1997. Expression of pancreatic digestive enzymes in normal and pathologic epithelial cells of the human gastrointestinal system. Virchows Archives 431, 195–203. Tobin, R.L., Nelson, R.W., Lucroy, M.D., Wooldridge, J.D., Feldman, E.C., 1999. Outcome of surgical versus medical treatment of dogs with b cell neoplasia: 39 cases (1990–1997). Journal of the American Veterinary Medical Association 215, 226–230. Tusher, V.G., Tibshirani, R., Chu, G., 2001. Significance analysis of microarrays applied to ionizing radiation response. Proceedings of the National Academy of Sciences of the United States of America 98, 5116–5121. Van der Loos, C.M., 2008. Multiple immunoenzyme staining: Methods and visualizations for the observation with spectral imaging. Journal of Histochemistry and Cytochemistry 56, 313–328. Vortmeyer, A.O., Huang, S., Lubensky, I., Zhuang, Z., 2004. Non-islet origin of pancreatic islet cell tumors. Journal of Clinical Endocrinology and Metabolism 89, 1934–1938. Yang, Y.H., Dudoit, S., Luu, P., Lin, D.M., Peng, V., Ngai, J., Speed, T.P., 2002. Normalization for cDNA microarray data: A robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Research 30, e15.
Contents lists available at SciVerse ScienceDirect
The Veterinary Journal journal homepage: www.elsevier.com/locate/tvjl
Gene expression profiling of primary canine insulinomas and their metastases Floryne O. Buishand a,⇑, Jolle Kirpensteijn a, Alexandra A. Jaarsma a, Ernst-Jan M. Speel b, Marja Kik c, Jan A. Mol a a b c
Department of Clinical Sciences of Companion Animals, Faculty of Veterinary Medicine, Utrecht University, Yalelaan 108, 3584 CM Utrecht, The Netherlands Department of Pathology, GROW-School for Oncology and Developmental Biology, Maastricht University Medical Centre, PO Box 5800, 6202 AZ Maastricht, The Netherlands Department of Pathobiology, Division Pathology, Faculty of Veterinary Medicine, Utrecht University, Yalelaan 1, 3584 CL Utrecht, The Netherlands
a r t i c l e
i n f o
Article history: Accepted 20 January 2013
Keywords: Insulinomas Canine Gene expression Microarrays Amphicrine Lipase
a b s t r a c t The gene expression profile of 10 primary canine insulinomas was compared with that of their accompanying metastases using microarray analysis and quantitative real time-PCR. Analysis of microarray data revealed 84 genes that were differentially expressed between primary insulinomas and their metastases, along with 243 genes differentially expressed between a low-metastatic and a high-metastatic subset of primary insulinomas. The genes differently expressed between primary insulinomas and their metastases clustered together in nine signalling pathways. Comparing the low-metastatic to the high-metastatic subset of primary insulinomas, 26 pathways appeared to be significantly influenced. The acinar enzymes pancreatic lipase (PNLIP) and chymotrypsinogen B1 (CTRB1) were amongst the most down-regulated genes in the malignant group of primary insulinomas and in metastases. Immunofluorescence demonstrated co-localisation of insulin and PNLIP in tumour cells. Different subsets of canine insulinomas can be identified on the basis of their gene expression profile. Canine insulinomas appear to contain amphicrine cells, which exhibit both endocrine and exocrine cell features. Ó 2013 Elsevier Ltd. All rights reserved.
Introduction Insulinomas, which cause clinical signs associated with hypoglycaemia, are the most common malignant pancreatic endocrine tumours in the dog. Although dogs treated by surgery survive longer than dogs treated with medication alone, the prognosis after surgery for insulinomas is guarded. Signs of hypoglycaemia often recur after surgery, because it may only be possible to achieve partial resection of tumour tissue or due to the presence of functional (micro)metastases (Tobin et al., 1999; Buishand et al., 2012). Since there are significant side effects in dogs with insulinomas following treatment with diazoxide, glucocorticoids, streptozocin or octreotide, new therapeutic strategies are warranted to improve prognosis in dogs with insulinomas (Steiner and Bruyette, 1996; Moore et al., 2002). In a previous study, we used quantitative real time-PCR (qPCR) to compare the expression of 16 candidate genes between primary canine insulinomas and their metastases (Buishand et al., 2012). Metastases of canine insulinomas had significantly increased expression of growth hormone (GH) and insulin-like growth factor-I (IGF-I). In addition to GH and IGF-I, other genes may be differen⇑ Corresponding author. Tel.: +31 30 2537563. E-mail address: [email protected] (F.O. Buishand). 1090-0233/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tvjl.2013.01.021
tially regulated between primary canine insulinomas and their metastases. The aim of the present study was to identify altered expression of genes and deregulation of gene signalling pathways in primary canine insulinomas and their corresponding metastases using a canine specific cDNA microarray representing 20,160 genes. Materials and methods Patients and tissue specimens Twenty-one primary canine insulinomas and 10 accompanying metastases (seven lymph node and three liver metastases) were resected at the Faculty of Veterinary Medicine, Utrecht University, from 1993 to 2009 (Table 1). Portions of tissue (5–15 mm diameter) for RNA extraction were removed from the central portion of the tumours, immediately frozen in liquid nitrogen and stored at 70 °C. Tissue fragments were fixed in 10% neutral-buffered formalin for 24–36 h for histopathology. After routine processing and paraffin embedding, sections were cut at 5 lm and stained with haematoxylin and eosin. RNA isolation and cRNA synthesis Total RNA was isolated using the RNeasy Mini Kit (Qiagen). To prevent contamination of the samples with genomic DNA, an on-column DNase treatment was performed. RNA concentrations were quantified by spectrophotometry (NanoDrop ND-1000, Isogen Life Sciences). Total RNA integrity was assessed using the Agilent BioAnalyzer 2100. Only RNA samples with an RNA integrity number (RIN) > 9.0 were selected for cRNA amplification. Two micrograms of total RNA from each sample were amplified by in vitro transcription using T7 RNA polymerase (Rao et al.,
193
F.O. Buishand et al. / The Veterinary Journal 197 (2013) 192–197 Table 1 Clinicopathological characteristics of insulinoma samples used in this study. Dog
Breeda
Sex
Glucose (mmol/L)
Specimen type
Diameter of tumour mass (cm)
TNM stageb
Microarray
ICH insulin (positive cells)c
ICH PNLIP (positive cells)c
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 9 11 13 14 15 16 17 19 20 21
Crossbreed Bouvier des Flandres Bearded collie Beagle West Highland white terrier German short haired pointer Boxer West Highland white terrier Bearded collie1 Irish setter Labrador retriever2 Belgian shepherd dog Crossbreed3 Rough collie4 German shepherd dog5 Anatolian shepherd dog6 Boxer7 Jack Russell terrier Rough collie8 Maltese9 Kooiker dog10 Bearded collie1 Labrador retriever2 Crossbreed3 Rough collie4 German shepherd dog5 Anatolian shepherd dog6 Boxer7 Rough collie8 Maltese9 Kooiker dog10
M Fx Mx Fx F M Fx M Fx Mx Mx Mx Fx Mx F Mx Fx Fx F M F Fx Mx Fx Mx F Mx Fx F M F
3.0 2.2 1.7 1.1 2.5 2.9 3.1 3.2 2.7 2.8 1.9 2.2 3.2 2.2 2.5 2.0 2.6 1.7 1.9 2.3 3.1 2.7 1.9 3.2 2.2 2.5 2.0 2.6 1.9 2.3 3.1
T T T T T T T T T T T T T T T T T T T T T N N N M M N N M N N
1.0 0.6 2.5 1.5 1.2 2.7 3.5d 1.5 1.5 4.0 4.0 1.5 0.8 5.0c 5.0 9.0d 2.5 3.5 4.0 1.0 1.5 1.0 4.0e 2.0 6.0f 5.0 2.0 4.0e 3.5 1.0 1.0
I III IV I I II II I III II III III III IV IV IV III II IV III IV – – – – – – – – – –
No No No No No No No No Yes No Yes No Yes Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
5 4 1 5 4 4 4 5 5 5 4 3 4 2 3 1 3 3 5 5 5 – 5 4 3 2 4 3 5 5 4
2 1 1 2 1 1 0 0 0 1 0 2 1 0 1 0 0 1 0 0 1 – 0 1 0 1 0 1 0 0 0
M, male; F, female, Mx, neutered male; Fx, neutered female; IHC, immunohistochemistry; PNILP, pancreatic lipase; T, primary tumour; N, lymph node metastasis; M, liver metastasis. a The same number indicates a metastasis (N or M) matched to a primary tumour from the same animal. b Staging was performed according to Buishand et al. (2010). c Grading proportion of IHC positive cells: 0, 0%; 1, <10%; 2, <24%; 3, 25–49%; 4, 50–74%; 5, 75–100%; –, not determined. d Two primary insulinomas present. e Multiple lymph node metastases present. f Multiple metastases present, diffusely spread through the liver.
2008). Three micrograms of amplified mRNA were coupled to Cy3 and Cy5 fluorophores (GE Healthcare). A pool of 60 lg amplified cRNA was formed, containing 3 lg of each amplified cRNA sample. Half of the cRNA samples from both the primary insulinoma group and the metastasis group were labelled with Cy5 and the other half from each group were labelled with Cy3. Equal quantities of pooled cRNA were also labelled with Cy5 and Cy3. ChromaSpin-30 columns (ClontechTakara Bio) were used to purify labelled cRNA and dye incorporation efficiency was measured by spectrophotometry (NanoDrop). Equal quantities of labelled cRNA samples with specific activity of 2–5% dye-labelled nucleosides were used for microarray hybridisation.
amplicons. A reaction solution was prepared for each gene using SYBR Green Supermix (Bio-Rad). The housekeeping genes HNRPH and RPL8 were used as reference genes for normalisation of target gene expression. qPCR was performed using the Bio-Rad My-iQ Single Color Real-Time PCR Detection System (Bio-Rad), with three-step reactions (denaturation, annealing and elongation) for RPL8 and two-step reactions (denaturation and annealing) for other primer sets (Brinkhof et al., 2006). All reactions were performed in duplicate and their efficiencies were assessed by a dilution series of pooled cDNA samples tested in each run. Efficiency cut-offs were 95–105%. Omission of the reverse transcriptase reaction showed no significant contamination with genomic DNA.
Microarray hybridisation
Immunohistochemistry
Gene expression profiling was performed using UltraGAPS slides (Corning) spotted with a canine specific collection of 20,160 non-redundant clones of 30 untranslated region cDNA fragments (Rao et al., 2008). Microarray hybridisation, high-stringency washing and scanning with an Agilent G2565AA DNA Microarray Scanner were performed according to Rao et al. (2008). The microarray specifications and derived data are accessible through National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO)1 accession numbers GPL5117 and GSE32934.
Paraffin-embedded blocks were available for immunohistochemistry for all 21 primary insulinomas and 9/10 metastases included in this study. Sections of normal canine pancreatic tissue were used as positive controls, while omission of primary antibodies served as negative controls. Rabbit polyclonal antibodies directed against canine PNLIP (a generous gift from Professor J.M. Steiner, College of Veterinary Medicine, Texas A&M University, Texas, USA) were used to detect expression of PNLIP protein. After deparaffinisation, sections were treated with 0.35% H2O2 in Tris buffered saline (TBS) for 15 min to block endogenous peroxidase activity. Slides were rinsed in TBS containing Tween-20 (TBST) and blocked for 30 min with 10% normal goat serum. Sections were then incubated for 30 min at room temperature with the primary antibody diluted 1:500 in TBST. The slides were rinsed with TBST and then incubated for 30 min with Envision peroxidase-conjugated anti-rabbit IgG (Dako). The slides were then rinsed with TBS and peroxidase activity was visualised following incubation with 3,3-diaminobenzidine. Sections of tissue were counterstained with haematoxylin, dehydrated in an ascending series of ethanol concentrations, incubated in xylene and mounted with Vectamount-permanent mounting medium (Vector Laboratories) under a coverslip. To detect expression of insulin protein, immunohistochemical staining was performed as described above, except that polyclonal guinea pig anti-human insulin (AR029-5R; Dako) diluted 1:1000 in phosphate buffered saline containing Tween-
Quantitative real-time PCR mRNA expression of four genes was evaluated by qPCR for a-1-antitrypsin (SERPINA1), pancreatic a-amylase precursor (PA), pancreatic lipase (PNLIP) and chymotrypsinogen B1 (CTRB1). cDNA was synthesised from 750 ng total RNA in 30 lL reaction volumes using the iScript cDNA Synthesis Kit (Bio-Rad). Primers were designed using OligoExplorer software version 1.2 (Gene Link) and PrimerSelect software version 5.05 (DNAStar) (Table 2). Sequencing confirmed specificity of the 1
See: http://www.ncbi.nlm.nih.gov/geo/.
194
F.O. Buishand et al. / The Veterinary Journal 197 (2013) 192–197
Table 2 Nucleotide sequences of primers used for quantitative real-time PCR. Symbol
Gene GenBank accession number
Primer sequences (50 –30 )
Ta (°C)
CTRB1
Chymotrypsinogen B1 XM_856586 Pancreatic a-amylase precursor XM_851071 Pancreatic lipase XM_535023 Serpin peptidase inhibitor, clade A NM_001080109 Ribosomal protein L8 XM_532360 Heterogeneous nuclear ribonucleoprotein H XM_538576
F: CCGGGGAGTTCGACCAGGGCT R: GGCAGGCACACGGCGGACAC F: GGTTCAGATTTCTCCACCC R: ACTCACAGCATTCCCACAC F: TGTGTGGACTGGAAGAGTGGC R: ACAAACTGGGCATCGCTGG F: CAACGCCACCGCCTTCTTCATC R: CCCATTTTGCTCAGGACGCTTTTC F: CCATGAATCCTGTGGAGC R: GTAGAGGGTTTGCCGATG F: CTCACTATGATCCACCACG R: TAGCCTCCATAACCTCCAC
77.5
PA PNLIP SERPINA1 RPL8 HNRPH
64.0 62.0 74.0 55.0 61.2
F, forward primer; R, reverse primer; Ta, annealing temperature. 20 (PBST) was used as the primary antibody. All slides were scored by determining the proportion of positive insulinoma cells: 0, 0%; 1, <10%; 2, <24%; 3, 25–49%; 4, 50–74%; 5, 75–100%.
Immunofluorescence Sections of insulinomas that demonstrated both insulin and PNLIP positivity on immunohistochemical staining, as well as normal pancreatic tissue sections, were evaluated for potential co-localisation by immunofluorescence using the insulin immunohistochemical staining protocol. Rabbit polyclonal anti-PNLIP (1:500) and mouse monoclonal anti-insulin (AB6995) (Abcam) together diluted 1:1000 in PBST were used as primary antibodies. Alexa Fluor-568 goat anti-mouse antibody (A11029) and Alexa Fluor-488 goat anti-rabbit antibody (A11036) (Invitrogen) were used as secondary antibodies in a mixture at a dilution of 1:70. Nuclei were counterstained with 40 ,6-diamidino-2-phenylindole dihydrochloride (DAPI) (D1306; Invitrogen). Sections were mounted with FluoroSave mounting medium under a coverslip (Calbiochem). Spectral imaging was performed as described by Van der Loos (2008) to exclude autofluorescence and to demonstrate potential co-localisation of insulin and PNLIP in insulinoma sections.
Statistical analysis Imagene 5.0 software (Biodiscovery) was used for microarray image analysis. Defective spots were flagged and data were normalised based on Lowess print-tip normalisation (Yang et al., 2002) and subjected to logarithmic (base 2) transformation. Significance analysis for microarrays (SAM) software was used for the identification of differences in gene expression at a false discovery rate (FDR) of 10%, with minimum fold changes of 2.0 (Tusher et al., 2001). Differentially expressed genes were identified by hierarchical clustering analysis and visualised as a two-dimensional dendrogram using Genespring version 7 (Agilent). The Enrichment Analysis algorithm of MetaCore Analytical Suite (GeneGo) was used to analyse gene regulatory networks among the differentially expressed genes identified by SAM. qPCR data were analysed with iQ5 software (Bio-Rad) and REST-384 version 1 (Pfaffl et al., 2002) was used to compare the Ct values by the pair-wise fixed reallocation test, with 2000 randomizations to correct for multiple testing. P values <0.05 were considered to be significant.
Differences in gene expression between two subsets of primary insulinomas Unpaired SAM analysis was used to compare the five primary insulinomas from the low-metastatic group with the five primary insulinomas from the high-metastatic group. SAM generated a list of 158 annotated down-regulated genes (see Appendix A: Supplementary Table 2) and 85 annotated up-regulated genes in the highmetastatic group compared to the low-metastatic group (see Appendix A: Supplementary Table 3). Validation of selected genes with quantitative real-time-PCR Changes in mRNA levels of 3/4 genes as assessed by qPCR were concordant with those observed by microarray analysis. There was significant down-regulation of PNLIP by 14.5 fold (P = 0.002), CTRB1 by 11.3 fold (P = 0.002) and PA by 10.0 fold (P = 0.002) in the highmetastatic group compared to the low-metastatic group. There was no significant difference in SERPINA1 mRNA expression between the high-metastatic group (+1.8 fold) and the lowmetastatic group (P = 0.44). Pathway analysis Pathway analysis of the down-regulated genes in metastases vs. primary insulinomas (data set 1) and the down-regulated and up-regulated genes in the high-metastatic group vs. the lowmetastatic group (data sets 2 and 3, respectively) was used to interrogate the microarray data for biological relevance. Using MetaCore, nine significantly changed pathways were detected in data set 1, 11 in data set 2 and 15 in data set 3 (see Appendix A: Supplementary Tables 4, 5 and 6).
Results
Immunohistochemistry
Genes differentially expressed in primary insulinomas and their metastases
All insulinoma sections were positive by immunohistochemistry for insulin (Table 1; Fig. 2A). Positive immunohistochemical staining for PNLIP was evident in 12/21 (57%) primary insulinomas and 3/9 (33%) metastases (Fig. 2B and C). Co-localisation of insulin and PNLIP was demonstrated in <10% insulinoma cells (Fig. 3).
Unpaired SAM analysis generated a list of 84 annotated genes that were significantly down-regulated in the group of metastases compared to the group of primary insulinomas (see Appendix A: Supplementary Table 1). A gene tree dendrogram revealed two distinct clusters; five primary insulinomas clustered together (lowmetastatic group), while the remaining five primary insulinomas (high-metastatic group) clustered together with metastases (Fig. 1). Four insulinomas from the low-metastatic group had metastasised only to the regional lymph nodes, whereas four insulinomas from the high-metastatic group had also metastasised to the liver (Table 1).
Discussion Gene expression profiles of primary insulinomas were compared with their metastases to identify genes associated with pancreatic b-cell tumorigenesis. Eighty-four annotated genes exhibited P2.0-fold down-regulation in metastases compared to primary insulinomas. Although insulinomas are tumours of endocrine cells, the 11 most down-regulated genes (fold changes 7.9 to 16.9) are
F.O. Buishand et al. / The Veterinary Journal 197 (2013) 192–197
195
Fig. 1. Gene tree dendrogram generated using significance analysis for microarrays (SAM) showing clustering of five primary insulinomas vs. 10 metastases and five additional primary insulinomas. T, primary tumour; M, metastasis. Sample numbers correspond to case numbers listed in Table 1.
Fig. 2. Immunohistochemistry of a primary insulinoma (sample 13 from Table 1). (A) Primary tumour: insulin. (B) Primary tumour: lipase. (C) Lymph node metastasis: lipase. Bar = 50 lm.
those normally expressed by exocrine pancreatic acinar cells rather than b-cells (Terada et al., 1997). Since care was taken to collect samples from the central portion of the insulinomas and since immunohistochemical staining of both primary insulinomas and their metastases demonstrated PNLIP expression, contamination of our insulinoma samples with exocrine pancreatic tissue is considered to be an unlikely explanation for detection of PNLIP expression in insulinomas. Moreover, co-localisation of insulin and PNLIP in insulinoma cells was demonstrated by immunofluorescence in insulinomas that did not show morphological features of acinar cell differentiation on histological examination. Therefore, it is concluded that acinar genes are expressed by b-cells in insulinomas and that primary neoplastic b-cells express acinar genes at a higher level than metastatic insulinomas. The presence of neoplastic cells with mixed endocrine–exocrine features has been reported in a small number of cases of human pancreatic endocrine tumours (Regitnig et al., 2001; Vortmeyer et al., 2004). Immunohistochemically, these so-called amphicrine cells are positive by immunostaining for endocrine markers, as well as pancreatic enzymes, and both neurosecretory and zymogen granules can be detected ultrastructurally (Klimstra et al., 1994; Kamisawa et al., 2002). Although, electron microscopy was not performed in the present study, co-localisation of insulin and PNLIP reactivity was demonstrated by immunofluorescence.
Therefore, it is concluded that canine insulinomas contain small subpopulations of amphicrine cells. Since acinar genes are downregulated in insulinoma metastases, it is further concluded that these amphicrine cells do not metastasise. Co-staining with insulin, synaptophysin and amylase, but not PNLIP, has been demonstrated in the rat amphicrine pancreatic cell line AR42J (Rosewicz et al., 1992; Huang et al., 2002). Five insulinomas (high-metastatic group) had a gene expression profile similar to that of their metastases, whereas the remaining five insulinomas (low-metastatic group) had a different gene expression profile. The high-metastatic group of insulinomas had a more malignant biological behaviour than the low-metastatic group: four tumours from the low-metastatic group had only metastasised to regional lymph nodes, while four tumours from the high-metastatic group had metastasised to both regional lymph nodes and the liver. Tumour-node-metastasis (TNM) stage is a prognostic marker in canine insulinomas; the survival time of dogs with distant metastases is significantly shorter than dogs in which insulinomas are restricted to the pancreas and regional lymph nodes (Buishand et al., 2010). Insulinomas from the high-metastatic group that metastasised to both lymph nodes and liver had similar gene expression profiles to lymph node and liver metastases. In contrast, insulinomas from the low-metastatic group that had only metastasised to the
196
F.O. Buishand et al. / The Veterinary Journal 197 (2013) 192–197
Fig. 3. Spectral imaging analysis of normal pancreas compared with a primary insulinoma (sample 13 from Table 1). (A) Fluorescence-like and pseudo-coloured composite of normal pancreas (DAPI in blue, insulin in green and pancreatic lipase in red). (B) Spectral imaging of (A) demonstrates the absence of co-localisation of insulin and pancreatic lipase. (C) Fluorescence-like and pseudo-coloured composite of a primary insulinoma. (D) Spectral imaging of (C) demonstrates co-localisation of insulin and pancreatic lipase in <10% of neoplastic cells in yellow. Bar = 50 lm.
lymph nodes had a different gene expression profile to their lymph node metastases. Neoplasms are composed of heterogeneous cell populations, from which only a small subset of cells has metastatic capacity (Fidler, 2002). On the basis of gene expression profiles generated in the present study, we hypothesise that insulinoma cells with metastatic capability have a similar gene expression profile, since lymph node and liver metastases cluster together. Primary insulinomas that metastasise only to the lymph nodes may contain a smaller subpopulation of these metastasising cells than primary insulinomas that also metastasise to the liver. Eighty-five genes were up-regulated P2.0 fold and 158 genes were down-regulated P2.0 in the high-metastatic group vs. the low-metastatic group of canine insulinomas. The 11 genes with the highest degree of down-regulation in the high-metastatic group (fold changes 99.5 to 33.1) were all acinar cell-related genes. This suggests that the high-metastatic group contains fewer amphicrine cells than the low-metastatic group. The existence of cells with mixed endocrine–acinar features supports the concept that endocrine and exocrine pancreatic cells may arise from a common progenitor stem cell (Ku, 2008; Puri and Hebrok, 2010; Pan and Wright, 2011). De Sá et al. (2007) speculated that insulinomas with a higher degree of differentiation have a higher level of expression of acinar cell markers than less differentiated insulinomas. On the basis of our microarray results, we suggest that insulinoma cells expressing acinar markers are less likely to metastasise. Whether this reflects a difference in differentiation potential between amphicrine cells and insulinoma cells without acinar gene expression remains to be elucidated. The acinar gene CTRB1 was down-regulated in the highmetastatic group of canine insulinomas in comparison with the low-metastatic group. In human insulinomas, CTRB1 was among the highest up-regulated genes in six well-differentiated endocrine tumours of benign behaviour (WDET-BB) compared to one well-
differentiated endocrine carcinoma (WDEC) and three metastases of endocrine carcinomas (MEC) (De Sá et al., 2007). Although De Sá et al. (2007) observed up-regulation of SERPINA1 in malignant insulinomas in humans, in our study there was no significant difference in SERPINA1 mRNA expression between the high-metastatic group and the low-metastatic group by qPCR. Down-regulation of the BRCA1 and ATM/ATR DNA damage response pathways was identified in the high-metastatic group of primary canine insulinomas and in metastases. Loss of control of cell cycle checkpoints and the DNA repair system are important in the pathogenesis of malignant cancers (Poehlmann and Roessner, 2010). Allelic loss of the tumour suppressor gene BRCA1 has been detected in human insulinomas (Hrasc´an et al., 2008). Since BRCA1/ BARD1 normally regulates ubiquitination during the DNA damage response, down-regulation of the BRCA1 pathway in the high-metastatic group of canine insulinomas and in metastases suggests that this pathway may have a role in tumour progression. Activation of the ATM/ATR pathway after DNA damage normally prevents cells from entering S phase of the cell cycle. Abundant data in humans and mice have demonstrated that the G1/S checkpoint in the cell cycle is critical in regulating pancreatic b-cell replication (Fiaschi-Taesch et al., 2009). Down-regulation of the ATM/ATR pathway in the high-metastatic group of insulinomas and in metastases suggests that down-regulation of this pathway in insulinomas contributes to ineffective cell cycle arrest and inadequate apoptotic responses after DNA damage. Aberrant expression of p53 protein, which is the central downstream mediator of this pathway, has been implicated in insulinoma tumorigenesis (Pavelic´ et al., 1995, 1996). Conclusions Differences in gene expression profiles were identified by microarray analysis in low-metastatic and high-metastatic subsets
F.O. Buishand et al. / The Veterinary Journal 197 (2013) 192–197
of primary canine insulinomas and their metastases. In particular, genes for the acinar enzymes pancreatic lipase (PNLIP) and chymotrypsinogen B1 (CTRB1) were down-regulated substantially in a more malignant subset of primary insulinomas and in metastases. Co-localisation of insulin and PNLIP was identified by immunofluorescence in insulinoma cells, supporting the hypothesis that canine insulinomas contain amphicrine cells. Conflict of interest statement None of the authors of this paper has a financial or personal relationship with other people or organisations that could inappropriately influence or bias the content of the paper. Acknowledgements The authors gratefully acknowledge the technical assistance of A. Slob, E.P. Timmermans-Sprang and M.E. van Wolferen. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.tvjl.2013.01.021. References Brinkhof, B., Spee, B., Rothuizen, J., Penning, L.C., 2006. Development and evaluation of canine reference genes for accurate quantification of gene expression. Analytical Biochemistry 356, 36–43. Buishand, F.O., Kik, M., Kirpensteijn, J., 2010. Evaluation of clinico-pathological criteria and the Ki67 index as prognostic indicators in canine insulinoma. The Veterinary Journal 185, 62–67. Buishand, F.O., van Erp, M.G., Groenveld, H.A., Mol, J.A., Kik, M., Robben, J.H., Kooistra, H.S., Kirpensteijn, J., 2012. Expression of insulin-like growth factor-1 by canine insulinomas and their metastases. The Veterinary Journal 191, 334– 340. De Sá, S.V., Corrêa-Giannella, M.C., Machado, M.C., Krogh, K., De Almeida, M.Q., Pereira, M.A.A., Siqeira, S.A.C., Patzina, R.A., Ibuki, F.S., Sogayar, M.C., Machado, M.C., Giannella-Neto, D., 2007. Serpin peptidase inhibitor clade A member 1 as a potential marker for malignancy in insulinomas. Clinical Cancer Research 13, 5322–5330. Fiaschi-Taesch, N., Bigatel, T.A., Sicari, B., Takane, K.K., Salim, F., Velazquez-Garcia, S., Harb, G., Selk, K., Cozar-Castellano, I., Stewart, A.F., 2009. Survey of the human pancreatic b-cell G1/S proteome reveals a potential therapeutic role for cdk-6 and cyclin D1 in enhancing human b-cell replication and function in vivo. Diabetes 58, 882–893. Fidler, I.J., 2002. Critical determinants of metastasis. Seminars in Cancer Biology 12, 89–96. Hrasc´an, R., Pec´ina-Slaus, N., Martic´, T.N., Colic´, J.F., Gall-Troselj, K., Pavelic´, K., Karapandza, N., 2008. Analysis of selected genes in neuroendocrine tumours: Insulinomas and phaeochromocytomas. Journal of Neuroendocrinology 20, 1015–1022. Huang, X., Sheu, L., Kang, Y., Eto, Y., Kojima, I., Gaisano, H.Y., 2002. Effects of selective endocrine or exocrine induction of AR42J on SNARE and Munc 18 protein expression. Pancreas 25, e56–e63.
197
Kamisawa, T., Tu, Y., Egawa, N., Ishiwata, J.I., Tsuruta, K., Okamoto, A., Hayashi, Y., Koike, M., Yamaguchi, T., 2002. Ductal and acinar differentiation in pancreatic endocrine tumors. Digestive Diseases and Sciences 47, 2254–2261. Klimstra, D.S., Rosai, J., Heffess, C.S., 1994. Mixed acinar–endocrine carcinomas of the pancreas. American Journal of Surgical Pathology 18, 765–778. Ku, H.T., 2008. Pancreatic progenitor cells – Recent studies. Endocrinology 149, 4312–4316. Moore, A.S., Nelson, R.W., Henry, C.J., Rassnick, K.M., Kristal, O., Ogilvie, G.K., Kintzer, P., 2002. Streptozocin for treatment of pancreatic islet cell tumors in dogs: 17 cases (1989–1999). Journal of the American Veterinary Medical Association 221, 811–818. Pan, F.C., Wright, C., 2011. Pancreas organogenesis: From bud to plexus to gland. Developmental Dynamics 240, 530–565. Pavelic´, K., Hrasc´an, R., Kapitanovic´, S., Karapandza, N., Vranes, Z., Belicza, M., Kruslin, B., Cabrijan, T., 1995. Multiple genetic alterations in malignant metastatic insulinomas. Journal of Pathology 177, 395–400. Pavelic´, K., Hrascan, R., Kapitanovic, S., Vranes, Z., Cabrijan, T., Spaventi, S., Korsic, M., Krizanac, S., Li, Y.Q., Stambrook, P., Gluckman, J.L., Pavelic, Z.P., 1996. Molecular genetics of malignant insulinoma. Anticancer Research 16, 1707– 1717. Pfaffl, M.W., Horgan, G.W., Dempfle, L., 2002. Relative expression software tool (REST) for 420 group-wise comparison and statistical analysis of relative expression results in real-time 421 PCR. Nucleic Acids Research 30, e36. Poehlmann, A., Roessner, A., 2010. Importance of DNA damage checkpoints in the pathogenesis of human cancers. Pathology – Research and Practice 206, 591– 601. Puri, S., Hebrok, M., 2010. Cellular plasticity within the pancreas – Lessons learned from development. Developmental Cell 18, 342–356. Rao, N.A., van Wolferen, M.E., van den Ham, R., van Leenen, D., Groot Koerkamp, M.J., Holstege, F.C., Mol, J.A., 2008. CDNA microarray profiles of canine mammary tumour cell lines reveal deregulated pathways pertaining to their phenotype. Animal Genetics 39, 333–345. Regitnig, P., Spuller, E., Denk, H., 2001. Insulinoma of the pancreas with insular– ductular differentiation in its liver metastasis – Indication of a common stemcell origin of the exocrine and endocrine components. Virchows Archives 438, 624–628. Rosewicz, S., Vogt, D., Harth, N., Grund, C., Franke, W.W., Ruppert, S., Schweitzer, E., Riecken, E.O., Wiedenmann, B., 1992. An amphicrine pancreatic cell line: AR42J cells combine exocrine and neuroendocrine properties. European Journal of Cell Biology 59, 80–91. Steiner, J.M., Bruyette, D.S., 1996. Canine insulinoma. Compendium on Continuing Education for the Practising Veterinarian 18, 13–23. Terada, T., Kitamura, Y., Ashida, K., Matsunaga, Y., Kato, M., Harada, K., Morita, T., Ohta, T., Nakanuma, Y., 1997. Expression of pancreatic digestive enzymes in normal and pathologic epithelial cells of the human gastrointestinal system. Virchows Archives 431, 195–203. Tobin, R.L., Nelson, R.W., Lucroy, M.D., Wooldridge, J.D., Feldman, E.C., 1999. Outcome of surgical versus medical treatment of dogs with b cell neoplasia: 39 cases (1990–1997). Journal of the American Veterinary Medical Association 215, 226–230. Tusher, V.G., Tibshirani, R., Chu, G., 2001. Significance analysis of microarrays applied to ionizing radiation response. Proceedings of the National Academy of Sciences of the United States of America 98, 5116–5121. Van der Loos, C.M., 2008. Multiple immunoenzyme staining: Methods and visualizations for the observation with spectral imaging. Journal of Histochemistry and Cytochemistry 56, 313–328. Vortmeyer, A.O., Huang, S., Lubensky, I., Zhuang, Z., 2004. Non-islet origin of pancreatic islet cell tumors. Journal of Clinical Endocrinology and Metabolism 89, 1934–1938. Yang, Y.H., Dudoit, S., Luu, P., Lin, D.M., Peng, V., Ngai, J., Speed, T.P., 2002. Normalization for cDNA microarray data: A robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Research 30, e15.