Expression of claudin 1, claudin 4, and claudin 7 in colorectal cancer and its relation with CLDN DNA methylation patterns

(1)

https://doi.org/10.1177/1010428317697569 Tumor Biology April 2017: 1 –10

sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1010428317697569 journals.sagepub.com/home/tub

Creative Commons Non Commercial CC BY-NC: This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 License (http://www.creativecommons.org/licenses/by-nc/4.0/) which permits non-commercial use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage).

Introduction

Targeting tumor suppressor genes by epigenetic silencing especially in the promotor region is one of the probable mechanisms of carcinogenesis.¹ However, differential methylation may also occur outside of the promoter region and previous studies have shown that gene expression and tissue differentiation are powerfully affected by such methylation sites.^1,2 So far, there are limited data available on claudin (CLDN), epigenetics, and the relation to claudin protein expression as well as its role in the development of colorectal carcinoma (CRC). While DNA methylation of claudin 6 has been shown to be involved in the invasion of breast cancer cells³ and hypermethylation of claudin 11 in malignant melanoma of the skin,⁴ there is

no information of claudin methylation and CRC. There are, however, several studies on gene and protein expression of claudins in CRC; several of the studies

Expression of claudin 1, claudin 4, and claudin 7 in colorectal cancer and its relation with CLDN DNA methylation patterns

Victoria Hahn-Strömberg

¹

, Shlear Askari

²

, Abrar Ahmad

²

, Rahel Befekadu

^2,3

and Torbjörn K Nilsson

⁴

Abstract

Altered claudin expression has been described in colon, prostatic, ovarian, and breast carcinoma. However, the role of epigenetic modifications in these genes and their role in colorectal cancer is unknown. We aimed our study to investigate whether claudin protein expression and methylation of CLDN can influence the tumorigenesis of colorectal cancer. A total of 31 patients diagnosed with colorectal carcinoma was used in this study. Immunohistochemical staining was used to study protein expression in both tumor and the adjacent nonneoplastic mucosa of claudin 1, 4, and 7. To detect the DNA methylation pattern of CLDN1, 4, and 7, genomic DNA was extracted from both the tumor and the adjacent nonneoplastic mucosa. Methylation analysis was carried out using bisulfite pyrosequencing. Cell membrane staining intensity of all claudins was found significantly lower in colorectal cancer tissues when compared to paired normal mucosa (p ≤ 0.001). For claudin 4, the percentage of cells staining positively was also significantly reduced (p = 0.04). In normal mucosa, cytoplasm showed no staining for claudins in any patient, whereas in paired colorectal cancer tissues, significant cytoplasmic staining appeared both for claudin 1 (p = 0.04) and claudin 4 (p = 0.01). Tumor samples were significantly hypomethylated in CLDN1 (p < 0.05). In conclusion, our results show that CLDN1 is significantly hypomethylated in tumor samples and that the membrane staining intensity for claudin 1, 4, and 7 is significantly lower in colorectal cancer tissues than in adjacent nonneoplastic tissue. Colorectal cancer cells showed dystopic cytoplasmic location of claudins.

Keywords

Tight junction, methylation, colon cancer, claudin

Date received: 17 August 2016; accepted: 23 December 2016

1 Department of Medical Cell Biology, Uppsala University, Uppsala, Sweden

2 Department of Clinical Research, Faculty of Medicine and Health, Örebro University, Örebro, Sweden

3 Department of Laboratory Medicine, Faculty of Medicine and Health, Örebro University, Örebro, Sweden

4 Division of Clinical Chemistry, Department of Medical Biosciences, Umeå University, Umeå, Sweden

Corresponding author:

Victoria Hahn-Strömberg, Department of Medical Cell Biology, Uppsala University, SE 75105 Uppsala, Sweden.

Email: Victoria.Hahn_Stromberg.2886@student.uu.se

Original Article

(2)

ated with an increasing risk of CRC development,⁸ making it interesting to study the DNA methylation pattern in CLDN1, 4, and 7 in CRC and find out whether there is any relation with claudin expression levels and/

or other clinical parameters. Since this group of genes has been implicated in tight junctions (TJs) disruption in several cancer forms, we hypothesized that CLDN1, 4, and 7 could be epigenetically changed in CRC. In this study, we investigated the claudin protein expression in colon carcinoma patients and related this to CLDN gene DNA methylation patterns in CRC tissue samples as well as matched samples of ANNM from CRC patients. For this purpose, we developed sequence- specific, quantitative pyrosequencing methods for ana- lyzing DNA methylation both in CpG islands and in shores of the CLDN1, CLDN4, and CLDN7 genes.

Materials and methods

In this study, 31 randomly selected formalin-fixed, paraf- fin-embedded (FFPE) tissue samples from CRC patients diagnosed between 2001 and 2008 from both men and women were used along with paired ANNM samples (mar- gin to the tumor at least 5 mm) from the same patient as controls. All cases were anonymized and given consecu- tive numbers along with the year of diagnosis.

Clinicopathological parameters such as age, tumor location, tumor stage, and tumor grade lymph node metastatis were extracted, as shown in Table 1. The Regional Ethics Review Board, Uppsala, Sweden, approved the study on condition that the samples were de-identified.

DNA extraction and bisulfite treatment

DNA was manually extracted from FFPE tissue samples of CRC patients and ANNM according to the manufacturers’

protocol using Viogen Blood and tissue genomic DNA extraction Miniprep Kit (Viogen, Gentaur, Kampenhout, Belgium). From the extracted DNA, 500 ng was used for the bisulfite treatment performed with EZ DNA Methylation Gold kit according to the manufacturer’s instruction and eluted in 20 µL elution buffer (Zymo Research, Orion Diagnostica, Sweden).

Polymerase chain reaction amplification and primer design

The genomic sequences of all CLDN genes were obtained from (National Center for Biotechnology Information

(NCBI). CpG island searcher (http://cpgislands.usc.edu/) was used to identify CpG sites. Polymerase chain reaction (PCR) primers were designed for CLDN1, CLDN4, and CLDN7 by using Primer3 (http://bioinfo.ut.ee/primer3/).

PCR primers, sequencing primers, and annealing temperatures are summarized in Table 2.

PCR was performed using HotStarTaq plus DNA polymerase kit (QIAGEN AB, Sollentuna, Sweden), containing 0.15 µmol/L of each primer, 0.25 U of Taq polymerase, 2.5 mM MgCl₂, and 0.1 mM each of deoxyguanosine triphosphate (dGTP), deoxyadenosine triphosphate (dATP), deoxythymidine triphosphate (dTTP), and deoxycytidine triphosphate (dCTP). A total of 20 ng of bisulfate-treated DNA was added to PCR reaction as template. PCR reaction was conducted as follows: starting at 94°C for 10 min, fol- lowed by 49 cycles (45 s at 94°C, 30 s at the annealing temperature, and 30 s at 72°C), and a final 7-min hold at 72°C.

Pyrosequencing assays

Pyrosequencing was used to analyze the methylation frac- tions of the CpG sites. Methylation analysis was carried out using PyroMark Q96 ID (Qiagen.Biotage AB, Uppsala, Sweden) according to the manufacturer’s protocol. Pyro Q-CpG software (Biotage, Uppsala, Sweden) was used to analyze the methylation status. The nucleotide addition order was optimized by the Pyro Q-CpG software version 1.0.9 (Qiagen). Briefly, 25 µL of the PCR product master Age (years)

≤70 17 55

>70 14 45

Localization

Distal colon 9 29

Proximal colon 18 58

Rectum 4 13

Tumor grade

High 13 42

Low 18 58

Tumor infiltration (T)

T1 + T2 14 45

T3 + T4 17 55

Lymph node (N)

Absent 18 58

Present 13 42

Metastasis (M)

1 0 0

X 31 100

X: no metastasis.

(3)

Table 2. PCR primer sequences, annealing temperatures, amplicon sizes, and sequencing primer sequences for CLDN methylation analysis by pyrosequencing. GeneCpG sitePCR primer sequence 5′–3′TmSize (bp)Sequencing primer 5′–3′ CLDN14–6Fw5′-biotin GGATGTTTAGAGATGGGTTTGAA55212ACAACTAATTTCTATCAACCT RwAACCCTTTTACTCTACATTACAAC 16–18FwGTTTGTTGTTTAGGTTGGAGTGTAG57341AGTTTTTTAAAGTGTTGGG Rw5′-biotin CATTTTCTTCTTTTTTTTATTCCAAACAT 23–26FwGAAGGAAGTTTTGGAGAAATATAGAGG5995GGAGAAATATAGAGGGAAATTA Rw5′-biotin TCCAACTTTTACTATAACTAACTATCTCAC 86–90Fw5′-biotin TTTTTGTATTTGTTATTTTTGAGTTAG56249AAAATAAAACCCAACAACT RwCAAAAACTTTACACTAAATCTACCC CLDN45–8FwGGGTTAAGGTATAGGAGTGTGA56275GGTTGGAGTGTAATGG Rw5′-biotin CCAACATAATAAAACCCCATCTCT 23–26Fw5′biotin TTTAGTATGGTTGGGAAGAGAGAT60215CTAAAACTCATCCCA RwACCTAAAATATCACCAAACCCTAAAA 37–41FwAGTAGAGAAGTTGGGTGATAAG55215GAGTGGGTGATGTATT Rw5′-biotin CCACCTTAATTATTATCATTTCTTATCCT 63–67Fw5′-biotin GTTTTTATGGGGTTATAGGTAATGGG55158AAATAACAATATTACTAC RwCAATTCATCCATAAACCCTCCCAAATAA 82–97Fw5′-biotin GTATTAATTGTTTGGAGGATGAAAG55246AAAATCTTAAATAATATTATAAACC RwAACAATTACAACAAAACAACCCCCAAAAAACAACAAAC 102–104FwGGGGGTTGTTTTGTTGTAATTGTTTAT59250GTTTTGTTGTTGTTAGTAAT Rw5′-biotin ACTACTACCTTCCTACCTAACTC CLDN716–18FwGAGGTGGGTTGGGATTATTTGAG58136ATTATTTGAGGTTAGGAATT RwACCTCCCAAATAACTAAAAATATAAATACC 58–64FwGAGGAGAGAGTATTTTGGATAGAAT58106GAGAGTATTTTGGATAGAATT RwCCTTCCTTTCTTCCCAACAA 95–101FwTTTTTATTTGGGTAAGGAGGG58263GTTTTATATATATTTAGG RwACAAATCCCAAAATATCCTAAACTATAGTATTTGTTGGTTTATT PCR: polymerase chain reaction; Tm: melting temperature; bp: base pair.

(4)

placed onto the streptavidin sepharose beads and then passed through one denaturation step and two washing steps (70% ethanol, denaturation using 0.2 mol/L NaOH, and washing with 10 mmol/L Tris–acetate and Milli-Q water).

The products were then transferred to a plate containing sequencing primer (0.4 µmol/L) in 40 µL annealing buffer (20 mmol/L Tris–acetate, 2 mmol/L magnesium acetate, pH 7.6).

The amplicons and sequencing primers were incubated at 80°C for 2 min, cooled down to room temperature to permit annealing of the sequencing primer before putting the plate into PyroMark Q96 ID. The substrate mixture, deoxynucleotide triphosphates (dNTPs), and enzymes were added to the cartridge according to the manufacturer’s protocol (PyroMark Q96 ID system; QIAGEN AB).

Immunohistochemical analysis

Slides preparation and staining. For immunohistochemical analysis, 4-µm sections were made from FFPE blocks containing material from colorectal cancer patients by using Leica microtome (Buffalo Grove, IL, USA). The sections were placed on super-frost slides and incubated for 60 min at 62°C. The slides were then pretreated for 1 h and 20 min with EDTA buffer saline solution with pH 9.3 at 97°C.

Primary antibodies for anti–claudin 1 and anti–claudin 7 were purchased from Invitrogen (Carlsbad, CA, USA) and anti–claudin 4 from Abcam (Cambridge, UK). All antibodies were polyclonal rabbit antibodies and were designed specifically for FFPE tissue staining. Staining was performed according to the manufacturer’s protocol (DAKO- Cytomation, Denmark) with incubation of the primary antibodies for 30 min at working dilutions of 1:200, 1:300, and 1:1000 for anti–claudin 1, 4, and 7, respectively.

Scoring of slides. Slides were scored on the scales of inten- sity and ratio for membrane and cytoplasm staining, with intensity scores of 0, 1, 2, and 3 where 0 = no staining, 1 = weak, 2 = moderate, and 3 = strong staining, whereas the percentage of the cellular population staining positively was scored with grade 0 representing <10% stained cells; grade 1, 10%−50% stained cells; grade 2, 50%–80%

stained cells; and grade 3, >80% stained cells.

Statistics

Statistical analysis was performed using SPSS version 15.0 software (SPSS, Inc, Chicago, IL, USA). Quantitative parameters were analyzed by mean and standard deviation, while qualitative parameters with percentage and frequencies.

Results

In this study, a total number of 62 FFPE samples were used from 31 patients diagnosed with colorectal cancer. Half of the samples were from paired ANNM and the other half from the tumor. The male/female sex ratio was 20:11. A total of 18 tumors (58.1%) were located in the proximal colon, 9 (29%) in the distal colon, and 4 (12.9%) in rectum. According to the World Health Organization’s (WHO) classification, out of 31 CRC samples, 13 (41.9%) were high-grade and 18 (58.1%) were low-grade carcinomas.

Membrane staining

Claudin 1 staining intensity was moderate in 10 CRC tumor tissue samples and strong in 21 CRC tissue samples. Conversely, all the 31 matched ANNM samples showed strong claudin 1 staining intensity (p = 0.001).

Claudin 4 intensity was moderate in 15 CRC samples and strong in 16 CRC samples, whereas intensity was strong in all 31 matched ANNM samples (p < 0.001). Claudin 7 intensity was moderate in 9 CRC samples and strong in 22 CRC samples. All the 31 matched ANNM samples showed strong intensity (p = 0.001) (Figures 1 and 2).

The ratios of positive staining for claudin 1, 4, and 7 showed that for claudin 1 there was no significant differ- ence between CRC tissue and ANNM tissue (p = 0.4).

Similar results were observed for claudin 7 as well (p = 1.0).

However, the ratio of cell staining for claudin 4 was sig- nificantly lower in CRC tissue than in ANNM (p = 0.04).

Cytoplasmic staining

Claudin 1 staining intensity in the cytoplasm was graded moderate in five and strong in one CRC samples. All 31 matched ANNM samples showed no staining (p = 0.04).

The ratio of cells staining positively in the cytoplasm was generally significantly higher in CRC tissue than in ANNM tissue (p = 0.01). Claudin 4 intensity was moderate in six and weak in two CRC samples, whereas staining was neg- ative in all 31 matched ANNM samples (p < 0.01). The ratio of cells staining positively in the cytoplasm was significantly higher in CRC tissue than in ANNM tissue (p = 0.001). Claudin 7 was weakly expressed just in one CRC sample, and the 31 matched ANNM samples all showed no cytoplasmic staining (p = 0.31). The ratio of cells staining positively was not statistically significantly different between CRC tissue and ANNM tissue (p = 0.31) (Figures 1 and 2).

(5)

Figure 1. Immunohistochemical staining for CLDN 1 in colorectal tissues: (a) Claudin 1 membrane staining—1: ANNM tissue;

2: low-grade tumor tissue; and 3 high-grade tumor tissue. No cytoplasmic staining found. (b) Claudin 1 membrane and cytoplasmic staining—1: low staining in tumor tissue and 2: moderate staining in tumor tissue.

Figure 2. Immunohistochemical staining for CLDN 4 in colorectal tissues: (a) Claudin 4 membrane staining—1: ANNM tissue;

2 low-grade tumor tissue; and 3: high-grade tumor tissue. No cytoplasmic staining found. (b) Claudin 4 membrane and cytoplasmic staining—1: low staining in tumor tissue and 2: moderate staining in tumor tissue.

(6)

DNA methylation in colorectal cancer tissue

Three methylation assays were designed for CLDN1, four for CLDN4, and five for CLDN7 covering a total number of 12, 18, and 21 CpG sites, respectively. The methods were designed to cover methylation status of the genes in both CpG islands and shores of all three genes. The selected locations are shown in Supplemental Figures 1–3.

In CLDN1, there was a significant hypomethylation upstream of the CpG island (CpG sites 23–26) between tumors and ANNM (p < 0.01), but no statistically signifi- cant difference was observed between high- and low-grade tumors (see Figure 3 and Table 3).

In CLDN4, there was a significant hypomethylation in CRC tissue as compared to paired ANNM for one CpG site (no. 24), while other CpG sites were non-significantly hypomethylated in CRC both upstream of the CpG island and in the CpG island itself (Figure 4, Table 4). There was frequent assay failure of the long amplicons (≈250 base pair (bp)) covering CpG sites 83–104, consuming up all available DNA from ANNM tissues, and thus we lack those data for these sites.

In CLDN7, a single CpG site (no. 42) showed significant hypomethylation in CRC tissue (p < 0.001) (Figure 5, Table 5).

Discussion

TJs have recently been targets of many studies regarding their role in the development of cancer.^8–12 We show here that cell membrane staining for claudin 1, 4, and 7 is stronger in normal mucosa than in CRC samples. In con- trast, cytoplasmic staining appeared higher for claudin 1

and 4 in several CRC cases, whereas ANNM cytoplasm showed no staining for claudins in any patient. Tumor tissues were significantly hypomethylated as compared to normal mucosa in CLDN1. In CLDN4 and 7, a single CpG site for each was significantly hypomethylated as compared to ANNM.

Current reports indicate that CLDN gene expression is regularly changed in several cancer forms, being up- or downregulated according to the tissue affected by cancer.^13,14 A previous study showed that high CLDN1 expression in CRC is associated with better prognosis.¹⁵ Similar results were reported by other researchers as well in different types of cancers,^9,16,17 while other studies claim that upregulated CLDN1 is associated with higher risk of CRC and hepatic metastasis.^18,19

Our results show that not only claudin 1 but also claudin 4 and 7 expression is significantly lower in tumor samples than in ANNM samples. Similar findings were reported by previous studies showing that claudin 4 and 7 are important for maintaining TJs in normal tissues.^8,19,20 However, opposite results for claudin 4 and 7 have been reported by other studies in different types of cancers,^21,22 suggesting that the roles of the various claudins may be cancer and tissue specific.

We also found that claudin 1, 4, and 7 were frequently expressed in cytoplasm of CRC samples but there was no cytoplasmic expression of claudins in ANNM samples which is in accordance with previous studies showing cytoplasmic expression of claudins in different types of cancers.^20,23,24 CLDN genes encode proteins that contains four transmembrane domains with their N- and C-terminal Figure 3. DNA methylation profile of the CLDN1 gene. Upper: Normal mucosa versus colorectal cancer tissue, CpG sites 23–25 were significantly different (p < 0.001). Lower: High-grade versus low-grade colorectal cancer tissue. The differences are not statistically significant. Lines show mean values.

(7)

Table 3. DNA methylated fraction of the CpG sites in the CLDN1 gene in normal tissues and high- or low-grade CRC.

CpG site High-grade tumor Low-grade tumor p¹ p² p³ p⁴

N Mean ± SD N Mean ± SD N Mean ± SD

4 1 99 12 96 ± 3.4 13 96 ± 2.2 0.63

5 1 88 12 95 ± 1.4 13 95 ± 1.6 0.00

23 26 56 ± 11 13 29 ± 15 18 22 ± 11.2 0.00 0.00 0.00 0.24

24 26 83 ± 12 13 52 ± 25 18 40 ± 23 0.00 0.00 0.00 0.22

25 26 78 ± 11 13 44 ± 24.4 18 34 ± 23.3 0.00 0.00 0.00 0.34

84 7 3.1 ± 6 10 3 ± 4.2 14 5.5 ± 17 0.84 1.00 0.91 0.85

85 7 3 ± 5 10 1.3 ± 3 14 7 ± 22 0.65 0.97 0.84 0.64

86 7 4.3 ± 4 10 2.2 ± 3 14 8 ± 22.4 0.66 0.96 0.87 0.65

87 7 6 ± 4 10 5.3 ± 2.5 14 8 ± 16 0.87 1.00 0.92 0.88

88 7 4.3 ± 5 10 3 ± 3.3 14 8 ± 23 0.78 0.97 0.90 0.77

89 7 1.1 ± 2 10 0.8 ± 2 14 6.9 ± 23 0.57 1.00 0.71 0.61

90 7 4.3 ± 4 10 0.9 ± 2 14 9.1 ± 22 0.43 0.90 0.78 0.41

CRC: colorectal carcinoma; SD: standard deviation; ANOVA: analysis of variance.

p¹: ANOVA between all three groups.

p²: Normal mucosa versus high-grade tumor tissue.

p³: Normal mucosa versus low-grade tumor tissue.

p⁴: High-grade versus low-grade tumor tissue.

Figure 4. DNA methylation profile of the CLDN4 gene. Upper: Normal mucosa versus colorectal cancer tissue. CpG site 24 is statistically significantly different. Lower: High-grade versus low-grade colorectal cancer tissue. The differences are not statistically significant. Lines show mean values.

ends located in the cytoplasm, one possible explanation for these results could be the presence of mutations rendering claudins more sensitive to being cleaved off from cell membranes. Another possibility could be aberrant activa- tion of multiple signaling pathways in colorectal cancer, causing the observed claudin mislocalization.

We found that CLDN1, 4, and 7 genes are hypomethyl- ated in CRC samples as compared to normal mucosa tis- sues. Among them, CLDN1 was significantly hypomethylated in CRC samples (p < 0.05), whereas in CLDN4, only CpG site, that is, 24 was significantly hypo- methylated in CRC samples while a trend was observed between other CpG sites. Moreover, in CRC samples,

CLDN1 and CLDN4 showed lower membranous expres- sion as compared to normal mucosa samples. Methylation status of CLDN1 and 4 has not been studied previously in CRC patients. However, there are studies on the methyla- tion of CLDN4 in bladder and ovarian carcinoma, report- ing that hypermethylation of CLDN4 is associated with lower expression of protein and vice versa.^25,26 Similar results were reported on CLDN1 in breast cancer.²⁷ Our results challenge the previous findings about CLDN1 and CLDN4 methylation and association with claudin expres- sion. Methylation of CLDN7 was not significantly differ- ent among CRC and ANNM samples in our study except for one CpG site, that is, 42 that was hypomethylated in

(8)

CRC compared to ANNM. Nakayama et al.²⁸ reported that hypermethylation can lead to lower expression of CLDN 7 in CRC and is associated with progression of tumors.

Similarly, downregulation of CLDN7 was reported in a breast cancer study showing that hypermethylation of promoter sequences for this gene leads to inactivation of the CLDN7 gene.²⁹

However, these reports do not consider the aberrant compartmentalization of claudin proteins that we describe here, and they did not use pyrosequencing to assess DNA methylation, which is now the gold standard method. Such methodological differences may explain the discrepancies between different studies on this topic. At any rate, in our hands, CLDN gene hypomethylation seems to be linked to

24 22 47 ± 20 11 36 ± 17 16 30 ± 16.5 0.02 0.26 0 0.68

25 22 36 ± 17.2 11 40 ± 21 16 32 ± 18.5 0.51 0.85 0.7 0.49

26 22 36 ± 19 11 29 ± 16 16 23 ± 16.5 0.91 0.5 0.1 0.71

63 17 48 ± 22.3 13 35 ± 21 18 33 ± 16 0.07 0.2 0.1 0.96

64 17 54 ± 24 13 42 ± 22 18 39 ± 16 0.1 0.27 0.1 0.92

65 17 53 ± 25 13 39 ± 22 18 38 ± 17 0.1 0.2 0.1 0.10

66 17 56 ± 26 13 45 ± 25 18 41 ± 18 0.17 0.41 0.2 0.90

67 17 51 ± 24 13 39 ± 22 18 36 ± 15.2 0.08 0.26 0.1 0.89

68 17 44 ± 20 13 38 ± 23.1 18 34 ± 15 0.33 0.7 0.3 0.83

83 − − 10 39 ± 30 13 33 ± 20 − − − 0.51

84 − − 10 31 ± 26.5 13 30 ± 32.3 − − − 0.91

85 − − 10 44 ± 33 13 30 ± 20 − − − 0.22

86 − − 10 43 ± 32.5 13 31 ± 20 − − − 0.28

87 − − 10 45 ± 35.5 13 34 ± 19.3 − − − 0.32

102 − − 13 38 ± 28 12 35 ± 27 − − − 0.74

103 − − 13 32 ± 28.3 12 24 ± 21.5 − − − 0.48

104 − − 13 21 ± 15.3 12 20 ± 14.5 − − − 0.94

Figure 5. DNA methylation profile of the CLDN7 gene. Upper: Normal mucosa versus colorectal cancer tissue. Lower: High- grade versus low-grade colorectal cancer tissue. Lines show mean values.

(9)

changes in compartmentalization of the claudin proteins rather than simple up- or downregulation. Such changes could be linked to differences in biological properties of the tumors such as invasiveness, metastasis, and perhaps even prognosis. Detailed IHC studies of claudin expression patterns at the invasive front of CRC tumors will be the next step to further explore this possibility.

To our knowledge, this is the first study in which expression of claudin 1, 4, and 7 along with their methylation status is analyzed in paired CRC and ANNM samples. Our results indicate that membranous expression of claudin 1, 4, and 7 is decreased in CRC samples, whereas expression in the cytoplasmic compartment is absent in ANNM but a fre- quent finding in CRC tissues. In addition, we found CLDN1 to be significantly hypomethylated in tumor as compared to ANNM tissues. Methylation of the CLDN1 gene may thus be an important factor in the tumorigenesis of colorectal carcinoma and claudin 1, 4, and 7 may be suited as IHC mark- ers for CRC. Future studies along these lines will resolve the potential clinical utility of these findings.

Acknowledgements

We thank Sanja A. Farkas for advice on the design of pyrosequencing methods.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This work was financially supported by Örebro Läns Landstings Forskningskommitté, Lions cancerfond, Uppsala-Örebro, and Nyckelfonden, Örebro.

References

1. Baylin SB. DNA methylation and gene silencing in cancer.

Nat Clin Pract Oncol 2005; 2: S4–11.

2. Bird A. DNA methylation patterns and epigenetic memory.

Genes Dev 2002; 16: 6–21.

3. Liu Y, Jin X, Li Y, et al. DNA methylation of claudin-6 pro- motes breast cancer cell migration and invasion by recruit- ing MeCP2 and deacetylating H3Ac and H4Ac. J Exp Clin Cancer Res 2016; 35(1): 120.

4. Walesch SK, Richter AM, Helmbold P, et al. Claudin11 promoter hypermethylation is frequent in malignant mela- noma of the skin, but uncommon in nevus cell nevi. Cancers 2015; 7(3): 1233–1243.

5. Ding L, Wang L, Sui L, et al. Claudin-7 indirectly regulates the integrin/FAK signaling pathway in human colon cancer tissue. J Hum Genet 2016; 61(8): 711–720.

Table 5. DNA methylated fraction of the CpG sites in the CLDN7 gene in normal tissues and high- or low-grade CRC.

CpG site Normal High-grade tumor Low-grade tumor p¹ p² p³ p⁴

N Mean ± SD N Mean ± SD N Mean ± SD

16 18 49 ± 26 13 48 ± 17.1 18 43 ± 17 0.64 0.1 0.64 0.77

17 18 30 ± 14.4 13 36 ± 14.4 18 33 ± 12 0.42 0.39 0.73 0.81

18 18 53 ± 23 13 54 ± 21 18 49 ± 16 0.75 0.96 0.86 0.74

25 28 42 ± 25.5 13 38 ± 15.2 18 35 ± 17 0.55 0.89 0.52 0.88

26 28 21 ± 12.3 13 20 ± 11 18 14 ± 7.4 0.07 0.92 0.06 0.27

27 28 31 ± 19.1 13 31 ± 11 18 27 ± 14 0.71 0.99 0.7 0.85

28 28 21 ± 15 13 16 ± 5.1 18 17 ± 7 0.32 0.36 0.53 0.92

41 25 4 ± 3 13 3.1 ± 2 18 3.2 ± 2 0.46 0.55 0.56 0.99

42 25 5.1 ± 3 13 1.2 ± 1.3 18 0.3 ± 1 0 0.00 0.00 0.59

43 25 3 ± 3.3 13 2.4 ± 2 18 1.8 ± 2 0.41 0.86 0.38 0.79

58 15 4 ± 5 13 3.7 ± 2.2 18 3.2 ± 2.1 0.84 0.1 0.85 0.9

59 15 6 ± 8 13 5.2 ± 4 18 4.3 ± 3.1 0.69 0.95 0.67 0.88

60 15 7 ± 9 13 8.8 ± 5 18 6.4 ± 3.6 0.49 0.95 0.99 0.51

61 15 7 ± 5 13 7.6 ± 5 18 4.9 ± 3.5 0.21 0.81 0.49 0.2

62 15 6.5 ± 7 13 6.1 ± 2.2 18 4.9 ± 3.3 0.58 0.96 0.57 0.77

63 15 7 ± 5.2 13 4.7 ± 2 18 4.2 ± 2 0.05 0.18 0.05 0.9

64 15 6 ± 8 13 6.5 ± 4 18 4.5 ± 3 0.57 0.9 0.81 0.55

98 18 3.4 ± 4 8 4.9 ± 4 14 2.6 ± 3.5 0.41 0.62 0.84 0.38

99 18 17 ± 16.1 8 8.3 ± 4 14 10 ± 15 0.24 0.31 0.38 0.94

100 18 6.2 ± 8 8 3.9 ± 5 14 8.2 ± 16.3 0.68 0.88 0.87 0.66

101 18 5 ± 6.3 8 1 ± 3 14 3 ± 5 0.2 0.19 0.55 0.68

(10)

8. Ahmad A BR, Askari S and Hahn-Strömberg V. Decreased expression of claudin 1, 3, 4, 5 and 7: a new prognostic marker in colon carcinoma. J Gastrointest Cancer Stromal Tumors 1: 105.

9. Ayrolles-Torro A, Vezzio-Vie N, Denis V, et al. Claudin-1 (CLDN1) as a new therapeutic target in colorectal cancer: inhibition of cell growth and survival by an anti-CLDN1 monoclonal antibody.

Mol Cancer Ther 2013; 12(11 Supplement): B245–B245.

10. Befekadu R and Hahn-Stromberg V. Polymorphisms in claudin 1 and claudin 7 in healthy individuals in Sweden.

Diagn Pathol 2016; 1: 113.

11. Kwon MJ. Emerging roles of claudins in human cancer. Int J Mol Sci 2013; 14(9): 18148–18180.

12. Singh AB and Dhawan P. Claudins and cancer: fall of the soldiers entrusted to protect the gate and keep the barrier intact. Semin Cell Dev Biol 2015; 42: 58–65.

13. Morin PJ. Claudin proteins in human cancer: promising new targets for diagnosis and therapy. Cancer Res 2005; 65(21):

9603–9606.

14. Swisshelm K, Macek R and Kubbies M. Role of claudins in tumorigenesis. Adv Drug Deliv Rev 2005; 57(6): 919–928.

15. Nakagawa S, Miyoshi N, Ishii H, et al. Expression of CLDN1 in colorectal cancer: a novel marker for prognosis.

Int J Oncol 2011; 39(4): 791–796.

16. Morohashi S, Kusumi T, Sato F, et al. Decreased expression of claudin-1 correlates with recurrence status in breast cancer. Int J Mol Med 2007; 20(2): 139–143.

17. Matsuoka T, Mitomi H, Fukui N, et al. Cluster analysis of claudin-1 and -4, E-cadherin, and β-catenin expression in colorectal cancers. J Surg Oncol 2011; 103(7): 674–686.

18. Kinugasa T, Akagi Y, Ochi T, et al. Increased claudin-1 protein expression in hepatic metastatic lesions of colorectal cancer. Anticancer Res 2012; 32(6): 2309–2314.

19. Tang W, Dou T, Zhong M, et al. Dysregulation of claudin family genes in colorectal cancer in a Chinese population.

BioFactors 2011; 37(1): 65–73.

carcinoma. Lab Invest 2011; 91(11): 1652–1667.

22. Rendon-Huerta E, Teresa F, Teresa GM, et al. Distribution and expression pattern of claudins 6, 7, and 9 in diffuse- and intestinal-type gastric adenocarcinomas. J Gastrointest Cancer 2010; 41(1): 52–59.

23. Rangel LB, Agarwal R, D’Souza T, et al. Tight junction proteins claudin-3 and claudin-4 are frequently overex- pressed in ovarian cancer but not in ovarian cystadenomas.

Clin Cancer Res 2003; 9(7): 2567–2575.

24. Dhawan P, Singh AB, Deane NG, et al. Claudin-1 regulates cellular transformation and metastatic behavior in colon cancer. J Clin Invest 2005; 115(7): 1765–1776.

25. Boireau S, Buchert M, Samuel MS, et al. DNA- methylation-dependent alterations of claudin-4 expression in human bladder carcinoma. Carcinogenesis 2007; 28(2):

246–258.

26. Litkouhi B, Kwong J, Lo CM, et al. Claudin-4 overexpres- sion in epithelial ovarian cancer is associated with hypomethylation and is a potential target for modulation of tight junction barrier function using a C-terminal fragment of Clostridium perfringens enterotoxin. Neoplasia 2007; 9(4):

304–314.

27. Di Cello F, Cope L, Li H, et al. Methylation of the claudin 1 promoter is associated with loss of expression in estrogen receptor positive breast cancer. PLoS ONE 2013; 8(7):

e68630.

28. Nakayama F, Semba S, Usami Y, et al. Hypermethylation- modulated downregulation of claudin-7 expression pro- motes the progression of colorectal carcinoma. Pathobiology 2008; 75(3): 177–185.

29. Kominsky SL, Argani P, Korz D, et al. Loss of the tight junction protein claudin-7 correlates with histologi- cal grade in both ductal carcinoma in situ and invasive ductal carcinoma of the breast. Oncogene 2003; 22(13):

2021–2033.