Impairment of peroxisomal biogenesis in human colon carcinoma

Peroxisomes and the activities of their enzymes have been reported to be significantly reduced in various types of tumors including the colon carcinoma. Therefore, the present study was designed to investigate the gene expression of several peroxisomal proteins in human colon carcinoma and additionally those of the peroxisome proliferator activated receptor α (PPARα) and PEX5, a receptor protein involved in the import of most peroxisomal matrix proteins. Samples from adenocarcinomas and adjacent normal colon were analyzed by immunohistochemistry and western blotting. The mRNA content was assessed by a novel sensitive dot blot RNase protection assay and northern blotting. By immunohistochemistry, peroxisomes were distinctly visualized in normal colonocytes but were not detected in colon carcinoma cells. The protein levels of catalase (CAT), acyl-CoA oxidase as well as the 22 and 70 kDa peroxisomal membrane proteins (PMP22 and PMP70) were all significantly decreased in carcinomas. The corresponding mRNAs for CAT and PMP70, however, were unchanged. In contrast, the mRNA of PEX5 was significantly increased. The expression of PPARα was not altered in tumors, neither at protein nor mRNA levels. These observations show that the reduction of peroxisomes and their proteins in colon carcinoma is not due to a generalized reduction of transcription of their genes. It seems more likely that this phenomenon is regulated at a post-transcriptional or translational level. Alternatively, and more likely, an impairment of the biogenesis of the organelle could account for the paucity of peroxisomes in colon carcinoma.

AOX, acyl-CoA oxidase, CAT, catalase, PMP22, 22 kDa peroxisomal membrane protein, PMP70, 70 kDa peroxisomal membrane protein, PPAR, peroxisome proliferator activated receptor.

Introduction

Peroxisomes are single membrane-bound organelles which are almost ubiquitous in eukaryotic cells. In mammals, they contain >50 different enzymes which play an important role in various metabolic pathways (1) such as the β-oxidation of very long chain fatty acids (2) and prostaglandins (3), and the synthesis of cholesterol (4,5) and plasmalogens (6,7). Additionally, peroxisomes contain oxidases which produce hydrogen peroxide, and catalase (CAT) which degrades it (8). It has been well known for many decades that CAT activity is reduced in human tumors (9–11). Moreover, recent investigations showed that there is also a reduction of the activity of other peroxisomal enzymes in primary tumors of different tissues (12–19). In colon carcinomas CAT and two enzymes of the peroxisomal β-oxidation sequence, the acyl-CoA oxidase (AOX) and the multifunctional protein, are decreased (19–22) and, following observations by electron microscopy, a reduction of the number of peroxisomes has been reported (19,22).

The proliferation of peroxisomes and transcription of their genes is regulated by a subfamily of the ligand-activated nuclear hormone receptor superfamily, the so-called peroxisome proliferator activated receptor (PPAR) (23). Recent reports indicate that PPARα, a subtype of PPAR, plays a crucial role in the biology of colon cancer. Independent studies have revealed that the growth of several different colon carcinoma cell lines is inhibited by the ligand activation of PPARα (24) and that transplantable tumors in mice derived from human colon cancer cells show significant reduction of growth when mice are treated with troglitazone, a PPARα ligand (25). On the other hand, Lefebvre et al. (26) have reported that ligand activation of PPARα in C57BL/6J–APC^Min/+ mice promotes the development of colon tumors, and Saez et al. (27) found that it accelerated the formation of colonic polyps in the same mice. Although the exact mechanism of those divergent observations remains to be elucidated, they all point to serious alterations of the peroxisomal compartment and its regulation in colon carcinoma.

The present study was designed to investigate in human colon carcinoma the change in expression of (i) peroxisomal proteins, (ii) the transcription factor PPARα involved in the regulation of peroxisomal β-oxidation and (iii) a peroxin gene involved in peroxisomal biogenesis. To this end, the peroxisomal matrix proteins CAT and AOX and the peroxisomal membrane proteins PMP70 and PMP22 were analyzed. Additionally, two factors involved in peroxisomal biogenesis were studied, PPARα and PEX5 [a receptor involved in the targeting of the majority of peroxisomal matrix proteins (28)].

Materials and methods

Tissue preparation

All samples were obtained from surgically resected cases of adenocarcinomas of colon as proven by pathologic diagnosis. The permission of the ethic commission of the Medical Faculty of the University of Heidelberg was obtained prior to the investigation. For comparative purposes in each case, the colonic tumor and the adjacent normal colonic mucosa from the same patient were analyzed. One portion of each sample was frozen in liquid nitrogen for protein and RNA isolation. For immunohistochemical studies, samples were fixed in Carnoy's fixative containing 60% ethanol, 30% chloroform and 10% glacial acetic acid (29). Tissue slices were dehydrated in graded series of ethanol and embedded in paraffin (Paraplast Plus; Monoject Scientific, Athy, Ireland) at 57°C.

Antibodies

Polyclonal antibodies to the peroxisomal matrix proteins CAT and AOX, and to the 22 and 70 kDa peroxisomal membrane proteins (PMP22 and PMP70) were prepared, and their specificity was assessed as described previously (30). The antibody to PPARα was a generous gift of Dr J.A.Gustafsson (Karolinska Institute, Stockholm-Huddinge, Sweden).

Immunohistochemistry

Paraffin sections were incubated for immunohistochemistry using antibodies against CAT, AOX or PMP70. The bound antibodies were visualized by a biotinylated secondary antibody followed by incubation with an avidin–peroxidase conjugate (ExtrAvidin peroxidase staining kit; Sigma, Munich, Germany) and aminoethylcarbazol as substrate (29,31). The sections were counterstained with haematoxylin.

SDS–PAGE and immunoblotting

The tissue was homogenized in an extraction buffer (10 mM Tris–HCl, 20 mM Na–molybdate, 1.5 mM EDTA, 0.6 M KCl, 1 mM phenylmethylsulfonyl fluoride) using an Ultra-Turrax (IKA Labortechnik, Germany). For immunoblotting, equal amounts of protein from homogenate were subjected to SDS–PAGE and, after electrotransfer onto nitrocellulose, the sheets were incubated overnight with polyclonal antibodies to CAT, AOX, PMP22, PMP70 and PPARα. After repeated washing, a peroxidase-conjugated goat anti-rabbit antibody (1:10 000; Sigma) was added for 1 h at room temperature. The immunoreactive bands were visualized by enhanced chemiluminescence (Amersham International, Amersham, UK) and analyzed by semi-quantitative densitometry.

RNA isolation, RNase protection assay and northern blotting

Probes. The mRNAs encoding the following proteins were studied: CAT, AOX, PMP 70, PPARα and Pex5p. Additionally, a probe for 28S RNA was used for normalization. The probes were prepared as described elsewhere (32).

RNA extraction, blotting and hybridization. Total RNA was extracted from the samples by means of guanidinium thiocyanate–phenol–chloroform using the Roti-Quick-Kit (Carl Roth GmbH, Karlsruhe, Germany). RNA was quantified by spectrophotometry at 260/280 nm and its integrity was assessed by denaturing 1% agarose gel electrophoresis (ratio of ribosomal 28S RNA versus 18S RNA). For the dot blot RNase protection assay (32,33), 1 μl (= 1μg) each of total RNA was dotted on a nylon membrane (`Qiabrane'; Qiagen GmbH, Hilden, Germany). In addition for PEX5, northern blots were prepared: Five micrograms of total RNA per lane were subjected to denaturated agarose gel (1%) electrophoresis and transferred by capillary blotting onto nylon membranes. After UV-crosslinking in a UV Stratalinker (Stratagene, La Jolla, CA), blots were prehybridized at 68°C for 2 h and hybridized overnight at the same temperature with digoxigenin (Dig)-labeled cRNA probes adjusted to a concentration of 100 ng/ml. After hybridization, membranes were incubated for 10 min at room temperature with ribonuclease A (1 μg/ml) followed by washing and incubation in blocking buffer (Blocking reagent; Boehringer Mannheim, Mannheim, Germany). Dig-labeled RNA hybrids were detected by chemiluminescence using alkaline-phosphatase-labeled anti-Dig Fab-fragments and a chemiluminescence substrate (CDP-Star; Boehringer Mannheim). The signals obtained were quantified by semi-quantitative densitometry.

Statistics

The densitometric data from protein and mRNA quantifications were analyzed by Bonferroni's method for pairwise multiple comparison (Sigma Plot; Jandel Scientific). For each mRNA or protein probe, tissues from five patients were analyzed and in each case the data from colon carcinoma and intact colon mucosa from the same patient were compared.

Results

Immunohistochemistry

Positive immunoreactivity was obtained with antibodies to CAT, AOX and PMP70 in the enterocytes of normal colonic mucosa with granular staining localized to the supranuclear region (Figure 1, normal). The strongest staining was observed in the enterocytes of the surface epithelium, with decreasing intensity towards the bottom of the crypts. The tumor epithelial cells, in contrast, showed no significant immunoreactivity, only occasionally some diffuse staining was observed (Figure 1, tumor).

Analysis of protein and mRNA levels

Protein. The western blot analysis revealed a significant decrease in colon carcinoma compared with controls of all peroxisomal proteins studied (Figure 2). The extent of the reduction, however, varied markedly between the different proteins. CAT was reduced only by 20%, PMP70 by 40% and AOX by 50%. The strongest effect was noted with PMP22 with a reduction of 65% compared with the normal mucosa. The protein level of PPARα was slightly decreased, but this effect was not statistically significant.

mRNAs. The amount of 28S RNA showed no alteration in colon carcinoma versus normal colon mucosa. The analysis of mRNA by dot-blot RNase protection (RNP) assay revealed that the mRNAs of CAT, PMP70 and PPARα were not significantly changed (Figure 3). Only the AOX mRNA showed a decrease of ~20%. The mRNA of PEX5, in contrast, showed a marked (2.4-fold) increase in tumor tissue both in the RNP assay (Figure 3) as well as in the northern blot (Figure 4).

Discussion

The present study has revealed significant alterations of peroxisomes and their proteins in colon carcinoma at the morphological, protein and mRNA levels. The lack of morphologically identifiable peroxisomes in our immunohistochemical preparations of colon carcinoma is in line with former reports on electron microscopic investigations of DAB-stained sections (22). Furthermore, the significant reductions of the peroxisomal enzyme proteins CAT and AOX correspond well to several earlier reports on a significant decrease of the activities of those enzymes in colon carcinoma (19–22). AOX catalyzes the first step and is the rate limiting enzyme of the peroxisomal β-oxidation sequence (34). The marked reduction of AOX (50%) was accompanied by a moderate reduction of the corresponding mRNA (20%). Moreover, the results of the present study showed that PPARα was not significantly altered in colon carcinoma although the PPARγ which is normally expressed predominantly in adipose tissue, immune system (35) and colon, is markedly elevated in colonic cancers (24–27,36). Overexpression of PPARγ has been proposed to inhibit the transcriptional activity of PPARα on the AOX gene (37). Thus, it is likely that the reduction of AOX mRNA in this study could be due to the elevation of PPARγ in colon carcinoma. This could also explain the reduced activity of the β-oxidation in colon carcinoma.

Unexpectedly, the significant reduction of the CAT and PMP70 protein was not associated with a concomitant reduction of the corresponding mRNAs, and even the 20% reduction of AOX mRNA did not match the marked 50% decrease of AOX protein. These findings strongly suggest that the reduction of those proteins in colon carcinoma may not be due to a reduced transcription of their genes. Possible explanations include a diminished translation rate for those mRNAs in carcinoma cells. Discordant levels of mRNAs and the corresponding proteins suggesting possible translational regulation have been described for many gene products in malignant as well as normal cells (38–42).

Another possible explanation for the discrepancy between the mRNA and protein levels in the present study could be the disturbance of the biogenesis of peroxisomes in colon carcinoma. Indeed, in sections of colon carcinomas no peroxisomes were detected by light microscopic immunohistochemistry, although by western blotting the proteins were present in the tumor tissue albeit in lower concentrations. Since peroxisomal proteins are imported post-translationally into peroxisomes (43,44), our results suggest that peroxisomal proteins may be synthesized but not transported into peroxisomes in colon carcinoma. Thus, in the absence of peroxisomes or defects in their import machinery, the peroxisomal proteins may accumulate in the cytoplasm, being subjected to proteolytic degradation. Indeed, a similar situation is observed in several genetic disorders due to peroxisomal biogenesis defects (45). Those disorders include cerebro-hepato-renal (Zellweger) syndrome and the less severe forms neonatal adrenoleukodystophy, infantile Refsum's disease and rhizomelic chondrodysplasia punctata. Patients with Zellweger syndrome are born with congenital neurological and other abnormalities and usually die within the first year of life. Several defects of peroxisomal protein import systems, among them the PEX5-associated PTS1 pathway, have been found in such patients (46). In cell lines from Zellweger patients and from PEX5-knockout mice (47), peroxisomes are absent or grossly deficient and peroxisomal proteins are synthesized but remain localized to the cytoplasm and are subjected to degradation by proteolysis.

Thus, possible explanations for the presumed non-incorporation of proteins into peroxisomes of colon carcinomas include disturbances in the formation of peroxisomes or failures in the protein import machinery. The first alternative is supported by the stronger reduction of PMP22 than of the other analyzed peroxisomal proteins. The exact function of PMP22 remains unknown, but as a major component of the peroxisomal membrane it has been suggested to play an important role in the peroxisomal biogenesis (48,49).

As part of the peroxisomal matrix protein transport system, the mRNA of the PEX5 gene was studied. The encoded protein Pex5p is the receptor for the peroxisomal targeting signal PTS1 (Ser–Lys–Leu) which is localized on the C-terminal end of many peroxisomal matrix proteins, and plays a crucial role in peroxisomal protein import (28,50). In contrast to all other mRNAs studied here, the mRNA of PEX5 was significantly increased in colon carcinoma, indicating that the diminuition of peroxisomes in colon carcinoma is not associated with a suppression of the PEX5 gene. Today, however, 19 PEX genes are known, and most of their gene products (peroxins) are thought to be involved in peroxisomal protein import and biogenesis (28). Thus, our results cannot exclude the possibility that other peroxins may be altered in colon carcinoma. The regulation of PEX5 is unknown. In the light of our results, however, the increase of PEX5 mRNA associated with the possible accumulation of ligands for Pex5p in the cytoplasm, suggests a ligand-dependent regulation of the PEX5 gene. The value of the marked increase of PEX5 mRNA in colon carcinoma as a possible diagnostic marker remains to be established.

The mechanisms of the reduction of peroxisomes and their enzymes in colon carcinoma are not well understood. The results of the present study suggest that it could be due to a failure of the peroxisomal biogenesis rather than to a down-regulation of the specific genes for peroxisomal matrix proteins. The two important aspects of peroxisomal biogenesis, the synthesis of peroxisomal membranes including the targeting of integral peroxisomal membrane proteins and the matrix protein import, are mediated by quite distinct pathways (28,51,52). The question which of these mechanisms is impaired primarily in colon carcinoma cannot be answered as yet. However, since the ligand activation of PPARα by thiazolidinediones has been found to inhibit the growth of colon carcinoma (25), it would be tempting to speculate that such treatment would also lead to the recovery of the peroxisomes and their enzymes. Further analysis of this problem could lead to the elucidation of the role of peroxisomes in biology of the colon carcioma.

Fig. 1.

Light micrographs from normal colonic mucosa (left) and tumor tissue (right) incubated with antibodies against CAT and PMP70. The insets show details from the mucosa near the surface. Note the specific (red) labeling of peroxisomes (some denoted by arrows) within the enterocytes on the surface of the outer part of the crypts. In the epithelial cells of tumor tissue the staining is weak and diffuse.

Fig. 2.

Western blots of peroxisomal proteins and their densitometric analysis. Immunoblots of corresponding probes of colon carcinoma (T) and normal colonic mucosa (N) from five patients are shown. Blots were incubated with antibodies to CAT, AOX, PMP70, PMP22 and PPARα. The immuncomplexes were visualized by the enhanced chemiluminescence technique (for details see Materials and methods). The diagram below shows the results of the densitometric analysis of immunoblots, the values from appropriate normal mucosa were set to 100%. *, significant differences to normal mucosa (P < 0.05).

Fig. 3.

Dot blot RNase protection assays of peroxisomal proteins and their densitometric analysis. Blots of corresponding probes of colon carcinoma (T) and normal colonic mucosa (N) from five patients are shown. The plot shows the results of densitometric analysis, the values from normal mucosa were set to 100%. *, significant differences to normal mucosa (P < 0.05).

Fig. 4.

Northern blot for PEX5 of corresponding probes of colon carcinoma (T) and normal colon mucosa (N) from five patients. Note the significant increase of PEX5 signal in all cases.

3

To whom correspondence should be addressed Email:

We wish to thank Prof. Dr H.F.Otto (Pathologisches Institut, University of Heidelberg) for providing the resection material. The study was supported by a grant from the Deutsche Forschungsgemeinschaft Be 1659/2-1.