SAP-1 is a microvillus-specific protein tyrosine phosphatase that modulates intestinal tumorigenesis

Authors

  • Hisanobu Sadakata,

    1. Laboratory of Biosignal Sciences, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-15 Showa-Machi, Maebashi, Gunma 371-8512, Japan
    2. Department of Obstetrics and Gynecology, Graduate School of Medicine, Gunma University, 3-39-22 Showa-Machi, Maebashi, Gunma 371-8511, Japan
    Search for more papers by this author
    • These authors have contributed equally to this work.

  • Hideki Okazawa,

    1. Laboratory of Biosignal Sciences, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-15 Showa-Machi, Maebashi, Gunma 371-8512, Japan
    Search for more papers by this author
    • These authors have contributed equally to this work.

  • Takashi Sato,

    1. Laboratory of Molecular Traffic, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-15 Showa-Machi, Maebashi, Gunma 371-8512, Japan
    Search for more papers by this author
    • These authors have contributed equally to this work.

  • Yana Supriatna,

    1. Laboratory of Biosignal Sciences, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-15 Showa-Machi, Maebashi, Gunma 371-8512, Japan
    Search for more papers by this author
  • Hiroshi Ohnishi,

    1. Laboratory of Biosignal Sciences, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-15 Showa-Machi, Maebashi, Gunma 371-8512, Japan
    Search for more papers by this author
  • Shinya Kusakari,

    1. Laboratory of Biosignal Sciences, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-15 Showa-Machi, Maebashi, Gunma 371-8512, Japan
    Search for more papers by this author
  • Yoji Murata,

    1. Laboratory of Biosignal Sciences, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-15 Showa-Machi, Maebashi, Gunma 371-8512, Japan
    Search for more papers by this author
  • Tomokazu Ito,

    1. Laboratory of Biosignal Sciences, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-15 Showa-Machi, Maebashi, Gunma 371-8512, Japan
    Search for more papers by this author
  • Uichi Nishiyama,

    1. Pharmaceutical Development Laboratories, Kirin Brewery Co. Ltd., Takasaki, Gunma 370-1295, Japan
    Search for more papers by this author
  • Takashi Minegishi,

    1. Department of Obstetrics and Gynecology, Graduate School of Medicine, Gunma University, 3-39-22 Showa-Machi, Maebashi, Gunma 371-8511, Japan
    Search for more papers by this author
  • Akihiro Harada,

    1. Laboratory of Molecular Traffic, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-15 Showa-Machi, Maebashi, Gunma 371-8512, Japan
    Search for more papers by this author
  • Takashi Matozaki

    Corresponding author
    1. Laboratory of Biosignal Sciences, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-15 Showa-Machi, Maebashi, Gunma 371-8512, Japan
    Search for more papers by this author

  • Communicated by: Yoshimi Takai

* Correspondence: matozaki@showa.gunma-u.ac.jp

Abstract

SAP-1 (PTPRH) is a receptor-type protein tyrosine phosphatase (RPTP) with a single catalytic domain in its cytoplasmic region and fibronectin type III-like domains in its extracellular region. The cellular localization and biological functions of this RPTP have remained unknown, however. We now show that mouse SAP-1 mRNA is largely restricted to the gastrointestinal tract and that SAP-1 protein localizes to the microvilli of the brush border in gastrointestinal epithelial cells. The expression of SAP-1 in mouse intestine is minimal during embryonic development but increases markedly after birth. SAP-1-deficient mice manifested no marked changes in morphology of the intestinal epithelium. In contrast, SAP-1 ablation inhibited tumorigenesis in mice with a heterozygous mutation of the adenomatous polyposis coli gene. These results thus suggest that SAP-1 is a microvillus-specific RPTP that regulates intestinal tumorigenesis.

Introduction

Intestinal epithelial cells of mammals are highly polarized. The apical surface of these cells consists of the brush border, which is formed by a large number of microvilli (Mooseker 1985; Holmes & Lobley 1989; Ross & Pawlina 2006), whereas the lateral membrane adjacent to the apical surface contains tight junctions, adherens junctions and desmosomes (in this order from the apical to the basal side), resulting in a relatively straight and tight interface with neighboring cells (Takeichi 1995; Gumbiner 2005; Miyoshi & Takai 2005; Ross & Pawlina 2006). This cell–cell adhesion is mediated by homophilic binding in trans of various adhesion molecules, including claudin at tight junctions as well as E-cadherin and nectin at adherens junctions (Takeichi 1995; Tsukita & Furuse 2002; Gumbiner 2005; Miyoshi & Takai 2005). Protein tyrosine phosphorylation plays a key role in regulation of cell–cell adhesion, with catenins, which are associated with the intracellular region of E-cadherin (Gumbiner 2005; Lilien & Balsamo 2005), being important targets for such phosphorylation and dephosphorylation. Indeed, receptor tyrosine kinases, such as the receptors for epidermal growth factor or hepatocyte growth factor, as well as Src family kinases and protein tyrosine phosphatases (PTPs) such as PTPµ are localized at sites of cell–cell adhesion in epithelial cells (Maratos-Flier et al. 1987; Takeda et al. 1995; Kamei et al. 1999; Sallee et al. 2006). Moreover, these enzymes contribute to the dynamic plasticity of cell–cell adhesion in epithelial cells (Takeda et al. 1995; Lilien & Balsamo 2005; Sallee et al. 2006). In contrast, it remains unknown whether tyrosine phosphorylation–dephosphorylation participates in regulation of brush border function or which tyrosine kinases or PTPs localize to and function at the brush border of epithelial cells.

Stomach cancer-associated protein tyrosine phosphatase-1 (SAP-1) (also known as PTPRH) was originally identified as a PTP expressed in a human stomach cancer cell line (Matozaki et al. 1994). It is a receptor-type PTP (RPTP, R3 subtype) (Andersen et al. 2001), with a single catalytic domain in its cytoplasmic region and eight fibronectin type III-like domains in its extracellular region (Matozaki et al. 1994). Although its expression was found to be markedly increased in human colon and pancreatic cancers (Matozaki et al. 1994; Seo et al. 1997), the expression and localization of SAP-1 in normal tissues have not been well defined. A “substrate-trapping” approach identified p130Cas, a prominent focal adhesion-associated component of the integrin signaling pathway, as a likely physiological substrate of SAP-1 (Noguchi et al. 2001). Forced expression of SAP-1 in cultured cells also resulted in the dephosphorylation of several additional focal adhesion-associated proteins, including focal adhesion kinase and paxillin, as well as in impairment of reorganization of the actin-based cytoskeleton (Noguchi et al. 2001), suggesting that SAP-1 regulates the latter process. Forced expression of this PTP also inhibited cell proliferation, an effect that was mediated in part either by attenuation of growth factor-induced mitogen-activated protein kinase activation or by induction of caspase-dependent apoptosis (Noguchi et al. 2001; Takada et al. 2002). The cytoplasmic region of SAP-1 binds directly to Lck, and over-expression of SAP-1 resulted in down-regulation of the kinase activity of Lck and consequent inhibition of T cell receptor-mediated T cell function (Ito et al. 2003). However, given that most of these observations were made with cultured cells, the physiological or pathological functions of SAP-1 in vivo remain unknown.

We have now determined the sequence of mouse SAP-1 cDNA and generated a mAb to the mouse protein. With the use of these materials, we have shown that SAP-1 is expressed predominantly in the gastrointestinal tract, most prominently in the small intestine, and that it is specifically localized at the microvilli of gastrointestinal epithelial cells. Moreover, our observations suggest that SAP-1 plays an important role in the regulation of intestinal tumorigenesis.

Results

Predicted structure of mouse SAP-1 and its predominant expression in the intestine

To examine the expression of SAP-1 in various tissues, we first cloned a cDNA encoding mouse SAP-1. Searching of mouse Expressed Sequence Tag (EST) and other DNA data bases revealed several mouse DNA or EST sequences with similarity to those of human SAP-1 cDNA. With the use of these sequences, we obtained partial cDNA fragments for mouse SAP-1 by reverse transcription-polymerase chain reaction (RT-PCR) as described in Experimental procedures. We further searched the mouse cDNA data base provided by the Functional Annotation of the mouse (FANTOM) project (Carninci et al. 2005) for sequences similar to those of the partial mouse SAP-1 cDNA fragments and thereby obtained a full-length mouse SAP-1 cDNA (approximately 3.5 kb) (DDBJ Accession No. AB217856). The predicted structure of the mouse SAP-1 protein, which comprises 971 amino acids, was found to be similar to that of human SAP-1 (Fig. 1A), with the exception that the mouse protein contains six fibronectin type III-like domains in its extracellular region, whereas the human protein contains eight such domains (Matozaki et al. 1994). The amino acid sequences of mouse and human SAP-1 proteins share an overall identity of 56%, with the corresponding values for the extracellular and intracellular regions being 43% and 77%, respectively (Fig. 1A). A search of the Novartis GeneAtlas <http://symatlas.gnf.org/symatlas> with the mouse cDNA sequence revealed that SAP-1 mRNA is abundant in the small intestine and colon. Indeed, Northern blot analysis showed that mouse SAP-1 mRNA was preferentially distributed to the small intestine, stomach and colon (Fig. 1B).

Figure 1.

Structure of mouse SAP-1 and its predominant expression in the gastrointestinal tract. (A) Schematic representation of the structures of human and mouse SAP-1. Numbers indicate amino acid residues. FN III, fibronectin type III; TM, transmembrane region; PTP, phosphatase region. The sequence identity between the extracellular or cytoplasmic regions of the human and mouse SAP-1 proteins is indicated. (B) Distribution of SAP-1 mRNA among mouse tissues. Polyadenylated RNA (2 µg) isolated from the indicated tissues was subjected to Northern hybridization with approximately 3.5-kb mouse SAP-1 (upper panel) or approximately 0.5-kb mouse glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (loading control, lower panel) cRNA probes. (C) Expression of SAP-1 protein in mouse tissues. Lysates of various tissues (20 µg of protein) were subjected to immunoblot analysis with mAbs to SAP-1 (upper panel) or to β-tubulin (loading control, lower panel). The positions of molecular size standards are indicated on the left. (D) Distribution of SAP-1 in the mouse gastrointestinal tract. Lysates of the stomach, duodenum, jejunum and colon were subjected to immunoblot analysis as in (C).

To investigate the expression and localization of SAP-1 protein, we generated a monoclonal antibody (mAb) (clone 123) to the extracellular region of mouse SAP-1. Immunoblot analysis with this mAb also revealed prominent expression of the approximately 250-kDa SAP-1 protein in the intestine, and a low level of the expression was observed in testis (Fig. 1C). Furthermore, the amount of SAP-1 in the duodenum or jejunum was markedly greater than that in the stomach or colon (Fig. 1D).

Specific localization of mouse SAP-1 to microvilli of gastrointestinal epithelial cells

We next examined the localization of SAP-1 in the small intestine. Immunohistofluorescence analysis with the mAb to SAP-1 revealed that SAP-1 was localized at the apical surface of intestinal epithelial cells (Fig. 2A), whereas β-catenin was present at sites of cell–cell adhesion as well as in the cytoplasm of epithelial cells, as described previously (Sena et al. 2006). The localization of SAP-1 was similar to that of ezrin or alkaline phosphatase (Fig. 2B), both of which are known to be localized at the microvilli of the intestinal epithelium (Narisawa et al. 2003; Saotome et al. 2004). Furthermore, the localization of SAP-1 overlapped with that of ezrin/radixin/moesin binding phosphoprotein of 50 kDa (EBP-50) at the apical surface of the intestinal epithelium (Fig. 2C). EBP-50 is a protein that links CD44 at the cell surface to ezrin and the actin cytoskeleton and is thus concentrated at the microvilli of the intestinal epithelium (Tsukita & Yonemura 1999). Each microvillus at the brush border of intestinal epithelial cells contains a bundle of actin filaments that extends the entire length of the microvillus as well as into the actin filament meshwork below it known as the terminal web (Mooseker 1985; Tyska et al. 2005; Ross & Pawlina 2006). SAP-1 immunoreactivity was detected immediately above the prominent staining of F-actin revealed by phalloidin, which may correspond to the terminal web, at the brush border of intestinal epithelial cells (Fig. 2C). Immuno-electron microscopy (EM) with the mAb to SAP-1 also revealed prominent staining at the microvilli of intestinal epithelial cells from wild-type mice (Fig. 2D), whereas such staining was virtually undetectable in sections from SAP-1-deficient mice, the generation of which is described below. These results thus indicated that SAP-1 is specifically expressed in intestinal epithelial cells and localizes to the microvilli of these cells.

Figure 2.

Localization of SAP-1 to the epithelium of the mouse small intestine. (A) A cryostat section of the jejunum of a 6-week-old male mouse was subjected to immunofluorescence staining with mAbs to mouse SAP-1 (left panels) and to β-catenin (middle panels). Merged images are also shown in the right panels. Boxed regions in the upper panels are shown at higher magnification in the lower panels. Scale bars, 200 µm (upper panels) and 50 µm (lower panels). (B) Sections of mouse jejunum were stained with mAbs to SAP-1 (left panel) and to ezrin (middle panel) as well as for alkaline phosphatase activity (right panel). Scale bar, 50 µm. (C) Sections of mouse jejunum were stained with a mAb to SAP-1 and either polyclonal antibodies to EBP-50 (upper panels) or rhodamine-conjugated phalloidin (lower panels), as indicated. Merged images are shown in the right panels. Scale bars, 10 µm. (D) Immuno-EM of the jejunum of wild-type (WT) or SAP-1 knockout (KO) mice with a mAb to mouse SAP-1. Scale bar, 200 nm.

SAP-1 was also localized at the apical surface of epithelial cells of the stomach, although the level of expression was markedly lower than that in the small intestine (Fig. 3A). Microvilli are more prominent in the small intestine than in the stomach (Ross & Pawlina 2006). Moreover, the expression of SAP-1 was absent in goblet cells (Fig. 3B), which were stained with Ulex europaeus agglutinin (UEA)-I and lack microvilli (Fischer et al. 1984). These observations were thus also consistent with the notion that SAP-1 is localized specifically at microvilli of the gastrointestinal epithelium.

Figure 3.

Expression of SAP-1 in various regions of the epithelium of the gastrointestinal tract and during development of the small intestine. (A) Sections of the jejunum (left panels) or stomach (right panels) of 6-week-old male mice were stained with both a mAb to mouse SAP-1 (upper panels) and rhodamine-conjugated phalloidin (middle panels). Merged images are shown in the lower panels. Scale bar, 50 µm. (B) A section of mouse ileum was stained with both a mAb to SAP-1 (upper panel) and FITC-conjugated UEA-I (middle panel). A merged image is shown in the lower panel. The arrowhead indicates a goblet cell in the ileal epithelium. Scale bar, 10 µm. (C) Lysates (20 µg of protein) of intestinal tissue from mice at the indicated stages of development were subjected to immunoblot analysis with mAbs to mouse SAP-1 (upper panel) or to actin (loading control, lower panel). (D) Sections of mouse intestinal tissue at the indicated developmental stages were stained with mAbs to mouse SAP-1 (upper panels) or to E-cadherin (lower panels). Scale bar, 200 µm. P1 and P8, postnatal days 1 and 8, respectively.

We next examined the expression of SAP-1 during development of the intestine in mouse embryos. Immunoblot and immunohistofluorescence analyses revealed that SAP-1 was virtually undetectable in the intestine at embryonic day (E) 14.5, whereas the expression of E-cadherin was apparent at this time (Fig. 3C,D). Expression of SAP-1 was detected at a low level at E16.5 and increased markedly after birth (Fig. 3C,D). These results suggested that expression of SAP-1 increases in parallel with the development of intestinal epithelial cells. We further examined the expression of SAP-1 during differentiation of intestinal epithelial cells. In adult mice, staining for SAP-1 was absent or present at only a low level in intestinal crypts (Fig. 4), where newly generated intestinal epithelial cells and their progenitors reside (Ross & Pawlina 2006; Blanpain et al. 2007). These results suggested that expression of SAP-1 increases in parallel with differentiation of intestinal epithelial cells.

Figure 4.

Expression of SAP-1 at the villus tip or crypts of mouse jejunum. (A) A cryostat section of the jejunum of a 6-week-old male mouse was stained with mAbs to mouse SAP-1 (red, right and left panels) and to β-catenin (green, right panel). (B) Regions of the section shown in (A) containing villus tips (panels a and b) or crypts (panels c and d) are shown at higher magnification. Scale bars, 200 µm (A) and 50 µm (B).

Normal morphology of intestinal epithelial cells in SAP-1-deficient mice

To investigate the physiological or pathological roles of SAP-1 in vivo, we generated SAP-1-deficient mice by targeted deletion of a fragment of genomic DNA containing exons 17 and 18 of the SAP-1 gene, which encode a region of the protein (amino acids 778–872) required for enzymatic activity. We generated a Sap1 mutant allele (Sap1flox) by introducing LoxP sites into the introns flanking exons 17 and 18 (Fig. 5A). Deletion of these two exons would be expected to introduce a frameshift mutation, resulting in loss of the COOH-terminal half of the catalytic domain as well as the downstream region of the protein. Crossing of Sap1flox/flox mice with CAG-Cre transgenic animals (Sakai & Miyazaki 1997) resulted in the generation of SAP-1-deficient (SAP-1 KO) mice. Southern blot analysis confirmed the presence of the deleted allele in SAP-1 KO mice (Fig. 5B). Immunoblot analysis with the mAb to SAP-1 also demonstrated the complete loss of SAP-1 protein or a reduction in its abundance of approximately 50% in the intestine of SAP-1 KO (Sap1−/–) mice and Sap1+/– heterozygotes, respectively (Fig. 5C). Immunohistofluorescence staining with this mAb also revealed the complete loss of SAP-1 in the apical brush border of the small intestine of SAP-1 KO mice (Fig. 5D). Given that the mAb to SAP-1 reacts with the extracellular region of the protein, we also performed immunoblot analysis with polyclonal antibodies to the cytoplasmic region of mouse SAP-1. This analysis also revealed the complete ablation of SAP-1 in the intestine of SAP-1 KO mice (data not shown). These results thus indicated that the entire SAP-1 protein is absent or undetectable in the intestine of SAP-1 KO mice, likely as a result of instability of the mutant SAP-1 mRNA or its protein product. We attempted to generate mice with microvillus-specific deficiency of SAP-1, but we found that the expression level of SAP-1 in Sap1flox/flox mice was reduced by > 50% relative to that in wild-type animals (data not shown). The SAP-1 KO mice were therefore backcrossed onto the C57BL/6 background for more than four generations, and offspring homozygous for the wild-type (WT) or mutant Sap1 alleles were studied in the following experiments.

Figure 5.

Generation of SAP-1 KO mice. (A) Schematic representation of the structures of the mouse SAP-1 protein and gene as well as of the targeting vector, targeted allele and deleted allele. The position of cysteine-871, which is essential for the PTP activity of SAP-1, is indicated by a closed circle. The portions of SAP-1 protein that are encoded by exons 14–21 of the wild-type allele are also indicated. The positions of the 5′ (S1) and 3′ (S2) probes for Southern blot analysis to confirm the homologous recombination or deletion events are shown by closed bars. The targeting vector was constructed with a 1.2-kb DNA fragment from exon 14 to intron 16 of Sap1 as a 5′ homologous region, a 6.4-kb fragment from intron 16 to intron 21 of Sap1 as a 3′ homologous region, a PGKβgeo cassette, a diphtheria toxin A chain (DTA) gene cassette, and two LoxP sites (one in intron 16 and one in intron 18, indicated by open triangles). The PGKβgeo cassette allowed positive selection of homologous recombinants, whereas the DTA cassette allowed negative selection of nonhomologous recombinants. Homologous recombination generated the targeted allele (Sap1flox), and crossing of Sap1flox/flox mice with CAG-Cre mice resulted in deletion of exons 17 and 18 of Sap1 as well as of the PGKβgeo cassette. S, Sac I; B, Bam HI. (B) Southern blot analysis of Sac I-digested (left panel) or Bam HI-digested (right panel) genomic DNA from wild-type (+/+) or heterozygous (+/–) or homozygous (–/–) SAP-1 mutant mice. The wild-type allele yielded a 3.7-kb fragment with the 5′ probe (probe S1, left panel) and a 4.9-kb fragment with the 3′ probe (probe S2, right panel), whereas the deleted allele yielded a 2.3-kb fragment with the 5′ probe (left panel) and a 5.3-kb fragment with the 3′ probe (right panel). (C) Lysates of the ileum of wild-type (+/+) or heterozygous (+/–) or homozygous (–/–) SAP-1 mutant mice were subjected to immunoblot analysis with mAbs to SAP-1 (upper panel) or to β-tubulin (lower panel). (D) Sections of the jejunum from wild-type (WT) or SAP-1 KO (Sap1−/–) mice were stained with mAbs to SAP-1 (red) and to β-catenin (green). The merged images are shown. Scale bar, 200 µm.

The SAP-1 KO mice were born apparently healthy, survived to adulthood, and fertile. Although intestinal epithelial cells play a central role in food digestion and in absorption of nutrients and electrolytes through the intestine, and SAP-1 localizes specifically to microvilli of the gastrointestinal epithelium, the body weight of male SAP-1 KO mice at 6 weeks of age was similar to that of WT controls (Table 1). Moreover, the blood glucose concentration as well as the serum concentrations of total protein, total cholesterol, iron, Na+, Cl and Ca2+ did not differ substantially between WT and SAP-1 KO male mice at 10 weeks of age (Table 1). The peripheral blood counts of SAP-1 KO mice were also normal (data not shown).

Table 1.  Body weight and biochemical analysis of peripheral blood of WT and SAP-1 KO mice
 BW (g)BG (mg/dL)TP (g/L)TChol (mg/dL)Fe (mg/dL)Na+(mEq/L)Cl(mg/dL)Ca2+(mg/dL)
  1. Body weight (BW) was determined for male SAP-1 KO (n = 8) or WT (n = 7) mice at 6 weeks of age. Peripheral blood samples were obtained from the femoral vein of 10-week-old male mice (WT, n = 6; SAP-1 KO, n = 8) anesthetized with ether in a glass jar. Blood glucose (BG) level was determined with a Glutest sensor (Sanwa Kagaku, Nagoya, Japan). Blood samples were immediately centrifuged for analysis of other biochemical markers in serum with a Clinical Analyzer 7180 (Hitachi Science Systems, Tokyo, Japan). Data are means ± SEM of values from the indicated numbers of mice; none of the measured parameters differed significantly between WT and SAP-1 KO mice. TP, total protein; TChol, total cholesterol.

WT20.3 ± 0.4150.5 ± 7.76.1 ± 0.2117.4 ± 7.8161.4 ± 10.4136.5 ± 5.598.2 ± 1.87.5 ± 0.6
KO21.0 ± 0.4142.9 ± 5.15.9 ± 0.295.2 ± 7.7174.8 ± 14.7145.0 ± 3.397.4 ± 2.08.4 ± 0.3

Protein tyrosine phosphorylation and dephosphorylation play a central role in organization of the actin cytoskeleton and hence in regulation of cell–cell adhesion and cell morphology. In particular, the proper regulation of actin filament networks is considered to be crucial for formation of intestinal microvilli (Fath & Burgess 1995). However, light microscopic examination of intestinal villi and epithelial cells revealed no marked difference in morphology between WT and SAP-1 KO mice (Fig. 6A). Moreover, immunohistofluorescence analysis showed that staining of junctional proteins, including E-cadherin, β-catenin and p120-catenin of adherens junctions (data not shown) as well as ZO-1 and claudin-15 of tight junctions (data not shown), was similar for WT and SAP-1 KO mice. The morphology of microvilli or of tight junctions, adherens junctions, or desmosomes between intestinal epithelial cells, as revealed by transmission EM, also did not differ substantially between mice of the two genotypes (Fig. 6B).

Figure 6.

Morphology of intestinal epithelial cells of SAP-1 KO mice. (A) Sections of the jejunum of male SAP-1 KO or WT littermates at 10 weeks of age were stained with hematoxylin–eosin. The images show villus-crypt units (upper panels) or the epithelium (lower panels). Scale bars, 200 µm (upper panels) or 20 µm (lower panels). (B) Transmission EM of the jejunum samples shown in (A). Images of the microvilli (upper panels) and of the upper region of cell–cell adhesion of the epithelium (lower panels) are shown. TJ, tight junction; AJ, adherens junction; DS, desmosome. Scale bars, 500 nm.

Reduced numbers of adenomas in ApcMin/+ mice with a Sap1−/– background

SAP-1 KO mice manifested normal morphology of intestinal epithelial cells and normal blood parameters. We next examined the possible effect of SAP-1 ablation on intestinal tumorigenesis, given that our previous studies with cultured cells suggested that SAP-1 regulates cell proliferation through its PTP activity (Noguchi et al. 2001; Takada et al. 2002). We analyzed intestinal tumorigenesis by crossing SAP-1 KO mice with mice of the Min/+ genotype for the adenomatous polyposis coli gene (Apc), a widely used model characterized by the spontaneous development of intestinal adenomas (Su et al. 1992; Rakoff-Nahoum & Medzhitov 2007). The numbers of visible adenomatous polyps (≥ 1 mm in diameter) that formed in the small intestine and colon were quantified by stereoscopic microscopy. At 20 weeks of age, the total number of adenomas in the small intestine, but not that in the colon, appeared to be reduced in ApcMin/+Sap1−/– mice compared with that in ApcMin/+Sap1+/+ mice (Fig. 7A), although such reduction was not statistically significant. However, the number of macroadenomas (≥ 2 mm in diameter) in the intestine was markedly reduced in ApcMin/+Sap1−/– mice compared with that in ApcMin/+Sap1+/+ mice (Fig. 7B). These results thus suggested that SAP-1 regulates intestinal tumor growth in ApcMin/+ mice.

Figure 7.

Effect of SAP-1 deficiency on adenoma formation in the intestine of ApcMin/+ mice. (A) The total number of visible adenomas (≥ 1 mm in diameter) in the small intestine or colon of ApcMin/+Sap1+/+ (WT) (n = 41) or ApcMin/+Sap1−/– (KO) (n = 29) mice at 20 weeks of age was quantified by stereoscopic microscopy. (B) Size range of intestinal and colonic adenomas for the animals examined in (A). Data in (A) and (B) are means ± SEM of values from the indicated numbers of mice. *P < 0.05, **P < 0.01. (C) Adenomas from BrdU-injected ApcMin/+Sap1+/+ (WT) (left panel) or ApcMin/+Sap1−/– (KO) (middle panel) mice were sectioned and evaluated for BrdU incorporation by immunohistochemistry with a mAb to BrdU. Scale bar, 100 µm. The percentage of BrdU-positive cells per 150 adenoma cells of the two mouse strains was determined (right panel); data are means ± SEM of values from eight mice of each group. (D) Adenomas from ApcMin/+Sap1+/+ (WT) (left panel) or ApcMin/+Sap1−/– (KO) (middle panel) mice were sectioned and evaluated for cell death by TUNEL staining. Arrowheads indicate TUNEL-positive cells. Scale bar, 100 µm. The percentage of TUNEL-positive cells per 150 adenoma cells of the two mouse strains was determined. Data are means ± SEM of values from eight or twelve ApcMin/+Sap1+/+ or ApcMin/+Sap1−/– mice, respectively.

Given that the size of tumors is influenced by the balance between cell proliferation and cell death within the tumor mass, we investigated the rates of cell proliferation and apoptosis in size-matched tumors from ApcMin/+Sap1+/+ and ApcMin/+Sap1−/– mice. Neither the percentage of 5-bromo-2′-deoxyuridine (BrdU)-positive (proliferating) cells (Fig. 7C) nor the percentage of TdT-mediated dUTP-biotin nick end labeling (TUNEL)-positive (apoptotic) cells (Fig. 7D) differed significantly between mice of the two genotypes. The reduced extent of intestinal tumorigenesis in ApcMin/+Sap1−/– mice thus did not appear to be attributable to a reduced rate of cell proliferation or an increased rate of apoptosis.

Discussion

We have here shown that mouse SAP-1 is localized at the microvilli of the brush border in gastrointestinal epithelial cells. SAP-1 is classified as an RPTP of the R3 subtype, as are PTP-RO, Vascular endothelial cell-specific phosphotyrosine phosphatase (VE-PTP) and DEP-1 (Andersen et al. 2001). All of these enzymes share similar structures, with a single catalytic domain in the cytoplasmic region and fibronectin type III-like domains in the extracellular region. PTP-RO (also known as GLEPP1) is specifically expressed in podocytes of the renal glomerulus (Thomas et al. 1994; Yang et al. 1996); in particular, it is localized at the apical surface of foot processes in the highly polarized visceral glomerular epithelial cells (Yang et al. 1996; Wharram et al. 2000). Although the morphology of foot processes is distinct from that of microvilli, both of these structures are rich in actin (Fath & Burgess 1995; Benzing 2004). In addition, the expression of VE-PTP is restricted to endothelial cells (Bäumer et al. 2006; Dominguez et al. 2007), with VE-PTP also being localized at the apical surface of these cells (Bäumer et al. 2006; Dominguez et al. 2007). Together, these observations suggest that the expression of RPTPs of the R3 subtype may be restricted to a single or limited number of cell types. In addition, these RPTPs may be expressed specifically at the apical surface of polarized cells.

The predominant expression and localization of SAP-1 at the microvilli of gastrointestinal epithelial cells implicate SAP-1 in functions specific to these cells. The proper regulation of the actin cytoskeleton is important for maintenance of the morphology of intestinal microvilli (Fath & Burgess 1995). Indeed, mice that lack either ezrin or EBP-50, both of which are localized at microvilli and associated with the actin cytoskeleton, exhibit shortened and irregular microvilli in enterocytes (Morales et al. 2004; Saotome et al. 2004; Tamura et al. 2005). Moreover, protein tyrosine phosphorylation, which is determined by protein tyrosine kinases and PTPs, plays an important role in reorganization of the actin cytoskeleton (Giannone & Sheetz 2006). However, SAP-1-deficient mice manifested no marked changes in the morphology of intestinal epithelial cells, including that of microvilli or of tight or adherens junctions between these cells. Consistent with this finding, the expression of SAP-1 in the intestine was shown to be minimal during embryonic development and to increase markedly after birth, suggesting that SAP-1 is not important for determination of cell architecture in the intestinal epithelium.

Intestinal epithelial cells play an essential role in food digestion and in the absorption of nutrients and electrolytes through the intestine. Indeed, several digestive enzymes as well as channels and transporters for electrolytes or nutrients, including those for glucose, amino acids and fatty acids, are associated with the gastrointestinal microvilli (Holmes & Lobley 1989; Ross & Pawlina 2006). However, general nutrition status appeared unaffected in SAP-1 KO mice, as were peripheral blood cell counts. SAP-1 may thus be dispensable for regulation of food digestion and absorption of nutrients and electrolytes in the intestine. We are not able to exclude the possibility, however, that SAP-1 indeed contributes to such functions and that other PTPs, such as other RPTPs of the R3 subtype, are able to compensate for the loss of SAP-1 activity in SAP-1 KO mice.

We also found that SAP-1 deficiency inhibited tumorigenesis in mice heterozygous for mutation of Apc. The number of large adenomas (≥ 2 mm in diameter) in SAP-1 KO mice was markedly reduced whereas that of small adenomas (< 2 mm) was similar to that in WT mice, suggesting that SAP-1 contributes to tumor expansion but not to the initial transformation of normal epithelial cells into dysplastic cells. Functional impairment of the APC protein in ApcMin/+ mice is considered to result in stabilization and marked accumulation of β-catenin, which initiates transformation of normal epithelial cells and promotes tumorigenesis through constitutive activation of the β-catenin-transcription factor 4 (TCF-4) transcriptional pathway (Clarke 2006). Our results thus suggest that SAP-1 regulates intestinal tumorigenesis through cooperation with this pathway, although the precise mechanism of such regulation remains to be determined. Indeed, the rates of cell proliferation and apoptosis in adenomas of ApcMin/+Sap1−/– mice did not differ significantly from those apparent in ApcMin/+Sap1+/+ mice. In addition, the cytoplasmic and nuclear accumulation of β-catenin in adenomas were similar in mice of these two genotypes (unpublished data), suggesting that SAP-1 likely does not regulate the β-catenin-TCF-4 pathway itself. Instead, SAP-1 may promote intestinal cell proliferation through activation of protein tyrosine kinases such as Src family kinases. Indeed, PTPα, another RPTP, was shown to induce transformation of Rat-1 fibroblasts as a result of activation of Src mediated by dephosphorylation of its COOH-terminal tyrosine residue (Zheng et al. 1992). However, forced expression of SAP-1 failed to increase the activity of Src in cultured fibroblasts (unpublished data). Moreover, the levels of c-Src activity (in either adenomas or normal tissues) of ApcMin/+Sap1−/– mice did not differ from those of ApcMin/+Sap1+/+ mice (unpublished data). We previously showed that the expression of SAP-1 is markedly increased in human colon cancers (Matozaki et al. 1994; Seo et al. 1997). Our present results thus support the notion that SAP-1 promotes the tumorigenic potential of intestinal epithelial cells. By contrast, Src kinase may not be responsible for promotion by SAP-1 of intestinal tumorigenesis. Recent studies also suggests the transforming properties of PTPs. Gain of function mutations in SHP-2 are implicated to cause juvenile myelomonocytic leukemia (Tartaglia et al. 2003). In addition, deletion of PTP1B activity delays mammary tumorigenesis and protects lung metastasis in transgenic mice with an active mutant of ErbB2 (Julien et al. 2007). Further studies are warranted to characterize the physiological and pathological roles of SAP-1 as an intestinal epithelium-specific PTP.

Experimental procedures

Antibodies and reagents

Mouse mAbs to β-tubulin (TUB 2.1) or to actin (AC-40) were obtained from Sigma (St. Louis, MO), those to β-catenin or to p120-catenin were from BD Biosciences (San Diego, CA), and that to E-cadherin (ECCD-2) was from Takara BIO (Otsu, Shiga, Japan). Rabbit polyclonal antibodies to EBP-50 were from Abcam (Cambridge, UK) and those to claudin-15 were kindly provided by M. Furuse (Kobe University, Kobe, Japan). A rat mAb to CD31 (MEC13.3) was from BD Biosciences and that to ezrin (M11) was kindly provided by S. Tsukita (Osaka University, Suita, Japan). Rabbit polyclonal antibodies to ZO-1, rhodamine-conjugated phalloidin, and secondary antibodies labeled with fluorescein isothiocyanate (FITC), Cy3 or Alexa488 for immunofluorescence analysis were obtained from Invitrogen (Carlsbad, CA). Horse radish peroxidase (HRP)-conjugated goat polyclonal antibodies to rat, mouse or rabbit IgG for immunoblot analysis were from Jackson ImmunoResearch (West Grove, PA). FITC-conjugated UEA-I was from Vector (Burlingame, CA).

Isolation of a full-length mouse SAP-1 cDNA

To obtain a full-length mouse SAP-1 cDNA, we searched mouse data bases for sequences similar to that of human SAP-1 cDNA with the use of the Basic Local Alignment Search Tool (BLAST) program (NCBI). Several EST fragments as well as genomic DNA fragments potentially corresponding to the mouse SAP-1 gene were identified and were used to obtain partial cDNAs for mouse SAP-1 by RT-PCR. The sequences of two RT-PCR products (226 and 431 bp) were then used to search the mouse DNA data base provided by the FANTOM project (Carninci et al. 2005), resulting in the identification of a cDNA clone (#9030616J04) that contained the full-length SAP-1 cDNA.

Generation of monoclonal and polyclonal antibodies to mouse SAP-1

A rat mAb to mouse SAP-1 (clone 123) was generated with the use of a recombinant fusion protein comprising the extracellular region of mouse SAP-1 (amino acids 1–604) and the Fc region of human IgG as the antigen. Rabbit polyclonal antibodies to mouse SAP-1 were also generated with the use of a glutathione-S-transferase fusion protein containing the cytoplasmic region of mouse SAP-1 (amino acids 648–971) as antigen, as previously described (Matozaki et al. 1994).

Northern blot analysis

A digoxigenin-labeled cRNA probe corresponding to nt –31 to 3454 of mouse SAP-1 cDNA was synthesized with the use of a DIG RNA Labeling Kit (Roche, Basel, Switzerland) and T7 RNA polymerase. Polyadenylated RNA (2 µg) fixed on a nylon membrane was subjected to hybridization overnight at 50 °C with 400 ng of the cRNA probe in DIG Easy Hyb (Roche). The membrane was then exposed to alkaline phosphatase–conjugated sheep polyclonal Fab fragments to digoxigenin (1 : 10 000 dilution; Roche), and hybridization complexes were visualized with the CDP-Star reagent (Roche).

Immunoblot analysis

Mouse tissues were homogenized on ice in lysis buffer (20 mm Tris–HCl (pH 7.6), 140 mm NaCl, 1 mm EDTA, 1% Nonidet P-40) containing 1 mm phenylmethylsulfonyl fluoride, aprotinin (10 µg/mL), and 1 mm sodium vanadate. The lysates were centrifuged at 10 000 g for 15 min at 4 °C, and the resulting supernatants were subjected to immunoblot analysis.

Histological and immunofluorescence analyses

For analysis of mouse tissues, mice were anesthetized by intraperitoneal (i.p.) injection of sodium pentobarbital at 25 mg/kg of weight and were then perfused transcardially with 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS). Tissues were then dissected, fixed again with 4% PFA in PBS for 1 h at room temperature, transferred to a series of sucrose solutions [7%, 20% and 30% (w/v), sequentially] in PBS, embedded in optimal cutting temperature (OCT) compound (Sakura, Tokyo, Japan), and rapidly frozen in liquid nitrogen. Frozen sections of 5 µm thickness were then prepared and subjected to immunofluorescence analysis with primary antibodies and secondary antibodies labeled with fluorescent dyes, as described previously (Ishikawa-Sekigami et al. 2006). Fluorescence images were obtained with a confocal laser-scanning microscope (LSM Pascal; Carl Zeiss, Jena, Germany). To examine the expression of alkaline phosphatase, we used a Vector Red Alkaline Phosphatase Kit I (Vector Laboratories, Burlingame, CA). Fixed tissue samples were also embedded in paraffin, sectioned, and stained with Mayer's hematoxylin–eosin.

Transmission EM

Mice were anesthetized as described above and were then perfused transcardially with 2% PFA and 2.5% glutaraldehyde in 0.1 m sodium cacodylate buffer (pH 7.4). Tissues were removed, immersed in the same fixative for 1 h at 4 °C, and then incubated for 1 h at 4 °C with 1% OsO4 in the same buffer. They were then dehydrated and embedded in Epon. For immuno-EM, tissue samples were fixed by immersion for 2 h at room temperature in PBS containing 4% PFA and 1% glutaraldehyde. Frozen sections of 10 µm thickness were then prepared and processed for immunostaining with a mAb to mouse SAP-1, and with Fab fragments of goat antibodies to rat IgG that were labeled with 1.4-nm gold particles (Nanoprobes, Yaphank, NY); signals were enhanced with the use of an HQ Silver Enhancement kit (Nanoprobes). The sections were fixed again with 1% OsO4 and 0.1% potassium ferrocyanide and were finally embedded in Epon. Ultrathin sections (90 nm) were prepared, stained with uranyl acetate and lead citrate, and examined with a JEM 1010 electron microscope (JEOL, Tokyo, Japan).

Generation of SAP-1-deficient mice

To generate SAP-1-deficient mice, we first obtained two bacterial artificial chromosome clones containing 129/Sv mouse Sap1 genomic DNA from BACPAC Resources Center (Children's Hospital Oakland Research Institute, Oakland, CA). The targeting vector (Fig. 5A) was constructed with a 1.2-kb DNA fragment spanning exon 14 to intron 16 as a 5′ homologous region, with a 6.4-kb fragment spanning intron 16 to intron 21 as a 3′ homologous region, with a construct (PGKβgeo) comprising the promoter of the mouse phosphoglycerate kinase 1 gene linked to a β-galactosidase–neomycin phosphotransferase fusion gene (Friedrich & Soriano 1991), and with a diphtheria toxin A chain (DTA) gene cassette (Takara BIO). A LoxP site was also inserted into intron 16 as well as into intron 18 of the targeting vector. The targeting vector was introduced into the J1 line of mouse embryonic stem cells by electroporation, and homologous recombination was confirmed by Southern blot analysis, as described previously (Harada et al. 1994), with two DNA probes (S1 and S2) (Fig. 5A). Chromosomal DNA was thus purified from each cell clone and was digested with Hind III (for the S1 probe) or Bam HI (for the S2 probe). We crossed transgenic mice that express Cre recombinase under the control of CAG promoter (CAG-Cre mice) (Sakai & Miyazaki 1997) with mice harboring a targeted Sap1 allele. The genotype of the resulting offspring was determined by either Southern blot (Fig. 5B) or PCR analysis. In PCR analysis, 396- and 496-bp DNA fragments corresponding to the wild-type and targeted alleles, respectively, were generated with the sense primer MSAP-MS (5′-ATGGAGAGACT GGCTCTTGC-3′) and the antisense primer MSAP LOX-AS (5′-CCATGCAGCTAGTAACATGG-3′) and a 300-bp DNA fragment corresponding to the deleted allele was generated with the primers Lox-g (5′-ACGGTATCGATAAGCTTGGC-3′) and MSAP LOX-AS. SAP-1-KO mice were then backcrossed onto the C57BL/6 background for four generations. Mice were bred and maintained at the Institute of Experimental Animal Research of Gunma University under specific pathogen-free conditions. All animal experiments were performed with the guidelines of the Animal Care and Experimentation Committee of Gunma University.

Adenoma formation in SAP-1-deficient mice with the ApcMin allele

ApcMin/+ mice (C57BL/6J-ApcMin/J) were obtained from the Jackson Laboratory (Bar Harbor, ME) and were crossed with SAP-1 KO mice. The resulting ApcMin/+Sap1+/– offspring were crossed to generate ApcMin/+Sap1−/– mice and their ApcMin/+Sap1+/+ littermates for comparison. The small intestine and colon were removed from mice at 20 weeks of age and were fixed in methacarn (methanol : chloroform : acetic acid, 4 : 2 : 1, v/v) for 3 min (Prokhortchouk et al. 2006). The number and diameter of adenomas were determined with the use of a stereoscopic microscope (SZX7; Olympus, Tokyo, Japan) at a magnification of 10×. Tissue was also embedded in paraffin for histological analysis.

BrdU incorporation assay

Mice were injected i.p. with BrdU (50 µg/g of body weight in PBS) and killed 1 h later. Intestinal tissue was fixed in methacarn, embedded in paraffin, sectioned, and stained with a mAb to BrdU with the use of a BrdU In-Situ Detection Kit (BD Pharmingen, Franklin Lakes, NJ). The BrdU labeling index was defined as the percentage of BrdU-positive cells per 150 adenoma cells.

Apoptosis assay

Intestinal tissue was fixed in methacarn, transferred to a series of sucrose solutions [7%, 20% and 30% (w/v), sequentially], embedded in OCT compound, and evaluated for apoptosis with an apoptosis detection kit (Takara BIO). The number of apoptotic cells per 150 adenoma cells was determined.

Statistical analysis

Data are presented as means ± SEM and were analyzed by Student's t test. A P value of < 0.05 was considered statistically significant.

Acknowledgements

We thank S. Tsukita and M. Furuse for antibodies, Y. Hayashizaki for a RIKEN full-length cDNA clone (#9030616J04), J. Miyazaki for CAG-Cre mice, T. Fujimoto for his suggestion and instruction for EM analysis, as well as K. Tomizawa, H. Kobayashi, Y. Hayashi, Y. Niwayama and T. Horie for technical assistance. This work was supported by a Grant-in-Aid for Scientific Research on Priority Areas Cancer, a Grant-in-Aid for Scientific Research (B), and a grant of the Global Center of Excellence Program from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Ancillary