PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Cancer Cell. Author manuscript; available in PMC 2012 April 12.
Published in final edited form as:
PMCID: PMC3077713
NIHMSID: NIHMS282662

Deletion of p120-catenin results in a tumor microenvironment with inflammation and cancer that establishes it as a tumor suppressor gene

Summary

P120ctn interacts with E-cadherin, but no formal proof that p120ctn functions as a bone fide tumor suppressor gene has emerged. We report herein that p120ctn loss leads to tumor development in mice. We have generated a conditional knockout model of p120ctn whereby mice develop pre-neoplastic and neoplastic lesions in the oral cavity, esophagus and squamous forestomach. Tumor derived cells secrete granulocyte macrophage colony stimulating factor (GM-CSF), macrophage colony stimulating factor (M-CSF), monocyte chemotactic protein-1 (MCP-1) and tumor necrosis factor-α (TNFα). The tumors contain significant desmoplasia and immune cell infiltration. Immature myeloid cells comprise a significant percentage of the immune cells present, and likely participate in fostering a favorable tumor microenvironment, including the activation of fibroblasts.

Introduction

Cell-cell junctions are critical for cell adhesion during normal epithelial homeostasis. The cardinal feature of the adherens junctions is the interaction between the cytoplasmic tail of classical type I (E-cadherin) and type II cadherins, and p120-catenin (p120ctn, also referred to as catenin, delta1 or Ctnnd1), which engenders stability of this specific junctions complex at the cell membrane (Davis et al., 2003; Ireton et al., 2002; Xiao et al., 2003). Recently, it was revealed by X-ray crystallography that p120 isoform 4A is in complex with the juxtamembrane domain core region (JMD(core)) of E-cadherin (Ishiyama et al., 2010). Loss of p120ctn or its phosphorylation on serine-threonine residues is able to destabilize E-cadherin (Fukumoto et al., 2008). Thus, p120ctn regulates cadherin stability and turnover. The balance between adhesion and migration is regulated further by p120ctn’s ability to regulate the activities of RhoA, Rac and Cdc42 through the amino-terminus of p120ctn, thereby orchestrating exquisite actin dynamics (Reynolds and Roczniak-Ferguson, 2004). Furthermore, the bridge between p120ctn and these effectors of migration involves p190RhoGAP (Wildenberg et al., 2006).

p120ctn’s structure is highlighted by an amino-terminal domain has regulatory functions, a central Armadillo Repeat domain (Arm), a nuclear export signal (NES), and a carboxyl-terminal domain whose function(s) remain to be elucidated (Anastasiadis et al., 2000). Isoforms of p120ctn result from differential splicing. Initiation at the first ATG start site (Isoform 1) results in preferential expression in the mesenchymal compartment, while isoforms that lack a coiled-coiled domain that utilize the third start site (Isoform 3) are expressed in the epithelial compartment. Not all isoforms express the NES.

It has become increasingly clear that loss, downregulation or mislocalization of p120ctn helps to define diverse tumor types in the prostate, breast, pancreas, colon, skin, bladder and endometrium (Reynolds and Roczniak-Ferguson, 2004; Thoreson and Reynolds, 2002; van Hengel and van Roy, 2007). Interestingly, decreased expression is not typically observed in the entire tumor but rather in regions of the tumor, similar to that observed for E-cadherin loss (Birchmeier, 1995). This loss or mislocalization of p120ctn frequently leads to E-cadherin destabilization, endowing a cancer cell an advantage in cell migration by virtue of abrogation of cell adhesion. Additionally, knockdown experiments using shRNA to p120ctn have demonstrated that p120ctn loss induces invasion in tumor cell lines with concomitant loss or down-regulation of E-cadherin (Macpherson et al., 2007), although collective invasion might involve also concurrent retention of membranous p120ctn and cadherin mediated (P- or E-cadherin) cell-cell contacts at least in vitro. In aggregate, these data suggest a potential tumor suppressor role for p120ctn in various cancer types, but a causal role in vivo has yet to be demonstrated unequivocally.

P120ctn loss in the mouse has been modeled in an attempt to phenocopy human disease, but to date, no published mouse model has yielded a cancer phenotype. Deletion of p120ctn by homologous recombination in mice results in embryonic lethality. Tissue-specific targeting of p120ctn has been utilized to study the role of p120ctn in development and tumorigenesis (Davis and Reynolds, 2006). For example, p120ctn loss results in impaired enamel development (Bartlett et al., 2010). Targeted p120ctn deletion in the embryonic salivary gland forces a loss of acinar development and an adoption of a ductal cell fate. Although these mice die immediately in the postnatal period, the salivary glands are distinguished by intraepithelial dysplasia but not cancer (Davis and Reynolds, 2006). P120ctn loss in the skin results in epidermal hyperplasia and chronic inflammation with loss of hair and body fat (Perez-Moreno et al., 2006). Furthermore, the epidermis has evidence of NFκB activation and mitotic defects such as aneuploidy but no overt cancer (Perez-Moreno et al., 2008). Recently, conditional p120ctn loss in the small intestine and colon was found to result in death by 21 days with evidence of mucosal erosion and bleeding, and recruitment of COX-2 expressing neutrophils, suggesting an underlying barrier defect (Smalley-Freed et al., 2010). Thus, to date, no tissue specific p120ctn knockout mouse models have been able to demonstrate that p120ctn loss results in the development of invasive cancer.

The overarching goal of this study was to generate and characterize mechanistically a genetic mouse model in which conditional p120ctn loss in the squamous oral cavity, esophagus and forestomach results in cancer that phenocopies precisely the histologic features of human oral and esophageal squamous cell carcinomas (OSCC, ESCC, respectively). In so doing, such a model might provide a platform for the consideration of combinatorial therapeutics in ESCC, OSCC and potentially other squamous cell cancers.

Results

P120ctn expression is lost or reduced in esophageal cancer tissues and cell lines

A human tissue microarray (TMA) with matched normal and tumor esophageal squamous cell cancer (ESCC) tissues was evaluated for p120ctn expression by immunohistochemistry (IHC) (Figure 1A). P120ctn has the expected membrane-associated localization in the normal esophageal squamous epithelium. By contrast, 100% of the 69 ESCC tumors have p120ctn loss or cytoplasmic mislocalization (Figure 1B). There is a statistically significant association (p=0.017) between decreased p120ctn and E-cadherin expression with odds ratio = 3.23 (95% CI: 1.13<O.R.<9.24). These data reveal that p120ctn and E-cadherin are coordinately decreased or lost in the majority of ESCC, and substantiate previous findings (Chung et al., 2007).

Figure 1
P120ctn and E-cadherin expression in ESCC

A panel of 14 human esophageal squamous tumor cell lines was analyzed for p120ctn and E-cadherin expression (Figure 1C). P120ctn expression was detected in all cell lines with different isoforms observed in a subset of them. E-cadherin expression was moderate to high in 11 of 14 cell lines, low in one (TE8) and absent in two cell lines (HCE4, HCE7). P120ctn and E-cadherin expression correlated well in the cell lines. Interestingly, the mesenchymal p120ctn isoform appears to be expressed in HCE-4 and HCE-7 cells along with N-cadherin, consistent with N-cadherin’s pattern of known mesenchymal expression.

Conditional deletion of p120ctn in the oral cavity, esophagus and forestomach results in invasive squamous cell cancer, desmoplasia, and inflammation

To elucidate the mechanistic roles of p120ctn in vivo, we generated spatially-targeted p120ctn deletion in the oral cavity, esophagus and squamous forestomach. We bred L2Cre and p120loxP/loxP mice together to generate L2Cre;p120loxP/loxP mice, where the L2 promoter is active specifically in the squamous oral cavity, esophagus and forestomach (Nakagawa et al., 1997; Opitz et al., 2002). LacZ (Gt(Rosa) 26Tm1Sor) reporter mice were crossed with L2Cre mice to determine the efficiency of Cre recombination in the oral cavity, esophagus and forestomach of mice. LacZ staining demonstrates effective recombination, and therefore Cre expression, in these specific tissues (Figure S1). To verify p120ctn loss in L2-Cre;p120loxP/loxP mice, p120ctn protein expression was assessed by immunofluorescence (IF). P120ctn was effectively lost in the epithelium in 2 month old L2-Cre;p120loxP/loxP mice compared to L2-Cre;p120+/+ control mice (Figure 2A).

Figure 2
Gross pathology of L2Cre;p120ctnloxP/loxP mice

Gross pathology examination of the oral cavities, esophagi and forestomachs of L2Cre;p120loxP/loxP mice revealed significant masses in each of these tissues, along with splenomegaly and local lymphadenopathy (Figures 2B, 2C). Histologic analysis at 4-6 months of age revealed epithelial dysplasia (data not shown). Histologic analysis at 9-12 months of age demonstrated severe dysplasia with invasive squamous cancer into the submucosa and muscle in L2Cre;p120loxP/loxP mice while all control mice (p120flox/flox, L2-Cre;p120+/+ and L2-Cre;p120+/loxP) had no abnormalities (Figure 3A). Approximately 70% of L2Cre;p120loxP/loxP mice develop tumors between 9-12 months of age while no tumors were detected in age-matched littermate control mice (Table S1, p<0.0001). Invasive esophageal cancer was detected as early as 4 months of age in one L2Cre;p120loxP/loxP mouse. Some esophageal tumors were poorly differentiated but displayed no evidence of any obvious distant metastasis, although we cannot exclude the possibility of micrometastasis in other organs (Figure 3A). The submucosa was altered with marked desmoplasia and the striking presence of immune cell infiltration (Figures 3A).

Figure 3
Histologic analysis of tumors from L2Cre;p120ctnloxP/loxP mice

Analysis of L2-Cre;p120loxP/loxP forestomachs revealed severe hyperplasia, severe dysplasia and invasive cancers into the underlying desmoplastic and inflammatory stroma. (Figure 3A). The forestomachs of all control mice (p120flox/flox, L2-Cre;p120+/+ and L2-Cre;p120+/loxP) revealed no morphologic changes (Figure 3A). Nearly identical results are evident in the oral cavities of L2-Cre;p120loxP/loxP mice, namely invasive squamous cell cancer with desmoplasia and immune cell infiltration (Figure 3A).

We next evaluated E-cadherin expression in L2-Cre;p120loxP/loxP mice. Significant loss of E-cadherin expression is noted in L2-Cre;p120loxP/loxP mice compared to L2-Cre;p120+/+ control mice with residual E-cadherin expression in some L2-Cre;p120loxP/loxP mice where it is cytoplasmic (Figure 2B). In some mice, while p120ctn is lost and E-cadherin is either lost or cytoplasmic, there are occasional focal areas within tumors that retain both p120ctn and E-cadherin at the cell membrane (Figure 2C), the latter consistent with the notion of collective invasion observed in A431 cells in vitro (Macpherson et al., 2007).

Alterations in proliferation and differentiation in L2-Cre/p120flox/flox mice

Ki-67 and keratin 14 (K14), markers of proliferation, were evaluated in esophageal tumors from L2-Cre;p120loxP/loxP mice. Normal proliferating basal cells express Ki-67 and K14 while normal differentiated suprabasal cells express keratins 4 and 13 (K4, K13) and harbor little proliferation. Ki-67 expression was significantly increased in the esophageal epithelia and in the invasive cancers in L2-Cre;p120loxP/loxP mice compared to control mice (Figure S2, p=3 × 10−6). K14 expression was increased in the esophageal epithelia at 5 months of age, which ultimately expanded to the luminal surface by 12 months of age (data not shown). Keratin 4 expression as evaluated by qPCR is decreased 2.5 fold (p=0.014) in L2-Cre;p120loxP/loxP mice compared to L2-Cre;p120+/+ control mice (data not shown). Invading tumor cells are K14 positive (Figure 3C). Interestingly, β-catenin expression was membranous in the esophagi of L2-Cre;p120loxP/loxP mice (Figure 3C), but lost in the oral cavity of the L2-Cre;p120loxP/loxP mice. Furthermore, no alterations were apparent for epidermal growth factor receptor (EGFR), c-myc or K-ras by IHC in the tumors (data not shown). However, p-Akt (but not p-Erk1/2 (p-Mapk)), was increased in the tumors, as was p-Stat3, which might contribute to tumorigenesis (Figure S2).

NFκB and inflammatory cells in the tumor microenvironment

Given the significant recruitment of immune cells in the tumor stroma, we hypothesized that NFκB activation may be present in these lesions as has been reported previously (Perez-Moreno et al., 2006; Perez-Moreno et al., 2008). Indeed, this was confirmed with robust NFκB induction in dysplasia and tumors, thereby suggesting this is an early event (Figure 4A). The NFκB induction was evident also in the immune cells found in the tumor stroma (Figure 4A).

Figure 4
Immature myeloid cells are recruited to the esophagi and squamous forestomachs of L2Cre;p120ctnloxP/loxP mice

We then established cells from an esophageal tumor (designated as 714ET) as well as esophageal cell lines from p120ctnloxP/loxP mice that in turn were infected with retroviral Cre recominbase or a control vector (designated as F2-Cre and F2-Tomato, respectively). Western blot analysis revealed complete p120ctn loss in both F2-Cre and 714ET cells and retained p120ctn expression in the control F2-Tomato cells (Figure 4B and quantified in Figure 4C-left panel). In F2-Cre and 714ET cells, there is activated p- NFκB (8-fold), p-Akt (4-fold) and p-Stat3 (2-fold) (Figure 4B, 4C). Additionally, 714ET cells were found to secrete significant amounts of GM-CSF, M-CSF and MCP-1 and TNFα (Figure 4D), but not other cytokines tested (IL-1β, IL-2, IL-4, IL-5, IL-6, or IFNγ, data not shown). GM-CSF has been suggested to contribute to immature myeloid cell recruitment (Marigo et al., 2010).

In order to characterize the immune cells in the tumor microenvironment, we undertook a number of experiments. Splenic weights in L2-Cre;p120loxP/loxP mice were 5-fold greater than in control mice (p=0.0004) and there was also leukocytosis with increased segmented neutrophils, band neutrophils, eosinophils and monocytes (Table S2). These findings suggested that there may be an elevated production of circulating myeloid lineage cells and that the invasive squamous cell cancers in the L2-Cre;p120loxP/loxP mice might contribute to the recruitment of these immune cells. Therefore, we analyzed the leukocyte fraction (CD45+ cells) for immature myeloid cells (defined phenotypically as Gr-1+ CD11b+ cells), macrophages (F4/80+), T cells (CD3+), B cells (CD19+), dendritic cells (CD11c+), and natural killer cells (CD49b+). The percentage of CD45+ cells was approximately the same for normal vs. dysplastic vs. tumor tissues and spleens (Figure 4E). However, there was a statistically significant increase (Figure S3) in immature myeloid cell infiltration in dysplastic and tumor tissues (Figure 4E). In dysplastic esophagi or forestomachs, the percentage of immature myeloid cells increased 4-8 fold compared to control tissues (<5% of CD45+ cells) (Figure 4F). The percentage of immature myeloid cells was higher also in esophageal and forestomach tumors (Figure 4F). Other types of immune cells (macrophages, T cells, B cells, dendritic cells, NK cells) did not increase with dysplasia and cancer (Figure 4E, Figure S3).

The fact that only one subpopulation of immune cells (immature myeloid cells) is specifically increased suggests p120ctn loss may lead to a highly selective process of immune cell recruitment. We next purified splenic Gr-1+ CD11b+ cells and prepared cytospins for cytology (Figure S3). Staining and quantification revealed that the Gr-1+ CD11b+ cells represent a range of cells, including myelocytes/metamyelocytes (11.7%), segmented neutrophils (48.1%), band neutrophils (12.5%) and monocytoid cells (7.7%). These cells have activated nuclear NFκB (Figure S3), consistent with what was found in the mouse tumors (Figure 4A).

In order to evaluate the functional properties of the immature myeloid cells, we again purified Gr-1+ CD11b+ cells from tumor-bearing L2-Cre;p120loxP/loxP mice and found that these cells inhibited peptide-specific T cell responses in vitro (using OVA specific TCR-transgenic T cells) (Figure 4G). Additionally, these immature myeloid cells produce nitric oxide, which have been suggested to suppress T cells (Figure 4H) (Bronte et al., 2003; Bronte and Zanovello, 2005). We next treated L2-Cre;p120loxP/loxP mice with dexamethasone and found a dramatic attenuation of esophageal tumor invasion to that of hyperplasia (Figure 5A, Figure S4). CD45+ cells and Gr-1+CD11b+ cells were quantified by flow cytometry and were reduced (Figure 5B and data not shown). We analyzed further these Gr-1+CD11b+ cells and found decreased IL-4Rα expression in the dexamethasone treated mice (Figure 5C). Notably, IL-4Rα expression on Gr-1+CD11b+ cells has been reported to correlate with suppression of CD8+ T cells (Gallina et al., 2006; Mandruzzato et al., 2009). These data suggest that inflammatory cells, including the Gr-1+CD11b+ cells, are fostering tumorigenicity.

Figure 5
The immune response is involved in the tumor microenvironment in L2Cre;p120ctnloxP/loxP mice

Desmoplasia is present in the tumor microenvironment

To evaluate the specific features of the microenvironment in the invasive cancers that develop in L2-Cre;p120loxP/loxP mice, a pronounced increase in fibrillar collagen was present in the tumor stroma consistent with the desmoplasia in human ESCC and OSCC (Figure 5D). By contrast, fibrillar collagen is localized to a narrow region below the epithelium (Figure 5D). Additionally, there was increased fibroblast specific protein-1 (Fsp1) expression in mouse OSCC and ESCC (Figure 5E). Similarly, alpha-smooth muscle actin (αSMA) was increased in the mouse tumors (data not shown). Interestingly, dexamethasone treated mice had resolution of the desmoplasia with little residual Fsp1+ fibroblasts (Figure 5E, 5F and Figure S4), suggesting a possible role for the immune system in regulating desmoplasia in our model system.

To evaluate a potential link between immature myeloid cell inflammation and desmoplasia, we next conducted co-culture of purified Gr-1+CD11+ cells from tumor-bearing L2-Cre;p120loxP/loxP mice with normal fibroblasts (from C57BL/6 mice) and cancer (OSCC, ESCC) associated fibroblasts (CAF) (from L2-Cre;p120loxP/loxP mice) that revealed activation of the normal fibroblasts and further augmentation of CAFs based upon αSMA and Fsp1 expression (Figure 6A-C). In fact, some of the activated CAFs have an interesting “dendritic” appearance in co-culture and display a similar morphology in mouse tumors (Figure 6C). Furthermore, the Gr-1+CD11+ cells survived better and expressed higher IL-4Rα levels in co-culture with CAFs (Figure 6D).

Figure 6
Crosstalk between immature myeloid cells and fibroblasts in the tumor microenvironment of L2Cre;p120ctnloxP/loxP mice

Given the different lines of evidence for the tumor-promoting role of Gr-1+CD11+ cells in L2-Cre;p120loxP/loxP mice, we next sought to determine a relationship between myeloid cells and p120ctn loss in human ESCC. Increased myeloperoxidase (MPO) staining in the tumor stroma for myeloid lineage cells correlates with p120ctn loss (Figure 7A, p=0.05). This raises the possibility that p120ctn loss might induce the recruitment of MPO+ cells in the pathogenesis of human ESCC.

Figure 7
Myeloperoxidase expression in human ESCC

Secondary alterations in invasive squamous cell cancers

Apart from enhanced activation of NF-κB, Akt and Stat-3 in the tumor cells, and the probability that GM-CSF, MCP-1 and TNFα are induced by NFκB in the tumor cells (Schreck and Baeuerle, 1990), we have begun to identify potential secondary alterations in p120ctn-deficient squamous epithelia that may contribute to cancer development. RNA microarray analysis was performed using isolated epithelia from the esophagi and forestomachs from 12 month old L2-Cre;p120loxP/loxP mice and L2-Cre/p120+/+ control mice. Unsupervised heirarchical clustering was performed and identified significantly altered genes (Figure S5). Several genes involved in cytokine signaling, IL-13Ra1, IL-1R2 and Pias1, are altered in p120-deficient squamous epithelia based upon microarray analysis and were confirmed by qPCR (Figure S5).

Discussion

P120-catenin loss results in invasive squamous cell cancer

The complex between E-cadherin and p120-catenin is critical for the formation and maintenance of the adherens junctions. Disruption of this complex is a hallmark feature of epithelially-derived cancers. P120ctn loss or mislocalization can destabilize E-cadherin. Definitive proof for p120ctn’s functional role as a tumor suppressor gene in vivo has been elusive to date. Targeted disruption of murine p120ctn in the salivary gland (MMTV-cre;p120ctnflox/flox), skin (K14-cre;p120ctnflox/flox) and intestine (Villin-Cre; p120ctnflox/flox) do not result in cancer (Davis and Reynolds, 2006; Perez-Moreno et al., 2006; Smalley-Freed et al., 2010). P120ctn may not be critical for tumor suppression in these specific tissues. Alternatively, a lack of tumor formation in these mice may be due to early death. Generation and characterization of L2-Cre;p120loxP/loxP mice result in a dramatic phenotype of invasive cancer in the oral cavity, esophagus and forestomach. These cancers may be either unifocal or multifocal and there is preceding severe dysplasia, underscoring a temporal progression from preneoplasia to neoplasia, mirroring human OSCC and ESCC progression.

How does p120ctn loss mechanistically cause invasive squamous cell cancers? Concordant p120ctn and E-cadherin loss is likely favoring cancer cell migration and invasion. Since p120ctn loss precedes E-cadherin loss in our model system, it is unclear to what extent E-cadherin destabilization or loss contributes to tumorigenesis. These squamous cancers induce pro-survival signals with NF-κB, Akt and p-Stat3 activation. In contrast to p120ctn loss in skin with activated MAPK (Perez-Moreno et al., 2006), p120ctn loss in oral and esophageal keratinocytes do not have enhanced MAPK signaling. These squamous cell cancers have striking features of desmoplasia and immune cell infiltration, with significant increases in immature myeloid cells but not in T cells, B cells, NK cells, dendritic cells and macrophages. The precise recapitulation of human ESCC (along with OSCC) by the conditional deletion of p120ctn is remarkable, and a definitive in vivo genetic demonstration of p120ctn deletion resulting in cancer.

The tumor microenvironment is critical in mediating the effects of p120-catenin loss

The tumor microenvironment is an array of diverse cell types that cooperate to foster tumor cell migration and invasion. The extracellular matrix includes invading tumor cells, fibroblasts (cancer associated fibroblasts), endothelial cells, pericytes, and immune cells (Andreu et al., 2010; DeNardo et al., 2010; Kalluri and Zeisberg, 2006; Lewis and Pollard, 2006).

Immature myeloid cell recruitment has been noted by IL-1β overexpression in the stomach with resulting inflammation and cancer (Tu et al., 2008), disruption of TGFβ signaling in mammary carcinomas (Yang et al., 2008), knock-in of mutant K-ras to the pancreas (Clark et al., 2007) and preparing the local pulmonary environment for metastasis (Yan et al., 2010). We observed a significant increase in immature myeloid cells in the tumor microenvironment and the spleen. It is conceivable that GM-CSF secreted by tumors contributes to the recruitment of immature myeloid cells (Marigo et al., 2010). We do not find evidence of increased IL-1β, IL-6, VEGF and prostaglandins from the tumor cells, which have been linked also to the recruitment of immature myeloid cells (Ostrand-Rosenberg and Sinha, 2009). Other cell types in the tumor microenvironment might serve as the sources for these other cytokines.

How do immature myeloid cells contribute to tumorigenesis? Immature myeloid cells have the ability to suppress antigen-specific T cell responses in vitro (Gabrilovich et al., 2001; Gallina et al., 2006; Sinha et al., 2005). In our model system, immature myeloid cells isolated from tumor-bearing mice also have the ability to inhibit T cell activation. Additionally, the immature myeloid cells activate normal fibroblasts and augment activation of CAFs based upon Fsp1 and αSMA expression, which are novel findings. In turn, CAFs in our system contribute to prolonged survival of immature myeloid cells. We cannot exclude other pro-tumorigenic functions of immature myeloid cells in our model system related to angiogenesis, secretion of chemokines/cytokines, and production of ECM degrading enzymes (DeNardo et al., 2010; DeNardo et al., 2009; Ferrara, 2010; Gabrilovich and Nagaraj, 2009; Ostrand-Rosenberg and Sinha, 2009; Yang et al., 2004; Yang et al., 2008). Since dexamethasone has protean effects, it will be important to ablate immature myeloid cells in a specific fashion in our model system and in those models where immature myeloid cells are thought to be critical.

In summary, conditional p120ctn loss in the esophagus, oral cavity and forestomach leads to invasive squamous cell cancers, involving NFκB, Akt, Stat-3 activation, increased proliferation, and production of GM-CSF, M-CSF, MCP-1 and TNFα from tumor derived cells. The accompanying tumor microenvironment is replete with desmoplasia and immature myeloid cells. These events likely contribute to cancer arising from p120ctn-deficient cells (Figure 7), which provide compelling evidence that p120ctn is indeed a bona fide tumor suppressor gene. Since the L2-Cre;p120loxP/loxP mice phenocopy human OSCC and ESCC, they serve as a model for the translational development of diagnostic and therapeutic agents, goals that have been hindered by the absence of robust, genetically based models. Furthermore, our model might be applicable to other squamous cancers given common genomic alterations between ESCC and lung SCC where SOX-2 is amplified (Bass et al., 2009). Several approaches merit consideration in this context. First, strategies designed to restore wild-type p120ctn may provide a venue to ameliorate tumor progression. Given the complex nature of the tumor microenvironment, targeting of components, either individually or in combination, might advance the field. For example, prodrugs have been demonstrated to be efficacious against fibroblast activation protein (FAP) in prostate cancer (Lebeau et al., 2009). Strategies to target immature myeloid cells might be attractive based upon our studies as well as in pancreatic cancer (Clark et al., 2007).

Experimental Procedures

Cell lines

TE and HCE cells, all established human esophageal squamous cancer cell lines, were cultivated in DMEM media supplemented by 10% fetal calf serum (Sigma-Aldrich) and 1% penicillin/streptomycin (Invitrogen) and grown in 5% CO2 conditions as described previously (Okano et al., 2000). Cell lines were established from mouse esophageal tumors (designated as 714ET) and from mouse fibroblasts (designated as OF3 from normal oral cavity) and mouse cancer associated fibroblasts (designated as 1136CF from OSCC and designated as 955EF from ESCC) and verified to be of epithelial or fibroblast origin. The mouse cell lines were cultivated in DMEM as above.

To generate pBABE-tdTomato-SV40-nlsCre (6982 bp), puromycin acetyltransferase of pBABE-puro was replaced with nlsCre cDNA derived from pHRnlsCre (Addgene), and then tdTomato cDNA (gift of Roger Y. Tisen, UCSD) was subcloned. The parental F2 cells were established through the isolation of esophageal epithelia (Opitz et al., 2002) from p120loxP/loxP mice, infected with pBABE-tdTomato-SV40-nlsCre (designated as F2-Cre) or pBABE-tdTomato (designated as F2-Tomato), and stable clones were FACS selected for tdTomato. The F2-Cre and F2-Tomato cell lines were cultivated in KSFM as described previously (Andl et al., 2003; Harada et al., 2003).

Histology and Immunohistochemistry/Immunofluorescence

Paraffin tissue sectioning and staining for hematoxylin and eosin (H&E) and IHC were performed as described previously (Stairs et al., 2008). Antibodies used for IHC were the following: p120ctn (610134, BD Transduction), E-cadherin (610182, BD Transduction), K14 (PRB-155P, Covance), pAKT (9271, Cell Signaling), pMAPK (9101, Cell Signaling), NFκB-p65 subunit (4764, Cell Signaling), Ki67 (NCL-KI67p, Novacastra). For IF, the following antibodies were used: Keratin 14 (PRB-155P, Covance), Fsp1 (DakoCytomation), β-catenin (9562, Cell Signaling), p120ctn (61634, BD Transduction), E-cadherin (3195, Cell Signaling), c-myc (1472-1, Epitomics), myeloperoxidase (45977, Abcam), p-Stat3 (9145, Cell Signaling), αSMA (A2547, Sigma). Secondary antibodies used: AlexaFluor 488-anti-rabbit, AlexaFluor 568 anti-rat or AlexaFluor 568 anti-mouse. Slides were coverslipped using VectaShield mounting medium with DAPI.

Tissue microarrays (TMA)

TMA were created using 69-matched paraffin blocks from de-identified patients with primary ESCC and adjacent normal esophageal tissues (with Institutional Review Board exempt status for the TMA due to the de-identification). Scoring of p120ctn, E-cadherin, and MPO IHC was based upon a quantitative evaluation of expression using a scale from 0 to 2 (0=absent, 1=mild-moderate, 2=high) and performed by Dr. Andres Klein-Szanto, pathologist, in a blinded fashion. Fisher’s exact test was used.

Generation of mouse models

The Institutional Animal Care and Use Committee (IACUC) at the University of Pennsylvania approved all animal studies. Mice were housed under a 12hr light/dark cycle and fed ad libitum. LoxP flanked p120ctn mice were generated previously (Davis and Reynolds, 2006). L2Cre transgenic mice were generated by cloning Cre recombinase under the control of the EBV ED-L2 promoter (Nakagawa et al., 1997; Opitz et al., 2002). Mice were genotyped for the LoxP allele as described previously (Davis and Reynolds, 2006). Mice were genotyped for L2Cre by PCR analysis using primers contained within the Cre (Forward: GGCCAGCTAAACATGCTTCATCGT, Reverse: ACGTAACAGGGTGTTATAAGCAAT). PCR conditions were 35 cycles at 94°C for 45sec, 61°C for 45sec, and 72°C for 1min, 30sec. Additionally, L2Cre mice were bred with Gt(Rosa)26Tm1Sor reporter mice, and tissues were subjected to LacZ staining as described previously (Brembeck et al., 2001).

Dexamethasone treatment

L2-Cre;p120loxP/loxP mice were injected intraperitoneally with either dexamethasone 10mg/kg/day or PBS (control) for 7 days. After 7 days, mice were sacrificed and the esophagi and spleens were excised for histology and FACS analysis.

Flow cytometry

Single-cell suspensions of fresh spleens, esophagi, and squamous forestomachs were prepared from genetically engineered mice and littermate controls. Spleens were crushed and passed through a 70-μm cell strainer, treated with ACK Lysis Buffer (Invitrogen) and washed twice with RPMI/10% FCS. Esophagi and forestomachs were minced into 1-2 mm pieces, incubated in 1 mg/mL collagenase (Sigma-Aldrich) in RPMI for 45 minutes at 37°C, and then were passed through a 70-μm cell strainer and washed once RPMI/10% FCS. Cells were stained in PBS/0.5% FCS with the following antibodies: CD45 (clone 30-F11), CD11c (clone HL3), Gr-1 (clone RBG-8C5), CD11b (clone M1/70), CD3ε (clone 145-2C11), CD49b (clone DX5), CD19 (clone 1D3), CD124 (clone mIL-4R-M1), and rat IgG2a, κ (clone R35-95) from BD Biosciences PharMingen and F4/80 (clone C1:A3-1) from AbD Serotec. Flow cytometry was done using a BD FACSCanto (BD Biosciences Immunocytometry Systems), and data were analyzed with BD FACSDiva software.

Gr-1+ CD11b+ cell isolation and functional T cell suppression assay

Gr-1+ CD11b+ cells were isolated from spleens using biotinylated anti-Gr-1 mAb, magnetic beads, and MiniMACS columns (Miltenyi Biotec). The purity was >90% Gr-1+ CD11b+ cells by flow cytometry. Antigen-specific suppression of CD8+ T cells was evaluated using splenocytes from OT-1 transgenic mice as responder cells. OT-1 splenocytes were seeded in triplicates in 96-well round-bottom plates (7 × 105/well) and stimulated with cognate antigen, OVA-derived peptide SIINFEKL (1 μg/mL; New England Peptide) and cultured for 4 days in the presence of increasing ratios of Gr-1+ CD11b+ cells. 18 hours before harvesting, cells were pulsed with [3H]-thymidine (1 μCi per well; Amersham Biosciences). [3H]-thymidine incorporation was assessed using a liquid scintillation counter and expressed as cpm. Student t-test was used.

NO production

NO production by Gr-1+ CD11b+ cells was evaluated in culture supernatant from T cell suppression assays using the Griess Reagent System (Promega). Equal volumes of culture supernatant (40 μL) were mixed with 1% sulfanilamide in 5% phosphoric acid and incubated at RT for 10-min, followed by addition of 0.1% N-1-naphthylethylenediamine dihydrochloride in water. After a 10-min incubation at RT, absorbance (550 nm) was measured using a microplate reader. Nitrite concentrations were determined by comparing the absorbance for the test samples to a standard curve generated by serial dilution of 0.1 mM sodium nitrite. Student t-test was used.

Co-culture of immature myeloid cells and fibroblasts

Immature myeloid cells were co-cultured with fibroblasts for 96 hours, and fibroblasts were evaluated for αSMA and Fsp1. Immature myeloid cells were evaluated for survival and IL-Rα expression. Student t-test was used.

Cytology of Gr-1+ CD11b+ cells

Gr-1+ CD11b+ cells were purified (>98%) from spleens of tumor-bearing mice using a MoFlo fluorescence activated cell sorter (DakoCytomation). Cytospin preparations were stained with Diff-Quick reagent and by IF for NFκB.

Cytokine analysis

The following cytokines were measured either by BD Cytokine bead array (BD Biosciences) or by ELISA (R&D systems): IL-1β, IL-2, IL-4, IL-5, IL-6, IFNγ, TNFα, GM-CSF, G-CSF, MCP-1, KC, M-CSF, SCF and VEGF.

RNA microarrays

The epithelia from mouse esophagi and squamous forestomachs were isolated by digesting the tissues with a 1:2 dilution dispase (BD) in deionized water at 37°C for 5-10 minutes. The epithelia were peeled and snap-frozen and stored at −80°C for RNA isolation. RNA preparation was done, cRNA was prepped (Affymetrix) and run on Affymetrix Mouse 1.0ST Affymetrix Arrays.

Statistical analysis of RNA microarray data

Gene expression differences were considered statistically significant if the p-value was less than 0.01. A global test was done as to whether the expression profiles differed between the classes by permuting the labels of which arrays corresponded to which classes. For each permutation, the p-values were re-computed and the number of genes significant at the 0.01 level was noted. The false discovery rate was estimated to be less than 10%. Cluster analysis was performed with Cluster and Treeview software (Eisen et al., 1998).

Significance

Oral squamous cell cancers (OSCC) and esophageal squamous cell cancers (ESCC) represent common human cancers worldwide. The molecular pathogenesis underlying these cancers involves p120-catenin (p120ctn), which is critical to epithelial homeostasis through the adherens junctions. P120ctn loss or mislocalization in these specific cancers served as a basis for the generation of a conditional knockout mouse model, revealing a dramatic invasive squamous cell cancer phenotype accompanied by induction of pro-survival signals, desmoplasia and recruitment of immature myeloid cells. Thus, p120ctn loss in this mouse model proves its tumor suppressor role and establishes a platform for strategies designed for early detection, molecular imaging and targeted therapeutics in these cancers that have been elusive to date.

Supplementary Material

01

Acknowledgements

This work was supported by the following NIH grants: NCI P01-CA098101 Mechanisms in Esophageal Carcinogenesis (DBS, JK, MV, JPK, AKS, PG, JAD, AKR), NCI F32-CA1308513 and NCI K99/R00-CA138498 (DBS), NCI T32-CA115299 (JK), NCI U01-CA143056 (AKR), NCI P30 Abramson Cancer Center pilot grant NCI P30-CA016520 (RHV), NIDDK R01-DK069984 (JK) and the NIDDK P30-DK050306 Center for Molecular Studies in Digestive and Liver Diseases. Work contributed by the Reynolds lab was supported by CA111947, CA055724, and the Vanderbilt GI SPORE (50 CA95103). We wish to thank the Morphology (S. Mitchell), Molecular Biology/Gene Expression, Transgenic/Chimeric Mouse, Microarray and Cell Culture Core Facilities as well as Amy Ziober in the Department of Pathology and Laboratory Medicine. We are also grateful to members of the Rustgi lab for helpful discussions.

Footnotes

Accession Numbers GEO Accession number is GSE20240 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE20240).

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  • Anastasiadis PZ, Moon SY, Thoreson MA, Mariner DJ, Crawford HC, Zheng Y, Reynolds AB. Inhibition of RhoA by p120 catenin. Nat Cell Biol. 2000;2:637–644. [PubMed]
  • Andl CD, Mizushima T, Nakagawa H, Oyama K, Harada H, Chruma K, Herlyn M, Rustgi AK. Epidermal growth factor receptor mediates increased cell proliferation, migration, and aggregation in esophageal keratinocytes in vitro and in vivo. J Biol Chem. 2003;278:1824–1830. [PubMed]
  • Andreu P, Johansson M, Affara NI, Pucci F, Tan T, Junankar S, Korets L, Lam J, Tawfik D, DeNardo DG, et al. FcRgamma activation regulates inflammation-associated squamous carcinogenesis. Cancer Cell. 2010;17:121–134. [PMC free article] [PubMed]
  • Bartlett JD, Dobeck JM, Tye CE, Perez-Moreno M, Stokes N, Reynolds AB, Fuchs E, Skobe Z. Targeted p120-catenin ablation disrupts dental enamel development. PLoS One. 2010;5 [PMC free article] [PubMed]
  • Bass AJ, Watanabe H, Mermel CH, Yu S, Perner S, Verhaak RG, Kim SY, Wardwell L, Tamayo P, Gat-Viks I, et al. SOX2 is an amplified lineage-survival oncogene in lung and esophageal squamous cell carcinomas. Nat Genet. 2009;41:1238–1242. [PMC free article] [PubMed]
  • Birchmeier W. E-cadherin as a tumor (invasion) suppressor gene. Bioessays. 1995;17:97–99. [PubMed]
  • Brembeck FH, Moffett J, Wang TC, Rustgi AK. The keratin 19 promoter is potent for cell-specific targeting of genes in transgenic mice. Gastroenterology. 2001;120:1720–1728. [PubMed]
  • Bronte V, Serafini P, Mazzoni A, Segal DM, Zanovello P. L-arginine metabolism in myeloid cells controls T-lymphocyte functions. Trends Immunol. 2003;24:302–306. [PubMed]
  • Bronte V, Zanovello P. Regulation of immune responses by L-arginine metabolism. Nat Rev Immunol. 2005;5:641–654. [PubMed]
  • Chung Y, Lam AK, Luk JM, Law S, Chan KW, Lee PY, Wong J. Altered E-cadherin expression and p120 catenin localization in esophageal squamous cell carcinoma. Ann Surg Oncol. 2007;14:3260–3267. [PubMed]
  • Clark CE, Hingorani SR, Mick R, Combs C, Tuveson DA, Vonderheide RH. Dynamics of the immune reaction to pancreatic cancer from inception to invasion. Cancer Res. 2007;67:9518–9527. [PubMed]
  • Davis MA, Ireton RC, Reynolds AB. A core function for p120-catenin in cadherin turnover. J Cell Biol. 2003;163:525–534. [PMC free article] [PubMed]
  • Davis MA, Reynolds AB. Blocked acinar development, E-cadherin reduction, and intraepithelial neoplasia upon ablation of p120-catenin in the mouse salivary gland. Dev Cell. 2006;10:21–31. [PubMed]
  • DeNardo DG, Andreu P, Coussens LM. Interactions between lymphocytes and myeloid cells regulate pro- versus anti-tumor immunity. Cancer Metastasis Rev. 2010;29:309–316. [PMC free article] [PubMed]
  • DeNardo DG, Barreto JB, Andreu P, Vasquez L, Tawfik D, Kolhatkar N, Coussens LM. CD4(+) T cells regulate pulmonary metastasis of mammary carcinomas by enhancing protumor properties of macrophages. Cancer Cell. 2009;16:91–102. [PMC free article] [PubMed]
  • Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci U S A. 1998;95:14863–14868. [PubMed]
  • Ferrara N. Role of myeloid cells in vascular endothelial growth factor-independent tumor angiogenesis. Curr Opin Hematol. 2010;17:219–224. [PubMed]
  • Fukumoto Y, Shintani Y, Reynolds AB, Johnson KR, Wheelock MJ. The regulatory or phosphorylation domain of p120 catenin controls E-cadherin dynamics at the plasma membrane. Exp Cell Res. 2008;314:52–67. [PMC free article] [PubMed]
  • Gabrilovich DI, Nagaraj S. Myeloid-derived suppressor cells as regulators of the immune system. Nat Rev Immunol. 2009;9:162–174. [PMC free article] [PubMed]
  • Gabrilovich DI, Velders MP, Sotomayor EM, Kast WM. Mechanism of immune dysfunction in cancer mediated by immature Gr-1+ myeloid cells. J Immunol. 2001;166:5398–5406. [PubMed]
  • Gallina G, Dolcetti L, Serafini P, De Santo C, Marigo I, Colombo MP, Basso G, Brombacher F, Borrello I, Zanovello P, et al. Tumors induce a subset of inflammatory monocytes with immunosuppressive activity on CD8+ T cells. J Clin Invest. 2006;116:2777–2790. [PMC free article] [PubMed]
  • Harada H, Nakagawa H, Oyama K, Takaoka M, Andl CD, Jacobmeier B, von Werder A, Enders GH, Opitz OG, Rustgi AK. Telomerase induces immortalization of human esophageal keratinocytes without p16INK4a inactivation. Mol Cancer Res. 2003;1:729–738. [PubMed]
  • Ireton RC, Davis MA, van Hengel J, Mariner DJ, Barnes K, Thoreson MA, Anastasiadis PZ, Matrisian L, Bundy LM, Sealy L, et al. A novel role for p120 catenin in E-cadherin function. J Cell Biol. 2002;159:465–476. [PMC free article] [PubMed]
  • Ishiyama N, Lee SH, Liu S, Li GY, Smith MJ, Reichardt LF, Ikura M. Dynamic and static interactions between p120 catenin and E-cadherin regulate the stability of cell-cell adhesion. Cell. 2010;141:117–128. [PubMed]
  • Kalluri R, Zeisberg M. Fibroblasts in cancer. Nat Rev Cancer. 2006;6:392–401. [PubMed]
  • Lebeau AM, Brennen WN, Aggarwal S, Denmeade SR. Targeting the cancer stroma with a fibroblast activation protein-activated promelittin protoxin. Mol Cancer Ther. 2009 [PMC free article] [PubMed]
  • Lewis CE, Pollard JW. Distinct role of macrophages in different tumor microenvironments. Cancer Res. 2006;66:605–612. [PubMed]
  • Macpherson IR, Hooper S, Serrels A, McGarry L, Ozanne BW, Harrington K, Frame MC, Sahai E, Brunton VG. p120-catenin is required for the collective invasion of squamous cell carcinoma cells via a phosphorylation-independent mechanism. Oncogene. 2007;26:5214–5228. [PubMed]
  • Mandruzzato S, Solito S, Falisi E, Francescato S, Chiarion-Sileni V, Mocellin S, Zanon A, Rossi CR, Nitti D, Bronte V, Zanovello P. IL4Ralpha+ myeloid-derived suppressor cell expansion in cancer patients. J Immunol. 2009;182:6562–6568. [PubMed]
  • Marigo I, Bosio E, Solito S, Mesa C, Fernandez A, Dolcetti L, Ugel S, Sonda N, Bicciato S, Falisi E, et al. Tumor-induced tolerance and immune suppression depend on the C/EBPbeta transcription factor. Immunity. 2010;32:790–802. [PubMed]
  • Nakagawa H, Wang TC, Zukerberg L, Odze R, Togawa K, May GH, Wilson J, Rustgi AK. The targeting of the cyclin D1 oncogene by an Epstein-Barr virus promoter in transgenic mice causes dysplasia in the tongue, esophagus and forestomach. Oncogene. 1997;14:1185–1190. [PubMed]
  • Okano J, Gaslightwala I, Birnbaum MJ, Rustgi AK, Nakagawa H. Akt/protein kinase B isoforms are differentially regulated by epidermal growth factor stimulation. J Biol Chem. 2000;275:30934–30942. [PubMed]
  • Opitz OG, Harada H, Suliman Y, Rhoades B, Sharpless NE, Kent R, Kopelovich L, Nakagawa H, Rustgi AK. A mouse model of human oral-esophageal cancer. J Clin Invest. 2002;110:761–769. [PMC free article] [PubMed]
  • Ostrand-Rosenberg S, Sinha P. Myeloid-derived suppressor cells: linking inflammation and cancer. J Immunol. 2009;182:4499–4506. [PMC free article] [PubMed]
  • Perez-Moreno M, Davis MA, Wong E, Pasolli HA, Reynolds AB, Fuchs E. p120-catenin mediates inflammatory responses in the skin. Cell. 2006;124:631–644. [PMC free article] [PubMed]
  • Perez-Moreno M, Song W, Pasolli HA, Williams SE, Fuchs E. Loss of p120 catenin and links to mitotic alterations, inflammation, and skin cancer. Proc Natl Acad Sci U S A. 2008;105:15399–15404. [PubMed]
  • Reynolds AB, Roczniak-Ferguson A. Emerging roles for p120-catenin in cell adhesion and cancer. Oncogene. 2004;23:7947–7956. [PubMed]
  • Schreck R, Baeuerle PA. NF-kappa B as inducible transcriptional activator of the granulocyte-macrophage colony-stimulating factor gene. Mol Cell Biol. 1990;10:1281–1286. [PMC free article] [PubMed]
  • Sinha P, Clements VK, Ostrand-Rosenberg S. Reduction of myeloid-derived suppressor cells and induction of M1 macrophages facilitate the rejection of established metastatic disease. J Immunol. 2005;174:636–645. [PubMed]
  • Smalley-Freed WG, Efimov A, Burnett PE, Short SP, Davis MA, Gumucio DL, Washington MK, Coffey RJ, Reynolds AB. p120-catenin is essential for maintenance of barrier function and intestinal homeostasis in mice. J Clin Invest. 2010;120:1824–1835. [PMC free article] [PubMed]
  • Stairs DB, Nakagawa H, Klein-Szanto A, Mitchell SD, Silberg DG, Tobias JW, Lynch JP, Rustgi AK. Cdx1 and c-Myc foster the initiation of transdifferentiation of the normal esophageal squamous epithelium toward Barrett’s esophagus. PLoS ONE. 2008;3:e3534. [PMC free article] [PubMed]
  • Thoreson MA, Reynolds AB. Altered expression of the catenin p120 in human cancer: implications for tumor progression. Differentiation. 2002;70:583–589. [PubMed]
  • Tu S, Bhagat G, Cui G, Takaishi S, Kurt-Jones EA, Rickman B, Betz KS, Penz-Oesterreicher M, Bjorkdahl O, Fox JG, Wang TC. Overexpression of interleukin-1beta induces gastric inflammation and cancer and mobilizes myeloid-derived suppressor cells in mice. Cancer Cell. 2008;14:408–419. [PMC free article] [PubMed]
  • van Hengel J, van Roy F. Diverse functions of p120ctn in tumors. Biochim Biophys Acta. 2007;1773:78–88. [PubMed]
  • Wildenberg GA, Dohn MR, Carnahan RH, Davis MA, Lobdell NA, Settleman J, Reynolds AB. p120-catenin and p190RhoGAP regulate cell-cell adhesion by coordinating antagonism between Rac and Rho. Cell. 2006;127:1027–1039. [PubMed]
  • Xiao K, Allison DF, Buckley KM, Kottke MD, Vincent PA, Faundez V, Kowalczyk AP. Cellular levels of p120 catenin function as a set point for cadherin expression levels in microvascular endothelial cells. J Cell Biol. 2003;163:535–545. [PMC free article] [PubMed]
  • Yan HH, Pickup M, Pang Y, Gorska AE, Li Z, Chytil A, Geng Y, Gray JW, Moses HL, Yang L. Gr-1+CD11b+ myeloid cells tip the balance of immune protection to tumor promotion in the premetastatic lung. Cancer Res. 2010;70:6139–6149. [PubMed]
  • Yang L, DeBusk LM, Fukuda K, Fingleton B, Green-Jarvis B, Shyr Y, Matrisian LM, Carbone DP, Lin PC. Expansion of myeloid immune suppressor Gr+CD11b+ cells in tumor-bearing host directly promotes tumor angiogenesis. Cancer Cell. 2004;6:409–421. [PubMed]
  • Yang L, Huang J, Ren X, Gorska AE, Chytil A, Aakre M, Carbone DP, Matrisian LM, Richmond A, Lin PC, Moses HL. Abrogation of TGF beta signaling in mammary carcinomas recruits Gr-1+CD11b+ myeloid cells that promote metastasis. Cancer Cell. 2008;13:23–35. [PMC free article] [PubMed]