Loss of LLGL1 Expression Correlates with Diffuse Gastric Cancer and Distant Peritoneal Metastases

Background Loss of LLGL1 has been associated with loss of cellular adhesion and dissemination of cells from colorectal cancer and malignant melanoma. Regulation and relevance of LLGL1 were analyzed in gastric cancer patients with lymphatic and distant dissemination. Furthermore, LLGL1 expression was analyzed in relation to the cellular adhesion protein E-cadherin. Methods LLGL1 and E-cadherin transcription levels were evaluated in 56 gastric cancer patients and five gastric cancer cell lines. IHC staining for LLGL1 was performed on 39 gastric cancer specimens. LLGL1 was stably transfected into LLGL1 negative gastric cancer cell line SNU16 (del(17) (p11.2)) for functional in vitro assays and a xenograft bioassay. Results Gastric cancer specimens and cell lines displayed LLGL1 and E-cadherin expression levels with variable intensity. In gastric mucosa, LLGL1 exhibited weak cytoplasmic and strong cortical staining. Loss of LLGL1 expression occurred in 65% of gastric cancers and significantly correlated with loss of E-cadherin expression (P=0.00009). Loss of LLGL1 expression was associated with the diffuse type of gastric cancer (P=0.029) with peritoneal carcinomatosis (M1; P=0.006) and with female gender (P=0.017). Stable reexpression of LLGL1 in SNU16 cells significantly increased both plastic surface adhesion and extracellular matrix proteins laminin and fibronectin, but had no impact on in vitro proliferation, apoptosis, or invasion or on in vivo proliferation or differentiation in our xenograft bioassay. Conclusion LLGL1 is coexpressed with E-cadherin. Loss of expression of either protein is associated with diffuse gastric cancer and peritoneal metastases. LLGL1 does not impact on proliferation or epithelial-mesenchymal transition (EMT) rather increasing cellular adhesion.


Introduction
Gastric cancer incidence has decreased steadily in industrialized countries over the last years. However, gastric cancer still ranks among the most common causes of cancer and its mortality rate remains high [1][2][3]. The current gold standard therapy with curative intention is radical surgical resection with standardized D2-lymphadenectomy. Despite considerable improvements achieving R0 resections patients still require (neo)adjuvant chemotherapeutic strategies as they are still at high risk for local recurrences and early lymph node or systemic metastases [4,5].
Tumor dissemination results from loss of cellular adhesion, chemotaxis, and neoangiogenesis. Junctions between epithelial cells have communicating functions such as gap junctions, are anchoring junctions such as desmosomes and adherens junctions, or are sealing junctions such as zonula occludens or tight junctions. Adherens junctions segregate the apical from the basolateral membrane domains. The predominant protein of adherens junctions is E-cadherin, a transmembrane protein stabilizing the basolateral cell-cell contact. Loss of E-cadherin expression has been linked to dissemination of various gastrointestinal malignancies [11,12]. As early as in 1994, loss of E-cadherin expression was correlated with diffuse type gastric cancer [13]. Since then, multiple reports have described the association between diffuse gastric cancer and metastatic disease and also linked the loss of E-cadherin expression with familial gastric cancer [14,15]. Loss of E-cadherin decreases cellular adhesion, resulting in a critical increase in cellular motility and migration [16].
Another relevant protein for cellular adhesion along the basolateral membrane domain is lethal giant larvae (l(2)gl) [17]. In Drosophila loss of l (2)gl results in loss of epithelial structure, malignant transformation of the brain hemispheres, and the imaginal discs and in growth of tumor masses resembling human cancers [18]. These tumors proliferate and migrate to distant sites upon transplantation into wild type Drosophila, thus acting like human metastatic cancers [19,20]. Homologues of l (2)gl have been identified in diverse species such as rat, insect, worm, and man [21][22][23]. Remarkably, the particular function of l (2)gl is conserved among species, as shown by rescue of the l(2)gl mutation in Drosophila with the human homologue LLGL1 [22,24].
Evidence has also been published that mammalian l(2)gl regulates epithelial cell polarity and migration as a member of the polarity complex consisting of Par6/Par3/atypical PKC and l(2)gl [25][26][27].
In humans, highly related homologues of l(2)gl, LLGL1, and LLGL2 have been identified, mapping to the short and long arm of chromosome 17. LLGL1 has been located in a critical pericentromeric region, 17p11.2-12 containing cancer susceptibility genes for primitive neuroectodermal tumors [21]. Furthermore, LLGL1 maps within the 17p interstitial deletion detected in mentally retarded children with Smith-Magenis syndrome [28,29].
Regarding LLGL2, reduced expression has been described in specimens of high grade pancreatic intraepithelial neoplasia, high grade gastric dysplasia, and carcinoma [37][38][39][40]. Interestingly, reduced basolateral LLGL2 expression was associated with diffuse type gastric cancer and reduced E-cadherin expression [38,41]. Taken together with the data presented in this paper, evidence is accumulating that both human homologues of Drosophila l(2)gl are involved in common human pathways, the inactivation of which promotes cancer dissemination.
The present study was performed to evaluate the role of LLGL1 in human gastric carcinogenesis and to analyze the association and shared regulation with E-cadherin expression. We screened the transcription profile of LLGL1 and Ecadherin in 5 human gastric cancer cell lines and 56 gastric carcinomas and performed additional IHC staining of 5 gastric mucosal samples and 39 gastric cancers. Functional in vitro assays with a stably LLGL1 transfected cell line were performed to characterize the biological features of LLGL1. We then used the cell lines to induce subcutaneous xenograft tumors and assessed size and grading with respect to LLGL1 expression.

Cell Culture.
For functional analyses, we studied the human gastric cancer cell lines AGS, NCI-N87, OE33, MKN45, and SNU16. All cell lines were cultured in DMEM supplemented with 10% FCS.

Tissue Source and Storage.
Following ethics committee approval and signed informed consent, samples from the center of the tumor were obtained from 56 patients undergoing elective surgery for gastric cancer at the Department of Abdominal-and General Surgery, Johannes Gutenberg University, Mainz, Germany. All tissues were stored in cryovials, shock frozen in liquid nitrogen immediately after extirpation and stored at -80 ∘ C until further processing.

Immunohistochemistry.
For IHC staining of paraffinembedded tissue sections, the avidin-biotin-complex method (LSAB+ System-HRP Kit, Dako Cytomation, Germany) was used to detect the proteins LLGL1 (1:50; 4 hours, mouse-antihuman monoclonal antibody, Clon 5G2, Abnova, Taiwan; Polyclonal rabbit-anti-human antibody, respectively) and Ecadherin (1:100, 1h, Dako Cytomation, M3162). Formalinfixed and paraffin-embedded tissues were deparaffinized and subsequently microwaved (600 W, 15 minutes) in citrate buffer (ph 6.0). After preincubation with hydrogen peroxide (LSAB+ System-HRP Kit, Dako Cytomation, Germany) and human AB plasma (Dept. of Transfusion, University of Mainz, Mainz, Germany) the primary antibodies were applied at room temperature. After incubation with the secondary antibody (LSAB+ System-HRP Kit, Dako Cytomation, Germany) the avidin-biotin complex was added and the enzyme activity was visualized with diaminobenzidine (LSAB+ System-HRP Kit, Dako Cytomation, Germany). Counterstaining was performed with haematoxylin (Roth, Karlsruhe, Germany). For negative controls of each sample, the secondary antibody was used alone. For positive controls, formalin-fixed and paraffin-embedded tissue samples of the human gastric mucosa were applied. Evaluation of the staining was performed semiquantitatively by three independent authors via light-microscopy. The intensity of staining was graded as negative: 0, weak: 1, medium: 2, and strong: 3.

Establishment of LLGL1-GFP Expressing Clones.
We established a SNU16 cell line clone stably expressing a GFP-LLGL1 fusion protein. The SNU16 gastric carcinomatosis cell line was selected for transfection, as it has been described as carrying a deletion on chromosome 17, p11.2, the locus of LLGL1. Therefore, SNU16 has lost LLGL1 expression and so was suited to investigate the effect of LLGL1 reexpression. The LLGL1 cDNA containing the complete open reading frame was cloned into the expression vector pcDNA3.1/NT-GFP (Invitrogen, Carlsbad, CA, USA), resulting in a GFP-LLGL1 fusion protein. SNU16 were seeded in six-well plates and transfected with either pcDNA3.1/NT-GFP-LLGL1 or pcDNA3.1/NT-GFP plasmid by lipofectamine 2000 reagent according to the recommendations of the manufacturer (Invitrogen, Carlsbad, CA, USA). The stably transfected SNU16-GFP and SNU16-GFP-LLGL1 cells were selected in medium containing G418 (400 g/ml). Stable clones grew after about 4 weeks of selection and were picked and analyzed by Western blot and RT-PCR.

Proliferation
Assays. 5x10 3 cells (SNU16-GFP-LLGL1 or SNU16-GFP) were seeded into 96-well plates. The number of cells per well was determined daily by luminescence (Celltiter-Glo, Cell Viability assay, Promega, USA). In brief, 50 l of Cell Titer Glo were added to 100 l serumfree medium per well, followed by incubation at room temperature for 15 minutes. Luminescence was then read with a luminometer after 10 minutes. Each procedure was performed in quadruplicate.

Apoptosis
Assay. 5x10 5 cells (SNU16-GFP-LLGL1 or SNU16-GFP) were plated in 6-well plates. Suspension cells were collected and adherent cells trypsinized prior to fixation with 70% ethanol, staining with propidium iodide and analysis by FACS, without gating. Cells in the G1 (n) and G2/M (2n) phases of the cell cycle could be distinguished. Apoptotic cells with DNA content lower than n were quantified. Each procedure was performed in quadruplicate.

Adhesion Assay.
For adhesion assays, SNU16-GFP-LLGL1 and SNU16-GFP cells were used. 96-well plates had been prepared with laminin (10 g/ml, 30 minutes, room temperature, Sigma, Germany), fibronectin (40 g/ml, 30 minutes, room temperature, Sigma, Germany), or PBS and were blocked with albumin (2%, over night, 4 ∘ C, Serva, Germany), respectively. After trypsinization, 80,000 cells were seeded per 96-well and allowed to attach for 24 hours. Thereafter the medium and none-attached cells were removed. Each well was washed twice with 100 l medium. The amount of attached cells per well was determined by luminescence assay (Celltiter-Glo, Cell Viability assay, Promega, USA). Luminescence was quantified with a luminometer. Again, each procedure was performed in quadruplicate.

Invasion Assays.
Invasion of SNU16-GFP-LLGL1 versus SNU16-GFP cells was assayed with 24-well HTS FluoroBlok Inserts in triplet approaches (8 M pore size; Becton Dickinson, USA). Membranes were covered with laminin (10 g/ml, 30 minutes, room temperature, Sigma, Germany) and blocked with albumin (2%, overnight, 4 ∘ C, Serva, Germany). In brief, 2x10 4 cells were resuspended in serum-free DMEM and added to the upper chamber, following which DMEM with 20% FCS and 70 ng/ml SDF-1alpha was added to the lower chamber. Chambers were incubated for 24h at 37 ∘ C in a humid atmosphere of 5% CO 2 . After incubation, the number of invaded and migrated cells in the lower chamber was determined by luminescence assay (Celltiter-Glo, Cell Viability assay, Promega, USA) according to the recommendations of the manufacturer. Luminescence was quantified with a luminometer, and each procedure was performed in triplicate.

Subcutaneous Tumor Xenograft.
Either SNU16-GFP-LLGL1 or SNU16-GFP expressing cells (5x10 6 ) were used to induce a subcutaneous tumor in 7-8 weeks old Nod-SCID mice. The mice were maintained in a laminar airflow cabinet under pathogen-free conditions. Mice were housed in microisolator cages with free access to laboratory chow and tap water. Nod-SCID mice were irradiated with 1.8 Gy one day prior to subcutaneous injection of tumor cells. Tumors grew for 6 weeks before the animals were sacrificed by carbon dioxide asphyxiation. Thereafter tumors were enucleated, embedded in paraffin, sectioned and immunostained. All animal experiments were performed in accordance with the German Animal protection Law and approved by the local responsible authorities.

Loss of LLGL1 Transcription in Human Gastric Cancer
Cell Lines. LLGL1 was expressed in gastric AGS, NCI-N87, OE33 and MKN cancer cell lines (Figure 1(a)). In contrast, LLGL1 was absent in SNU16 derived from human gastric peritoneal carcinomatosis, resulting from a deletion of p11.2 on chromosome 17.   (Table 2). TNM classification revealed a significant correlation between loss of LLGL1 expression and distant peritoneal metastases (M1; P=0.006). In contrast, loss of LLGL1 impacted neither on T-nor on N-status. In addition, loss of LLGL1 showed a significant association with female gender (P=0.017) but had no relevance for the resection status (R-Status). Patients whose tumors revealed a loss of LLGL1 showed a trend toward a shorter survival (575 days) compared to those with LLGL1 expressing tumors (856 days; n.s.). These results revealed a significant association between loss of LLGL1 in gastric cancer samples and distant dissemination.

Immunohistochemical Analysis of LLGL1 Expression in
Gastric Cancer Samples. To further examine LLGL1 expression in vivo, five healthy gastric mucosa samples and 39 gastric adenocarcinoma specimens (62% diffuse and 58% intestinal type according to Lauren classification) were immunostained with an anti-LLGL1 antibody. In human gastric mucosa, LLGL1 immunohistochemistry exhibited weak cytoplasmic and strong cortical staining along the basolateral membranes (Figure 1(b)). Interestingly, LLGL1 expression of gastric epithelial cells was most intense at the apical foveolar segments and absent in the basal segments of the gland. Gastric carcinoma samples revealed varying expression intensities of LLGL1 ranging from strong to absent (Figure 1(c)). Loss of LLGL1 expression was significantly correlated with the diffuse type of gastric cancer (  summary, these data reveal that loss of LLGL1 protein staining is associated with the diffuse type gastric cancer.

Loss of E-Cadherin versus Tumor and Patient Characteristics.
Loss of E-cadherin expression occurred in 68% of gastric carcinoma samples (Table 3). TNM classification showed a trend between loss of E-cadherin expression and distant metastases (P=0.07). In contrast, loss of E-cadherin impacted on neither T-nor N-status. However, loss of Ecadherin revealed a significant association with female gender (P=0.0017) but had no relevance for the resection status (R-Status). Patients whose tumors revealed loss of E-cadherin showed a trend to reduced survival (614 days) compared to those with E-cadherin expression (798 days; n.s.). These   (Figure 2(a)). Two different LLGL1-GFP expressing clones were selected, SNU16-GFP-LLGL1 and SNU16-GFP-LLGL1(2). Expression of LLGL1 did not modify the transcription or the protein expression level of E-cadherin, implicating that both proteins are independent downstream targets of a common regulator. SNU16 cells expressing GFP-LLGL1 revealed an intense submembranous accumulation of GFP-LLGL1 indicating a cortical localization of LLGL1, which was enhanced in regions of cell-cell contact (Figure 2(b)). In contrast SNU16-GFP cells depicted a cytoplasmic localization of GFP (Figure 2(b)).
In summary, these functional assays demonstrate that LLGL1 expression has no impact on cell proliferation, apoptosis, or invasion but does significantly increase cell adhesion. These observations are in accordance with the hypothesis that loss of LLGL1 expression contributes to cancer dissemination and progression by loss of cell-to-cell junction mediating adherence

Subcutaneous Tumor Growth of SNU16 LLGL1-GFP Cells
Stably Expressing SNU16 Cell Line in a Xenograft Model. SNU16-GFP-LLGL1 and SNU16-GFP expressing cells were used to induce subcutaneous tumors in Nod-SCID mice (Figure 3(b)). Immunohistochemistry revealed a predominantly membranous staining of LLGL1 in GFP-LLGL1 expressing tumors, in contrast to GFP only expressing tumors. Expression of LLGL1 did not alter the expression intensity of Ecadherin, but increased membranous redistribution of Ecadherin. However, LLGL1 impacted on neither tumor size (LLGL1-GFP versus GFP; 11mm versus 10mm) nor differentiation of the tumor, indicated by tumor grading (G3, respectively). These data confirm that LLGL1 does not impact on proliferation or on epithelial-mesenchymal transition (EMT), but increases adhesion as depicted in our functional analyses.

Discussion
We initiated this study to investigate the relevance of LLGL1 expression for gastric cancer development and progression. Specifically, we were interested to know whether LLGL1 expression is lost in gastric cancer and if so whether loss of LLGL1 expression occurs in a larger context of cellular deadhesion. Therefore, we analyzed the expression and regulation of E-cadherin in parallel.
We have previously described the loss of LLGL1 expression in a large cohort of colorectal cancer patients and its impact on tumor cell dissemination in vivo and in vitro [30]. Matching our current observations in gastric cancer, LLGL1 expression did not impact on proliferation, cell cycle, or apoptosis in colorectal cancer. Further studies revealed that loss of LLGL1 expression is lost in various cancers [24,30,31]. In addition, Tsuruga and colleagues described loss of LLGL1 expression in endometrial cancer and reported a correlation with metastatic disease [32]. Furthermore, loss of LLGL1 expression is correlated with reduced overall survival in pancreatic and squamous lung cancers [34,36].
Our current data are supported by these reports, and prove an interesting link between LLGL1 and gastric cancer, underlining the relevance of cellular deadhesion in the context of tumor cell dissemination for the following reasons: (1) We found that LLGL1 transcription was lost in 65% of all gastric cancers and that its loss correlated significantly with distant dissemination, particularly with peritoneal carcinomatosis in patients.
(2) Loss of LLGL1 expression significantly correlated with the diffuse type of gastric cancer as compared to the better differentiated intestinal type according to the Lauren classification. These results match the findings of the second human Drosophila homologue, LLGL2, as was recently reported [41].
(3) We found a highly significant correlation between loss of LLGL1 and loss of E-cadherin expression, respectively. Loss of E-cadherin expression had previously been correlated with the diffuse type gastric cancer in a landmark paper by Becker and colleagues back in 1994 [13]. Since then, multiple groups described this clinical association and linked the loss of E-cadherin expression with familial diffuse gastric cancer [15]. Downregulation or loss of E-cadherin decreases the strength of cellular adhesion within a tissue and induces activation of the ß-catenin pathway, resulting in increased cellular motility and invasion [16]. A similar association was found for LLGL2 in other studies [41].  ATCC, USA). Reexpression of LLGL1 enabled these cells to grow in clusters with an epithelial phenotype, reflecting increased cellular adhesion. These findings are in accordance with descriptions in mammary epithelial cells. Knockdown of LLGL1 expression was correlated with mesenchymal phenotype and reduced acinar formation [42]. Therefore, a role of LLGL1 in reinforcement of epithelial junctions or desmosomes should be postulated, demanding further analyses [17,43].
Our results point toward recent mechanistic findings from Drosophila's LLGL1 homologue l(2)gl. It has been shown that basolateral l(2)gl is part of the cortical membrane cytoskeleton stabilizing epithelial structures. Here, l(2)gl forms a complex with Dlg and cribble crucial to the formation of epithelial junctions such as tight junctions in mammalian epithelial cells [17]. In contrast, apical l(2)gl plays a critical role in induction of migration [25,27] Among the strongest inductors of chemotaxis-mediated migration are chemokine receptors and their ligands, such as CXCR4 and CXCL12 [44,45]. Activation of diverse chemokine receptors results in activation of the PI3K pathway which again results in activation of aPKC and phosphorylation of apical l(2)gl [44,45]. Phosphorylated l(2)gl dissociates from the apical cytoskeleton in order to become a member of the polarity complex (L(2)gl, Par6, and aPKC) [25][26][27]. For cell migration, the polarity complex concentrates integrin clusters in the anterior aspect of the cell, resulting in polarized adhesion and transmigration.
In summary, the development of gastric cancer is associated with progressive loss of epithelial structure, cell polarity, and decreased cell-to-cell contact. The available information on LLGL proteins from studies in Drosophila and humans supports the theory that LLGL1 contributes to maintenance of epithelial integrity. The coregulation with E-cadherin implicates a relevant role for LLGL1 in epithelial junctions or desmosomes. Taken together with the results presented in this paper, a role for LLGL1 in diverse human malignancies is predicted, thus warranting further investigations.

Data Availability
All experimental data used to support the findings of this study are included within the article

Ethical Approval
All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1964 and later versions. All institutional and national guidelines for the care and use of laboratory animals were followed.

Consent
Informed consent or substitute for it was obtained from all patients for being included in the study.