The cytoprotective protein clusterin is overexpressed in hypergastrinemic rodent models of oxyntic preneoplasia and promotes gastric cancer cell survival

The cytoprotective protein clusterin is often dysregulated during tumorigenesis, and in the stomach, upregulation of clusterin marks emergence of the oxyntic atrophy (loss of acid-producing parietal cells)-associated spasmolytic polypeptide-expressing metaplasia (SPEM). The hormone gastrin is important for normal function and maturation of the gastric oxyntic mucosa and hypergastrinemia might be involved in gastric carcinogenesis. Gastrin induces expression of clusterin in adenocarcinoma cells. In the present study, we examined the expression patterns and gastrin-mediated regulation of clusterin in gastric tissue from: humans; rats treated with proton pump (H+/K+-ATPase) inhibitors and/or a gastrin receptor (CCK2R) antagonist; H+/K+-ATPase β-subunit knockout (H/K-β KO) mice; and Mongolian gerbils infected with Helicobacter pylori and given a CCK2R antagonist. Biological function of secretory clusterin was studied in human gastric cancer cells. Clusterin was highly expressed in neuroendocrine cells in normal oxyntic mucosa of humans and rodents. In response to hypergastrinemia, expression of clusterin increased significantly and its localization shifted to basal groups of proliferative cells in the mucous neck cell-chief cell lineage in all animal models. That shift was partially inhibited by antagonizing the CCK2R in rats and gerbils. The oxyntic mucosa of H/K-β KO mice contained areas with clusterin-positive mucous cells resembling SPEM. In gastric adenocarcinomas, clusterin mRNA expression was higher in diffuse tumors containing signet ring cells compared with diffuse tumors without signet ring cells, and clusterin seemed to be secreted by tumor cells. In gastric cancer cell lines, gastrin increased secretion of clusterin, and both gastrin and secretory clusterin promoted survival after starvation- and chemotherapy-induced stress. Overall, our results indicate that clusterin is overexpressed in hypergastrinemic rodent models of oxyntic preneoplasia and stimulates gastric cancer cell survival.


Introduction
In the gastric oxyntic mucosa, glands are divided into different zones containing characteristic cell lineages that normally differentiate from immature progenitor cells in isthmus [1][2][3]. During carcinogenesis, the typical differentiation pattern is disrupted and the mucosa undergoes step-wise transformation, which for the intestinal type gastric adenocarcinoma is thought to progress through oxyntic atrophic (loss of acid-secreting parietal cells) gastritis, intestinal metaplasia and dysplasia before emergence of cancer [4,5]. In addition, spasmolytic polypeptide-expressing metaplasia (SPEM), which possibly evolves by transdifferentiation of mature chief cells, may develop prior to intestinal metaplasia and play a central role in the early phases of the cascade [6][7][8].
Gastrin is a key secretagogue for gastric acid, and regulates cell proliferation, apoptosis and migration, making it essential for normal growth and maturation of the oxyntic mucosa [9][10][11]. Hypergastrinemia might promote gastric carcinogenesis, particularly when combined with oxyntic atrophy and chronic inflammation due to Helicobacter-infection [11,12]. The pro-survival cytoprotective protein clusterin (CLU) is gastrin-responsive in rat pancreatic adenocarcinoma cells, and is involved in the anti-apoptotic effect of gastrin [13]. In oxyntic mucosa of rats, proton pump (H+/K+-ATPase) inhibitor (PPI)-induced hypergastrinemia led to increased CLU expression and a prominent shift in the CLU expression pattern [13].
CLU is nearly ubiquitously expressed in different tissues and associated with regulation of cell survival, migration/invasion, differentiation, cellular stress responses, and resistance to cancer therapy [14,15]. The secretory CLU (sCLU) isoform is characterized as a stress-responsive extracellular chaperone with functional similarities to small heat shock proteins [14,16]. Dysregulation of CLU is found during malignant progression in several tissues and might follow a dysplasia-dependent U-shaped curve [14,17]: in normal tissues, CLU can inhibit tumorigenesis by sensing and counteracting cellular stress [18], while dysplastic cells can temporarily downregulate CLU expression to allow malignant transformation [14]. Then, in established malignancies, upregulation of CLU can promote anti-apoptotic signaling, resistance to chemotherapy, and increased metastatic spread [14,16,18].
Still, expression and function of CLU in normal gastric physiology, hypergastrinemic conditions and during gastric carcinogenesis have not been widely examined. In the present study, we characterize the expression and regulation of CLU in gastric oxyntic mucosa of hypergastrinemic rodent models and in humans, and elucidate the function of sCLU in human gastric cancer cells during stress.

Gene expression analysis of clusterin in human gastric adenocarcinomas
The RNA isolation and microarray analysis of the expression profile of CLU mRNA followed standard protocols, analyzing 300 ng total RNA per sample with the HumanHT-12 Expression BeadChips (Illumina, San Diego, CA) (ArrayExpress E-MTAB-1338). Analyses of CLU mRNA expression in human gastric adenocarcinomas were done using our in-house dataset and the Oncomine database (www.oncomine.org), as previously described [3].

In situ hybridization
In situ hybridization (ISH) was performed with the RNAscope 2.0 HD Reagent Kit (Brown) for FFPE tissue (310035, Advanced Cell Diagnostics (ACD) Inc., Hayward, CA) according to manufacturer's instructions. After pretreatment, sections were incubated with custom transcript-and species-specific probes against clusterin mRNA for 2 hours in a humid chamber at 40˚C, followed by a series of amplifications and visualization with HRP and DAB. Sections were counterstained with hematoxylin. Target-specific probes against bacterial RNA (DapB) and ubiquitin C were used as negative and positive controls, respectively. alkylation with iodoacetamide (1 μmol/mg protein) for 30 min in the dark. Proteins were precipitated using a methanol-chloroform method as described [30] and submitted to another round of protein reduction and alkylation by resuspension in 50 μl 50 mM NH 4 HCO 3 , 5 mM TCEP, incubation for 30 min and subsequent incubation with 1 μmol/mg protein of iodoacetamide for 30 min in the dark. Trypsin (Thermo Scientific, Waltham, MA) was added at 1:50 ratio (w/w, enzyme:protein) prior to overnight digestion at 37˚C in a shaker. Subsequently, formic acid was added to all samples (final concentration 0.1%) followed by centrifugation for 10 minutes at max. speed (16 000 g) for removal of insoluble particles prior to mass spectrometry analysis.

Targeted mass spectrometry
All parallel reaction monitoring (PRM)-based targeted mass spectrometry methods were designed, analyzed, and processed using Skyline software version 3.6.0.10162 [31]. In silico selection of proteotypic peptides was performed via Skyline using the Rattus norvegicus reference proteome available at www.uniprot.org to exclude non-unique peptides. Synthetic light peptides (Thermo Scientific) were used as standards for targeted mass spectrometry analysis. Peptide standards were first analyzed on a Thermo Scientific Q Exactive HF mass spectrometer coupled to an Ultimate 3000 RSLC system (Thermo Scientific, Sunnyvale, California, USA) in PRM mode. Precursor ions of higher intensity (charge state 2+ or 3+) were selected for further analysis. Information on retention time and fragmentation pattern of the standard peptides was used for identification and to build a scheduled method with a retention time window of 5 min. The method was then employed for detection and quantification of corresponding peptides in rat samples. The same instrument parameters described below for the analysis of rat samples were adopted for the establishment of the PRM method with standard peptides. Peptides (2 μg) were separated during a biphasic ACN gradient from two nanoflow UPLC pumps (flow rate of 200 nL/min) on a Acclaim PepMap100 C18 column (75 μm i.d. × 2 cm nanoviper, 3 μm particle size, 100 Å pore size) (Thermo Scientific) and further separated on a PepMap RSLC C18 analytical column (50cm x 75 μm i.d. EASY-spray column, packed with 2μm C18 beads) (Thermo Scientific). Solvent A and B were 0.1% TFA (vol/vol) in water and 100% ACN respectively. The gradient composition was 5%B for 5 min followed by 5-8%B over 0.5 min, 8-24%B for the next 109.5 min, 24-35%B over 25 min, and 35-90%B over 15 min. Elution of very hydrophobic peptides and conditioning of the column were performed during 15 minutes isocratic elution with 90%B and 20 min isocratic conditioning with 5%B.
The peptides eluting LC-column were ionized in the electrospray and analyzed by the Q-Exactive HF in Parallel Reaction Monitoring (PRM) mode. The spray and ion-source parameters were as follows. Ion spray voltage = 1800V, no sheath and auxiliary gas flow, and capillary temperature = 250˚C. Instrument control was through Q Exactive HF Tune 2.4 and Xcalibur 3.0 MS spectra were acquired in the scan range 375-1500 m/z with resolution R = 15,000 at m/z 200, automatic gain control (AGC) target of 3e6 in and maximum injection time (IT) of 15ms. Target MS2 spectra (Top15) were acquired with a resolution R = 15,000, AGC target of 1e5, IT of 100 ms and normalized collision energy of 28%. The isolation window was set to 1.6 m/z for the selection of the precursor. Lock-mass internal calibration was used.
Quantification of peptides detected in rat samples was achieved by summing the integrated peak areas of the most intense fragments. Peptide areas for multiple peptides of the same protein were summed to assign relative abundance to that protein. A minimum of two peptides per protein was used for quantification. Endogenous β-actin levels were used for data normalization.
Expression and secretion of CLU in human gastric cancer cells AGS-GR cells were treated with gastrin (Gastrin 17) (G9020, Sigma-Aldrich) 5 and 10 nM or cisplatin (479306, Sigma-Aldrich) 10 or 20 μM for 24 or 48 hours, and conditioned medium and whole cell lysates were harvested.
For immunocytochemistry, serum-starved (7 hours) AGS-GR cells were treated with gastrin (G9020, Sigma-Aldrich) 10 nM for 24 or 48 hours, before they were washed with cold PBS and fixed 20 minutes at RT using freshly made 3.7% paraformaldehyde + 4% sucrose in PBS. Following permeabilization and blocking of unspecific binding, cells were incubated with rabbit anti-CLU (H-330, sc-8354, Santa Cruz Biotechnology Inc.) 1:50 dilution at 4˚C overnight, followed by re-blocking. Then incubation with secondary antibody goat anti-rabbit A488 (A11008, Invitrogen) diluted 1:400 was done for 60 minutes at RT. Lastly, nucleic DNA and actin was sequentially stained with DAPI (D3571, Invitrogen) (0.1 μg/ml in PBS) and then Rhodamine Phalloidin (R415, Invitrogen) diluted 1:100 both for 5 minutes at RT, and replaced with PBS and kept in the dark at 4˚C. Non-immunized rabbit IgG or omitting the primary antibody, were used as negative controls.

Morphometrics, statistics and imaging
Enterochromaffin-like (ECL) cell hyperplasia was defined as presence of histidine decarboxylase (HDC)-positive cells in linear or micronodular patterns [34]. The total number of CLUpositive cells and the number of CLU-positive cells also expressing PGA5, vesicular monoamine transporter 2 (VMAT2) or proliferating cell nuclear antigen (PCNA) was counted in >40 glands per rat (n = 4-8 per group). The number of CLU-positive cells also expressing Ki67 or VMAT2 was counted in similar basal mucosal areas (0.44 mm 2 ) in control and H/K-β KO mice of all ages (n = 4 per group) (Ki67) and all Mongolian gerbils (n = 4-7 per group) (Ki67 and VMAT2). All dual CLU-and chromogranin A (CgA)-positive cells were counted in pinch biopsies (n = 9) of normal human gastric mucosa. To calculate relative migration and apoptosis ratios, results from xCelligence and Caspase assays were normalized to median of the untreated control (or non-immunized goat IgG) in each individual experiment. Statistically significant differences (p value<0.05) were analyzed using analysis of variation (ANOVA) with Bonferroni's or Tukey's multiple comparison test or Student's t-test using Prism 7 (GraphPad Software, San Diego, CA) and Microsoft Excel 2013 (Redmond, WA). Chromogenic images were captured using Nikon E400 microscope, DS-Fil U2 camera and NIS-Elements BR imaging software (Nikon Co., Tokyo, Japan). Immunofluorescent images of tissue were captured using Olympus IX71 inverted microscope, digital monochrome XM10 camera and P^cell software (Olympus Co. Tokyo, Japan) and further processing using ImageJ (Wayne Rasband, National Institutes of Health, USA), and of cells with Leica SP8 inverted microscope (LeicaMicrosystems, Mannheim, Germany) equipped with an HC PL APO 63x/ 1.20 W and further processing using Fiji [35].

Neuroendocrine cells in normal oxyntic mucosa express clusterin
We have previously indicated using serial section staining that neuroendocrine cells (presumably ECL cells) in normal rat oxyntic mucosa and human carcinoids express CLU [13]. Here, we use double immunofluorescence staining, with antibodies against CLU and known markers for different neuroendocrine cell types, to confirm and further elucidate CLU expression in neuroendocrine cells in normal oxyntic mucosa from three different rodent models and humans (Fig 1). In rat oxyntic mucosa, the single cells expressing high levels of CLU were ECL cells (HDC-positive) and A-like cells (ghrelin-positive) (Fig 1A). In oxyntic mucosa of wildtype mice, CLU expression pattern was different from other rodents, with CLU expressed in ECL cells (HDC-positive) (Fig 1C) as well as in mucous neck cells (GSII-positive) (Fig 1D), as previously reported [17]. Also, in oxyntic mucosa of normal Mongolian gerbils, CLU was expressed mainly in ECL cells (VMAT2-positive) (Fig 1E). In human oxyntic mucosa, the CLU mRNA and protein expression patterns were similar to rodents, with highly CLU-positive single cells partly co-expressing the neuroendocrine cell marker CgA (26.5%), and other gland cells showing more diffuse expression (Fig 1F). There was no CLU expression in parietal cells in any of the examined species (Fig 1B and S1A Fig). Taken together, these results show that, in normal oxyntic mucosa in different species, neuroendocrine cells, particularly ECL cells, express high levels of CLU; in addition, there is less prominent expression of CLU in cells of the mucous neck cell-chief cell lineage.
Gastrin/CCK2 receptor signaling contributes to the regulation of clusterin expression in oxyntic mucosa of rats Previously, we have found that gastrin regulates expression of CLU in vitro, and PPI-induced hypergastrinemia increases the level of CLU in oxyntic mucosa of rats [13]. In the present study, ISH revealed that the change in CLU expression pattern was due to increased expression of Clu mRNA de novo (Fig 2A), and was not attributable only to increased secretion of CLU from neuroendocrine cells. In fact, there were significantly fewer neuroendocrine cells that highly expressed CLU in oxyntic mucosa of hypergastrinemic PPI-rats compared with controls ( Fig 2B and 2C). On the contrary, in hypergastrinemic PPI-rats, the main cell type expressing high levels of CLU in oxyntic glands were basal chief cells, co-expressing either MIST1 or PGA5 (chief cell-markers) (Figs 2D and 3G) [36]. However, not all chief cells expressed high  To investigate whether those changes were dependent upon CCK2R signaling, rats were treated with PPI and/or the CCK2R antagonist netazepide. Elevated plasma concentrations of gastrin, and attenuated ECL cell and mucosal hyperplasia, confirmed efficient blockade of the CCK2R (Table 2 and Fig 3A) [34]. Levels of CLU protein in oxyntic mucosa lysates ( Fig 3B) and plasma (Fig 3C) were increased by PPI-induced hypergastrinemia and decreased by blocking the CCK2R, reaching statistical significance in tissue lysates when comparing hypergastrinemic PPI-rats with or without CCK2R antagonist (Fig 3B).
In most hypergastrinemic PPI-rats also receiving CCK2R antagonist, the main CLU expression pattern was still shifted to basal groups of chief cells (mRNA 5/5, protein 7/8) ( Fig  3E). However, there was a trend towards fewer CLU-positive cells (Fig 3F), and less coexpression of CLU and chief cell markers (Fig 3G), and there were significantly fewer actively dividing CLU-positive cells (Fig 3H). In addition, despite also being hypoacidic and hypergastrinemic, the number of CLU-positive cells decreased (Fig 3D and 3F) after treatment with the CCK2R antagonist alone, compared with the controls. Overall, these findings show that expression of CLU in rat oxyntic mucosa is partly regulated by gastrin signaling through the CCK2R.

Hypoacidity and hypergastrinemia causes upregulation of clusterin expression in the mucous neck cell-chief cell lineage
We wanted to test whether the pattern of CLU overexpression that we observed in rat oxyntic mucosa was a general feature of hypergastrinemic animal models. H/K-β KO mice are, like PPI-treated rats, hypo/anacidic and therefore hypergastrinemic [37,38]. In H/K-β KO mice aged 3-14 months, the morphology of the oxyntic mucosa was fundamentally altered [26], and we found coincident, massive upregulation of CLU mRNA and protein (Fig 4A and 4B). Since CLU was co-localized with the mucous neck cell markers GSII ( Fig 1D) and TFF2 (spasmolytic polypeptide) (Fig 4D) (in addition to neuroendocrine markers) in wild-type mice [17], we examined GSII and TFF2 also in H/K-β KO mice. In distinct areas of their hyperplastic mucosa, there were mucous cells expressing high levels of GSII (Fig 4C), TFF2 (Fig 4D) (Fig 4E), and significantly more numerous proliferating CLU-positive cells than in wild-type controls (Fig 4F and S2D Fig). Interestingly, SPEM is characterized by expression of TFF2 or GSII in chief cell marker-positive (hybrid) cells in the base of oxyntic glands, often with increased proliferation [6,39], and those changes, together with overexpression of CLU, can be used as markers for SPEM [17,23].
In hypergastrinemic H. pylori-infected Mongolian gerbils, the CLU expression pattern also shifted compared with control (Fig 5A and 5C). CLU was highly expressed in the basal half of several metaplastic and invasive glands (Fig 5E), particularly co-expressed with TFF2 and Ki67 (Fig 5B and 5F), similar to previous findings in human and rodent SPEM [17,23]. In contrast, CLU expression in oxyntic mucosa of CCK2R-antagonized H. pylori-infected gerbils was restricted to non-proliferative single cells, including ECL cells (Fig 5A, 5B and 5D), and there were no visible signs of SPEM (Fig 5G), as the mucosa in general was unchanged from uninfected controls [27]. Taken together, these results indicate that, in different animal models with hypergastrinemia due to diminished parietal cell proton pump function or H. pyloriinfection, CLU is overexpressed in basal groups of (metaplastic) cells from the mucous neck cell-chief cell lineage, and that overexpression can be inhibited partially by antagonizing the CCK2R.

Human gastric adenocarcinoma cells express and secrete clusterin
Given our findings in normal and premalignant rodent mucosa, and since CLU has been found by immunohistochemistry to be expressed in human SPEM and gastric cancer [17,19,21], we did further examinations in human gastric cancer material. Our in-house gene expression dataset (96 samples) showed no difference in CLU mRNA expression in adenocarcinomas (of both intestinal and diffuse type) compared with normal or adjacent non-tumor mucosa. However, further sub-analyses revealed significantly higher CLU expression in diffuse tumors containing SRCs compared with diffuse tumors without SRCs (fold change 1.935) (Fig 6A). Likewise, Oncomine analyses showed mainly unchanged CLU mRNA expression in all types of gastric adenocarcinomas compared with normal mucosa (5 sets, 478 samples), and significantly increased expression mainly in diffuse tumors (without specification of SRCs) [40][41][42] versus intestinal [42] and mixed [43] tumors when comparing histological subtypes (5 sets, 534 samples). In SRCs tumors, we observed by immunohistochemistry apparently stronger expression of CLU protein in extracellular matrix than in tumor cells (Fig 6B). However, ISH revealed that mainly tumor cells expressed CLU mRNA, indicating that CLU protein was secreted from tumor cells into extracellular matrix. Both CLU protein and mRNA expression were seen in some, but not all, SRCs (Fig 6B).
Next, we performed western blot analysis to show expression of CLU in three human gastric cancer cell lines (Fig 6C). In AGS-GR cells, CLU was localized to seemingly small vesicles in the perinuclear area and in plasma membrane extensions ( Fig 6E). Furthermore, expression and secretion of CLU from AGS-GR cells increased in response to 24 or 48 hours' treatment with the known stress-inducer cisplatin, and increased even more after treatment with the hormone gastrin or the two in combination (S3 Fig and Fig 6D). Combined, our results show that CLU is increased in diffuse SRC adenocarcinomas, and that gastric cancer cells display signalinduced production and secretion of CLU.

Control (n = 9) PPI (n = 6) CCK2R antagonist (n = 8) PPI + CCK2R antagonist (n = 8)
Gastrin start (pmol/L) (RIA) 58   Gastrin showed a reproducible anti-apoptotic effect on 48 hours' serum starvation-induced apoptosis (Fig 7B and 7C) [45]. Neutralization of sCLU with anti-CLU antibodies (C-18) [33] had a significant pro-apoptotic effect, and partly reversed the anti-apoptotic effect of gastrin (Fig 7B). Recombinant sCLU and gastrin and sCLU together yielded a strong anti-apoptotic effect (Fig 7C and S7 Fig). AGS cells have low sensitivity to the cytostatic cisplatin (IC50 value 13.6 μM [47]), normally used in perioperative treatment of gastric cancer [48]. Neutralizing sCLU during cisplatin treatment of AGS-GR cells had a pro-apoptotic effect compared with cisplatin alone, although the effect was not significant by post hoc analysis (p value = 0.10) ( Fig  7D). These findings suggest that secretion of CLU can promote increased survival of gastric cancer cells after gastrin stimulation and prolonged starvation-or chemotherapy-induced stress.

Discussion
In this study, we investigated the expression and regulation of CLU in oxyntic mucosa of hypergastrinemic rodent models and humans, and elucidated the function of sCLU in gastric cancer cells during stress. In normal rat, mouse and gerbil oxyntic mucosa, we identified the prominent single cells that highly expressed CLU as ECL cells and A-like cells. We found a similar pattern in human oxyntic mucosa, which contrasts with a previous report describing expression of CLU only in the isthmus [17]. Nevertheless, the picture in that report seems to show basal CLU-positive single cells as well. Also, CLU is expressed in neuroendocrine cells in colon and pancreas [49,50], and is overexpressed in some gastric and pancreatic neuroendocrine tumors [50,51]. The CLU expression pattern in oxyntic mucosa of wild-type mice was different from other rodents and humans, with prominent expression in both neuroendocrine cells and mucous neck cells. The significance of this difference is not known. However, it is intriguing that CLU consistently is present in cells from the neuroendocrine or mucous neck cell-chief cell lineages, and not in parietal cells, indicating that CLU is differentially expressed by peptide-and/or mucus-secretory cell types.
In rats, after PPI-induced acid inhibition with subsequent hypergastrinemia, a few neck cells and several basal groups of chief cells massively increased their expression of CLU. Some gland bases with highly CLU-positive cells expressed chief cell markers (MIST1 and PGA5) at low or undetectable levels, distinguishing them from surrounding chief cells and suggesting that they might represent a subtype. In addition, some of the CLU-positive chief cells appeared to be actively dividing. Downregulation of MIST1, loss of PGA5, increased expression of CLU, and emergence of proliferating cells at the base of oxyntic glands are associated with chief cell transdifferentiation and development of SPEM, which are normally induced by oxyntic atrophy (parietal cell loss) [17,36,39]. Pharmacological inhibition (PPI) of parietal cell proton pump function and subsequent hypergastrinemia could mimic some features of parietal cell loss, and one might speculate that the proliferating CLU-positive chief cells represent a variant of chief cell transdifferentiation. Importantly, we did not observe definite SPEM with (hybrid) cells co-expressing chief and mucous neck cell markers in the base of oxyntic glands.
The changes in oxyntic CLU expression were partly reversed in CCK2R-antagonized PPIrats and were absent or downregulated in rats given CCK2R antagonist alone. The CCK2R oxyntic mucosa from wild-type control mice and H/K-β KO mice aged 8 and 14 months. (E) Double immunofluorescence staining showing co-expression of GSII (green) and PEPII (red) in oxyntic mucosa from H/K-β KO mice aged 3 months. (F) Number of dual CLU-positive/Ki67-positive cells per area (0.44 mm 2 ) of oxyntic mucosa (n = 4 mice per group). Data presented as means ± SEM. The basal zone (~100 μm from the gland bottom) is highlighted with a dotted line. *ANOVA with Tukey-adjusted p value < 0.05. Nuclei were counterstained with hematoxylin (blue) or DAPI (blue). Scale bars = 100 μm.
https://doi.org/10.1371/journal.pone.0184514.g004 antagonist alone also causes hypoacidity and hypergastrinemia, but less so than PPI, and the parietal cell proton pump is not directly affected. Thus, our findings show that blocking CCK2R signaling affects oxyntic CLU expression in vivo and that, in hypergastrinemic PPIrats, CLU expression is upregulated and shifted to chief cells by both gastrin-dependent and gastrin-independent pathways. In cells expressing the CCK2R, gastrin can induce expression of Clu through the AP-1 transcription factor complex; as the clusterin promoter contains an AP-1 responsive element and gastrin regulates expression of the AP-1 complex members c-fos and junB [13,52]. In CCK2R-negative cells, like chief cells are normally reckoned to be [53], the gastrin-dependent pathway for regulation of CLU expression is most likely mediated indirectly through factors released from ECL cells or parietal cells [54,55], such as epidermal growth factor receptor (EGFR) ligands, which indeed can induce expression of Clu [56]. Independent of gastrin, it seems likely that hypoacidity per se, diminished proton pump function, alteration of CLU-repressive or -inducing signals from PPI-targeted parietal cells, abnormal activity of chief cells (altered pepsinogen cleavage), bacterial overgrowth, and more, could also influence CLU expression in oxyntic mucosa, as several of these factors, in addition to hypergastrinemia, are present in the rodent models where CLU is overexpressed.
H/K-β KO mice are a non-pharmacological model of potent acid inhibition and hypergastrinemia, with abnormal development of oxyntic glands [26,37,38,57]. In this study, we thoroughly confirm that CLU is highly upregulated in oxyntic mucosa of H/K-β KO mice of different ages [58]. CLU-positive cells in H/K-β KO mice proliferated and co-localized with mucous neck cell markers located in the base of elongated mucinous glands, next to (hybrid) cells co-expressing GSII and pepsinogen II, a pattern typical of SPEM [6,7]. Intriguingly, in these mice, removal of a functional subunit of the proton pump, rather than loss of parietal cells per se, seems to trigger emergence of those oxyntic mucosal changes. Thus, our results suggest that H/K-β KO mice develop CLU-positive SPEM, in addition to the severe mucosal disruption previously described [26,37,38], which supports the notion that the H/K-β KO mouse is a novel spontaneous SPEM model [57].
Another SPEM model is Mongolian gerbils infected with H. pylori [8,23]. Within that model, we identified the expression pattern of Clu mRNA and confirmed that these animals developed CLU-positive SPEM. Sørdal et al. [27] reported that the CCK2R antagonist netazepide prevented development of gastritis and subsequent pathological changes in oxyntic mucosa, after H. pylori infection. Similarly, in oxyntic mucosa of CCK2R-antagonized H. pylori-infected gerbils, we found no areas of SPEM and the CLU expression pattern remained identical to non-infected controls, indicating that gastrin signaling through the CCK2R plays an important role in development of H. pylori-induced SPEM in this model.
Overall, in three animal models with hypergastrinemia due to diminished parietal cell proton pump function or H. pylori-infection, CLU was overexpressed in basal groups of proliferative (and metaplastic) cells in the mucous neck cell-chief cell lineage, and that overexpression could be partially inhibited by antagonizing the CCK2R. Medically-or H. pylori-induced oxyntic atrophy induces SPEM [6,39]. However, a recent report indicates that targeted apoptosis of  parietal cells is insufficient to induce metaplasia [59], thus, the exact mechanisms are unknown. A common factor is affection of parietal cell proton pump function and absence of gastric acid secretion, which leads to hypergastrinemia. One study shows that gastrin protects against development of SPEM [60], while others report that gastrin promotes metaplastic transformation [61,62]. Indeed, H+/K+-ATPase β-subunit/gastrin double-KO mice did not develop SPEM-like "mucus-rich" cell hyperplasia [57]. That, together with our findings, indicates that, although gastrin is not essential for the development of SPEM, it might influence the metaplastic cascade, possibly through upregulating CLU expression in hypoacidic oxyntic glands. Upregulation of CLU could make metaplastic cells more resistant to harmful stimuli, such as oxidative stress due to chronic inflammation [18], and oxidative stress may play a central role in gastric tumorigenesis [63].
CLU is dysregulated in multiple cancers, including gastric ones, in which overexpression seems to correlate with cancer progression [17,19,21]. The suggested U-shaped correlation pattern probably explains why overexpression could be hidden in large-scale analysis of heterogeneous samples. Interestingly, we identified a difference in CLU expression between histological subtypes of gastric adenocarcinomas, with higher CLU expression in diffuse tumors containing SRCs compared with diffuse tumors without SRCs. The diffuse SRC tumors represent a subtype with possible distinct clinicopathologic characteristics and prognosis than other diffuse cancers [64][65][66]. Our findings suggest that CLU could be one of several mediators contributing to these characteristic features. Whether there is a link between the expression patterns of CLU in gastric normal mucosa and metaplasia, and in diffuse gastric cancer with SRCs, is not known. The diffuse SRC tumors are not typically thought to develop on the background of metaplastic changes. Still, SPEM (and clusterin expression) and intestinal metaplasia have been found in adjacent gastric mucosa of diffuse tumors [17,19,67,68] and another SPEM-marker (WAP four-disulfide core domain protein 2 (WFDC2)) is also strongly expressed in diffuse SRC tumors [69]. CLU was mainly localized to tumor cells but the protein apparently was readily secreted. Furthermore, gastric cancer cell lines expressed and secreted CLU upon stimulation with stressors like gastrin and cisplatin. Both gastrin and CLU might promote cell migration/invasion [16,33,44,46], and, even though sCLU did not affect migration of AGS-GR cells, our findings do not exclude a pro-migratory role of CLU in other gastric cancer cell lines or in vivo.
Both gastrin and sCLU alone promoted an anti-apoptotic effect on gastric cancer cells, and it is noteworthy that gastrin and sCLU in combination enhanced survival even further, particularly on early apoptosis signaling. We confirmed that sCLU is directly involved in the antiapoptotic effect of gastrin [13], and a possible indirect involvement could be mediated by CLU's activation of autophagy [45,70,71]. sCLU functions as an extracellular molecular chaperone, enabling it to bind a wide array of peptides, including misfolded or denatured proteins, cellular debris, lipids and other potentially harmful molecules; thereby preventing that they form insoluble aggregates, bind to receptors, or impede damage on neighboring cells [14,15,18]. This scavenging function of sCLU is cytoprotective, and is highly relevant in stressed cells, where damaging agents accumulate and need to be cleared away to avoid cell death [15,18].
CLU might also reach the cytoplasm, either by alternative transcription or splicing, failed translocation, or retrotranslocation from the endoplasmic reticulum (ER)-Golgi pathway [15]. Intracellularly, CLU has been found to bind and inactivate apoptosis regulator BAX (BAX), thereby altering the ratio between pro-apoptotic BAX and anti-apoptotic B-Cell CLL/Lymphoma 2 proteins towards a pro-survival level [72,73]. Interestingly, we observed an increase in putative intracellular forms of CLU (60 and 75 kDa precursors of sCLU) in gastric cancer cells in response to gastrin and/or cisplatin, but without further comprehensive analyses, we do not know whether these peptides are exclusively located in the ER-Golgi-pathway, waiting to be secreted, or also locate to the cytoplasm, possibly interacting with mitochondrial and apoptosis-related proteins. Indeed, immunocytochemistry of gastric cancer cells showed that CLU was mainly located in the perinuclear area and towards the cell membrane, suggestive of secretory pathway localization. Additionally, sCLU decreased the cytotoxicity of cisplatin on gastric cancer cells, similarly as gastrin [45], and both might be involved in treatment resistance. Taken together, these results suggest that gastrin and sCLU can make gastric cells more resistant to stress-induced cell death.
In summary, we have shown that CLU is highly expressed in oxyntic mucosa of hypergastrinemic rodent models, particularly in glands containing putative metaplastic cells from the mucous neck cell-chief cell lineage, and that CLU expression is partly regulated by gastrin in vivo. Furthermore, cisplatin and gastrin made gastric cancer cells express and secrete CLU, leading to increased survival and possibly treatment resistance. Overall, our results indicate that CLU could be involved in gastrin-induced pro-survival signaling and remodeling of the oxyntic mucosa; and might therefore influence both gastric homeostasis and cancer risk.  to the examination of clusterin expression in oxyntic mucosa of H+/K+-ATPase β-subunit knockout mice. We thank Dr. Steve Warrington for language editing the manuscript. We thank the animal unit Comparative Medicine Core facility (CMC), NTNU for assistance with animal models. The imaging analyses were performed in collaboration with Dr. Bjørnar Sporsheim at Cellular & Molecular Imaging Core Facility (CMIC), NTNU. Proteomic analyses were performed in collaboration with Proteomics and Metabolomics Core Facility (PRO-MEC), NTNU and The Proteomics Unit at the University of Bergen (PROBE). The microarray and bioinformatics analysis were provided in collaboration with Arnar Flatberg and the Genomics Core Facility (GCF), NTNU. All core facilities (CMC, CMIC, PROMEC and GCF) at NTNU are funded by the Faculty of Medicine and Health Sciences at NTNU and the Central Norway Regional Health Authority.