Gastric Corpus Mucosal Hyperplasia and Neuroendocrine Cell Hyperplasia, but not Spasmolytic Polypeptide-Expressing Metaplasia, Is Prevented by a Gastrin Receptor Antagonist in H+/K+ATPase Beta Subunit Knockout Mice

Proton pump inhibitor use is associated with an increased risk of gastric cancer, which may be mediated by hypergastrinemia. Spasmolytic polypeptide-expression metaplasia (SPEM) has been proposed as a precursor of gastric cancer. We have examined the effects of the gastrin receptor antagonist netazepide (NTZ) or vehicle on the gastric corpus mucosa of H+/K+ATPase beta subunit knockout (KO) and wild-type (WT) mice. The gastric corpus was evaluated by histopathology, immunohistochemistry (IHC), in situ hybridization (ISH) and whole-genome gene expression analysis, focusing on markers of SPEM and neuroendocrine (NE) cells. KO mice had pronounced hypertrophy, intra- and submucosal cysts and extensive expression of SPEM and NE cell markers in the gastric corpus, but not in the antrum. Numerous SPEM-related genes were upregulated in KO mice compared to WT mice. NTZ reduced hypertrophia, cysts, inflammation and NE hyperplasia. However, NTZ neither affected expression of SPEM markers nor of SPEM-related genes. In conclusion, NTZ prevented mucosal hypertrophy, cyst formation and NE cell hyperplasia but did not affect SPEM. The presence of SPEM seems unrelated to the changes caused by hypergastrinemia in this animal model.


Introduction
Proton pump inhibitors (PPIs) are widely used in the management of acid-related gastrointestinal diseases such as peptic ulcers and gastro-esophageal reflux. The increase in PPI use has been well Stomach weight was higher in the KO/PEG group than in all other groups (p = 0.001), as expected from the known effects of hypergastrinemia. KO/NTZ mice had lower stomach weight than KO/PEG mice, reflecting the effects of gastrin receptor antagonism, but still higher than both WT groups (p = 0.001) ( Figure 1C). Oxyntic mucosal thickness was lower in KO/NTZ mice compared to KO/PEG, whereas there was no difference between the WT/NTZ and WT/PEG groups ( Figure 1D). Antral mucosal thickness did not differ between the groups ( Figure 1F).

NTZ Reduced Intramucosal Cysts and Invasions below the Muscularis Mucosae in KO Mice
In addition to pronounced hyperplasia of the oxyntic mucosa, KO mice had intramucosal cysts and half of the animals had invasions of the muscularis mucosae with benign appearance, often in the proximity of vascular structures also penetrating the muscularis mucosae. The histopathological

NTZ Reduced Intramucosal Cysts and Invasions below the Muscularis Mucosae in KO Mice
In addition to pronounced hyperplasia of the oxyntic mucosa, KO mice had intramucosal cysts and half of the animals had invasions of the muscularis mucosae with benign appearance, often in the proximity of vascular structures also penetrating the muscularis mucosae. The histopathological Int. J. Mol. Sci. 2020, 21, 927 4 of 18 changes are presented in Table 1. NTZ reduced the number of intramucosal cysts as well as submucosal invasions in KO mice. NTZ also reduced inflammation in KO mice. Representative HE histologic appearance of oxyntic mucosa in the two KO groups compared to WT/PEG are presented in Figure 2.
Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 4 of 18 changes are presented in Table 1. NTZ reduced the number of intramucosal cysts as well as submucosal invasions in KO mice. NTZ also reduced inflammation in KO mice. Representative HE histologic appearance of oxyntic mucosa in the two KO groups compared to WT/PEG are presented in Figure 2.    were overexpressed in KO/PEG versus WT/PEG mice, including tff2, clu, muc6, cd44 and wfdc2, with no significant change in KO mice receiving NTZ (Table 2). It was confirmed by both IHC and ISH that KO/PEG mice had pronounced expression of the SPEM markers TFF2 and clusterin in the gastric corpus [23] (Figures 3 and 4), and that the SPEM did not seem to be affected by NTZ. There was also overexpression of markers of intestinalizing transcripts such as cftr and muc4. Only two (muc13 and muc5b) of the SPEM associated genes overexpressed in KO/PEG mice compared to WT/PEG mice were reduced by NTZ. Expression of the mouse chief cell markers pgc (pepcinogen C), gif (intrinsic factor), anpep (Alanyl Aminopeptidase) and Bhlha15 (MIST1) was substantially lower in KO/PEG than WT/PEG mice but was also not affected by NTZ with the exception of anpep. Gene expression analysis was performed to examine the relative expression of NE and SPEM related genes. This showed that numerous genes previously reported to be associated with SPEM were overexpressed in KO/PEG versus WT/PEG mice, including tff2, clu, muc6, cd44 and wfdc2, with no significant change in KO mice receiving NTZ (Table 2). It was confirmed by both IHC and ISH that KO/PEG mice had pronounced expression of the SPEM markers TFF2 and clusterin in the gastric corpus [23] (Figures 3 and 4), and that the SPEM did not seem to be affected by NTZ. There was also overexpression of markers of intestinalizing transcripts such as cftr and muc4. Only two (muc13 and muc5b) of the SPEM associated genes overexpressed in KO/PEG mice compared to WT/PEG mice were reduced by NTZ. Expression of the mouse chief cell markers pgc (pepcinogen C), gif (intrinsic factor), anpep (Alanyl Aminopeptidase) and Bhlha15 (MIST1) was substantially lower in KO/PEG than WT/PEG mice but was also not affected by NTZ with the exception of anpep.    Table 3). Expression of the general NE markers eno2 (NSE) and uchl1 (PGP9.5) and the enterochromaffin (EC) cell marker tph1 (tryptophan hydroxylase) were, however, not affected by NTZ in KO mice. Conversely, NTZ significantly reduced the volume density of CgA-positive cells in KO mice ( Figures 1F and 5), demonstrating the effect of NTZ on NE cells. Gene expression analysis demonstrated that NTZ reduced expression of the general NE marker chga (CgA) as well as the ECL cell markers hdc (HDC), slcl18a2 (VMAT-2) and cckbr (CCKBR) in KO mice, whereas expression of the D cell marker sst (somatostatin) was increased ( Table 3). Expression of the general NE markers eno2 (NSE) and uchl1 (PGP9.5) and the enterochromaffin (EC) cell marker tph1 (tryptophan hydroxylase) were, however, not affected by NTZ in KO mice. Conversely, NTZ significantly reduced the volume density of CgA-positive cells in KO mice ( Figures 1F and 5), demonstrating the effect of NTZ on NE cells. Gene expression analysis demonstrated that NTZ reduced expression of the general NE marker chga (CgA) as well as the ECL cell markers hdc (HDC), slcl18a2 (VMAT-2) and cckbr (CCKBR) in KO mice, whereas expression of the D cell marker sst (somatostatin) was increased ( Table 3). Expression of the general NE markers eno2 (NSE) and uchl1 (PGP9.5) and the enterochromaffin (EC) cell marker tph1 (tryptophan hydroxylase) were, however, not affected by NTZ in KO mice.  Table 2. Expression of spasmolytic polypeptide-expressing metaplasia (SPEM)-related genes in the gastric corpus mucosa of KO/PEG versus WT/PEG mice and in KO/NTZ versus KO/PEG mice. Numerous of these genes were overexpressed in KO/PEG mice, but most of these genes were not affected by NTZ. Green: significant change (adjusted p-value (q-value) < 0.05); red: non-significant change (q-value > 0.05).

Global Gene Expression Profiles Were Influenced More by Genotype Than Administration of NTZ
The global gene expression illustrated by a principal component analysis ( Figure 6) displayed separation of both KO groups from the WT groups, with only minor separation between the NTZ and PEG groups. Thus, the global gene expression was influenced by genotype to a larger extent than by NTZ. The 20 most differentially expressed genes are shown in Table 4. Only three genes were differentially expressed in WT/PEG mice versus WT/NTZ mice (Supplementary Table S1).

Global Gene Expression Profiles Were Influenced More by Genotype Than Administration of NTZ
The global gene expression illustrated by a principal component analysis ( Figure 6) displayed separation of both KO groups from the WT groups, with only minor separation between the NTZ and PEG groups. Thus, the global gene expression was influenced by genotype to a larger extent than by NTZ. The 20 most differentially expressed genes are shown in Table 4. Only three genes were differentially expressed in WT/PEG mice versus WT/NTZ mice (Supplementary Table S1).

Discussion
Various animal models with gastric hypoacidity and/or hypergastrinemia have been used to study carcinogenesis of the human gastric corpus and fundus [24]. Animal models have the potential to delineate disease mechanisms and numerous rodent models have also been used to study presumed premalignant changes of gastric corpus [16]. In the current study, we have examined the effects of the gastrin receptor antagonist NTZ on the oxyntic mucosa of anacidic and hypergastrinemic H + /K + ATPase beta subunit KO mice with particular focus on NE cell hyperplasia and SPEM. We found that NTZ reduced the pronounced hypertrophy of the oxyntic mucosa as well as intramucosal cysts and invasions below the muscularis mucosae. An antral gastrin receptor has been reported [25], but a trophic effect of hypergastrinemia or effects of gastrin receptor blockade with NTZ on antral mucosal thickness was not seen. NTZ reduced the volume density of NE cells as well as gene expression of several NE markers in the oxyntic mucosa of KO mice. Gene expression of the ECL cell markers HDC (gene hdc), VMAT-2 (gene slcl18a2) and CCKBR (gene cckbr) were reduced, as expected from studies localizing the CCKB-R on the ECL cell [26]. The expression of tryptophan hydroxylase (gene Tph1), which is expressed in EC cells, was not affected by NTZ in agreement with previous studies showing that EC cells are not stimulated by gastrin [27]. Interestingly, the expression of the D cell marker somatostatin (gene sst) was increased by NTZ in KO mice. This is in concordance with gastrin having a negative trophic effect on the D cells in the stomach [27,28] and may reflect simultaneous augmentation of a compensatory mechanism. Previously, we have found that NTZ prevents development of neoplasia in the gastric corpus in hypergastrinemic cotton rats [29]. Similarly, carcinogenesis in M. natalensis is enhanced by the histamine 2 receptor antagonist (H2RA) loxtidine [30], but inhibited by NTZ [31] and likewise gastric carcinogenesis in hypergastrinemic transgenic INS-GAS mice is inhibited by NTZ [32]. In patients with hypergastrinemia due to chronic atrophic gastritis, we have reported that NTZ can eradicate (type 1) ECL cell NE tumors (NETs) [33][34][35]. Interestingly, patients with chronic atrophic gastritis have an increased risk of both gastric adenocarcinomas and NETs [36][37][38]. Hypergastrinemia has also been found to be a risk factor for subsequent development of gastric cancer [8]. More recently, several epidemiological studies reported that patients using PPIs area at increased risk of gastric adenocarcinoma [5][6][7]39], whereas H2RA users did not. The studies combined therefore suggest that hypergastrinemia, a common factor in the mentioned animal models and human conditions with increased cancer risk, is pivotal in carcinogenesis of the gastric corpus and fundus [8,9].
Adenocarcinomas of the intestinal type after Lauréns classification [40] decline in Western populations [41] and the incidence is closely related to the prevalence of Helicobacter pylori (H. pylori) infection. IM is a proposed precursor lesion of intestinal type adenocarcinomas [42] and there is considerable interest for IM in risk stratification of patients undergoing gastroscopy, as well as in animal models to study mechanisms in the premalignant mucosa. More recently, SPEM has also been suggested to be a precursor lesion of gastric adenocarcinomas of intestinal type [43], which may be viewed as either an alternative or a supplementary hypothesis to the previously proposed Correa cascade [42]. In epidemiologic studies, antral H. pylori infection protects against gastric cancer [44] and it therefore seems paradoxical that H. pylori-induced IM in the antrum could be a precursor lesion of cancer. However, H. pylori infection of the oxyntic mucosa with atrophy, which leads to hypoacidity and pronounced hypergastrinemia, increases the risk of cancer considerably [45]. A corresponding risk consideration concerning isolated antral IM has gained acceptance and endoscopic surveillance is not recommended in the most recent clinical guidelines in Europe [46,47].
Within the described context, it was of particular interest to further examine the role of gastrin in SPEM development. We have previously found that the SPEM marker clusterin is highly expressed in ECL cells of normogastrinemic rats, but in rats with hypergastrinemia due to PPI administration, clusterin expression was considerably increased, through a dominant shift in the expression towards cells of the mucous neck-chief cell lineages. [23]. Moreover, clusterin was found to be highly upregulated in mucus neck and SPEM cells of the H + /K + ATPase beta subunit KO mice of different ages [23]. However, the finding of accelerated SPEM development in gastrin KO mice given DMP-777 [17] demonstrates that SPEM development in mice cannot depend entirely on gastrin. In the current study, KO/PEG mice had pronounced expression of SPEM in the gastric corpus and numerous genes previously reported to be associated with SPEM were differentially overexpressed compared to WT/PEG mice. The genes overexpressed included SPEM-defining clu, tff2 and muc6, as well as wfdc2 [48] and cd44 [49] which also have been identified as markers of SPEM and do not occur in the normal gastric corpus. However, when comparing KO/PEG and KO/NTZ groups NTZ neither affected the existence of SPEM nor the expression of previously reported SPEM associated genes, as assessed both histologically and on protein and transcriptional level.
We have previously described SPEM in oxyntic mucosa of 3-month-old H + /K + ATPase beta subunit KO mice [23]. Others have reported mice of age 35 days to be depleted of mature chief cells based on morphological criteria in HE stained sections [50]. The phenomenon was independent of hypergastrinemia since the same was found in mice double KO for gastrin and H + /K + ATPase beta subunit [50]. Loss of morphologically normal mature chief cells is one of the characteristics of SPEM, which might indicate that SPEM is present very early in oxyntic mucosa of this mouse model. Exactly when this occurs, and whether it is distinguishable from glands of the embryonic and juvenile stomach described by others [51,52], has not been investigated. In the current study, we found that the KO/PEG mice (at age 13 months) expressed only negligible amounts of several chief cell markers compared to WT/PEG mice, and the expression was unaffected by NTZ. Since KO mice depleted of mature chief cells still develop SPEM, mature chief cells may not be the origin of SPEM in the current mouse model. Expressions of Xbp1 and Mist1 (gene Bhlha15) were low in KO mice and these are important for chief cells to form zymogen-containing vesicles [53,54]. In the current mouse model, one could speculate that SPEM develops from the large proportion of cells previously described as immature cells in young KO mice [50], possibly represented chief cells devoid of granules. Alternatively, the observations could support the theory that SPEM is derived from isthmus stem cells [55].
The term SPEM is viewed by many as synonymous with the older terms "pyloric metaplasia", "pseudopyloric metaplasia" or "antralization" and the actual alterations may be considered a common response to glandular injury observed in both mice and humans. Furthermore, ulcer-associated cell linage (UACL) has also been viewed as similar reparative linages and is observed in reparative processes along the gastrointestinal tract, including at the edges of gastric ulcers [56] and also in inflammatory bowel disease [57,58]. Also, one of the highly expressed markers of gastric SPEM, clusterin, seems to have subtle protective roles on oxyntic mucosa [59]. These observations combined suggest that SPEM should not be considered merely a preneoplastic lesion. Accordingly, it has been argued that SPEM could be an phenomenon associated with carcinogenesis, as a form of mucosal injury, not a precursor of cancer [60], and that SPEM and IM could be commensals for a neoplastic process rather than true direct precursors [43]. Reported findings from animal models of IM and SPEM have led to conflicting hypotheses of the cellular origin of the lesions, which may be either chief cells [61] or isthmus stem cells [55]. It is notable that the ability of gastric metaplasia to progress to invasive cancer seems to be lacking in the more than 20 mouse models utilized to study this, which also questions the validity of studies aiming to examine premalignant changes [16]. We propose that further studies of dysplasia and SPEM in patients with chronic atrophic gastritis treated long-term with NTZ [33,35] could provide more information about the roles of SPEM and hypergastrinemia in human carcinogenesis.
In conclusion, the gastrin receptor antagonist NTZ prevented mucosal hypertrophy, intramucosal cysts, invasions of the muscularis mucosae as well as NE cell hyperplasia in the gastric corpus of H + /K + ATPase beta subunit KO mice. However, SPEM is found in oxyntic mucosa independently of gastrin signaling and the presence of SPEM seems unrelated to the presumed premalignant changes caused by hypergastrinemia in this animal model. The clinical implications of the current findings could be clarified through further studies of patients treated with NTZ.

Animals and Genotyping
H + /K + ATPase beta subunit KO mice, originally with a BalbC/black6 [22] background and later BalbC [62], were back-crossed 8 times onto BalbC mice (Møllegaard, Skensved, Denmark) before study start. DNA was isolated from tail samples using the high pure PCR template preparation kit, according to the manufacturer's instruction (11796828001, Roche Diagnostics, Indianapolis, IN, USA). A PCR assay was set up to genotype mice and was used to distinguish between homozygous KO mice (-/-), heterozygous mice (+/-) and homozygous WT mice (+/+). For detection of H + /K + ATPase WT alleles, specific primers (forward primer: GGACCAACTGACTTCTGGGA and reverse primer: ACCTGCATGGCAGTCTCTCT, product length: 472 bp) were based on H + /K + ATPase beta subunit (Atp4b) sequence (Ensembl: ENSMUSG00000031449) and located approximately 100 bp upstream and downstream of exon 1 (the target for the homologous recombination). A PCR product with size of 472 bp would appear only if the WT allele was present. Inserted construct (PKG-neo) alleles were detected using specific primers selected from the PKG-neo sequence [22] (forward primer: AGACAATCGGCTGCTCTGAT and reverse primer: ATACTTTCTCGGCAGGAGCA, product length: 261 bp) and a PCR product with the size 261 bp would only appear if the KO allele was present. All variants were genotyped by combining these two PCR reactions using the FastStart PCR Master kit according to the manufacturer´s instruction (Roche Diagnostics), and visually inspected after gel electrophoresis separation and ethidium bromide staining.
For the experiments, we used female H + /K + ATPase KO mice (n = 21) and WT controls (n = 21) 1 month of age. They were housed in cages with aspen woodchip bedding (B&K Universal Ltd., Hull, UK). The room temperature was 24 ± 1 • C with a relative humidity of 40-50% and a 12h light/dark cycle. RM1 (E) diet (SDS, Essex, UK) and tap water were provided ad libitum. The study was approved by the Norwegian National Animal Research Authority.

Study Design
The drug netazepide (NTZ), previously named YF476 (Hammersmith Medicines Research, London, UK), has been shown to be a potent and highly selective competitive antagonist of the gastrin/CCKB receptor (CCKBR) [63]. It was dissolved to a concentration of 12 mg/mL in polyethylene glycol 300 (PEG) and given as subcutaneous injections at a dose of 40 mg/kg (80 µmol/kg) to n = 11 KO mice (KO/NTZ) and n = 10 WT mice (WT/NTZ) from age 1 month, once every two weeks for 12 months. Animals in the two control groups received an equivalent volume of PEG (KO/PEG, n = 10 mice and WT/PEG, n = 11 mice). The toxic dose of NTZ have been studied by Ferring and in 13-week studies, the no-observable-adverse-effect level was 100 mg/kg/day in rats and in dogs [64].

Intragastric pH and Plasma Gastrin at Termination
The intragastric pH was measured by using a pediatric pH catheter with the stomach in situ. The lowest pH obtained after searching the entire corpus and antral mucosa was recorded for each animal. The animals were killed by exsanguination from the inferior cava while anesthetized with 5% isoflurane with O 2 and N 2 in a ratio of 3:2 as carrier gas with total gas flow of 1 L/min. Plasma was separated and kept at −20 • C until gastrin analysis by radioimmunoassay as previously described [65].

Histopathology, Immunohistochemistry (IHC) and in situ Hybridization (ISH)
After measuring intragastric pH, the stomach was removed, rinsed in saline, weighed, opened along the greater curvature and inspected macroscopically. Tissue samples for histopathology were taken longitudinally from the rumen to the pylorus on the major curvature and fixed in 4% formaldehyde before processing and paraffin embedding. Sections (4 µm) were cut and stained with hematoxylin and eosin (HE) and examined microscopically. Dysplasia was graded as none, mild, moderate or severe, and inflammation was graded according to the Updated Sydney Classification [66]. To assess trophic changes, mucosal thickness was determined using an ocular grid. Measurements were performed in areas without macroscopic abnormalities and where gastric crypts were visible in their full length. Corpus and antral mucosal thickness were measured at five locations in each animal and the means recorded. The number of cystic dilations and invasions through the muscularis mucosa per mm horizontal mucosa was counted in areas where the gastric glands were cut longitudinally and visible in their whole length.
ISH for clusterin mRNA was done using the RNAscope 2.0 HD Reagent Kit (Brown) for FFPE tissue (310035, Advanced Cell Diagnostics Inc, CA, USA) and a custom probe (427891, Mm-Clu) for clusterin, according to the manufacturer's protocol. All IHC and ISH sections were counterstained with hematoxylin. Images were captured using Nikon E400 microscope, DS-Fil U2 camera and NIS-Elements BR imaging software (Nikon, Melville, NY, USA), and further processed using ImageJ.

Whole-Genome Gene Expression Analysis
Biopsies from the ventral part of the gastric corpus mucosa were frozen in liquid N 2 and then homogenized in lysis buffer using an Ultra-Turrax rotating knife homogenizer. The RNeasy Mini kit (Qiagen, Germantown, MD, USA) was used for total RNA extraction, according to the manufacturer's instructions. RNA concentration was measured using Qubit ® RNA HS Assay Kit on a Qubit ® 3.0 Fluorometer (Thermo Fisher Scientific Inc., Waltham, MA, USA). Integrity was assessed using Agilent RNA 6000 Pico Kit on a 2100 Bioanalyzer instrument (Agilent Technologies, Santa Clara, CA, USA).
RNA sequencing libraries were prepared using TruSeq Stranded mRNA kit (Illumina, San Diego, CA, USA) according to the manufacturer's instructions. In brief, 2 µg total RNA was used as starting material. First, mRNA was purified from the total RNA using poly-T oligo-attached magnetic beads, follows by random fragmentation using divalent cations at 94 • C for 4 min. First and second strand cDNAs were synthesized using random oligonucleotides and SuperScript II, followed by DNA polymerase I and RNase H. Exonuclease/polymerase was used to produce blunted overhangs. Illumina duel index adapter oligonucleotides were ligated to the cDNA after 3' end adenylation. DNA fragments were enriched by 15 cycles of PCR reaction. The libraries were purified using the AMPure XP (Beckman Coulter, Inc., Indianapolis, IN, USA), quantitated by qPCR using KAPA Library Quantification Kit (Kapa Biosystems, Inc., Wilmington, MA, USA) and validated using Agilent High Sensitivity DNA Kit on a Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). The size range of the DNA fragments were measured to be in the range of app. 200-1000 bp and peaked around 285 bp.
Libraries were normalized and pooled to 2.0 pM and subjected to clustering on two NextSeq 500 high output flow cells. Finally, single read sequencing was performed for 75 cycles on a NextSeq 500 instrument (Illumina, Inc. San Diego, CA, USA), according to the manufacturer's instructions.
Base calling was done on the NextSeq 500 instrument by RTA 2.4.6. FASTQ files were generated using bcl2fastq2 Conversion Software v2.17 (Illumina, Inc. San Diego, CA, USA). For each sample, kallisto (v0.42.4) was used to quantify GRCm38 transcripts and Sleuth was used for differential expression analysis [68]. Statistical significance was defined with a Wald test and Benjamini-Hochberg false discovery rate adjusted P-value < 0.05. The microarray experiments are Minimum Information About a Microarray Experiment (MIAME) compliant and have been deposited in the Gene Expression Omnibus (GEO) repository and assigned the GEO accession number GSE142513. The expressions of genes reported to be associated with SPEM [69,70] and NE cells [26,[71][72][73][74] in previous publications were manually identified in the gene expression analysis.  Acknowledgments: The authors thank Bjørn Munkvold for preparation of histopathological specimens, Britt Schulze for gastrin RIA, and staff at the Comparative Medicine Core facility (CoMed) at NTNU for assistance with drug administration and animal caretaking. The RNA library prep, sequencing and bioinformatics analysis were performed in close collaboration with the Genomics Core Facility (GCF), Norwegian University of Science and Technology (NTNU). GCF is funded by the Faculty of Medicine and Health Sciences at NTNU and Central Norway Regional Health Authority.

Conflicts of Interest:
The authors declare no conflict of interest.