Molecular characterization of mouse gastric zymogenic cells*

time quantitative RT-PCR studies of laser capture microdissected populations harvested from gastric cryosections. One finding from the analysis is that ZCs are a source of platelet derived growth factor (pdgf) A and B. Our previously published molecular profile of PCs (11) revealed that they are a source of another angiogenic factor, vascular endothelial growth factor or B (Vegfb). The results of quantitative three-dimensional imaging studies of the capillary networks that surround gastric units in the proximal and distal glandular epithelium of adult germ-free transgenic mice with an engineered ablation of their PCs and ZCs, and in their normal germ-free littermates, provide direct evidence that these lineages contribute to the regulation of angiogenesis. a discrete block in terminal differentiation of neck cells to mature ZCs (21). The changes are gastric unit autonomous: in mice with a mosaic pattern of transgene expression, only units with tox 176-mediated PC ablation lack mature ZCs. Neighboring units that do not support transgene expression have a normal complement of PCs, and ZCs (21). units. findings support an theme that epithelial-mesenchymal cross talk plays an important role in modulating angiogenesis in the adult mouse gut: removal of Paneth cells, which differentiate at the base discrete mucosal the small crypts results in


Introduction
The human gastric epithelium is renewed continuously throughout life with an estimated 500,000 cells shed per minute in adults (1). Dysregulated renewal results in a number of disease states, including gastric adenocarcinoma, the second most common cause of cancer-related deaths worldwide (2). The mouse has been used as a model organism to characterize the cellular and molecular features of this renewal, and to decipher its regulation.
The proximal third of the mouse stomach (forestomach) is lined with a squamous epithelium, and its distal two thirds with a glandular epithelium. The glandular epithelium contains thousands of tubular mucosal invaginations known as gastric units. In the central or corpus region of the stomach, each unit has a steady state population of ~200 epithelial cells, representing three principal lineages (pit, parietal, zymogenic) and two minor lineages (enteroendocrine, caveolated) (3). All lineages are derived from multipotent stem cells located in the mid-portion (isthmus) of each unit ( Fig. 1A-C) (4,5).
Karam has used in vivo pulse labeling with [ 3 H]thymidine, followed by electron microscopic autoradiography, to obtain morphologic descriptions of the presumptive multipotent stem cell and its immediate committed oligopotential daughters, as well as the pathways their descendants follow during terminal differentiation (5)(6)(7)(8)(9)(10). Mucus-producing pit cells, derived from one of the stem cell's committed daughters, differentiate during a rapid (3 d) migration from the Stem cell niche located in the middle portion of the unit (isthmus) through the upper portion time quantitative RT-PCR studies of laser capture microdissected populations harvested from gastric cryosections. One finding from the analysis is that ZCs are a source of platelet derived growth factor (pdgf) A and B. Our previously published molecular profile of PCs (11) revealed that they are a source of another angiogenic factor, vascular endothelial growth factor or B (Vegfb). The results of quantitative three-dimensional imaging studies of the capillary networks that surround gastric units in the proximal and distal glandular epithelium of adult germ-free transgenic mice with an engineered ablation of their PCs and ZCs, and in their normal germ-free littermates, provide direct evidence that these lineages contribute to the regulation of angiogenesis.

Experimental Procedures
Animals -Conventionally raised, specified pathogen-free normal FVB/N mice were maintained under a strict 12 h light cycle and fed a standard chow diet (Picolab rodent diet 20, Purina Mills). Members of a pedigree of germ-free FVB/N transgenic mice that express an attenuated diphtheria toxin A fragment (tox176) under the control of nucleotides -1035 to +24 of the mouse gene encoding the b-subunit of H + /K + ATPase (14), and their germ-free normal littermates, were maintained in plastic gnotobiotic isolators (15). All experiments involving mice were conducted using protocols approved by the Animal Studies Committee of Washington University.

Isolation of ZCs by elutriation and lectin panning -For each cell preparation, 10
conventionally raised mice were sacrificed at eight weeks of age. Their stomachs were rapidly removed, and put in sterile 150 mM NaCl (pH 7.4). Ligatures were placed at the junction between the esophagus and forestomach, and at the pylorus. A 2 mm incision was made in the forestomach, and the entire stomach everted through the incision. Another ligature was placed at the junction between the forestomach and corpus, and 500 µl of a solution of 1.0 mg/ml Streptomyces griseus Pronase E (7 U/mg; Roche Applied Science) in medium A (70 mM NaCl, 5 mM KCl, 50 mM HEPES, 1 mM Na 2 HPO 4 , 0.5 mM NaH 2 PO 4 , 20 mM NaHCO 3 , 10 mM EDTA, by guest on March 24, 2020 http://www.jbc.org/ Downloaded from 11 mM glucose, 5 mg/ml BSA; pH 7.8) was instilled with a 27 gauge needle. Filled gastric sacs were placed in a flask containing 75 ml of medium A and incubated for 30 min at 37!C (16,17).
The everted sac was then put into 75 ml of medium B (medium A with 100 mM CaCl 2 , 15 mM MgCl 2 and no EDTA; pH 7.4) and incubated for 30 min at 37!C with constant gentle magnetic stirring. The medium, containing shed epithelial cells, was decanted and passed though a nylon mesh (60 µm pore diameter; Millipore). Cells in the flow-thru were recovered by centrifugation (150 ¥ g, 5 min, 23!C) and re-suspended in 5 ml of medium C (140 mM NaCl, 12 mM MgSO 4 , 10 mM CaCl 2 , 15 mM HEPES, 11mM glucose, 0.5 mM DTT, 10 mg/ml BSA; pH 7.4). Cellular viability was defined by staining an aliquot with a solution containing 0.1% neutral Red, 0.1% Brilliant Cresyl Blue, and 0.01% Janus Green in PBS.
Cellular concentration was adjusted to 1.5x10 7 /ml medium C and 5 x 10 7 cells were subjected to counter-flow elutriation (JE-6 elutriator rotor with standard chamber; Beckman Instruments; injection flow rate 10 ml/min medium C containing 1 mg/ml BSA). Details of the elutriation protocol are provided in Table 1S of the on-line supplemental material. Fractions 1-5 (Table 1S)  Cells in fractions 4 and 5 that remained after DBA panning (Table 2S) were pooled to generate a "ZC+" fraction. The purified ZC+ population, the purified PC+ population, and the cells in fractions 1-2 remaining after DBA panning (ZC-/PC-population) were each collected by centrifugation (150 ¥ g; 5 min; 23!C), re-suspended in RTL lysis buffer (Qiagen), and RNA was extracted (RNeasy Mini kit with on-column DNAase treatment; Qiagen). of ZCs); the top ~25% (pit; predominantly pit cells); and the mesenchyme immediately below the gastric unit (n=300 gastric units microdissected/cryosection; 4-5 cryosections sections/mouse).
RNA was isolated from ~5000 cells/fraction/mouse using the PicoPure RNA Isolation kit (Arcturus) with on-column DNase treatment (Qiagen). Equal amounts of RNA from each area were pooled from three mice, and the material used for qRT-PCR studies.
Transmission EM -Cells recovered by elutriation/lectin panning were subjected to transmission EM analysis using protocols described in a previous publication (11).
Three-dimensional imaging studies of the gastric microvasculature -Eight-week-old germ-free normal and tox176 mice were anesthetized, and 200 µl of a 20 mg/ml aqueous solution of high molecular weight (2000 kDa) fluorescein isothiocyanate (FITC)-labeled dextran (Sigma) was injected in their retro-orbital plexus using a 30 gauge needle attached to a 1 ml syringe. Three minutes after the 10-15 sec infusion, animals were sacrificed, their stomachs removed, and perfused with fixation solution (0.5% paraformaldehyde, 15% picric acid, and 0.1 M sodium phosphate buffer; pH 7.0). Each stomach was opened with an incision along its greater curvature, pinned on wax, and shaken at 4 o C for 12h in fixation solution. Following three washes in ice-cold PBS (5 min/cycle), the stomach was incubated for 3 h in 10% sucrose/PBS (4 o C), and then overnight in 20% sucrose/10% glycerol/PBS (4 o C). After freezing the tissue in O.C.T., 60 µmthick cryosections were cut along the cephalocaudal axis. Sections were air-dried (2 h, room temperature in the dark), re-hydrated in ice-cold PBS (1 min), incubated overnight (4 o C) in 3% deoxycholic acid (Sigma), rinsed with deionized water (2 cycles, 5 min each, room temperature), followed by a PBS wash (5 min, room temperature) and stained with either Syto61 (Molecular Probes; 1:1000 dilution in PBS; 1h at room temperature) or DBA. Following three more PBS washes (5 min/cycle; room temperature), sections were mounted in 50 % glycerol/PBS, viewed under a LSM 510 confocal microscope (Zeiss), and scanned at 3 µm-thick intervals. Scans were projected in three dimensions by taking 20 serial images, aligning them at 7-10 degree intervals, and compiling/rotating them around the y-axis using LSM 510 software.

Purification of zymogenic cells from adult FVB/N mouse stomachs
ZCs were purified from the stomachs of conventionally raised 8 week-old mice belonging to the FVB/N inbred strain. Mucosa from the corpus region of the stomach was dissociated by Pronase E digestion, and the cellular suspension size-fractionated by counter-flow elutriation.
Five fractions were recovered, and each fraction was subjected to lectin panning with DBAmagnetic bead conjugates to remove parietal cells. This approach yielded two fractions enriched for ZCs (fractions 4 and 5 in Tables 1S and 2S of the on-line supplemental material). Histo-and immunohistochemical analysis using a previously described panel of lineage-specific lectins and antibodies (11,12,18,21) disclosed that ZCs represented 71-73% of the cells in these fractions, compared to 11% in the starting material (Table 2S; n = 6 independent preparations). Overall cell viability in the two fractions, which together composed the 'ZC+ population', was >90%, as defined by vital dye exclusion (Table 1S). Transmission EM studies confirmed that the purified  Tables 1S, and   2S). This 'ZC-/PC-' fraction contained an average of 50% pit cells, 20% neck cells, <1% PCs, <2% ZCs and 27% 'other' cell types (GEPs, enteroendocrine cells, mesenchymal components) (Table 2S; Fig. 2D).

GeneChip-based dataset of transcripts enriched in zymogenic cells
Total cellular RNA was extracted from the ZC+, PC+ and ZC-/PC-populations obtained from each of the six preparations. Equivalent amounts of RNA from a given  Table 5S). The union of these two datasets ( Fig. 2A) produced a list of 57 known genes and 14 uncharacterized genes from what amounted to quadruplicate comparisons (Table 6S). The qRT-PCR study confirmed that the levels of these 12 mRNAs were 10±1 to 421±97fold higher in ZC+ versus PC+ RNA ( Table 1) Stomachs were serially sectioned, and cells harvested from cryosections that had been briefly stained with methyl green and eosin Y. Dissection was directed (navigated) using an electronic image template of the immediate previous section that had been stained with Griffonia simplifolica II lectin to mark neck cells, and DBA to tag PCs (Fig. 3A). The quality of the microdissection was confirmed by noting that the level of the ZC marker, Gif mRNA, was 760±33-fold higher in n-LCM epithelial cells harvested from the base of gastric units compared to epithelial cells retrieved from the pit region. The pit cell marker, trefoil factor 1 (Tff1) mRNA, was 14.3-fold higher in pit compared to base epithelium, while vimentin mRNA was 67±2 and 6±0.8 fold higher in mesenchyme versus the pit and base fractions, respectively (Fig. 3B).
Having established the quality of the dissection, subsequent analysis of Cckar, Fes, Anpep, and Spp1 expression revealed that the concentrations of their mRNAs were 18.6±0.3 to 903±15-fold higher in n-LCM ZC-rich base epithelium compared to the other compartments (Fig.   3C).

The impact of ZCs on gastric unit homeostasis
The functions and fates of PCs and ZCs appear to be intertwined. Gastric acid secretion first evolved in Elasmobranches ~350 million years ago (31). In non-mammalian vertebrates, the gastric oxyntopeptic cell secretes acid and pepsinogen, while in mammals, PCs export acid and Growth factors -Our previous GeneChip profiling of PC gene expression revealed that these cells are enriched for insulin-like growth factor binding protein 2 (Igfbp2) mRNA (11).
Igfbp2 decreases the bioavailability of IGFs (33). Isthmal GEPs, in turn, preferentially express transcripts encoding Igf-I and the Igf-I receptor (recognizes both Igf-I and Igf-II) (12). Based on these findings, we proposed that GEPs produce and respond to Igfs through an autocrine loop, while PCs may regulate GEP proliferation by sequestering Igfs (12). In this scheme, tox176mediated ablation of PCs would increase Igf bioavailability in the isthmal stem cell niche (through loss of Igfbp2), thereby contributing to the stimulation of GEP proliferation that is observed in these mice (12,21,34).
The present study disclosed that furin mRNA is enhanced in ZCs relative to PCs or the ZC-/PC-fraction (see Tables 4S-6S  Angiogenesis -The GeneChip study revealed that ZCs have enriched levels of Pdgfa mRNA. Follow-up qRT-PCR analysis showed that Pdgfb mRNA was also augmented ( Table 1).
These findings were confirmed in vivo by qRT-PCR assays of LCM cell populations (Fig. 3D).
qRT-PCR assays disclosed that the levels of mRNAs encoding Pdgf receptor subunits, Pdgfra and Pdgfrb, were higher in the ZC-/PC-fraction compared to the ZC+ population (17±5 and 2.7±0.3-fold respectively; Table 1). This ZC-/PC-fraction is enriched in mucus-secreting pit and neck cells, GEPs and mesenchymal components. Therefore, to assess Pdgf expression in these populations in vivo, we used germ-free tox176 mice to procure these cells from the pit, isthmal and neck regions by n-LCM. Germ-free rather than conventionally raised tox176 mice were used since the latter develop bacterial overgrowth with gastritis from loss of the acid-barrier to colonization (12,14). To further test the hypothesis that ZCs and PCs affect angiogenesis, we compared capillary network density in the same regions of the stomachs of germ-free normal mice and ageand gender-matched tox176 littermates that lack PCs and ZCs in all regions of their glandular epithelium. ZC/PC ablation from the corpus region of tox176 animals reduced gastric unit microvascular network density 2-fold (p<0.01) without affecting network density in antral units ( Fig. 5B-F). Indeed, antral gastric unit capillary density in tox176 mice was not significantly different from that of normal mice, suggesting that the effects of PC/ZC ablation are confined to the region of the stomach where these cells are lost (i.e., the effects on appear to be gastric unit autonomous). In all regions of the tox176 stomach, vessels run parallel to the base-to-pit axis of gastric units, with few encircling branches -a pattern similar to that encountered in the antral units of normal mice (Fig. 5C-E).       Table 6S for gene names.