Serca2 deletion in the mouse adult Bmi1+ compartment induces a lethal phenotype involving a severe gastric dysfunction

Abstract

BACKGROUND

Cytoplasmic calcium ions (Ca²⁺) are well-known intracellular signalling molecules, being involved in the integration of a variety of metabolic, proliferative and survival signals. They have been demonstrated an important role in embryonic and most adult stem cells (ASC). In the gut, it has been long recognized that Ca²⁺ handling is critical for the generation of pacemaker activity by Cajal interstitial cells. More recently, some studies in have shown that cytosolic Ca²⁺ plays a central role in intestinal stem cell (ISC) regulation, while this connection remains undefined in mammalian ISC populations.

AIM

To evaluate whether cytosolic Ca²⁺ could play an equivalent role in murine ISC population as in Drosophila melanogaster, and compare this effect to the lymphohematopoietic compartment.

METHODS

To this aim, we disrupted the expression of the calcium transporter Serca2, one of the crucial players in Ca²⁺ storage, within murine ISC population. Murine ISC compartment was selected using Bmi1, a well-established marker for reserve ISCs crucial in regeneration upon injury, and Serca2 conditional deletion was evaluated in the gut epithelium using the Serca2^Bmi1homo mouse strain. Main results were compared with the Serca2^villinhomo strain. Metabolic parameters were evaluated, including food and water intake, body weight evolution, glucose response, serum biochemical markers and fecal fat content. Histological analysis along the gastric tract was used to assess Bmi1⁺ ISC proliferative state. Intestinal crypts were purified and analyzed for proliferation, cell death and gene expression. Total and segmented intestine was analyzed for differential gene expression. Due to the severe phenotype found, the impact of Serca2 conditional deletion was later evaluated in the lymphohematopoietic system, to eliminate a generalized effect.

RESULTS

Conditional deletion of Serca2 in Bmi1⁺ adult ASC (DSerca2^Bmi1homo mice) resulted, unexpectedly, in a lethal phenotype. Important weight loss suggested severe malabsorption/malnutrition as the most probable cause of deaths. Results were further confirmed in DSerca2^Villin mice, that developed a fulminant phenotype during the first week after tamoxifen-induction. Blood biochemistry showed a reduced glucose uptake, while feces content showed an increased total amount of fat in DSerca2^Villin mice. Histological analysis along gastric tract revealed major extension of Bmi1⁺ cell compartment in DSerca2^Bmi1homo mice. In the gut, Bmi1⁺ ISC population showed a substantial increase (3-fold) in their proliferative rate upon Serca2 deletion. Differential gene expression did not identify major changes in intestinal crypts, but key calcium regulatory players such as Orai-1, Pmca-1 and Pmca-4 expression were affected, as well as endoplasmic reticulum stress response genes. Finally, important thymic alterations were characterized in DSerca2^Bmi1 mice, plausibly associated to the malnourished state of the animals.

CONCLUSION

As in Drosophila melanogaster, cytosolic Ca²⁺ is a key element of murine ISC regulation, affecting their proliferation rate and the proper differentiation equilibrium. Severe lethality provoked by Serca2 deletion in both gut Bmi1⁺ and Villin⁺ compartments raises the possibility that dysregulation of cytosolic Ca²⁺ could play some role in some gastric pathologies.

Key Words: Bmi1; Adult stem cell; Serca2; Calcium ions regulation; Small intestine; Crypts; Intestinal stem cell

Core Tip: Although cytoplasmic [calcium ions (Ca²⁺)] plays an important role in regulating several critical cellular processes, low information is available regarding intestinal stem cells (ISC). Recent seminal results, obtained in Drosophila melanogaster, have demonstrated that the reduction of Ca²⁺ in ISC increases their proliferation rate; downregulation of Serca is one of the main genes involved. Based on these relevant results, and taking advantage of a dedicated mouse models, we have demonstrated that the deletion of Serca2 in the intestinal Bmi1⁺ populations provokes, unexpectedly, a lethal phenotype, associated with a severe malnutrition. Furthermore, because this promotes ISC proliferation rate a moderate reduction of Serca2 activity could be involved in some neoplasic conditions.

INTRODUCTION

Intracellular [calcium ions (Ca²⁺)] controls several critical processes including cell growth and survival reviewed in[1,2] playing a key role in governing cell-signaling pathways in embryonic and different adult stem cells (ASC)[3], including long-term hematopoietic stem cells[4]. Intracellular Ca²⁺ impact, however, showed significant heterogeneity, event contrary, among the different ASC [Lin^-Sca⁺ckit⁺ (LSK) population], probably related to the level of Bmi1⁺ cells. In the gut it has been long recognized that Ca²⁺ handling is critical for the generation of pacemaker activity by Cajal interstitial cells[5], but less is known about the involvement of Ca²⁺ transients in mammalian intestinal stem cell (ISC) homeostasis. More recently, a clear relationship between gut microbiota and intracellular calcium homeostasis has been demonstrated[6].

An important role for Ca²⁺ has been recently demonstrated in the proliferative adaptation of ISC to diet in Drosophila melanogaster, as high cytosolic Ca²⁺ induces ISC proliferation[7]. Indeed, target perturbations that showed impaired Ca²⁺ oscillations with a parallel increase in cytosolic Ca²⁺, both knocking down Sarco(endo)plasmic reticulum calcium ATPases (SERCA) or plasma membrane calcium ATPase, confirmed a significant interference with ISC proliferation[7]. Authors concluded that the dynamic of intracellular Ca²⁺ allows the integration of diverse mitogenic signals in ISC, adapting their proliferative activity to the tissue needs; in addition, it was proposed that this mechanism might be conserved in mammals and other ASC[7,8].

SERCAs are the main players in Ca²⁺ transport from the cytosol to the endoplasmic reticulum (ER) to ensure a releasable Ca²⁺ store. In mammals, three paralogs (SERCA1-3) have been identified and are expressed at various levels in different cell types[9]. Serca2 encodes Serca2a and Serca2b isoforms; the former is the so-called “cardiac isoform”, highly expressed in cardiac, slow-twitch skeletal muscle and smooth muscle cells, whereas the latter, Serca2a, is ubiquitously expressed[10].

Given these findings, we aimed to interrogate the relevance of cytosolic Ca²⁺ distortion in tissue homeostasis by deleting Serca2 in adult murine ISC, using Bmi1 as a general marker for ASCs, and specifically ISC, aiming us to test the conclusions previously derived from Drosophila melanogaster. Our results indicate that Serca2 conditional deletion in adult mouse Bmi1⁺ compartments differentially affects the homeostatic regulation of distinct ASC populations, being the intestinal Bmi1⁺ progenitor cells to be the most sensible to chronic augmented cytosolic Ca²⁺; DSerca2^Bmi1homo provoking a distorted proliferative rate of Bmi1⁺ ISC and key calcium regulators expression; ultimately this generates an altered proliferation/differentiation equilibrium and malabsorption that cause a severe and early lethal phenotype.

MATERIALS AND METHODS

Mouse models and in vivo procedures

Two different transgenic models (Serca2^Bmi1 and Serca2^Villin), based on previously described modules (see below), were generated for the conditional deletion of exons 2-3 of Serca2 in adult Bmi1⁺ lineages upon induction with tamoxifen (Tx). Along the text Serca2 deleted animals (after Tx-induction) are indicated by the “D” prefix (i.e., DSerca2^Bmi1 or DSerca2^Villinhomo), and regarding DSerca2^Bmi1, it is indicated whether we are working with homozygote animals (DSerca2^Bmi1homo) or heterozygote (DSerca2^Bmi1hetero) mice. Bmi1CreERT[11] was a kind gift of Dr. Capecci (Salt Lake City, UT, United States) and floxed-Serca2 strain[12] by Dr. Christensen (University of Oslo, Oslo, Norway), that allow the conditional deletion of exons 2-3 of Serca2 upon Tx-induction. VillinCreERT[13], Bmi1^tm1Ilw Thy1a/J (Bmi1^GFP/+)[13,14], B6.129X1-Gt(ROSA)26^{Sortm1(EYFP)Cos/J} (Rosa26^YFP/+) and B6;129S6-Gt(ROSA) 26^{Sortm14(CAG-tdTmt)Hze}/J (Rosa26^Tmt/+) are mouse strains obtained from The Jackson Laboratory (ME, United States), that allow the conditional expression of YFP or Tmt, respectively, when the recombinase Cre fused to the estrogen receptor domain, is activated by the administration of Tx, and acts on loxp sites strategically located flanking translational stop sequences (specific trinucleotide codons UAA, UAG or UGA that signal the termination of protein synthesis). Finally, the floxed-diphtheria toxin A (DTA) strain, a kind gift of Dr. Riethmacher (University of Southampton, Southampton, United Kingdom), to obtain the composed Bmi1CreERT^+/-/floxedDTA^+/- strain[13], that allows the highly controlled expression of a DTA, upon administration of Tx, along the Bmi1⁺ lineages.

When indicated, heterozygote mice (DSerca2^Bmi1hetero) were also analyzed. Serca2^Bmi1homo mice, after induction with Tx (DSerca2^Bmi1homo), were compared with age-matched mice treated with corn oil (controls) or double control transgenic mice Tx-induced. For evaluation of the Bmi1⁺ stem cell compartments, a triple transgenic mouse Bmi1CreERT^+/-/floxedSTOP-Tmt^+/-/floxedSerca2homo mouse (shorten as Serca2^Bmi1homo) was generated and compared with double control transgenic mice (Bmi1CreERT^+/-/Rosa26-floxedSTOP-Tmt ^+/-), shorten as Bmi1-DTmt.

Tx-induction of Serca2^Bmi1homo mice provokes the combined expression of the fluorescence protein Tmt⁺ and deletion of Serca2 (exons 2,3). Control Bmi1-DTmt mice, upon Tx-induction, deletes some STOP sequences allowing the expression marker Tmt⁺ from the Rosa 26 Locus (Bmi1-DTmt). Finally, to evaluate the phenotype in small intestine epithelium, we studied the DSerca2^Villinhomo model (VillinCreERT⁺/floxedSerca2homo), that deletes Serca2 (exons 2,3) in Villin⁺ cells, generating the DSerca2^Villinhomo mice.

Animal studies were approved (Ref. 23002) by the Ethics Committee of the National Center for Biotechnology (CEEA-CNB). All mouse strains were bred on the C57BL/6 background. Tx (Sigma, MO, United States) was dissolved in corn oil (Sigma) at 20 mg/mL and injected intraperitoneally (9 mg per 40 g body weight), during three consecutive days. Experiments were carried out in male and female mice as recommended by the United States National Institutes of Health since preliminary analysis revealed no differences between males and females. 8-12-week-old mice were used as adult mice unless otherwise indicated in the text. When indicated, the negative controls were equivalent batches of animals but treated with solvent (corn oil) or Tx-induced control doble transgenic mice (Bmi1-DTmt).

Monitorization of animal food and water intake, weight evolution, biochemical parameters in serum and fat in feces

Batches of adult (8-12 weeks old) Serca2^Bmihomo and Serca2^Villinhomo animals were Tx-induced and compared with their corresponding control groups; when indicated Serca2^Bmi1hetero mice were evaluated. All animals were maintained individually and were compared along the indicated periods. Each animal was manually monitored daily for food and water intake, weight evolution and changes in feces consistency (e.g., loose stool, diarrhea). Perimortem DSerca2^Bmihomo and DSerca2^Villinhomo mice were sacrificed for histopathology and serum analysis and were compared with appropriate controls. The amount of fat in feces was determined by near-infrared reflectance spectroscopy after sample homogenization and according to[14].

Glucose tolerance testing

Randomized mice batches, previously induced with Tx (DSerca2^Bmi1homo), were subjected to 18 hours of fasting and 2 hours of water deprivation before the test. Young adult animals were weighted and injected D-glucose (20% solution) intraperitoneally at time 0, with a total dose of 2 g of glucose/kg of body weight, essentially as described[15]. Right after injection, blood glucose concentration at basal level was monitored using small tail incisions with a FreeStyle Lite glucometer (Abbott, Chicago, IL, United States). Those tests were repeated at different time points: 10’, 20’, 30’, 60’, 120’ and finally 180’. Resulting values were compared with Serca2^Bmi1 control mice treated with corn oil (n = 3, each group).

Immunofluorescence analyses

Detailed protocols can be found in the Supplementary material. Immunofluorescence analysis in thymus (Thy): 8 μm-thick thymic cryosections from DSerca2^Bmi1homo and control mice were stained with specific antibodies for keratin 5 (K5), pan cytokeratin (PanCK) and Aire for epithelial cell detection. Image acquisition was performed in a Leica SP8 confocal microscope equipped with a HC PL APO CS2 20x/0.75 DRY objective and analyzed using LasX software. Aire⁺ cells were determined counting the number of positive cells in the K5⁺ medullary area (pixels²) and referred as Aire⁺/K5⁺ area using Image J software. Immuno-fluorescence analysis in gastric samples: Tissues of Tx-induced [5 days (d5) and d7] adult DSerca2^Bmi1homo mice (Serca2^Bmi1homo control mice, induced with corn oil) were obtained and processed as described in the Supplementary material. When indicated in vivo proliferation was evaluated by Ki-67 staining following the provider’s instructions (BD Pharmingen; San Diego, CA, United States). All microscope images were taken with a multispectral confocal system Leica STELLARIS 5, with 4 laser lines, 3 Power HyD S spectral detectors and using the objective Leica HC PL APO CS2 20x/0.75 DRY. All antibodies used in the study are summarized in the Supplementary Table 1.

Histological analysis of DSerca2^Bmi1homo and DSerca2^Villinhomo mice

Gut isolated from DSerca2^Bmi1homo and DSerca2^Villinhomo mice, as well as the corresponding controls (injected with corn oil), were washed, fixed with 4.5% paraformaldehyde in phosphate-buffered saline (PBS) for 2 hours and stored in 1% paraformaldehyde in PBS until embedding. Then, samples were dehydrated twice for 5 minutes in increased ethanol proportions (50% to 100%), clearance in xylene and embedding for 2 hours in paraffin (60 °C). Thick sections (8 μm) were obtained in a microtome, stretching in a bath at 50 °C and collected on slides treated with poly-L-Lysine, and stained with Hemalum-Picro Indigo Carmine and Alcian Blue solution. Finally, samples were washed in water for 1 minute, dehydrated, mounted in distyrene, plasticizer and xylene and examined in a microscope Zeiss Axioplan-2 microscope, equipped with a Retiga 4000R camera and acquired with Metamorph program from the Cytometry and Fluorescence Microscopy Centre of the Complutense University of Madrid.

Electron microscopy analysis

Small intestine samples from controls and DSerca2^Bmi1homo mice were fixed in 2.5% glutaraldehyde/0.1M sodium cacodylate at 4 °C for 1 hour, washed in cacodylate/sucrose buffer overnight at 4 °C, then fixed in 1% osmium tetroxide for 2 hours and stained with uranyl acetate. Then, samples were dehydrated in increasing gradients of acetone (30% to 100%) and finally embedded in resin PB 812 (PolySciences, Warrington, PA, United States). Intestinal semithin sections were stained with toluidine blue/borax for identifying under a light microscope the most interesting areas. Ultrathin sections were double stained with uranyl acetate for 20 minutes and lead citrate for 8 minutes and examined in a JEOL 10.10 electron microscope at the National Center for Electron Microscopy of Complutense University of Madrid.

Flow cytometry

Bone marrow (BM), spleen (Sp), lymph nodes (LN) and thymus (Thy) samples were dissected and single-cell suspensions were prepared and analyzed essentially as described[16]. Primary and secondary antibodies used are described in Supplementary Table 1; detailed protocols can be found in the Supplementary material.

For the in vivo proliferation assay 5-ethynyl-2’-deoxyuridine (EdU) (EdU; Sigma) was dissolved in 0.9% sodium chloride and stored at 10 mg/mL; mice (DSerca2^Bmi1hetero and controls; age matched) received EdU at 10 μg/g (injected intraperitoneally), once daily during 5 days[13]. Then, the animals were sacrificed, and tissues were collected. The incorporated EdU was detected using the “Click-iT” reaction (Thermo Fidher Scientific; Alcobendas, Madrid, Spain), which uses a fluorescent azide dye to specifically label the incorporated EdU.

In vivo proliferative status was evaluated for the BM LSK population and Bmi1⁺ cells in the gut of DSerca2^Bmi1homo mice and compared to control animals, as described, following the manufacturer’s protocol. Samples were analyzed with a Beckman Coulter Moflow XDP cell sorter, Beckman Coulter GALLIOS Analyzer, BD FACSCanto II and FACSAria III (BD Biosciences) cytometers. FlowJo vX1 (TreeStar, Williamson Way Ashland, OR, United States) and FlowLogic (Inivai Technologies, Mentone, Victoria, Australia) software were used for data analysis.

Intestinal crypt cell purification

Small intestine samples were cut longitudinally, microvilli were carefully removed by gentle scraping and the remaining tissue was chopped into smaller pieces that were incubated in 8 mmol/L of ethylenediaminetetraacetic acid in hank’s balanced salt solution, for 20 minutes on ice. Samples were vigorously shaken obtaining a supernatant enriched for crypts. The incubation/shaking step was repeated twice, and all supernatants were centrifuged for 5 minutes at 1200 rpm.

Pellets from the intestinal crypts were also evaluated by flow cytometry; approximately 500000 cells for each condition. Annexin-V-APC antibody and 4’,6-diamidino-2-phenylindole staining were diluted in Binding Buffer (ThermoFisher Scientific, ref: V13246) and stored 15 minutes at 4 °C in darkness, then passed through the cytometer. For the proliferation evaluation cells were stained following the protocol CytoFix/CytoPerm (BD Biosciences, ref: 554714) according to the manufacturer’s instructions. The antibody against Ki-67 (Supplementary material) was diluted at 1:50 in PBS with 10% fetal bovine serum and incubated overnight at 4 °C in darkness. The next day cells were processed and analyzed by the flow cytometer. All experiments were performed in a Flow cytometer analyzer Beckman Coulter GALLIOS. Data was processed with KALUZA software.

Quantitative real-time polymerase chain reaction analysis

RNA was purified using the Zymo research kit (Irvine, CA, United States). Complementary DNA (cDNA) was obtained by reverse transcription with the high capacity cDNA reverse transcription kit from Thermo Fisher Scientific, using total RNA (500 ng). cDNAs were analyzed by real-time quantitative polymerase chain reaction (RT-qPCR) using the Power SYBR Green PCR Master Mix (Thermo Fisher Scientific); all primers used are indicated in Supplementary Table 2. Amplification, detection and data analysis were carried out with an ABI PRISM 7900HT Sequence Detection System and normalized to GusB, Hprt or Gapdh expression, as indicated.

Statistical analysis

Comparisons between groups were made using unpaired t-tests; black lines summarize P values (^aP < 0.05, ^bP < 0.01, ^cP < 0.002) (one-way analysis of variance followed by the Bonferroni correction for multiple comparison). Animal statistical analysis was carried out attending the supervision by Sorzano CÓ, specialized member of institutional ethical committee, to our whole project when indicated, statistical analysis was performed using 2-way analysis of variance test.

RESULTS

Serca2 conditional deletion in Bmi1⁺ adult cells provokes an early and lethal phenotype of malabsorption

Aiming to evaluate the functional consequences of chronic deregulation cytosolic Ca²⁺ in ASC compartments, and specifically in the ISC population, we took advantage of the Serca2^Bmi1 mouse strain (Supplementary Figure 1), that allow the conditional deletion of exons 2 and 3 of Serca2 in Bmi1⁺ cells (DSerca2^Bmi1homo) upon Tx administration.

Unexpectedly, early after Tx-induction we found that DSerca2^Bmi1homo mice suffer a lethal phenotype: Analysis of survival rates revealed that survival was considerably lower in DSerca2^Bmi1homo mice compared with controls. Specifically, DSerca2^Bmi1homo mice died starting on day 7, being all animals dead by day 20 (Figure 1A). In the current study, we confirmed by expression analysis that Serca2 levels were significantly decreased (> 70%) in whole BM in DSerca2^Bmi1homo mice (d5 post-Tx) as compared with control mice (corn oil-treated Serca2^Bmi1homo mice) (Supplementary Figure 1). Moreover, histopathology analysis revealed no overt findings in perimortem DSerca2^Bmi1homo mice; only limited peripheral lymphoid infiltration in the pancreas and moderate hyperkeratosis in the stomach and esophagus were evidenced (data not shown). In contrast to homozygous mice, Tx-induction of heterozygous Serca2^Bmi1 mice (DSerca2^Bmi1hetero) did not provoke a lethal phenotype, although mean animal survival rate was moderately decreased (Figure 1B). Previous results suggested that the lethal phenotype of DSerca2^Bmi1homo mice was not the consequence of the elimination (95%) of Bmi1⁺ cells perse (by expression of DTA) in homeostasis (Supplementary Figure 1), did not have significant impact on animal survival[13], but, it was, presumably due to a severe functional interference provoked by the lack of SERCA2 activity. Indeed, analysis of the BM at 4 months post-depletion demonstrated an important, but not complete, recovery of many Bmi1⁺ BM cell subpopulations with a clear over-representation of the B-cell lineage (Supplementary Figure 1).

Figure 1 Impact of conditional Serca2 deletion on survival, blood biochemistry, weight evolution and glucose absorption. A and B: Survival curves of homozygous (homo) (A) and heterozygous (hetero) (B). DSerca2^Bmi1homo mice compared with their respective controls (n = 15; each; age-matched animals); C and D: Complete blood count tests of DSerca2^Bmi1homo [10 days post-tamoxifen (Tx)] and control mice (n = 3 each; age-matched animals); E: Blood biochemical analysis of DSerca2^Bmi1homo (10 days post-Tx) and control mice (n = 3 each; age-matched animals); F and G: Weight evolution of DSerca2^Bmi1homo mice (F) and DSerca2^Bmi1hetero mice (G), for 12 days or 6 weeks, respectively (n = 9 each; age-matched animals); H: Glucose tolerance test (10 days post-Tx) comparing DSerca2^Bmi1homo and controls (n = 3 each; age-matched animals). Data are expressed as mean ± SD. ^aP < 0.05. ^bP < 0.01. Homo: Homozygous; hetero: Heterozygous; RBC: Red blood cell; HCT: Hematocrit; HGB: Hotal hemoglobin; MCV: Mean red blood cell size; MCH: Mean corpuscular hemoglobin; MCHC: Mean corpuscular hemoglobin are indicated in C; WBC: White blood cells; PLT: Platelet counts; MPV: Mean platelet volume; NEUT: Neutrophils; MONO: Monocytes; LYM: Lymphocytes and included in D; Tx: Tamoxifen.

Then, we looked deeply into the possible causes of the lethal phenotype induced by Serca2 conditional deletion. Compared with control animals, DSerca2^Bmi1homo mice showed a significant increase in red blood cell (RBC) counts, hematocrit (HCT) and total hemoglobin (HGB) levels (Figure 1C). No differences were found for RBC size (mean red blood cell size), mean corpuscular HGB and the concentration of HGB per RBC (Figure 1C). In addition, no significant differences were found for white blood cells, platelet counts and for mean platelet volume (Figure 1D). DSerca2^Bmi1homo mice showed a significant over-representation of neutrophils, without changes in monocytes, and under-representation of lymphocytes rates (Figure 1D). Furthermore, serum biochemistry revealed that several parameters were significantly lower in DSerca2^Bmi1homo mice than in controls, including glucose (97 mg/dL vs 242 mg/dL), total cholesterol (61 mg/dL vs 105 mg/dL), triglycerides (49 mg/dL vs 92 mg/dL), creatinine (0.17 mg/dL vs 0.3 mg/dL) (Figure 1E) and alanine aminotransferase/glutamic pyruvic transaminase (130 U/L vs 600 U/L) (Supplementary Figure 1); the latter showed the largest difference (4.6-fold reduction in DSerca2^Bmi1homo mice) although this was not statistically significant. Among the parameters analyzed, only bilirubin (direct and total) was higher (2- and 2.4-fold, respectively) in DSerca2^Bmi1homo mice than in controls (Figure 1E). The serum biochemical profile of DSerca2^Bmi1homo mice resulted consistent with a low-to-moderate hepatic or renal impairment or suggestive of malnutrition/malabsorption[17,18]. According to this, increments in HCT (Figure 1C) have been also associated with dehydration[17].

These results, together with the previous finding of moderate keratinization in stomach and esophagus, prompted us to evaluate weight evolution and glucose uptake. Compared with controls, DSerca2^Bmi1homo mice exhibited progressive and significant weight loss, up to 30% at d12 post-Tx (Figure 1F). By contrast, Tx-induction in Serca2^Bmi1hetero mice did not induce significant weight reduction, even after longer post-Tx periods (up to 6 weeks) (Figure 1G). This pronounced weight loss found at d12 post-Tx can be considered critical as it could severely compromise mice survival[19,20].

In addition, glucose tolerance test in DSerca2^Bmi1homo and control mice revealed clear functional differences between the two groups (Figure 1H): DSerca2^Bmi1homo mice showed a less efficient (1.5-2-fold) absorption of glucose in peripheral blood (t = 10-20 minutes, respectively) compared with control mice, and a faster clearance (t = 20-120 minutes). These results strongly suggest that Serca2 deficiency in Bmi1⁺ adult cells courses with malabsorption of glucose in small intestine, reducing body weight to critical levels in a short period of time (< 12d post-Tx).

Bmi1⁺ intestinal cells show increased proliferation in small intestine crypts after Serca2 depletion

As biochemical and physiological analyses indicated a plausible malnutrition/malabsorption phenotype in DSerca2^Bmi1homo mice, we next characterized the Bmi1⁺ populations throughout the digestive tract. We employed the reporter Tmt to label the Bmi1⁺ population both in DSerca2^Bmi1homo mice and in Bmi1-DTmt animals, used as controls (Supplementary Figure 2). The moderate keratinization evidenced in stomach and esophagus prompted us to analyze first both organs by immunofluorescence. In the esophagus, Bmi1⁺ (Tmt⁺) cells were scarce at d5 post-Tx (Figure 2A) in both mouse strains (DSerca2^Bmi1homo and Bmi1-DTmt control mice). In stomach, Bmi1⁺ cells were located in the gastric crypts, as described[21], and the numbers were moderately higher in DSerca2^Bmi1homo mice than in controls, although they seem to present a similar proliferation rate (Ki-67⁺ cells) (Figure 2A). We next focused on the small intestine Bmi1⁺ population (Bmi1⁺ ISC). Immunofluorescence analysis after d5 post-Tx showed that most of the intestinal epithelium was replaced by Bmi1-derived cells (Tmt⁺) in Bmi1-DTmt control mice (Figure 2B). Proliferating cells (Ki-67⁺) were located at the transit-amplifying domain, and no major differences were apparently found between the two groups, or after analyzing individual small intestine segments (Supplementary Figure 2). However, at d7 post Tx-induction, a wider region positive for Ki-67 was observed in DSerca2^Bmi1 intestine compared with Bmi1-DTmt control mice, which was more pronounced in some small intestine sections including duodenum and ileum (Supplementary Figure 3). These results suggest a net and progressive increase in Bmi1⁺ ISC spreading after Serca2 depletion.

Figure 2 Inducible deletion of Serca2 in the Bmi1⁺ populations in the gastric tract. Cryosections prepared from DSerca2^Bmi1homo and control (Bmi1-DTmt) mice [0 day-7 days post-tamoxifen (Tx)]; samples were stained for Ki67 analysis and counterstained with 4’,6-diamidino-2-phenylindole. A: Cryosections from esophagus and stomach, were prepared from 5 days post-Tx mice (n = 4); B: Samples from small intestine, 0 day-7 days post-Tx (n = 3) were prepared (several sections were analyzed for each animal). Representative images. Scale bars: 100 μm.

To further analyze the aforementioned changes in proliferation, we first quantified and characterized Bmi1⁺ cells in isolated crypts. After confirmation of the quality of the crypt-enriched fractions by expression profiling (Supplementary Figure 4), we quantified the Bmi1⁺ population in the crypts of both models by flow cytometry at d5 post-Tx. We found an 8% increase in the Tmt⁺ (Bmi1⁺) population in DSerca2^Bmi1homo compared with the Bmi1-DTmt control mice (Figure 3A); furthermore, Ki-67⁺ cells analysis confirmed a significant increase in the proliferation rate (3-fold) of the Tmt⁺ (Bmi1⁺) population in the small intestine crypts of DSerca2^Bmi1homo mice compared to controls (Figure 3B). According to these results, no significant differences in apoptosis or necrosis rate were found in the total crypt or in the Tmt⁺ (Bmi1⁺) population (Figure 3C).

Figure 3 Impact of conditional Serca2 deletion in the small intestine. A-C: Analysis of crypt-enriched fractions isolated in parallel from DSerca2^Bmi1homo and control animals [5 days post-tamoxifen (Tx)]. A: Estimation of Bmi1⁺ cells (Tmt⁺) by flow cytometry (n = 3 each; independent animals); B: Proliferation analysis (Ki-67⁺) by flow cytometry (n = 3 each; independent animals); C: Evaluation of apoptosis/necrosis rate by flow cytometry (4’,6-diamidino-2-phenylindole/Annexin-V) (n = 3 each; independent animals); D and E: Expression analysis (real-time quantitative polymerase chain reaction) in small intestine segments (duodenum; jejunum; ileum) from DSerca2^Bmi1homo and control mice; Control (n = 5 each; independent animals) and DSerca2^Bmi1homo 5 days (n = 3 each; independent animals) and 7 days (n = 2 each; independent animals) mice. All samples were normalized to levels of the housekeeping gene Hprt. Analysis of a panel of functionally relevant genes in intestinal crypts in crypt-enriched fractions of the indicated animals (n = 3 each; independent animals). D: Analysis of a panel of relevant genes involved in calcium ions regulation in total tissue of the indicated groups of animals (n = 3 each; independent animals); E: Assays were performed three times and data are expressed as mean ± SD. ^aP < 0.05. ^cP < 0.001. FSC: Forward scatter; SSC: Side scatter; DAPI: 4’,6-diamidino-2-phenylindole; Ca²⁺: Calcium ions.

Based on previous studies, Serca2 cardiomyocyte-specific deletions using the myosin heavy chain 6 (Myh6 or αMHC) regulatory sequences to drive Cre expression led to embryonic lethality or severe cardiac dysfunction due to endoplasmic and sarcomeric reticulum stress and apoptosis[12]; in contrast, the Bmi1-CreER driver shows minimal activity in mature cardiomyocytes. Consistent with this, our analyses of DSerca2^Bmi1homo mice at 10 days post-Tx revealed no apparent cardiac abnormalities (nor changes in electrocardiogram analysis) and negligible generation of new cardiomyocytes from the Bmi1⁺ cardiac progenitor cells (B-CPC) population. These data, unlikely underlie the lethal phenotype; in any case, we know that the found phenotype is not the mere consequence of the elimination of B-CPC cells based on Bmi1⁺ depletion with diphtheria toxin[13], but a more complex mechanism that seems to be the result of the interference of DSerca2 with the normal regulation in some target cells/organs. Due to this lack of important evidence it seemed that Serca2 deficit is not essential for our model in the analysis of neuromuscular alterations, although we fully agree that it would be of interest to carry out additional neuromuscular assessments in future studies.

Serca2 deficiency in adult Bmi1⁺ cells alters the calcium regulatory system in the small intestine

In order to define the molecular basis of the phenotype associated with Serca2 deletion, we first compared the expression of a panel of genes in crypt samples along the three regions of the small intestine (duodenum, jejunum and ileum) at d5 post-Tx. No significant differences were found, with the exception of a significant increase in Cdx2 expression in ileum crypts (Figure 3D); in this scenario, the evident increase in proliferation in the Bmi1⁺ population in DSerca2^Bmi1homo mice (Figure 3B) could be associated with the upregulation of Cdx2. Other specific gene markers for crypts such as EphB2(transit amplifying compartment) and Lyz (Paneth cells) were unchanged between groups. Analysis at d7 post-Tx rendered similar results (Supplementary Figure 4). We also tested the expression of a panel of genes (Nhe3, Asbt and Apoa-1) related to the final differentiation of enterocytes and absorptive cell markers; Muc2 and ChgA were used as markers for goblet and enteroendocrine cells, respectively (Supplementary Figure 4). No major changes were globally consistent along the three regions of the small intestine, although some variations were observed between the different segments: This was the case for the upregulation of Villin in duodenum, Apoa-1 and ChgA in jejunum, and a clear increased expression of EphB2 in ileum in DSerca2^Bmi1homo mice compared to Bmi1-DTmt controls. By contrast, Lyz (Paneth cells) was clearly downregulated in ileum, and compatible profile in duodenum and jejunum, in DSerca2^Bmi1homo mice compared to controls (Supplementary Figure 4). A significant up-regulation of Bmi1 expression, albeit transient, was found in jejunum and ileum at d5 post-Tx in DSerca2^Bmi1homo mice, but no significant changes in Cdx2 expression (Supplementary Figure 4).

SERCA2 is the only calcium pump that transports Ca²⁺ from the cytosol to the ER, which represents the main reservoir of intracellular Ca²⁺. To assess the effect of Serca2 depletion on Ca²⁺ homeostasis, we analyzed the expression of key genes involved in the Store Operating Calcium Entry system, including Orai-1 and genes involved in regulation of intracellular Ca²⁺ such as the calcium pump genes Pmca-1 and Pmca-4, which are specific to the small intestine. Differential gene expression showed an upregulated Pmca-1 expression in all gut segments (total tissue) and Orai-1 only in duodenum (clear at the early time point; d5 post-Tx) in DSerca2^Bmi1homo mice as compared with controls (Figure 3E). Conversely, a trend for a reduction was found in Pmca-4 expression levels, more pronounced in duodenum (Figure 3E), in DSerca2^Bmi1homo compared with control mice; this apparent reduction was also observed in the crypts of all gut segments at early analyzed times (d5 post-Tx) in DSerca2^Bmi1homo mice compared with the Bmi1-DTmt control mice (Supplementary Figure 4). Evidence from many cell lineages indicates a strong association between altered Ca²⁺ homeostasis and ER stress response, with SERCA2 reported as a main partner[2]. Then, we analyzed the expression levels of genes related to ER stress in the duodenum, which was the gut region showing the strongest phenotype in Ca²⁺ homeostasis in DSerca2^Bmi1homo mice compared with the Bmi1-DTmt controls. Results showed a trend for downregulation in Bcl2 and Pmaip1 expression (d7 post-Tx), after a transient and moderate up-regulation (d5) in DSerca2^Bmi1homo mice. By contrast, Fatp4, Abcg5 and Abcg8 were overexpressed (Figure 4A). Fatp4 is a member of a family of fatty acid transport proteins that are involved in the translocation of long-chain fatty acids cross the plasma membrane and appears to be the principal fatty acid transporter in enterocytes. Abcg5 and Abcg8 form a heterodimer that mediates magnesium ion- and adenosine triphosphate-dependent transport across the cell membrane, where Abcg5 contributes to limit intestinal absorption, promoting biliary excretion of sterols. Overall, these results suggest that after 5-7 days post-Tx, when the progeny of Bmi1⁺ cells have replaced a substantial part of the gut epithelium, the conditional knockout of Serca2 provokes a moderate ER stress in duodenum; these results would be compatible with the lack of effect on apoptosis/necrosis (Figure 3C).

Figure 4 Serca2 deletion in the intestinal epithelium provokes a fulminant lethal phenotype. A: Expression analysis (real-time quantitative polymerase chain reaction) of a panel of relevant genes involved endoplasmic reticulum stress in total duodenum tissue isolated in parallel from control (n = 3, each); DSerca2^Bmi1homo (5 days; n = 3, each) and DSerca2^Bmi1homo (7 days; n = 2, each) mice. All data were normalized to levels of the housekeeping gene Hprt; B: Electron microscopy analysis of microvilli length in DSerca2^Bmi1homo compared with control mice [5 days-post-tamoxifen (Tx)]. Representative images (50000 ×) are shown and representative example, microvilli are indicated with an orange line, with the estimated length in μm. Below, statistical analysis (n = 4 each); C: Scheme illustrating the Serca2^Villinhomo mice; after Tx-induction exons 2 + 3 of Serca2 are deleted in the adult Villin⁺ lineages (n = 4 each); D: Survival curves of DSerca2^Villinhomo mice (n = 10) compared with corn oil-treated controls (n = 10); E: Serum biochemical analysis of DSerca2^Villinhomo (3 days post-Tx; open circles) or corn oil-treated Serca2^Villin control samples (close circles) (n = 4 each); F: Determination of total fat in feces (% of total weight; n = 4 each, independent animals); G and H: Histological analysis of the esophagus and the esophagus-stomach union of DSerca2^Villinhomo mice (3 days post-Tx) and corn oil-treated controls (n = 3 each; independent animals, several sections were analyzed for each animal). Bar sizes are included. Assays were performed three times and data are expressed as mean ± SD. ^aP < 0.05. ^bP < 0.02. ^cP < 0.002. The online version of the manuscript includes the following information: Electron microscopy analysis of small intestine after conditional deletion of Serca2 in the Bmi1⁺ stem cell compartment (Supplementary Figure 5). ER: Endoplasmic reticulum; Tx: Tamoxifen; SERCA: Sarco(endo)plasmic reticulum calcium ATPases; Lox P: Locus of X-over P1.

Moreover, to address whether additional canonical Ca²⁺ signaling pathways [e.g., calcineurin-nuclear factor of activated T-cells (NFAT), Ca²⁺/calmodulin-dependent protein kinase II, ER stress-unfolded protein response responses] are modified upon Serca2 deletion, we analyzed the expression of representative downstream targets of the calcineurin-NFAT (Rcan1, Cox2) and and CaMK-cyclic adenosine monophosphate-response element-binding protein (CREB) (cFos, Egr1) Ca²⁺ signaling pathways, as well as targets for the unfolded protein response (Hspa5). RT-qPCR analyses were performed in total duodenum fractions at 5 and 7 days after Tx administration. A general downregulation of those transcripts was found in DSerca2^Bmi1homo animals compared with controls, with the most pronounced decreases in Rcan1, Cox2 (Ptgs2), cFos, and Egr1 although only cFos regulation was statistically significant (Supplementary Figure 4). These results suggest, under the conditions analyzed, that Serca2 depletion does not elicit Ca²⁺ handling compensatory effects. This exemplified by the parallel reduction of SERCA2 and Rcan1, although they play a critical role in the regulation of intracellular calcium, but acting through different pathways. All together, these findings confirmed deregulation of Ca²⁺ homeostasis and a moderate ER-stress in the gut after Serca2 deletion. Interestingly, electron microscopy comparative analysis of DSerca2^Bmi1homo mice small intestine revealed no gross alterations as compared with control mice (d5 post-Tx) (Supplementary Figure 5) but significant shortening (25.5%) of microvilli length was evidenced: 0.74 ± 0.13 vs. 0.98 ± 0.15 in DSerca2^Bmi1homo and Bmi1-DTmt control mice, respectively (Figure 4B).

Taken together, these results suggest that conditional Serca2 deletion in the adult Bmi1⁺ intestinal population affects Ca²⁺ regulation of Bmi1⁺ ISC, the transient amplifying compartment (TAC) and the more mature cell types, although with differential impact in the distinct small intestine segments. Accordingly, alteration of TAC regulation is the most likely source of microvilli shortening. Mature Serca2-knockout cells show moderate ER stress, which most probably alters their differentiation dynamics, provoking, in combination of shorter microvilli, a significant deficit in full absorption competence. By contrast, Bmi1⁺ ISC could be proliferatively activated due to the incremented intracellular Ca²⁺, as found in Drosophila melanogaster ISC[7].

Serca2 deletion in Villin⁺ cells aggravates the gastric phenotype

Although the phenotype described seems to be mainly related to the perturbation of the intestinal ISC compartment (because of Bmi1 expression, related to ASC compartments and, at low levels, some mature cell lineages), we attempted to discard a systemic effect. To this aim, we generated a novel mouse strain for a preferential inducible deletion of Serca2 in the gut epithelium (DSerca2^Villinhomo mice) using a well-established and highly preferential gut (Villin) promoter (Figure 4C). Upon Tx induction, we confirmed an efficient reduction (> 75%) of Serca2 expression in the gut epithelium of DSerca2^Villinhomo mice compared to corn-oil induced controls (Supplementary Figure 5). DSerca2^Villinhomo mice batches showed a survival curve comparable to that of DSerca2^Bmi1homo mice, but more severe and accelerated (Figure 4D); DSerca2^Villinhomo mice started dying from day 2 post-Tx and 80% of the animals were dead at day 4, however some surviving animals (20%) remained alive and stable at least for 2 months.

Analysis of food and water intake in DSerca2^Villinhomo mice showed a 10%-15% reduction compared with controls (Supplementary Figure 5). Blood biochemical analyses of perimortem DSerca2^Villinhomo animals (d3-d4 post-Tx) revealed a similar profile to that found in DSerca2^Bmi1homo animals with regards to glucose, total cholesterol and serum proteins, but no reduction in triglycerides; additionally, DSerca2^Villinhomo mice showed a moderate reduction in high density lipoprotein- and low density lipoprotein-cholesterol (Figure 4E). These findings suggested that DSerca2^Villinhomo animals could also be experiencing severe malnutrition/malabsorption. Accordingly, when we evaluated the total fat in feces, we found a significant (2.4-fold) increase (3d post-Tx) in DSerca2^Villinhomo mice compared with controls (Figure 4F). Finally, histopathological analysis of DSerca2^Villinhomo and control mice (3d post-Tx) revealed no major structural alterations in esophagus, esophagus/stomach junctions or stomach and adipose tissue in the peri-stomach/pancreas space (Figure 4G and H). In addition, no significant changes were observed both in proximal and distal intestine compared with control animals (Supplementary Figure 5). Again, the more consistent finding in DSerca2^Villinhomo mice was a moderate keratinization, specifically localized to the esophagus/stomach junctions (Figure 4G and H, Supplementary Figure 5), coincident with the initial findings in DSerca2^Bmi1homo animals.

To sum up, Serca2 deletion in the gastric epithelium (Villin⁺ cells) provokes a fulminant lethal phenotype likely associated with major regulatory alterations mediated by chronic deregulation of cytosolic Ca²⁺ in gastric mature cells. This would be compatible with the results obtained in DSerca2^Bmi1homo animals, which presented with moderate alterations in the proliferation/differentiation equilibrium of Bmi1⁺ ICS. Likewise, after the generation of a large proportion of Serca2-deficient progeny, after 5-7 days post-Tx animals (DSerca2^Bmi1homo mice) start dying most likely due to a significantly reduced absorption capacity associated, in addition, with the appearance of shorter microvilli; the combined factors seem to contribute to the lethal phenotype.

Impact of Serca2 deletion in adult Bmi1⁺ cells in the lymphohematopoietic system

Aiming to evaluate whether the gastric epithelium phenotype found after the conditional Serca2-knockout (DSerca2^Bmi1homo mice) is specific or a preferential one, we studied the lymphohematopoietic system, aiming to reveal whether the conditional deletion of Serca2 could promote an inflammatory state that could contribute to the lethal phenotypes. Supplementary Figure 6 illustrates the distribution of GFP⁺ cells among the different cell lineages in BM of control animals Bmi1^GFP/+. Conditional deletion of Serca2 (d10 post-Tx) induced a moderate but non-significant decrease in total BM cellularity (Figure 5A). Comparative analysis of the major BM cell lineages indicated that granulocytes were significantly under-represented (17%) in DSerca2^Bmi1homo mice and a similar trend, although non-significant, for immature myeloid cells (Figure 5B, Supplementary Figure 6). By contrast, mature T-cell receptor (TcR) β⁺ cells appeared unmodified in DSerca2^Bmi1homo mice (Figure 5C), whereas pre/pro-B cells, and immature and mature B cells showed a clear trend (non-statistically significant) to be over-represented (Figure 5D). This phenotype was similar to that found after recovery, when adult Bmi1⁺ cells were depleted (Supplementary Figure 1)[13] and in the initial analysis of Bmi1 knockout mice. Finally, analysis of BM populations enriched for more primitive progenitors and stem cells (LSK) revealed a moderate, but, again, non-significant increase, in the percentage of LSK with respect to total cellularity in DSerca2^Bmi1homo mice compared to Bmi1-DTmt control animals (Figure 5E); however, after in vivo EdU administration DSerca2^Bmi1homo, LSK did not show an incremented proliferation rate (Figure 5F).

Figure 5 Evaluation of conditional Serca2 deletion in adult bone marrow. A: Bone marrow (BM) cellularity in samples from DSerca2^Bmi1homo [10 days post-tamoxifen (Tx) induction] and control (Bmi1-GFP) animals (n = 2 each); B: Flow cytometry analysis of different BM populations (defined as in Supplementary Figure 6) in DSerca2^Bmi1homo (10 days post-Tx) and control animals (n = 2 each); C and D: Flow cytometry analysis of relevant T (C) and B (D) cell populations in DSerca2^Bmi1homo mice (10 days post-Tx) and control animals (n = 2 each); E and F: Evaluation of the immature Lin^- Sca1⁺ ckit⁺ (LSK) population representation (per femur) (E) and its in vivo rate of proliferation (5-ethynyl-2’-deoxyuridine +) (F) in DSerca2^Bmi1homo LSK (10 days post-Tx) and control LSK samples (n = 3 each; age-matched animals). Assays were performed two times and data are expressed as mean ± SD. ^aP < 0.05. The online version of this article includes the characterization, by flow cytometry, of the GFP⁺ population (Bmi1-GFP) distribution in bone marrow. RBC: Red blood cell; TCR: T-cell receptor; NK: Natural killer; CD: Cluster of differentiation; IgM: Immunoglobulin M; EdU: 5-ethynyl-2’-deoxyuridine.

Therefore, we concluded that Serca2 deletion does not seem to promote a significant chronic proliferation of BM stem cells and early progenitors (LSK populations) but alters their differentiation profile, mainly B-cells (over-represented) and granulocytes (less frequent). On the contrary to gut, depletion of Serca2 in BM stem cell enriched population (Bmi1⁺ cells) does not promote their (LSK) proliferative status. Given the initial anatomopathological finding of moderate peripheral infiltration in the pancreas of DSerca2^Bmi1homo mice, we also analyzed the peripheral lymphoid organs (Sp and LN) and the Thy. After induction (10d post-Tx), global cellularity in Sp and LN was lower (34% and 29%, respectively) in DSerca2^Bmi1homo mice than in Bmi1-DTmt control mice; this was even more pronounced in Thy (78% reduction) (Figure 6A). We found variations in some thymocyte subsets (5.5- and 3.2-fold upregulation for cluster of differentiation 4 (CD4⁺) and 8 (CD8⁺), respectively, and 4-fold over-representation in CD8^- [double-negative (DN)] in DSerca2^Bmi1homo mice compared with controls; on the contrary, CD4⁺ CD8⁺ [double-positive (DP)] population showed a very clear percental reduction (45%) in DSerca2^Bmi1homo mice (Figure 6B). Then, the reduced cellularity in DSerca2^Bmi1homo Thy with respect to controls (Figure 6A) seemed to be mainly the consequence of the important reduction in DP T cells (45%), the main thymocyte compartment in size terms (Figure 6B). The reduction in DP cells increased the percentage of DN cells, provoking and important increment in single positive and double negative (SP/DN) ratios (Figure 6C). In addition, DN1 (CD44⁺ CD25^-), which is the ontogenically the more primitive subpopulation within the DN subpopulations, showed over-representation (not statistically significant), in DSerca2^Bmi1homo Thy (Figure 6D). Conversely, no significant differences were found for mature T cells (both CD4⁺ or CD8⁺) or for B cells (B220⁺), in Sp or LN (Figure 6E).

Figure 6 Impact of Serca2 conditional gene deletion in the lymphohematopoietic adult Bmi1⁺ cell lineages. A: Total cellularity in spleen (Sp), lymph nodes (LN) and thymus (Thy) of DSerca2^Bmi1homo [10 days post-tamoxifen (Tx) induction] (orange bars in all analyses) and control mice (blue bars in all analyses); B-D: Flow cytometry analysis of different thymic subpopulations including single-positive (SP) [cluster of differentiation 4 (CD4⁺) or 8 (CD8⁺)], double-positive (DP) (CD4⁺ CD8⁺) and double-negative (DN) (CD4^- CD8^-); B: Percental representation of SP, DP and DN populations; C: Percental representation of SP/DN ratios; D: Percental representation of DN subpopulations; E: Evaluation of T (CD4⁺ or CD8⁺) and B (B220⁺) cells in Sp and LN (n = 3 each; independent animals); F and G: Representative histograms (F) and mean fluorescence intensity (G) of CD5 expression in DP, SP and DN thymocyte subsets; notice the absence of CD5^high and reduced CD5^low cells in DSerca2^Bmi1homo Thy compared with controls. All comparisons were carried out in DSerca2^Bmi1homo (10 days post-Tx) compared with control mice (n = 3 each; independent animals); H: Proportions of total thymic T-cell receptor-αβ^high CD69⁺ positive selected cells; I: Gated dot plots for evaluating negatively selected thymocytes (Caspase3⁺ CD5^high CD69⁺) in both control and DSerca2^Bmi1homo thymuses; J: Thy histology of both control and DSerca2^Bmi1homo mice using an anti-pan cytokeratin (PanCK) marker for detecting cortical and anti-keratin 5 (K5) cytokeratin mAb for medullary epithelium; note the reduced cortical area and the enlarged medullary compartment in DSerca2^Bmi1 Thy compared with controls. Scale bar: 50 μm; K: Proportions of different thymic epithelial cells (TEC) subsets defined by the expression of major histocompatibility complex (MHC) II and ulex europaeus agglutinin (UEA)-1 cells markers in the total TEC (epithelial cell adhesion molecule⁺ CD45^-): Cortical TEC^low (UEA-1^- MHCII^low), medullary TEC^low (UEA-1⁺ MHCII^low), cortical TEC^high (UEA-1^- MHCII^high) and medullary TEC^high (UEA-1⁺ MHCII^high). Animals used for F-I and K were n = 3, controls; n = 2, DSerca2^Bmi1homo; age-matched animals; L: Numbers of Aire⁺ cells respect to K5⁺ area in thymic sections of DSerca2^Bmi1homo thymuses compared with controls ones (n = 2 each; age-matched animals). Data are expressed as mean ± SD. ^aP < 0.05. ^bP < 0.02. TEC: Thymic epithelial cell; Sp: Spleen; LN: Lymph nodes; Thy: Thymus; CD: Cluster of differentiation; DN: Double-negative; DP: Double-positive; TcR: T-cell receptor; K5: Keratin 5; PanCK: Pan cytokeratin; mTEC: Medullary thymic epithelial cell; cTEC: Cortical thymic epithelial cell.

We also monitored the expression of CD69 (C-type lectin) for early activation and CD25 (IL2Ra subunit) (Supplementary Figure 7), as T- and B-cell markers for early and intermediate/Late activation, respectively. Both in Sp and LN (clearer in LN) we found moderate early overactivation in CD4⁺, CD8⁺ and B220⁺ cells; intermediate/Late activation showed a low level of overactivation. Finally, a clear trend for an increase in monocytes and monocyte-derived macrophages (F4/80^high/Low CD11b⁺) was found in DSerca2^Bmi1homo mice, which was more pronounced in Sp (Supplementary Figure 7); in addition, no significant alterations in eosinophils/neutrophils (F4/80^- CD11b⁺) were found both in LN and Sp (Supplementary Figure 7).

Together with the reduced thymic cellularity and the altered T-cell development profile DSerca2^Bmi1homo mice exhibited a strong reduction in CD5 expression in all thymocyte subsets, affecting both the proportion of CD5⁺ cells (Figure 6F) and their mean expression level (Figure 6G). In developing thymocytes CD5 negatively modulates T-cell signaling strength through the TcR[22]; in addition, it has been demonstrated that CD5 and TcR show mutual co-regulation[23,24]. Accordingly, we analyzed the positive and negative T-cell selection processes in DSerca2^Bmi1 Thy. The proportion of positively-selected thymocytes (TcRab^high CD69⁺) was significantly higher in DSerca2^Bmi1homo compared with in control animals (Figure 6H), whereas negatively-selected thymocytes (Caspase3⁺ CD5^high CD69⁺) were practically non-existent (Figure 6I) due to the minimal proportion of CD5^high cells in DSerca2^Bmi1 Thy (Figure 6F and G).

Because similar altered thymocyte differentiation profiles, dependent on an altered thymic epithelial microenvironment, have been previously described in mouse models[25], we evaluated the thymic epithelium in DSerca2^Bmi1homo mice, compared with controls. Immunohistochemistry analysis revealed a significant disorganization and reduction of the thymic cortical area (PanCK⁺) and an enlarged medullary area (K5⁺) consisting of separated and slightly disorganized epithelial network (Figure 6J) in DSerca2^Bmi1homo Thy. Remarkably, no significant changes were observed in the proportions of total cortical thymic epithelial cells (cTECs) vs medulary thymic epithelial cells (mTEC) (Supplementary Figure 7), but in both areas, immature (cTEC^low, mTEC^low) TEC were increased and the proportions of both cTEC^high and mTEC^high reduced (Figure 6K). Also, the number of Aire⁺ mTEC is significantly diminished (Figure 6L, Supplementary Figure 7).

In conclusion, despite the profound thymic alterations found, involving a clear stroma dysfunction, peripheral B and T cells do not show any critical dysfunction. Although further studies are necessary, these results globally indicate that no acute/aggressive inflammatory process is triggered by Serca2 deletion in the lymphohematopoietic compartment that could significantly contribute to the lethal phenotype described. On the contrary, malnutrition/malabsorption has also been described to be involved in the changes observed in T-cell differentiation[26-28].

DISCUSSION

A key role for high cytosolic Ca²⁺ in promoting ISC proliferation during adaptation to diet and in response to oxidative stress has been demonstrated in Drosophila melanogaster, with the calcium transporter SERCA defined as one of the key regulatory players[7,27]. Similarly, in vertebrates, high plasma L-glutamate levels and the glutamate receptor (GluR) homolog are strongly associated with Ca²⁺ signaling, being regulated by diet[28]. Moreover, glutamine metabolism has been proposed for a relevant role in malignancy[26], and Ca²⁺ signaling distortion has been described in association with many human tumors. In addition, alterations of intracellular Ca²⁺ management as well as dysfunction of Serca2 has been associated to transformation. These data would be consistent with the roles identified in Drosophila melanogaster[7,27] and suggest that chronic metabotropic GluR activation in stem and progenitor-like cells could drive deregulated proliferation, contributing to transformation in some cancers. Here, we aimed to explore whether high cytosolic Ca²⁺ could play a similar role in ISC in mammals, evaluating the impact of the Serca2 deletion in murine adult Bmi1⁺ compartments.

Bmi1 is a widely expressed marker of adult stem and progenitor cell populations in adult mice. Specifically in the gut, Bmi1 identifies an ISC population located at the + 4 position in the crypts of the small intestine[29]. Bmi1⁺ ISC has demonstrated to be particularly important for maintaining intestinal homeostasis, acting as reservoir, and for regenerative processes[30-35], being proliferative activated upon injury[31]. We previously demonstrated that ablation of Bmi1⁺ cells in adult mice, using the same activation module in homeostasis, had no significant impact on animal survival[13], probably based on the significant recovery capacity of highly proliferative organs such as BM and the gut. It was therefore unexpected that targeted deletion of Serca2 in Bmi1⁺ cells resulted in the severe lethal phenotype described. This phenotype suggested that the deficit in SERCA2 critically interferes with the homeostatic regulation of certain adult Bmi1⁺ stem and progenitor cell compartments, putatively altering also progeny generation.

Our results conclusively indicate that conditional deletion of Serca2 in intestinal Bmi1⁺ ISC promote their proliferation but provoking a severe deficit in small intestine absorption, concomitant with a significant microvillus shortening. Altered blood biochemistry parameters and Thy composition suggested that animals develop severe malabsorption, which is the most probable cause for early death due to the critical weight reduction (up to 30% at d12 post-Tx) of DSerca2^Bmi1homo mice compared to appropriated controls. Malnutrition, as a consequence of a severe deficit in intestinal absorption, could also contribute indirectly to the important changes observed in T-cell differentiation[28], although no in the periphery. Main results were further corroborated in DSerca2^Villin mice, which, upon Tx-induction, provokes a highly preferential deletion of Serca2 along the whole digestive epithelium. DSerca2^Villinhomo mice developed a fulminant lethal phenotype (deaths starting at d2 post-Tx) accompanied by compatible blood biochemistry alterations and impaired lipid absorption. Taking together, the development of a severe gastric phenotype seems to be generated when mature cells, derived from Bmi1⁺ ISC, and deficient in SERCA2, colonize a high proportion of the gut epithelium with a severe alteration in their absorption capabilities. These results point to the gastric epithelium as the most sensitive to chronic alterations of cytosolic Ca²⁺ adult compartments.

It is widely known that Leucine rich repeat containing G protein-coupled receptor 5 (Lgr5⁺) and Bmi1⁺ ISC are the main ISC subpopulations characterized in mice, acting in functional equilibrium[31-35]. Most recently, it has been demonstrated that in the early phase of murine development, Bmi1⁺ ISC are the predominant epithelial population in the gut, representing a highly proliferative population; during maturation, Bmi1⁺ ISC transition to quiescent states as Lgr5⁺ ISC emerge[32]. In adults, a remarkable plasticity has been ascribed to both ISC populations: Bmi1⁺ ISC are able to transdifferentiate to Lgr5⁺ ISC when the intestinal epithelium is damaged or when Lgr5⁺ ISC are ablated[33,34]. Using Bmi1-DTmt animals we confirmed that Tmt⁺ (Bmi1⁺)-ISC progeny (> d5 post-Tx induction) replaced a very significant proportion of the crypts, and that the conditional deletion of Serca2 in mice (DSerca2^Bmi1homo) did not seem to provoke significant alterations in small intestine turnover. However, direct in vivo evaluation of the proliferation rate (Ki-67⁺) in crypts showed that Serca2 deletion significantly enhanced (3-fold) the proliferation rate of Tmt⁺ (Bmi1⁺)-ISC. A moderate increase in total Tmt⁺ (Bmi1⁺)-ISC numbers (approximately 8%) was also confirmed in DSerca2^Bmi1homo mice (d5 post-Tx). This robust proliferative rate of Tmt⁺ (Bmi1⁺)-ISC has been previously reported in the developing intestine and upon injury[33,34], being non-canonical Wnt pathway the key activator in the first scenario[32]. In this sense, Wnt pathways and Ca²⁺ signaling are coupled in mammalian cells, which suggest that high cytosolic Ca²⁺ could be a missing link of all these proliferative activation states associated with human pathology; further research will be necessary to address this issue.

Similar to the role defined in the Drosophila melanogaster gut, our results strongly suggest that high cytosolic Ca²⁺ is associated with enhanced adult mouse Bmi1⁺ ISC proliferation rate. However, although the role of cytosolic Ca²⁺ is critical for ISC homeostasis, it seems not to be a general mechanism in mouse ASCs. In this sense, it has recently been demonstrated that quiescent hematopoietic stem cells (HSC) are mainly regulated by cyclin-dependent kinases 4/6 and cytosolic Ca²⁺, where high concentrations of cytoplasmic Ca²⁺ are linked to HSC quiescence[36]. Our results are compatible with this notion, as high cytoplasmic calcium Ca²⁺ neither promote proliferation of the LSK population nor augmented the LSK pool. Most likely, in mammals, cytosolic Ca²⁺ is an important regulator for many ASC, but it must be coordinated with additional compartment specific regulation layers. In addition, found phenotype in gut epithelium could reflect those associated to tissues/organs with higher turnover (dermal epithelia, gut epithelia, etc.) and whose stem cell compartment would be associated to Bmi1 positiveness.

Furthermore, we carried out a comparative analysis of the molecular and cellular mechanisms involved in the lethal gastric phenotype of DSerca2^Bmi1homo mice using specific markers for duodenum, jejunum and ileum samples. No relevant differences were found for enterocytes, selected absorptive functions, globet and enteroendocrine cells; only Apoa-1 and ChgA showed a modest upregulated expression in jejunum. Analysis of other general markers including Villin, EphB2 (transit amplifying zone) and Lyz (Paneth cells) demonstrated a moderate up-regulation of EphB2 and Villin in duodenum and ileum, respectively. According to this, a recent study has established a correlation between the increased levels of long-chain polyunsaturated fatty acids and the expression of EphB2 in ileum[34]. Therefore, we hypothesized that the observed up-regulation of EphB2 could be related to the altered lipid metabolism in DSerca2^Villinhomo mice.

Finally, analysis of key calcium regulatory genes revealed increased expression Orai-1 and Pmca-1 in different gut segments, as well as ER stress response genes up-regulation in duodenum. Up-regulation of specific small intestinal calcium pump as Pmca-1 could be a compensatory mechanism to Serca2 deficiency. The absence of the expected induction of some of the canonical Ca²⁺-responsive transcripts (Rcan1, Ptgs2, cFos, Egr1 and Hspa5) in DSerca2^Bmi1homo duodenum can be explained by several, non-mutually exclusive mechanisms: First, loss of SERCA2 may not simply increase cytosolic Ca²⁺ but deeply distort Ca²⁺ dynamics abolishing the oscillatory patterns required for efficient nuclear activation of NFAT and CaMK/CREB signaling. Second, chronic ER and metabolic stress may induce global transcriptional and translational repression, leading to reduced levels of many immediate-early genes. Third, Serca2 does not apparently elicit Ca²⁺ handling compensatory effects (d5/d7-post-Tx). On the contrary, concomitant reduction of SERCA2 and Rcan1, although they both play a critical role in the regulation of intracellular calcium, but acting through different pathways. Fourth, a transient early induction of these transcripts cannot be excluded, as the present analysis was performed only at 5 and 7 days post-induction. Taken together, these findings indicate that Serca2 deficiency disrupts the temporal and spatial pattern of Ca²⁺ signaling rather than activating some of the Ca²⁺-dependent pathways, that could compensate the phenotype. Future studies assessing nuclear localization of NFAT and phospho-CREB, splicing of X-box binding protein 1, and real-time Ca²⁺ dynamics in crypt cells will be required to go deeply in the evaluation of this hypothesis and delineate more precisely the mechanisms linking disturbed Ca²⁺ homeostasis DSerca2^Bmi1 with transcriptional reprogramming of the intestinal epithelium.

There are several relevant examples of mouse models that develop similar early deleterious phenotypes associated with severe malnutrition/malabsorption but apparently involving different pathways. The combined phenotype defined in DSerca2^Bmi1 and DSerca2^Villin mice (enhanced proliferation of ISC combined with a lethal malabsorption) could be partially comparable to the deletion of the proendocrine transcription factor Ngn3 in the small intestine, which provokes hyperproliferation in crypt populations, more active turnover and with shorter microvilli, what was related to impaired lipid absorption and reduced weight gain[19]. Therefore, this strongly suggests that Ngn3 and cytosolic Ca²⁺ should be involved in similar signaling pathways. Similar to Ngn3 knockout model, other severe gastric phenotypes originate from an imbalance in the proliferation-differentiation axis, at different levels, but impairing physiological absorption of nutrients. None of them have described a significant alteration of Ca²⁺ homeostasis, therefore the current study could provide a new insight into alternative targets.

Finally, we conclude that the phenotype found in BM could be the mere consequence of BM repopulation recovery dynamics after damage (Bmi1⁺ cells alterations or Ca²⁺ deregulation), but the thymic phenotype is also compatible with malnutrition, associated with severe deficit in intestinal absorption[33]. Further studies must broach if the described thymic alterations are a consequence of direct effects of the SERCA lack or due to the malnutrition observed in DSerca2^Bmi1homo mice. There are, however, other plausible factors associated with cytosolic Ca²⁺ dysregulation in DSerca2^Bmi1 mice, that could also contribute to T cell dysregulation. TcRs and major histocompatibility complex loaded with peptides formed complexes, which induce the increase of cytosolic Ca²⁺ that return to basal values with the concourse of SERCA[37,38]. Therefore, the increased chronic cytosolic Ca²⁺ in DSerca2^Bmi1homo mice has a marked negative impact in CD5 expression, as demonstrated for Rag expression; then reduced CD5 expression affects the thymic negative selection procedure, favoring the increase the proportion of positively-selected cells[39]. These findings are similar to the phenotype described for CD5 knockout mice[40]. Furthermore, the pronounced depletion of DP thymocytes, associated with a noteworthy reduction in the thymic cortical area and altered TEC development, has been described in other mutant mice[40]. Overall, our results indicated the absence of an acute/aggressive inflammatory process triggered by Serca2 conditional knockout in the lymphohematopoietic compartment that would importantly contribute to the lethal phenotype.

Strengths and limitations

The combined used of Serca2^Bmi1 and Serca2^Villin strains has allowed us to demonstrate a critical role for cytoplasmic Ca²⁺ regarding the regulation of the Bmi1⁺ ISC proliferation rate and the equilibrium of proliferation/differentiation, in good agreement with Drosophila melanogaster. Serca2 conditional deletion in the adult Bmi1-lineages showed a substantially increased sensitivity in the gut, inducing a lethal phenotype, mainly associated with severe malnutrition/malabsorption. In clear contrast with previous mouse models of malnutrition, Serca2 deficiency does not seem to trigger equivalent pathways, so further research would be needed for a deeper comprehension of the specific pathways impacted by high cytoplasmic Ca²⁺ in correlation with the severe enterocyte absorption deficit and the regulation of thymic populations.

CONCLUSION

Chronic high cytoplasmic Ca²⁺ disrupts the homeostatic control of gut mouse Bmi1⁺ ISC and thymic immature populations, altering their proliferation/differentiation equilibrium. This is the main cause of the severe deficit in absorption described, that provokes the dead of animals. Our findings support that cytoplasmic Ca²⁺ homeostasis critically regulates intestinal Bmi1⁺ stem cell behavior, while its impact on hematopoietic appears limited or context-dependent, suggesting that this mechanism cannot be generalized to all stem cells compartments.

ACKNOWLEDGEMENTS

We want to thank all the present and past members of our laboratories for their dedicated work and discussion aimed at elucidation of the mechanisms involved by Serca2 in small intestine homeostasis. We wish also to thank to Fischer T and García MA (DIO-CNB) for help with the glucose tolerant evaluation, and McCreath K for English editorial work.