Functional and molecular profiling of fasted piglets reveals decreased energy metabolic function and cell proliferation in the small intestine

Abstract

The small intestine requires energy to exert its important role in nutrient uptake and barrier function. Pigs are an important source of food and a model for humans. Young piglets and infants can suffer from periods of insufficient food intake. Whether this functionally affects the small intestinal epithelial cell (IEC) metabolic capacity and how this may be associated with an increased vulnerability to intestinal disease is unknown. We therefore performed a 48-h fasting intervention in young piglets. After feeding a standard weaning diet for 2 wk, 6-wk-old piglets (n = 16/group) were fasted for 48 h, and midjejunal IECs were collected upon euthanasia. Functional metabolism of isolated IECs was analyzed with the Seahorse XF analyzer and gene expression was assessed using RNA-sequencing. Fasting decreased the mitochondrial and glycolytic function of the IECs by 50% and 45%, respectively (P < 0.0001), signifying that overall metabolic function was decreased. The RNA-sequencing results corroborated our functional metabolic measurements, showing that particularly pathways related to mitochondrial energy production were decreased. Besides oxidative metabolic pathways, decreased cell-cycle progression pathways were most regulated in the fasted piglets, which were confirmed by 43% reduction of Ki67-stained cells (P < 0.05). Finally, the expression of barrier function genes was reduced upon fasting. In conclusion, we found that the decreased IEC energy metabolic function in response to fasting is supported by a decreased gene expression of mitochondrial pathways and is likely linked to the observed decreased intestinal cell proliferation and barrier function, providing insight into the vulnerability of piglets, and infants, to decreased food intake.

NEW & NOTEWORTHY Fasting is identified as one of the underlying causes potentiating diarrhea development, both in piglets and humans. With this study, we demonstrate that fasting decreases the metabolism of intestinal epithelial cells, on a functional and transcriptional level. Transcriptional and histological data also show decreased intestinal cell proliferation. As such, fasting-induced intestinal energy shortage could contribute to intestinal dysfunction upon fasting.

INTRODUCTION

The gastrointestinal tract is the gateway to the body: it enables digestion and selective uptake of nutrients, while simultaneously maintaining a barrier against the outside milieu. The intestine’s absorptive and barrier functions are actualized by intestinal epithelial cells (IECs), a type of cell that is continuously generated from stem cells in the crypts, which migrate to the villus tip as they mature, in a 5- to 7-day cycle (1). The intestinal epithelium forms a physical barrier to protect the host, with IECs comprising a tight layer with actively maintained cell-cell contact and specialized cells that secrete mucus to create a protective layer on top of the IECs (2). To perform its complex role, the intestine constitutes an enormous surface area, divided into various longitudinal regions with distinct functions. Fermentation of poorly digestible nutrients occurs distally, in the colon, which is also important for water and electrolyte transport (3). Digestion and uptake of nutrients mainly occur in the small intestine (3). Much of digestion takes place in the first part of the small intestine, the duodenum, whereas nutrient uptake primarily happens in the jejunum, the middle part of the small intestine, and this is therefore an important site to study functional effects of nutritional interventions (3).

Sustaining intestinal functions requires an enormous amount of energy, which is mainly produced in the mitochondria. Mitochondria consume oxygen in the oxidative phosphorylation (OXPHOS) pathway to oxidize reducing equivalents that are generated from substrates, to ultimately generate the cellular energy carrier: ATP. Energy demand in the intestine doubles upon intake of a meal, as can be seen from the 25% increase in resting metabolic rate for humans (4), and 15–20% for growing pigs (5). Inability to produce sufficient energy to answer to intestinal demands can impair intestinal barrier function. For example, inhibition of mitochondrial ATP production was shown to induce internalization of the tight junction protein claudin 7, resulting in decreased barrier function (6). Intestinal permeability was also shown to be increased by the widely used drugs NSAIDs, as a likely consequence of induced mitochondrial uncoupling, decreasing energy efficiency of ATP production per mole substrate (7–9). Furthermore, based on gene expression profiling, an increased ATP utilization has also been proposed to explain a decreased intestinal barrier integrity (10). Together this indicates that an adequate mitochondrial energy production is essential to maintain intestinal barrier and IEC function.

To produce energy, mitochondria require substrates that in the intestine can originate either from the arterial blood supply or directly from the diet (11). That luminal substrates are essential for intestinal mitochondrial energy production is evidenced by prolonged fasting in mice, which was found to decrease expression of electron transport chain (ETC) and TCA genes in the intestine (12). Reduced availability of luminal substrates may impair intestinal function by reducing the ability of IEC mitochondria to produce sufficient ATP, which is needed for nutrient uptake and barrier integrity. In animal husbandry, a limited availability of substrates from the diet is frequent in weaning, because it is mostly associated with a check in food intake (13, 14). This is accompanied with an increased susceptibility for intestinal disease, as evidenced by the high incidence of diarrhea at this time (15). Globally, pig diarrhea accounts for large losses of lives and industrial profits each year, as indicated by several local studies (16–18). Similarly in humans, childhood diarrhea accounts for nearly 1.7 billion cases every year, leading to over half a million deaths in children under 5 yr of age (19). Malnutrition is identified as one of the underlying causes potentiating diarrhea development (19, 20). However, it is unclear whether limited luminal substrate availability, and consequent decreased mitochondrial ATP production, causally contributes to the increased susceptibility of piglets and infants to develop intestinal diseases, because the effects of weaning and malnutrition are mostly studied as a whole, without isolating the specific contribution of fasting to these complex phenomena. In addition, the evidence for decreased mitochondrial function upon fasting is currently primarily obtained using indirect measurements of mitochondrial function in the intestine, using gene expression analysis as a tool (e.g., see Refs. 12 and 21). However, gene expression does not necessarily accurately reflect altered mitochondrial oxidative capacity and metabolic flux. Furthermore, it is not clear whether fasting impacts IEC ATP production as a whole, or whether mainly mitochondrial ATP production is impacted, and glycolytic ATP production largely compensates for this. Finally, it is unclear which IEC functions are primarily altered in association with decreased mitochondrial function and whether these may explain an increased vulnerability of the intestine to infection and diarrhea. Thus, we studied whether and how fasting functionally affects IEC energy metabolism in vivo and with which altered molecular processes this is primarily associated. For this, we weaned piglets and fed them a standard weaning diet for 2 wk, and then exposed them to a 48-h fasting challenge to isolate the effects of fasting from other weaning-associated challenges. Midjejunal IECs were isolated to measure mitochondrial respiration and glycolytic flux, using a previously optimized method (22). In addition, molecular processes affected by the 48-h fast were investigated in midjejunal scrapings using RNA-sequencing and confirmed using immunohistochemistry.

MATERIALS AND METHODS

Animals

All experiments and methods were performed in accordance with the local and (inter)national guidelines. The Central Authority for Scientific Procedures on Animals (CCD), and the Animal Welfare Body (IvD) approved the protocol of the experiment (AVD1040020209884), which was in accordance with the Dutch law on animal experimentation and the European Directive 2010/63/EU on the protection of animals used for scientific purposes. The experiment was carried out in compliance with the ARRIVE guidelines (http://www.nc3rs.org.uk/page.asp?id=1357). The study described here was started together with another study, which ran in parallel until weaning. Thirty-two gilts (Topigs Tempo x TN20) were selected from 14 litters on a commercial farm, from sows with a parity range between one and seven (parity 1, 2 sows; parity 3, 4 sows; parity 4, 3 sows; parity 5, 2 sows; parity 6, 2 sows; parity 7, 1 sow). Cross-fostering was minimized and if needed only applied to male piglets. Apart from Lianol basdiar (Ardol BV, Susteren, The Netherlands) in the first postnatal week, the piglets did not receive creep feed. Piglets were weaned between postnatal day 22 and 26 and were transferred to CARUS, the research facility of Wageningen University. To select gilts, all female piglets in the litter were weighed, and only the piglets with ±1 standard deviation (SD) from the mean were selected. At CARUS, piglets were allocated to the pens based on the sow’s parity, the farrowing date, the genetic background, and their body weight at weaning, to minimize variation between pens and treatments. Piglets were pair-housed in 2 m² partially slated pens, and temperature was kept between 24 and 26.5°C with a humidity of 65%, lights and radio were switched on between 0700 and 1900. Lights were dimmed to 5%, and radio was off from 1900 till 0700. Health and welfare were assessed visually thrice a day during feeding, and fecal consistency score was assessed (score of 1–5, 1 = liquid diarrhea, 5 = hard feces). The feeds were produced by Research Diet Services (RDS BV, Wijk bij Duurstede, The Netherlands) in a single batch in the form of pellets (Table 1). Piglets were fed at 2.4 times the maintenance energy requirement (NEm = 0.29 MJ NE/kg body wt^0.75), divided over three equally portioned meals a day at 0730, 1230, and 1700. Piglets were weighed at the start of the experiment and then every week to adjust feed intake to their current weight. Water was available ad libitum throughout the entire study period. Forty-eight hours before the end of the experiment, piglets either continued to receive feed thrice a day at 2.4 times NEm, or were fasted until the end of the study period. At the end of the study period, piglets were sedated using intramuscular injection of zoletil + xylazine (5:2 ratio, 1 mg/10 kg body wt) and euthanized by lethal injection with pentobarbital in the ear vein (24 mg/kg body wt).

Table 1.

Ingredient and calculated nutrient composition of the starter diet

Components	Starter Diet†
Ingredient composition (%)
Soybean meal	16.5
Sunflower meal	6
Wheat	27.1
Barley	25
Corn	20
Soya oil	1.3
l-Lysine HCl	0.52
l-Threonine	0.16
dl-Methionine	0.15
l-Tryptophan	0.035
l-Valine	0.02
Premix‡	0.5
CaCO3	1.3
Ca(H2PO4)2	0.9
NaCl	0.4
Citric acid	0.1
Phytase	0.0025
Total	100
Nutrient composition, g/kg
Dry matter	871.6
Crude protein	181
Crude Fat	39.3
Crude ash	52.9
Starch	433.6
Sugars	35.7
NE, MJ/kg	10.1

NE, net energy.

^†Nutrient composition was calculated based on ingredient composition and table values for the composition of the ingredients (23); ‡Supplied per kilogram of feed: retinyl acetate, 10,000 IU; cholecalciferol, 2,000 IU; dl-α-tocopherol, 40 mg: menadione, 1.5 mg; thiamine 1.0 mg; riboflavin, 4 mg; pyridoxin-HCl, 1.5 mg; cyanocobalamin, 20 µg; niacin, 30 mg; d-pantothenic acid, 15 mg; choline chloride, 150 mg; folic acid, 0.4 mg; biotin, 0.05 mg; iron(II)sulfate monohydrate, 331 mg; copper(II)sulfate pentahydrate, 80 mg; manganese(II)oxide, 49 mg; zinc sulfate monohydrate, 194 mg, potassium iodate, 1 mg; sodium selenite, 0.56 mg.

Small Intestinal Cell Isolation

The method for IEC isolation was previously optimized for colonocyte isolation (22), but here successfully applied to small intestinal IEC isolation. After excision from the abdominal cavity, the entire intestine was separated from the mesentery, and the small intestine was identified. The small intestine was spread out and 50% of the intestinal length was identified, which corresponds to midjejunum. A 16-cm segment distal of the 50% point was used for histological sample preparation and mucosal scrapings. A 30-cm segment proximal to the 50% point was removed and placed in aerated modified Krebs–Henseleit buffer (KHB) containing 5 mM glucose (No. K3753, Sigma Aldrich, Sigma-Aldrich, St. Louis, MO, hereafter referred to as modified-KHB), supplemented with 2.5 g/L BSA (No. A7906, Sigma-Aldrich, St. Louis, MO). When all samples were obtained, intestines were rinsed thoroughly with modified KHB and inverted. Using dialysis clamps (No. Z371092, Sigma Aldrich, St. Louis, MO) a sac was created by filling the inverted intestine with modified-KHB, which was then placed in 118 mM NaCl, 4.7 mM KCl, 1.2 mM MgSO₄, 1.2 KH₂PO₄, 10 mM HEPES (Ca²⁺-free KHB, pH 7.4), supplemented with 10 mM DTT, 20 mM EDTA, and 2.5 g/L fatty-acid free BSA (No. 3117057001, Sigma-Aldrich, St. Louis, MO) to further wash the mucus away. The intestines were incubated in this wash buffer for 20 min in an oscillating water bath at 37°C. The buffer was then discarded, and the jejunal sacs were reverted and filled with Ca²⁺-free KHB buffer containing, 10 mM DTT, 400 U/mL hyaluronidase type IV (No. 3884, Sigma-Aldrich, St. Louis, MO), and 2.5 g/L fatty acid-free BSA. After a 15-min incubation, the intestinal sacs were gently massaged for 15 s, after which the content of the sacs was collected in 50-mL tubes. The cells were then passed through a 40-μm cellulose filter to remove debris and large tissue pieces. The collected cells were washed twice using modified-KHB containing 2.5 g/L BSA and once with pH-balanced XF DMEM assay medium supplemented with pH-balanced 10 mM XF glucose, 2 mM XF glutamine, and 1 mM XF pyruvate. Cells were spun down at 400 g for 5 min. Then cells were counted using the Bürker chamber, and cell viability was assessed by staining cells with ViaStain (No. CS2-0106, Nexcelom Bioscience, Lawrence, MA) and imaging them using the Cellometer K4 (Nexcelom Bioscience, Lawrence, MA). Cell yield did not differ between groups, and cell viability was above 80% for all isolated cell populations (Supplemental Fig. S1; see https://doi.org/10.6084/m9.figshare.21303750).

Metabolic Flux Analysis

Isolated IECs were plated in an XF96 cell plate that was coated with Cell-Tak (No. 354240, Corning, New York, NY) according to the manufacturer’s protocol, no longer than 1 wk before the assay. Per XF Extracellular Flux analysis run, IECs isolated from eight piglets were included. A standard curve was included on every plate to allow better normalization within and between plates, according to a previously published protocol (24). To generate the standard curve, IECs from all eight pigs were pooled and plated at concentrations ranging from 75,000 to 300,000 cells/well in 50 µL pH 7.4 balanced XF DMEM assay medium supplemented with 10 mM XF glucose, 2 mM XF glutamine, and 1 mM XF pyruvate. The IECs isolated from individual pigs were plated at 140,000 cells/well in eight replicates in the same medium. Cells were then left to settle for 5 min before spin-down (200 g for 2 min with zero breaks). After spin-down, cell plates were imaged as described in Cell Number Quantification for Metabolic Flux Assay, while kept at 37°C. After imaging, an additional volume of 130 µL assay medium was added, and cell plates were incubated for another 20 min in a non-CO₂ 37°C incubator. Extracellular flux analyses (XF assays) were performed using the Seahorse XFe96 (Seahorse Bioscience, Agilent Technologies, Santa Clara, CA). XF assays were performed using serial injections of first a combination of 1.5 µM Oligomycin (No. O4875, Sigma-Aldrich, St. Louis, MO) and 1 µM FCCP (No. C2920, Sigma-Aldrich, St. Louis, MO), followed by a combination of 1.25 µM Rotenone (No. R8875, Sigma-Aldrich, St. Louis, MO) and 2.5 µM antimycin A (No. A8674, Sigma-Aldrich, St. Louis, MO) and finally 50 mM 2-deoxyglucose (2-DG, No. D8375, Sigma-Aldrich, St. Louis, MO). The XF assay protocol consisted of 12 measurement cycles of 3 min, with 2 min of mixing in between measurements.

Cell Number Quantification for Metabolic Flux Assay

Bright-field images of the inner probe area of each well in the XF96 cell plates were obtained before the XF assay run using a 37°C equilibrated Cytation 1 Cell Imaging Multi-Mode Reader (BioTek Instruments, Inc., Winooski, VT) using a ×4 objective. LED intensity was kept at 5 and integration time at 80 ms for all cell plates. To ensure optimal image quality, focus height was adjusted by visual inspection for each cell plate. The bright-field images obtained before the XF assay run were then processed and quantified using an in-house generated R-script (24), which uses the EBImage package available for Bioconductor (25). A low-pass Gaussian blur filter was applied to generate a background image, which was subsequently subtracted from the original image. The background corrected image was then inverted and cropped by 5% to remove potential boundary noise from the XF assay plate molded stops. The final processed images were used to calculate the total pixel intensity for each image. To convert pixel intensities back to cell numbers, the values of the internal plate standard curves were fitted with a second-order polynomial regression. The obtained cell numbers were finally used for normalization of the Seahorse XF assays. Per pig, eight replicate wells were included during Seahorse XF analysis and normalization, and mean of all eight replicates was then used for further processing.

Object Detection for Cell Size Quantification

To quantify cell population distributions using only bright-field images, we developed algorithms that segment images of pre-Seahorse imaged wells and capture the sizes of individual cells. For this, we used image analysis tools from the R packages imager, imagerExtra, EBImage, and magick. Before the image segmentation was performed, histograms of grayscale images were equalized to lower background and enhance contrast, followed by slight blurring. Next, we used a Hessian matrix blob detection approach (26) to localize and segment individual cells in our relatively low-resolution bright-field images. For each of the identified cells, we used the radius in pixels as an estimate of its size. The cell radii were binned into different categories, as a proxy for different cell populations. We classified cells as being “small” (0–2 pixels), “medium” (2–4 pixels), or “large” (4–6 pixels) cells (Supplemental Fig. S5; see https://doi.org/10.6084/m9.figshare.22811192).

Tissue Collection and RNA Isolation

Mucosal scraping samples were obtained from midjejunal section. Mucosal scrapings were snap-frozen in liquid nitrogen and stored at −80°C until use. To isolate RNA, mucosal scrapings were first crushed to powder in liquid nitrogen with a pestle and mortar. Total RNA was extracted using the RNeasy Mini Kit (No. 74106, Qiagen, Hilden, Germany), according to the manufacturer’s protocol. For each sample, 5–10 mg of powdered tissue was dissolved in a mixture of RLT buffer (Qiagen, Hilden, Germany) and β-mercaptoethanol (M6250, Sigma Aldrich, St. Louis, MO). Tissue samples were subsequently lysed and homogenized by shaking at maximum speed in a thermomixer (Eppendorf, Hamburg, Germany) at room temperature for at least 45 min. Then, 70% ethanol was added, and total RNA was bound on the RNeasy spin column. After DNA digestion with DNase I treatment, the column was washed three times with RW1 and RPE buffer, and total RNA was eluted in RNase free water. Quality of the RNA was assessed using Nanodrop spectrophotometer (IsoGen Life Science, de Meern, The Netherlands) and Agilent 2200 Tapestation (Agilent Technologies, Inc., Santa Clara, CA). Samples met the criteria with a ratio higher than 1.8 for both 260/280 nm and 260/230 nm or an RNA integrity number (RIN) above 7.

Semiquantative Real-Time Polymerase Chain Reaction

From the RNA samples, cDNA was synthesized with the iSCRIPT cDNA synthesis kit (170-8891, Bio-Rad, Hercules, CA) in the Eppendorf Mastercycler (5 min 25°C, 30 min 42°C, 5 min 85°C, 10°C ∞). Gene expression was measured using the CFX96 Touch Real-Time Polymerase Chain Reaction (RT-qPCR) Detection System (Bio-Rad, Hercules, CA) and SYPBR green master mix (1725006CUST, Bio-Rad, Hercules, CA). The cycling program was set as follows: 3 min 95°C, 40 cycles of 15 s 95°C and 45 s 58°C, 1 min 95°C, and 1 min 65°C, followed by melt curve analysis by increasing temperature every 10 s with increments of 0.5°C. Primers were designed using NCBI Primer BLAST. Normalized expression was calculated according to the ΔΔCq method, by making use of multiple reference genes (RPL19, RPL4, and RPS18), using the CFX maestro software (Bio-Rad, Hercules, CA). An overview of the primers used can be found in Table 2.

Table 2.

Details of primers

Symbol	RefSeq	Forward Primer*	Reverse Primer*	bp
RPS18	NM_213940.1	CCCTGAGAAGTTCCAGCACAT	CCGTCCTACACCCTTAATCGC	100
RPL4	XM_005659862.2	TTCAAGGCTCCCATTCGACC	GCACTGGTTTGATGACCTGC	110
RPL19	XM_003131509.4	ATATGGGCATCGGTAAGCGG	ACGGTATCTTCTGAGCAGCC	107
CFTR	NM_001104950.1	TAGATGTGGATAGCTTGATGCGA	CTGTTAGGTTGATCTCCTTCTGC	82
SGLT1	NM_001164021.1	GTCATCTACTTCGTGGTGGTGATG	CCACCCAAATCAGAGCATTCCA	232

bp, fragment length.

^*From 5′ to 3′.

RNA-Seq and Data Analysis

RNA preparation, library construction, sequencing on the DNB-sequencing platform, and read clean-up were performed at Beijing Genomics Institute (BGI, Hong Kong). Quality check of the clean reads was performed using FASTQC (27). Reads were aligned to the pig genome (Sscrofa11.1.104) using STAR2.7 (28), and counts were quantified using HTSeq (29). The average sequencing depth was 24M paired-end reads, of which at least 92.9% were uniquely mapped. After read alignment and counting, data analysis and statistical testing were performed in R 4.1, using appropriate Bioconductor packages.

Genes with a total sum of less than 10 counts were removed. Then, differentially expressed genes (DEGs) between fed and fasted piglets were identified using the DE2seq package, using Benjamini–Hochberg correction for multiple testing (30). An adjusted P value of below 0.05 was considered significantly regulated. Principal component analysis was performed using variance stabilizing transformed (VST) data. Gene set enrichment analysis (GSEA) (31) was performed using clusterProfiler (32) for the Reactome annotated gene sets (33) that were extracted using the Molecular Signatures Database [MSigDB (34)]. Gene sets were considered enriched with a Benjamini–Hochberg (BH) adjusted P value <0.05. Only the pathways with a P value <3.5 × 10⁻³ were included for further analysis and clustering. The 25 pathways that remained were then clustered based on semantic similarities in the gene ontology (GO) description of the genes in each pathway, using the GoSEMSim Package (35). The human MitoCarta 3.0 gene set was used as a reference inventory for mitochondrial genes (36). The marker gene sets for intestinal cell populations as identified by Burclaff et al. (37) and Franzén et al. (38) were used to identify shifts in intestinal cell populations. The gene set for barrier genes as assembled by Vancamelbeke et al. (39) was used to identify significantly regulated barrier genes. Volcano plots and heatmaps were generated using the EnhancedVolcano and ComplexHeatmap packages (40, 41).

Immunohistochemistry

Midjejunal tissue sections were cut longitudinally, washed in cold PBS, rolled up proximally to distally, and placed inside a histology cassette. Tissue was then fixed in 4% paraformaldehyde (1.040.051.000, Merck, Kenilworth, NJ) and stored at 4°C for 24 h, after which tissues were dehydrated as follows: 4 h 70% EtOH, 4 h 80% EtOH, 4 h 90% EtOH, 4 h 100% EtOH, 4 h 100% EtOH, 4 h 100% EtOH, 3 h xylene, 2 h xylene, 2 h xylene, and 4 h paraffin, and embedded in paraffin blocks.

Tissue sections of 5 µm were cut using a Leica microtome and subsequently placed at ∼42°C to stretch the samples, before mounting them on a glass slide and drying them overnight at 37°C. Before being stained, slides were deparaffinized (2 × 5 min xylene, 2 × 3 min 100% EtOH, 3 min 96% EtOH, 3 min 70% EtOH, 2 × 3 min demi water). For Ki67 staining, slides then underwent a 10-min heat-mediated antigen retrieval in 0.1 M sodium citrate buffer (pH 6.0) at subboiling temperatures by microwaving the slides. Slides were then cooled to room temperature and rinsed in Tris-buffered saline (TBS) (pH 7.4). To block aldehyde residues, slides were then incubated in 0.75% glycine in TBS for 20 min. After being rinsed with TBS, sections were preincubated with 5% (vol/vol) normal goat serum (Vector Laboratories, Peterborough, UK) in TBS for 60 min at room temperature in a humidifying box. Slides were subsequently incubated overnight at 4°C in a humidifying box with primary rabbit polyclonal anti-Ki67 (1:500, ab15580, Abcam, Cambridge, UK) antibody diluted in TBS-BSA-c (Aurion, Wageningen, The Netherlands). Slides were then rinsed with TBS and treated with a fluorochrome-labeled goat-anti-rabbit (Alexa Fluor 546, Invitrogen, CA) diluted 1:200 (vol/vol) in TBS-BSA-c for 1 h at room temperature. Finally, nuclear counterstaining was performed for 10 min using DAPI (1 μg/mL; Sigma-Aldrich, St. Louis, MO). Sections were imaged at ×10 magnification using fluorescence microscopy (Leica DM6B), a digital camera (DFC365 FX), and imaging software (LasX; all Leica Microsystems, Amsterdam, The Netherlands). Per slide, 10 representative images were obtained and included in the analysis. Images were analyzed according to a previously published protocol (42). First, images were loaded into CellProfiler software (Broad Institute, Cambridge, MA) to select the appropriate objects. Briefly, DAPI objects were identified that ranged between 2 and 20 pixels. Then, only Ki67-stained objects that overlapped with the previously identified DAPI-stained objects were selected. Data for the identified objects were then exported both as a spreadsheet and as images, and those files were further processed using the FCS express software (De Novo software, Pasadena, CA). First, large DAPI objects, often due to artifacts in the tissue during processing, were removed. Then, low-intensity and high-intensity DAPI images were excluded. Lower bound of intensity was identified as median image intensity – 2SD, and upper bound of image intensity was identified as median image intensity + 3SD. Subsequently, the low-intensity Ki67-stained object, which could be artifacts or background staining, was removed by taking median image intensity +1 SD as the lower bound, and median image intensity + 10 SD as the upper bound. The percentage of cells in the Ki67 gate over the total amount of cells in the DAPI gate was taken as the final ratio of Ki67 over DAPI-stained cells. See Choudhury et al. (42) for further details about image processing.

Histochemistry

After deparaffinization, slides were oxidized for 10′ using 0.5% perioding acid solution, after which they were placed in a Schiff reagent bath for 12 min, washed three times for 2 min in a sulfite bath, rinsed under running tap water for 5 min. Slides were then incubated in Mayer’s hematoxylin for 1.5 min, followed by 10 min rinsing under running tap water and dehydration (3 min 70% EtOH, 3 min 96% EtOH, 2 × 3 min 100% EtOH, 2 × 5 min xylene). Sections were imaged at ×5 magnification using bright-field microscopy (Leica DM6B), a digital camera (DFC450C), and imaging software (LasX). Per slide, at least three representative images were obtained and included in the analysis. Per pig, at least three intact crypts and villi were identified. Using ImageJ software (43), crypt length and crypt area of a full crypt were measured. For the villi, its length was measured and the surface area of half a villi. Goblet cells were manually counted in the full crypt and half the villi.

Statistical Data Analysis

Data are presented as means ± standard deviation. Piglets were pair-housed for the duration of the experiment, but because of the nature of the intervention and analysis performed, individual pigs were analyzed as experimental units. For the body-weight measurements, eight data points were missing because they were not assessed, whereas all other analyses were performed with n = 16/group. Histology was performed on n = 8/group, except for the fed group in goblet cell analysis, where n = 7 because not enough intact villi could be identified in one sample. Normality of model residuals was checked using Shapiro–Wilk test, and if assumptions were not met, data was transformed as indicated in the figure legends. If all assumptions for normality were met, statistical testing was performed using Student’s t test or repeated-measures ANOVA, as indicated in figure legends. A P value of <0.05, or a BH-adjusted P value of <0.05 was considered statistically significant. Repeated-measures ANOVA was performed in SPSS Version 25.0 (IBM Corp., Armonk, NY), and all other statistical analyses and data visualizations were performed using GraphPad Prism v.9 (GraphPad Software, CA) and R version 4.1.

RESULTS

All piglets received an equal amount of a standard weaning diet throughout the 2-wk period following weaning, resulting in equal growth in both fed and fasted groups in the 2 wk preceding the fasting intervention (Fig. 1A). No difference in diarrhea score was observed in the 2-wk intervention period, and fasting did not result in higher diarrhea incidence (data not shown). As expected, the 48-h fasting intervention led to a significant decrease in weight in the fasted group (Fig. 1, A and B, P < 0.001), whereas the fed group continued to eat and gain weight throughout the 48-h intervention period (Fig. 1B). Fasting also significantly decreased small intestinal (SI) length (Fig. 1C, P < 0.01), but no significant effect of fasting on SI length relative to bodyweight was observed (Fig. 1D).

Figure 1.

Macroscopic effects of 48-h fasting in 6-wk-old piglets. A: piglet growth during the study period, n = 12/group. B: piglet weight for the fed and fasted groups before and after the start of the 48-h fasting period, n = 12/group. C and D: small intestinal length at the end of the study period, raw values, and corrected for bodyweight, n = 16/group. Data are presented as means ± SD, and each dot in the bar-graphs represents an individual piglet. Significance was determined using repeated-measures ANOVA (piglet weight), or Student’s t test (intestinal length).

Metabolism of isolated IECs was functionally assessed using seahorse XF analysis. Fasting decreased basal mitochondrial oxygen consumption rate (OCR) OCR by 51 ± 9.9% (Fig. 2A, P < 0.0001), and likewise reduced mitochondrial capacity, as was shown by a reduction of maximal mitochondrial OCR by 43 ± 9.8% (Fig. 2B, P = 0.0001). In addition, fasting decreased basal glycolytic function by 45 ± 10.3% [basal glycolytic proton efflux rate (glycoPER), Fig. 2C, P < 0.0001] and compensatory glycolytic function by 35 ± 14.8% (compensatory glycoPER, Fig. 2D, P < 0.05). Overall, fasting thus simultaneously decreased oxidative and glycolytic IEC extracellular flux, leading to decreased metabolic oxidative capacity during fasting (Fig. 2E).

Figure 2.

Metabolic function of isolated jejunal piglet small intestinal epithelial cells (IECs) following 48 h of fasting. Mitochondria flux parameters represented by: basal oxygen consumption rate (OCR) (A) and maximal respiration (B). Glycolytic flux parameters represented by: basal glycolytic proton efflux rate (glycoPER, C) and compensatory glycoPER (D). E: metabolic phenotype graph, showing basal OCR vs. basal glycoPER. Data are presented as means ± SD, n = 16/group. Significance was determined using Student’s t test. Maximal respiration and Compensatory glycoPER were log-transformed to meet normality assumptions.

Intestinal epithelial gene expression was analyzed using RNAseq analysis. PCA analysis on transformed counts from RNAseq data using variance stabilizing transformation revealed a clear separation between the fed and fasted 6-wk-old piglets; principal component 1 explained 33% of the variation in the data (Fig. 3A). Gene set enrichment analysis (GSEA) using the Reactome curated gene sets showed that 34 pathways were significantly regulated upon fasting. To select the most regulated pathways, we only included the pathways with a P value below 3.5 × 10⁻³, which left us with 25 pathways. These 25 most regulated pathways were grouped in five clusters, using an algorithm based on semantic similarities in the gene ontology (GO) description of the genes in the pathways [Fig. 3B; (35)]. Fasting downregulated the pathways in clusters 2 through 5, whereas the pathways in cluster 1 were upregulated upon fasting. Cluster 1 consisted of five pathways primarily related to protein translation. The top 10 genes in the cluster were involved in protein translation initiation (EIF3A, EIF3G, EIF4B, and EIF3H) or were part of the ribosome machinery (RPL30, RPS16, RPS11, RPS28, RPS8, and RPL11) (Fig. 3C). Cluster 2 consisted of ten pathways that were primarily related to cell cycle. The top 10 genes in the cluster were related to cell-cycle progression (ANAPC7 and HSP90AA1), microtubular organization (TUBB4B, TUBA4A, CETN2, and DYNC1LI1), chromatin structuring (WAPL), and organelle assembly (GORASP2). Cluster 3 consisted of three pathways related to mitochondrial function. Top 10 genes in the cluster were mainly TCA-cycle enzyme genes (DLTS, SUCL2A, IDH3A) and ETC complex genes (COX6B, NDUFS1, NDUFS3, NDUFA6, and NDUFB5) but also other metabolically important genes such as malic enzyme (ME1, converts malate to pyruvate) and basigin (BSG, plays a role in correct localization of glucose transporters on the cell-surface). Cluster 4 consisted of three pathways related to stress response. The top 10 genes in the cluster were involved in proteosome assembly and stability (PSME3, PSMD14, PSMD11, and HSPA8), signal transduction (MAP2K3 and HSP90AA1), nuclear protein import (KPNA4), and protein translation initiation and quality control via protein (un)folding (EIF4E, CANX, and HSP90B1). Cluster 5 was the least well-defined cluster, containing four pathways related to protein folding, extracellular matrix organization, and transporters. The top 10 regulated DEGs in cluster 5 are involved in protein quality control and extracellular matrix organization (COL8A1, CANR, CANX), lipid metabolism (SCD and ELOVL6), glycoprotein, and glycolipid production (GFPT1, GMPPB, and GNPNAT1) or are transporters (SLC10A2 and VDR).

Figure 3.

Jejunal epithelial gene expression analysis upon 48 h fasting in 6-wk-old piglets. A: PCA plot of Euclidean distance between samples on variance stabilizing transformed data. Samples were colored based on group. B: Treeplot of Reactome pathway analysis of the differentially expressed genes (DEGs), clustered on similarity using the Jaccard similarity coefficient. C: heatmaps of the top 10 significantly regulated genes in each cluster. Scaled normalized counts are plotted for each individual gene and piglet, n = 16/group.

One of the primary regulated gene expression clusters consisted of the downregulated mitochondrial energy metabolism pathways in cluster 3, which is in full accordance with the downregulation of basal and maximal mitochondrial OCR upon fasting in our functional metabolic analysis in freshly isolated pig IECs (Fig. 2). To gain an even deeper understanding of the molecular changes that occur within the mitochondria upon fasting, we specifically analyzed mitochondrial gene expression. We used the MitoCarta 3.0 annotated gene set: a gene set containing 1,139 genes with strong support for mitochondrial localization (36). Fasting piglets led to altered expression of 424 of the MitoCarta genes, which was 37% of all MitoCarta genes (Fig. 4A). Of those, 147 DEGs were upregulated, whereas 278 DEGs were downregulated (Fig. 4B, Supplemental Table S1; see https://doi.org/10.6084/m9.figshare.21303780). Especially the expression of PDK4, a critical regulator of the pyruvate dehydrogenase complex, stood out as being highly upregulated (Fig. 4B). We then performed GSEA analysis using the pathway annotation of MitoCarta 3.0 (Fig. 4C). Fasting of the piglets resulted in a significant regulation of 11 mitochondrial pathways, four of which were upregulated and seven that were downregulated. The genes with the highest log₂FC for each pathway are listed in Fig. 4D and are given below in brackets after each pathway. Upregulated pathways were fatty acid oxidation (ACAD11 and ACAA2), ketone metabolism (BDH1 and ACAA2), carbohydrate metabolism (PDK4 and GLYCTK), and pyruvate metabolism (PDK4 and PDK2). When looking specifically at the genes regulated within the carbohydrate and pyruvate metabolic pathways, we observed that mainly the gluconeogenic genes were upregulated, whereas overall carbohydrate metabolism was downregulated (Supplemental Fig. S2; see https://doi.org/10.6084/m9.figshare.21303771). Downregulated pathways were mainly related to expression of TCA cycle enzymes (SUCLA2 and IDH3A) and electron transport chain complex I (NDUFAB1 and NDUFA5) and complex IV (COX7A2 and NDUFA4). In addition, fasting led to decreased expression of mitochondrial ribosomal protein expression (MRPS28 and MRPS27).

Figure 4.

MitoCarta analysis of differentially expressed genes (DEGs) in jejunum upon 48 h of fasting in 6-wk-old piglets. A: pie chart showing that 37% of the 1,126 genes in the MitoCarta 3.0 set are significantly regulated upon 48 h of fasting in piglets. B: volcano plot displaying the DEGs of the MitoCarta 3.0 gene set. Eight hundred eighty-six of the 1,126 genes in the MitoCarta 3.0 set are present in our dataset and plotted in the volcano plot. Of those 886 genes, 424 genes were differentially regulated between fed and fasted piglets. C: Gene Set Enrichment Analysis (GSEA) of the MitoCarta 3.0 annotated mitochondrial pathways, showing which mitochondrial pathways are significantly regulated with a nominal P value <0.05. D: list of the two genes in each pathway with the highest Log2FC upon 48 h of fasting in piglets.

In addition to regulation of mitochondrial genes, we found that fasting had the most profound effect on cell cycle, since the largest of the five clusters, cluster 2, consisted of ten pathways that were all related to cell cycle and proliferation (Fig. 3B). This suggests that small intestinal jejunal cell turnover is lower in fasted piglets. To verify this, we analyzed cell proliferation immunohistochemically using the marker Ki67, which is expressed only in the nucleus of dividing cells. Indeed, immunostaining showed that fasting decreased the ratio of the proliferating cells over total number of cells by 43 ± 5.3% (Ki67/DAPI, Fig. 5, A and B, P < 0.05), confirming the finding that fasting significantly decreased cell proliferation in the jejunum of 6-wk-old piglets. Furthermore, we analyzed villi length and crypt depth to characterize changes in intestinal architecture. While both villi length and crypt depth were significantly reduced upon fasting, the ratio between villi length and crypt depth remained unaltered (Fig. 6, C and D). The altered proliferation and intestinal architecture may also point toward an altered cellular composition of the intestine. We therefore analyzed the expression of marker gene sets for various intestinal cell populations, as identified using single-cell sequencing by Burclaff et al. (37) and Franzén et al (38). The differential expression of these markers in our data set suggests an increase in stem cell markers and a decrease in goblet cells and transit amplifying cell markers, whereas tuft cell, enteroendocrine cell, and enterocyte cell markers were similarly up- and downregulated (Supplemental Fig. S3; see https://doi:10.6084/m9.figshare.24183537). In addition, the expression of two often used markers for intestinal maturation (CFTR and SGLT1) also decreased as assessed by qPCR (Supplemental Fig. S4; see https://doi:10.6084/m9.figshare.24183603). Notably, SGLT1 is in fact also responsive to low luminal glucose levels (3). Its lower expression in fasted piglets is therefore in line with our other metabolic phenotypes. Together, these results indicate changes in intestinal architecture upon fasting. We therefore further assessed the number of goblet cells present in crypts and villi, as they have a characteristic histochemical morphology. Although there was no absolute change in goblet cell numbers in crypts and villi of fasted piglets (Fig. 6, E and F), goblet cell numbers relative to the surface area increased in both villi and crypts (Fig. 6, G and H). Although the relative number of goblet cells increased, their functionality might be decreased as shown by the marker gene set analysis.

Figure 5.

Cell proliferation analysis of 6-wk-old piglet’s jejunum following 48 h of fasting. A: representative images of processed immunohistochemical analysis of fed and fasted piglets stained with Ki67. Red objects are DAPI-stained objects that are larger than the defined size. Scalebar is 200 µm. B: bar-graph of the Ki67 stained cells corrected for the total number of cells per crypt. Data are presented as means ± SD, n = 8/group. Significance was determined using Student’s t test.

Figure 6.

Morphological analysis of 6-wk-old piglet’s small intestine upon 48 h of fasting. A: representative images of Periodic Acid-Schiff and Hematoxylin (PASH) PASH stained fed and fasted piglet jejunum. Scale bar is 200 µm. Crypt depth (B), villi length (C), and villi length:crypt depth ratio in µm (D). E and F: total number of goblet cells per crypt or villi. G and H: number of goblet cells relative to crypt or villi area in µm². Data are presented as means ± SD, n = 7 for fed piglets and n = 8 for fasted piglets. Significance was determined using Student’s t test (*P < 0.05; **P < 0.01; ***P < 0.001).

To further assess whether the observed changes in both proliferation, intestinal architecture, and relative goblet cell numbers impact intestinal function, we investigated the effect of fasting on intestinal barrier function. For this, we used a gene set described by Vancamelbeke et al. (39). This gene set includes genes classified as being involved in one of eight categories: mucus layer, tight junctions, adherence junctions, desmosomes, hemidesmosomes, cytoskeleton, extracellular matrix, and regulating proteins. We found that of the 128 genes in this gene set, 93 were present in our data set (Fig. 7A). Out of these 93 genes, 26 were significantly regulated, the majority of which were downregulated (Fig. 7B). The most significantly regulated genes were mucus genes (TFF1, TFF2, and EMCN, including also MUC2, the most prominent excreted mucin in goblet cells), tight junction genes (CLDN2, MARVELD3), an adherence junction gene (CDH1, also known as E-cadherin), regulating genes (MAGI3 and GNA12), and a cytoskeleton gene (MYL9). These data suggest an overall decreased intestinal barrier function in response to fasting.

Figure 7.

Analysis of barrier function and detoxification gene sets upon 48 h of fasting in 6-wk-old piglets. A: volcano plot of barrier function genes; 93 of 128 genes are present and plotted in the volcano plot, and 26 are significantly regulated. The top ten significantly regulated genes are labeled. B: heat map of the 26 significantly regulated genes in the barrier function gene set, with scaled normalized counts per pig and an additional column depicting the Log2FC of each gene in response to fasting (log2FC value plotted in the square). C: volcano plot of mitochondrial detoxification genes as annotated by MitoCarta 3.0. Forty-two of the 51 genes are present and plotted in the volcano plot, and 15 are significantly regulated. The eleven redox detoxification specific genes are labeled. D: heatmap of the 15 significantly regulated genes in the detox gene set, with scaled normalized counts per pig and an additional column depicting the Log2FC of each gene in response to fasting (log2FC value plotted in the square).

Fasting is common during the weaning period in pigs. During the weaning transition, increased oxidative stress has been observed in the intestine (21, 44–47). Because mitochondria are a major source of reactive oxygen species (ROS) (48), we specifically analyzed the MitoCarta 3.0 annotated mitochondrial detoxification pathway. Out of the 42 genes present in the data set, 15 were regulated upon fasting (Fig. 7, C and D), nine of which were downregulated and six were upregulated. Out of the 15 significantly regulated detoxification genes, 11 were specific redox detoxification genes (see labeled genes in Fig. 7C). Eight redox genes were downregulated (most prominently GSR, TXNRD1, PRDX3, and SOD2), and three were upregulated (CAT, HAGH, and NIT1). Thus, we observe both an up- and downregulation of mitochondrial redox genes, suggesting an adaptation of redox homeostasis rather than a change in a specific direction.

DISCUSSION

Intestinal tissues rely on sufficient energy production to maintain homeostasis. Although previous studies have observed a decreased expression of key energy-producing genes in fasted mice and weaned piglets (12, 21), which often also display low or no food intake, concomitant functional data were lacking. In this study, we showed that 48-h fasting decreased both the mitochondrial and the glycolytic extracellular flux of isolated jejunal IECs in 6-wk-old piglets. Since metabolic extracellular flux analysis of the isolated IECs was performed in the presence of sufficient energy substrates (pyruvate, glucose, and glutamine), this indicated that the metabolic function of IECs is not only decreased because of a lack of available substrates but due to long-lasting changes in mitochondrial function. With this study, we are, to our knowledge, the first to directly link the decreased expression of metabolic genes in the intestine upon fasting to a functional decrease in the metabolism of isolated IECs. These findings substantiate literature findings by providing functional data, but also validated our own transcriptome analysis, where mitochondrial pathways were among the top regulated Reactome gene sets upon fasting. We found that both glycolytic and mitochondrial metabolism was reduced, indicating that there was an overall reduction of metabolic function. As such, there seems to be no adequate systemic compensation for the loss of metabolizable substrates from the diet. This notion is strengthened by the concomitant decreased cell proliferation, which seems to be the most prominently regulated pathway upon fasting in these piglets. This decreased proliferation, together with decreased expression of important barrier function genes, suggests that fasting may indeed lead to increased susceptibility to disease development. These findings are important to better understand how fasting, as a common consequence of weaning, can exacerbate weaning-associated intestinal dysfunction.

In this study, 48-h fasting-induced significant metabolic changes in the intestine of young, 6-wk-old piglets, with PDK4 standing out. PDK4 was increased 3.4-fold and was previously shown to be induced by fasting in the intestine of mice (12, 49). PDK4 is a well-known metabolic regulator that drives cells away from glucose oxidation and toward fatty acid oxidation (50). Indeed, higher PDK4 expression has been suggested as marker of increased fatty acid oxidation (51). The higher PDK4 expression is thus expected to induce fatty acid oxidation, which was indeed the most upregulated mitochondrial pathway in our data (Fig. 4). Increased fatty acid oxidation, and likely lipid breakdown, is strengthened by the simultaneous decreased expression of key lipogenic genes such as ACLY and FASN (Supplemental Table S1; see https://doi.org/10.6084/m9.figshare.21303780). Mechanistically, the increased expression of PDK4 may be also linked to the observed upregulation of ketogenesis. The ketone body β-hydroxybutyrate has been identified as a post-translational histone lysine modification (52), and histone β-hydroxybutyrylation induced PDK4 expression in the intestine of fasted mice (53). Via β-hydroxybutyrylation, ketogenesis could thus potentially contribute to the induction of lipolysis in the intestine of fasted piglets. This, however, requires experimental confirmation. Because PDK4 is expected to increase lipid oxidation and simultaneously decrease carbohydrate metabolism, the apparent upregulation of carbohydrate metabolism was unexpected (Fig. 4). Close inspection, however, revealed that the key gluconeogenic genes were upregulated, whereas overall carbohydrate metabolism was indeed decreased, as expected (Supplemental Fig. S1; see https://doi.org/10.6084/m9.figshare.21303750). Increased gluconeogenesis is concomitant with increased PDK4 expression (54) and is in line with the increased recognition that intestinal gluconeogenesis upon fasting is important for maintaining whole body glucose levels (55–57). The importance of gluconeogenesis for piglet health is furthermore highlighted by the finding that decreased gluconeogenesis contributed to postnatal growth retardation in pigs (58). Together, our data underpin the role of PDK4 as a central regulator of metabolism upon fasting, also in the small intestinal epithelium. Even though the intestine plays a key role during fasting by providing ketone bodies and glucose to support whole body metabolic function, this metabolic shift appears insufficient to maintain nutrient flux in the IECs, as is apparent from the overall decreased mitochondrial and glycolytic function of isolated IECs upon fasting as measured using extracellular flux analysis in freshly isolated IECs. This can possibly be explained by the fact that other organs could be prioritized over the intestine, as occurs physiologically during intensive exercise to particularly support muscles (59) or during starvation to ensure brain energy supply (60). In addition, mitochondria are key regulators of apoptosis (61). Lowering mitochondrial metabolism could serve to prevent the induction of apoptosis (62). Our gene expression data do not indicate that apoptosis is regulated or that cell death is a prevalent response to fasting, possibly because of the decreased mitochondrial metabolism.

The decreased overall mitochondrial metabolic capacity may impact other key functions of the intestine, either as a direct consequence of energy shortage or otherwise. We therefore investigated which key intestinal functions are further affected by fasting in piglets, and found that proliferation is markedly affected by 48-h fasting. This has also been described in the intestine of mice (12) and agrees with the decreased cell proliferation that is often observed in response to the weaning-associated decrease in feed intake in piglets (63). The IECs typically have a very high turnover rate, with renewal every 5–7 days (1) and as many as 300 cells are being produced in a single adult mouse crypt (64). This high level of proliferation and renewal is needed to withstand constant exposure to the abrasive luminal environment. Failure to maintain intestinal proliferation could contribute to intestinal dysfunction. Thus, a decreased proliferation could possibly result in a decreased barrier function. Although relative goblet cell numbers increased in our study (Fig. 6, G and H), other important markers for barrier function were decreased upon fasting [Fig. 7; (39)]. Not only tight junction genes and adherence genes (e.g., CLDN2 and CDH1), but also mucus genes were downregulated (e.g., TFF1, TFF2, and EMCN). Thus, goblet cell numbers seem to be increased, but they may not be as functional as in the fed piglets. Barrier function maintenance, upheld by, among others, tight junction gene expression and mucus production, is a crucial, but energy-demanding process, that is often reported to be decreased upon insufficient feed intake upon weaning (65, 66). Although these findings are not in themselves abundant proof for decreased barrier function upon fasting, they do suggest the importance of maintaining sufficient feed intake throughout the postweaning period to support key intestinal processes such as cell proliferation and possibly barrier function.

Weaning has been shown to induce cortisol levels (67), possibly enhanced by fasting (68). However, increased cortisol levels were shown not to be a primary cause of increased intestinal permeability upon weaning (69). On the other hand, cortisol is known to stimulate gluconeogenesis to improve glucose availability in the body (70), which is in line with the increased expression of gluconeogenic genes we observed Supplemental Fig. 2. In this way, cortisol may provide survival benefit, as intestinal gluconeogenesis is increasingly recognized as a key survival mechanism (71). Thus weaning and fasting stress may induce cortisol to perhaps counteract energy insufficiency which seems the primary cause of the observed decreased expression of barrier function genes.

A striking finding in our study was that translation initiation factors and ribosomal gene expression were increased in response to fasting (Fig. 3). Possibly, this is an example of “translation on demand,” meaning that mRNA transcription is maintained to ensure rapid translation upon reintroduction of feed (72). Indicative of the “translation on demand” mechanism is increased mRNA abundance observed in the intestine of fasted young rats to prepare for elevated protein synthesis upon acutely increased energy availability (73). Alternatively, the initiation factors that are upregulated may be involved in the translation of mitochondrial genes, as was recently found for EIF3 (74). Nonetheless, the increased expression of the eukaryotic initiation factors and the ribosomal proteins appears inadequate to maintain critical intestinal functions such as maintaining metabolic function, cell proliferation, and barrier function upon 48 h of fasting in piglets.

Several studies have found that weaning induced oxidative stress in piglets. For example, ROS levels were found to be increased in serum and jejunal mucosa of weaned piglets (45, 46). It is often hypothesized that weaning is additionally accompanied by decreased antioxidant capacity, further exacerbating oxidative stress. However, the results in the literature vary greatly. For example, the expression of antioxidant genes such as CAT and GPX2 (45), GPX2 and GPX4 (21), and GPX2, SOD3, and CAT (47) were found to be decreased upon fasting. But some of those studies simultaneously report antioxidant genes that are not regulated upon weaning, e.g., SOD1 and SOD2 (45). In addition, and in contrast to the previously mentioned studies, Novais et al. (44) reported increased expression of antioxidant defense genes such as TXNRD2 and GPX2. Interpretations of these findings is further complicated by the fact that studies often use very targeted analysis of only a small set of selected genes. Because mitochondria are a major source of ROS (48), especially when they become dysfunctional, we investigated the impact of fasting on specifically on mitochondrial redox defense, and found no evident regulation in one direction (Figs. 4 and 6). Thus, although there is both significant up- and downregulation of specific genes, there is no clear pattern in the mitochondrial redox homeostasis alterations. There may still be increased ROS formation upon fasting. However, additional pathways could be used to mitigate them, such as decreased mitochondrial oxidation to prevent ROS formation (75). Thus, the observed decreased mitochondrial function may be a way to alleviate redox stress.

The findings presented in this study are of obvious relevance for pigs but also have potential ramifications for humans. Malnutrition frequently occurs in lower-income countries, where food availability is not guaranteed (76). Because research in infants is not ethically justified, alternative animal models with close similarity to humans are needed to broaden our understanding of how the intestine of young individuals responds to periods of fasting. Pigs are a good model for humans, especially for large organ systems such as the intestine (77). In addition, pigs have similar responses as humans upon nutritional interventions (78). This makes the findings of our study interesting to also translate to infants, highlighting the importance of adequate enteral feed intake to prevent the onset of disease.

Finally, we want to discuss methodological aspects of this study that could have impacted our results. First, we have not extensively characterized the isolated cell populations. For more insight in potential influences of differences in percentages of cell populations in the observed large metabolic difference between fed and fasted isolated IECs, we have attempted to characterize cell populations in our isolates by analyzing the images obtained before Seahorse XF Analysis. Cells were classified as being “small,” “medium,” or “large” based on visual observation of the images (Supplemental Fig. S5A; see https://doi.org/10.6084/m9.figshare.22811192), but no significant difference between the fed and fasted cell populations in terms of cell sizes was observed (Supplemental Fig. S5B). No differences were observed across the radius of objects between the cells from the fed and fasted condition (Supplemental Fig. S5C). Although it remains difficult to assess how this translates to cell types, it suggests that indeed no major shifts occur (49, 79–81). As we do observe that fasting decreased cell proliferation, crypt depth, and villi length and increased relative goblet cell numbers, it is possible that fasting impacted the proportion of cell types present in the intestine (49, 79, and Supplemental Fig. S3; see https://doi:10.6084/m9.figshare.24183537). Fasting is indeed often found to decrease villus height, instead favoring a stem-cell phenotype (49, 80). As a result, the proportion of enterocytes could be decreased in our isolated cell population relative to other cell types. Different cell types are expected to harbor distinct metabolic phenotypes. Enterocytes usually display high metabolic rates, and especially higher mitochondrial metabolism, compared with stem cells and goblet, Paneth and enteroendocrine cells (81, 82), part of the observed decreased respiration could be explained by an altered cell composition. For example, in Foxo1/3 knockout mice and organoids, the differentiation of the IEC population was affected, with a concomitant decreased mitochondrial function, whereas glycolysis was unaffected (82). However, as we observed both decreased mitochondrial and glycolytic function, and we observed only limited changes in cell population based on our cell population marker analysis and goblet cell counts, this indicates that an altered cell population composition is not the sole explanation for our altered metabolic phenotype. Second, the functional metabolism of isolated IECs was assessed in a nutrient-rich environment, as we wanted to eliminate analysis differences between the two intervention groups and standardize the nutritional environment to better observe the cell effects of fasting. The IECs of both fed and fasted animals were thus exposed to nonphysiological nutritional environments for functional analysis that could have impacted the overall results. However, as we observe evident differences between the fed and the fasted conditions and a good correlation between the transcriptome data and the functional data, we believe that the functional data still adequately reflect the experimental fasting conditions.

In conclusion, we found that 48 h of fasting reduced the capacity of the intestinal epithelium to generate sufficient ATP and that this coincides with decreased cell proliferation, altered intestinal architecture, and intestinal barrier function. This could contribute to enhanced susceptibility to infection and diarrhea as a direct consequence of decreased energy availability in IECs caused by insufficient entral nutrient availability, which is common in young piglets and humans. Our functional analysis of the metabolic function of IECs provides a tool to further study and substantiates this. In addition, specific gene expression sets related to metabolic responses in the intestinal epithelium can be developed using the tools presented in this study. However, this will require further studies to develop sets that are robust to other influences than fasting. Finally, our data indicate that targeted support of mitochondrial energy metabolism may provide a strategy to prevent the negative consequences of limited intestinal epithelial substrate availability. This may help to combat disease susceptibility in the weaning period in pigs, but it may also have a role in preventing or alleviating infant diarrhea.

DATA AVAILABILITY

The RNA-sequencing data has been deposited in the GEO database under ascension number GSE203439. The R-script for image analysis is available from GitHub (https://github.com/vcjdeboer/seahorse-data-analysis-PIXI).

SUPPLEMENTAL DATA

Supplemental Fig. S1: https://doi.org/10.6084/m9.figshare.21303750.

Supplemental Fig. S2: https://doi.org/10.6084/m9.figshare.21303771.

Supplemental Fig. S3: https://doi.org/10.6084/m9.figshare.24183537.

Supplemental Fig. S4: https://doi.org/10.6084/m9.figshare.24183603.

Supplemental Fig. S5: https://doi.org/10.6084/m9.figshare.22811192.

Supplemental Table S1: https://doi.org/10.6084/m9.figshare.21303780.

GRANTS

All authors were funded by the research partnership program TTW-DSM with Project Number 14942, which is partly financed by the Dutch Research Council (NWO) and partly financed by DSM Nutritional Products.

DISCLAIMERS

The funding partners had no role in the design of the project, nor data curation and analysis or in the writing of the manuscript.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

A.F.B., V.C.d.B., W.J.J.G., and J.K. conceived and designed research; A.F.B. performed experiments; A.F.B. analyzed data; A.F.B., V.C.d.B., W.J.J.G., and J.K. interpreted results of experiments; A.F.B. prepared figures; A.F.B. drafted manuscript; A.F.B., V.C.d.B., W.J.J.G., and J.K. edited and revised manuscript; A.F.B., V.C.d.B., W.J.J.G., and J.K. approved final version of manuscript.