Main

Enteral feeding markedly affects the intestine in preterm neonates, and the changes may not only be important for the adaptation to enteral nutrient intake but may also provoke severe intestinal complications, such as necrotizing enterocolitis (NEC) (1). In our previous proteomic studies on the preterm pig intestines, aspects such as antioxidation and detoxification, nutrient metabolism, and some signal transduction pathways were disturbed after newborn pigs were fed with formula that led to variable degrees of NEC lesions (2,3). These feeding associated proteome changes after just 1 d of feeding may have been induced by feeding alone or by the combined effects of feeding and microbial colonization, physiological changes at birth, and the progression of NEC inflammatory lesions. However, it seems that it is not factors such as mode of birth (4), in utero growth restriction (5), or bacterial flora (6) but rather the amount and type of feeding that determine the intestinal response just after birth (7). Feeding per se may be the main drive for maturation of neonatal intestines. Hence, we hypothesized that feeding alone would stimulate the growth and affect the proteome of the immature fetal intestine, independently of the later birth transition, bacterial colonization, and feeding-induced NEC lesions.

To prove our hypothesis, we used short-term oral feeding of fetal pigs in utero, having the same GA and feeding regimen (diet, amount, and frequency) as in the previous studies on postnatal preterm pigs (1–3,5,6,8). Gel-based expression proteomics was used to assess the expression of hundreds of different intestinal proteins simultaneously. We could thereby indicate that molecular pathways in the immature intestine were most clearly affected by feeding alone, in addition to some standard parameters of intestinal mucosal structure and digestive enzyme function.

MATERIALS AND METHODS

Animals.

Three pregnant sows (Large White × Landrace, 105 d gestation, term = 116 d) were anesthetized (5–10 mg/kg i.v. thiopental sodium; Abbott Laboratories, North Chicago, IL). After exposure of the uterus, two fetuses per sow were chosen randomly for surgery and later formula feeding. A small incision (5 cm) was made in a relatively nonvascular area of the uterus. A silastic catheter (vinyl tube, inner diameter of 0.86 mm, outer diameter of 1.52 mm; Dural Plastics and Engineering, Auburn, Australia) was passed down the esophagus of fetuses with the tip reaching the stomach. Antibiotics (2 mg/kg enrofloxacin, Baytril; Bayer, Leverkusen, Germany) were administered to the amniotic cavity of each fetus. After the surgery, all of the sows were given antibiotics (20 mL/d i.m. streptocillin; Boehringer Ingelheim, Copenhagen, Denmark) and progesterone (50 mg/d i.m.; University Pharmacy, University of Copenhagen, Denmark). After return of the sow to its crate, the two catheterized fetuses were fed with 15 mL enteral feeding formula every 3 h for 24 h (EF group, n = 6). On the basis of our observations in several previous studies on intestinal structure and function after feeding fetal pigs in utero (9–11), we did not include a separate group fed with saline. The effects of fetal surgery alone on intestinal physiology in similar aged pig fetuses are minimal, especially when the time from fetal surgery to later cesarean section is short (only 1 d in this study). All procedures were approved by the Danish National Committee on Animal Experimentation.

The formula was designed to match the macronutrient composition of sow's milk during lactation and identical to that used previously to feed preterm pigs with a similar GA at delivery (2,3,6). Each liter of the formula contained 80 g Peptide 2–0, 70 g Maxipro, and 75 mL Liquigen-MCT (SHS, Liverpool, United Kingdom; all products were donated by SHS). Major ingredients in the formula were hydrolyzed dried maize syrup (maltodextrins), bovine whey protein, hydrolyzed meat and soya protein, and oils from maize and coconut (medium-chain triglycerides). Macronutrient composition was (per liter formula) as follows: 64 g protein, 47 g total carbohydrate, 3 g lactose, and 61 g fat. Before feeding, the formula was sterilized by γ-irradiation (2 × 5.0 kGy; LR Plast, Glostrup, Denmark).

Tissue collection.

After feeding, the fed fetuses (EF group, n = 6) were delivered by cesarean section (6), together with some random unfed control fetuses (CTR group, n = 6). The small intestine from the pyloric sphincter to the ileocecal junction was rapidly excised by cutting along the mesenteric border. The small intestine was carefully emptied of its contents and divided into three equally long segments, designated proximal, middle, and distal small intestine. The middle part (2–3 cm) of middle small intestine was snap frozen in liquid nitrogen and saved at −80°C for later proteomic studies or fixed in 4% paraformaldehyde for 48 h and then transferred to 70% ethanol to be stored at 4°C for later histological preparation. The proportion of mucosa was recorded by gently scraping off the mucosal layer of a 10-cm intestinal piece (proximal, middle, and distal) and determining its proportional dry weight of the intact intestine after drying both at 65°C.

Intestinal morphology and enzyme activities.

The paraformaldehyde fixed samples were embedded in paraffin and sectioned at 3 μm before mounted on the slides for staining with hematoxylin and eosin. All slides were checked with a light microscope (Orthoplane, Leitz) and analyzed by NIH Image J software (National Institutes of Health, MD).

Frozen middle intestinal sections were extracted in 1.0% Triton X-100 (10 mL/g tissue) and homogenized (0°C, 20 s). The homogenates were assayed for activities of lactase, sucrase, maltase, dipeptidyl peptidase IV (DPPIV), amino peptidase A (ApA), and peptidase N (ApN) as described previously (1). For all enzymes, a hydrolytic rate of 1 μmol substrate per minute at 37°C was regarded as one unit of enzyme activity.

Protein extraction.

Extraction of intestinal protein was performed as previously described (2,3). Briefly, the tissue samples were disrupted with a tissue teaser (BioSpec Products, Bartlesville, OK) in a cocktail buffer (1% Triton X-100, 25 mm HEPES, 150 mm NaCl, 1 mm EDTA disodium salt, and 1 mm DTT) with Protease Inhibitor Cocktail Set III (Bio-Rad, Hercules, CA). The protein extracts were obtained by centrifuging the homogenates at 15,800 × g for 30 min at 4°C. The superfluous salt in the extracts were removed by precipitation with trichloroacetic acid-acetone solution and acetone contained 20 mm DTT. The protein pellet was resuspended in a buffer containing 7 M urea, 2 M thiourea, 4% Chaps, 100 mm DTT, and 5% glycerol. The protein solution was stored at −80°C until further analysis of two-dimensional gel electrophoresis (2-DE). Protein concentration of the samples was determined by Bio-Rad Protein Assay (Bio-Rad).

2-DE.

The protein sample (100 μg) was mixed with rehydration buffer (9.5 M urea, 2% Chaps, 0.28% DTT, and 0.5% IPG buffer pI 3–10) and applied onto one ReadyStrip IPG Strip (18 cm, pI 3–10 NL, Bio-Rad). The isoelectric focusing was carried out on an Ettan IPGphor III IEF System (GE Healthcare) after an active rehydration for 7 h, after the program, i.e. linearly running up to 500 V in 1 h, holding at 500 V for 3 h, after linearly increasing up to 10,000 V in 3 h, and holding at 10,000 V to reach a total of 90,000 Vh (V × h) (2,3). The focused gel strips were placed onto 1.0-mm-thick 12.5% PAGE gels, and SDS-PAGE was carried out in a PROTEAN xi II cell (Bio-Rad), after a two-step equilibration with a buffer (6 M urea, 30% glycerol, and 2% SDS in 50 mm pH 8.8 Tris-HCl) containing 1% DTT and then with 2.5% iodoacetamide successively.

Gel image analysis and protein identification.

After electrophoresis, gels were stained with SYPRORuby Protein Stain (Bio-Rad) according to the manufacturer's guide. The stained gels were scanned with a Molecular Imager PharosFX Plus System (Bio-Rad) and analyzed by PDQuest 8.0 (Bio-Rad) according to the following procedures: background subtraction, spot detection, and spot matching. Matched spots were assigned with numbers automatically. The in-gel trypsin digestion of the manually cut-out gel spots was performed as described previously (3). The resulting peptide mixtures were applied on a 4800 MALDI TOF/TOF Analyzer (Applied Biosystem, Carlsbad, CA) for MALDI-TOF/TOF MS (matrix-assisted laser desorption ionization-time of flight/time of flight mass spectrometry). The identity searching was based on the MS spectrum obtained is a combination of peptide mass fingerprint and MS/MS search using in-house searching engine, Mascot, with searching taxonomy limited as Mammalia (mammals) against NCBInr database (ftp://ftp.ncbi.nih.gov/blast/db/README). A protein match with a protein score >71 was considered significant.

Western blot analysis.

Briefly, 25 μg protein extracted was resolved by electrophoresis. Anti-glutathione synthetase, anti-caspase 1, and anti-calreticulin (Abcam, Cambridge, United Kingdom) were used to detect the expression of glutathione synthetase, caspase 1, and calreticulin as three examples of proteins from the proteomics analysis that were markedly affected by feeding in utero. The protein bands were visualized, and the density of the protein bands was detected by Quantity One (Bio-Rad). The lack of monospecific antibodies for most porcine protein variants precluded a large- scale Western blot analysis of differentially regulated proteins.

Statistical analysis.

All data are expressed as mean ± SEM. Two-tailed t test with Levene's test for equality of variances was used to compare the expression level of proteins by proteomics and Western blot analysis. SPSS 11.5 was used for all statistical analyses with p value <0.05 considered as significant.

RESULTS

Intestinal structure and enzyme function.

All pigs were well and alive when delivered by cesarean section 24 h after the start of enteral feeding. No pigs showed any macroscopic signs of NEC. Representative hematoxylin and eosin stained histological graphs of the intestine in EF and CTR groups are showed in Figure 1. The examinations showed that there were no significant differences in villus height and crypt depths between groups, whereas villus width showed a small increase in the EF pigs relative to the controls (p < 0.05). Furthermore, the EF piglets showed higher weights of stomach (+24%, p < 0.05), intestine (+55%, p < 0.01), and specifically the mucosa part (+75%, p < 0.001) than CTR piglets (Fig. 2A). The EF piglets also showed a tendency to a longer intestine (303 versus 240 cm/kg, p = 0.07) and a larger circumference (10.7 versus 8.3 mm, p < 0.001). The activities of ApA (+54%, p < 0.001) and maltase (+40%, p < 0.05) were found higher, whereas activities of lactase (−40%) and DPPIV (−60%) were found lower in EF piglets compared with CTR piglets (both p < 0.001, Fig. 2B).

Figure 1
figure 1

Representative histopathology of the intestine in CTR (A) and EF (B) pigs (magnification = ×200). Hematoxylin and eosin stained jejunum sections show normal intact villus structures of both groups of pigs with no signs of histopathology or mucosal damage. The EF group showed a trophic responses to feeding involving some villus swelling (marginal increase in villus width and length, see also Fig. 2) and empty vacuolated cells.

Full size image
Figure 2
figure 2

Gut morphology values (A) and intestinal brush border enzyme activities (B) in formula-fed (EF) pig fetuses relative to values in CTR fetuses (=1.0). Enteral feeding for 24 h significantly increased intestinal length (cm/kg body weight) and weight (g/kg body weight), circumference (mm), villus height and width (μm), and total amount of mucosa determined on a dry matter basis (g/kg body weight). Similarly, in utero feeding increased maltase and aminopeptidase A activities (U/g tissue), while decreased lactase and DPPIV activities (U/g tissue) (*p < 0.05; **p < 0.01; §p < 0.001).

Full size image

Identified proteins with differential expression.

Representative 2-DE gel graphs of the intestinal proteome from EF and CTR piglets are shown in Figure 3. The pI range of the gels was 3–10 and that of Mr 10–150 kD; ∼500 spots were identified on each gel with the expression analyzing software. Among the detected spots, 38 proteins were successfully found with definite identity and the corresponding spots on the 2-DE gel graph were labeled with numbers as shown in Figure 3. Descriptive information of the identified proteins, including spot number showing the position of spots on gels, protein name, GenInfo identifier, protein score, searched and apparent pI and Mr, and expression change between EF and CTR groups, are showed in Table 1. Among the identified proteins, 18 proteins showed higher expression in EF piglets than that in CTR piglets with two of them (ferritin L subunit, spot 4010; cathepsin Y, spot 5208) having no detectable expression in the CTR piglets. The other 20 identified proteins showed lower expression in EF piglets, with one of these (caspase 1, spot 3208) showing expression below the detection limit in EF piglets. A detailed description of the most relevant groups of identified proteins is given in the Discussion.

Figure 3
figure 3

2-DE gels of intestinal proteome map of CTR (A) and EF (B) piglets. Spot number was assigned by analysis software and correlated with the ones presented in Table 1.

Full size image
Table 1 Identified proteins with differential expression between EF and CTR pigs
Full size table

Western blot.

The relative abundance of three selected proteins were further tested by Western blot analysis (Fig. 4). Glutathione synthetase, caspase 1 p45, and calreticulin showed lower levels in EF pigs than in CTR pigs (p < 0.05).

Figure 4
figure 4

Western blot analysis of caspase 1 (A), glutathione synthetase (B), and calreticulin (C). Abundances of protein bands are presented as mean ± SEM *p < 0.05 in EF vs CTR.

Full size image

DISCUSSION

The intestinal response to the first enteral feed plays a key role in neonatal adaptation and growth, for both preterm and term neonates. Soon after birth, enteral food tolerance is required to achieve adequate nutrient intakes for body growth, and a suboptimal adaptation may lead to maldigestion, poor mucosal protection from colonizing bacteria, and increased risk of NEC, particularly for preterm neonates. In this study, we documented that the immature intestine was highly responsive to enteral feeding, even before birth, and that marked growth and brush border enzyme maturational changes (in the range of ±50%) occurred within just 24 h of fetal enteral feeding. These responses in fetuses were quantitatively and qualitatively similar to those reported previously after feeding preterm newborn pigs with the same quantity of the same formula (6). The changes do not arise from an effect of fetal surgery itself because such treatment have been documented in our earlier fetal studies not to induce any marked changes on fetal intestinal growth and function, relative to those by enteral feeding (1). However, our proteomic analyses indicate that the functional responses to formula feeding in utero differ at a number of points, from those reported previously for postnatal preterm pigs (2,3,12). The proteome changes in this study can be regarded as specific for the first feeding, independent of factors related to the birth transition (microbial colonization, metabolic and cardiovascular changes, thermoregulation, respiration, and a series of endocrine changes). Furthermore, the feeding related proteome changes in utero will not be connected with possible pathological proteome changes related to progression of NEC lesions, as others and we have shown that NEC in preterm neonates does not develop in the absence of bacterial stimuli (1,6).

Six proteins that showed an expression change similar to that in our previous studies on postnatal preterm pigs (2,3) are glyoxalase 1 (spot 0205), GDI 2 (spot 3509), carbamoyl-phosphate synthetase 1 (spot 6902), galactokinase 2 isoform 2 (spot 4507), aldehyde dehydrogenase, mitochondrial precursor (spot 3609), and chain A, medium-chain acyl-CoA dehydrogenase with 3-thiaoctanoyl-CoA (spot 6401). These proteins represent cellular functions at least partly related to detoxification, energy, and nutrient metabolism, apparently required to facilitate the tissue adaptation response to feeding. Glyoxalase 1 converts cytotoxic α-oxoaldehyde generated from the oxidation of carbohydrates and glycolysis into noncytotoxic d-lactate (13,14). Galactokinase 2 involves in the conversion of galactose to glucose (15), whereas aldehyde dehydrogenase is related to oxidation of aldehydes to carboxylic acids (16). Chain A, medium-chain acyl-CoA dehydrogenase with 3-thiaoctanoyl-CoA (spot 6401) catalyzes the mitochondrial fatty acid beta-oxidation (17), whereas pyruvate dehydrogenase α subunit (spot 5409), succinate-CoA ligase, ADP-forming, β unit (spot 4403), and chain A, structure of glutamate dehydrogenase-Apo form (spot 7508) are all enzymes involved in the citric acid cycle (18). Regulation at all these steps of energy metabolism may be required to facilitate the rapid tissue protein synthesis, as reflected also in increased elongation factor expression [eIF5A (eukaryotic translation initiation factor 5A, spot 1010), elongation factor 1-γ (spot 1002)], and the associated decrease in tissue proteolysis reflected by reduced cathepsin Y (spot 5208) and CNDP dipeptidase 2 (spot 2502) expression. Even before the birth transition, the immature intestine is capable of activating all these metabolic pathways to facilitate growth and functional adaptation. This observed adaptation could in part be specific to the specific formula composition used, i.e. the change in medium-chain acyl-CoA dehydrogenase could reflect a high level of vegetable medium-chain triglycerides in our formula, relative a diet such as mother's milk.

In postnatal preterm pigs, formula feeding rapidly increased apoptosis in the intestine and this could be partly mediated by the proinflammatory effects of formula feeding in combination with microbial actions (19). In this study, formula feeding markedly decreased the proenzyme of caspase 1 (45kD, spot 3208, Fig. 4A; 20), a key marker of apoptosis, suggesting that the drive for apoptosis is less after feeding in utero. This could again be related to the lacking interaction of feeding with bacteria as the cytoskeleton rearrangement of enterocytes has been found to be related to bacterial colonization (21). Nevertheless, there is no doubt that feeding in utero induced significant changes in cellular structure-related proteins as well, as indicated by the differential expression of keratin 9 (spot 1202), 19 (spot 1302), and 10 (spot 3815) and actin γ2 (spot 0210), α2 (spot 4305), and PRO 2619 (spot 4714).

Differentially regulated proteins present in our earlier studies (2,12), but not in this study, may represent proteins influenced by the birth transition, bacterial colonization, and early development of NEC. Among the proteins absent in this study are the stress proteins: heat shock protein (HSP) A5 and 27 (confirmed by Western blot, data not shown), which appeared in our previous studies with the existence of microbial colonization (2,12) showing that the differentiated expression of these proteins might depend on microbial colonization and/or inflammation. These changes are likely some of the early signs of formula-induced mucosal lesions, which, together with the action of bacterial stimuli and reperfusion injuries, may lead to NEC (22). In the current study, the trophic effect of formula in utero might be so mild that the fetal intestine did not have to recruit the HSPs to cope with the stress. Along the same lines, it is interesting that we found increased expression of guanine nucleotide binding protein (G protein), betapolypeptide 2-like 1 (GNB2L1, spot 8204) that may cause increased degradation of hypoxia-inducible factor-1α (HIF-1α; 23), whereas in postnatal preterm pigs formula feeding increased the HIF-1α expression (24). Moreover, proteins related to antioxidation, such as peroxiredoxin 5 (spot 6002) and glutathione synthetase (spot 3504) (25), were down-regulated after fetal formula feeding suggesting a limited drive for antioxidation and detoxification after fetal formula feeding. Finally, carbamoyl-phosphate synthetase 1 (spot 6902) that catalyzes the production of citrulline, arginine, and NO (26) were down-regulated in the intestine of formula-fed fetuses. The amino acids citrulline and arginine are known to play key roles for neonatal gut adaptation and mucosal homeostasis, partly because they are linked to NO synthesis, a pivotal molecule in maintaining tissue perfusion (27–29). Collectively, the above results indicate that fetal formula feeding does not up-regulate to the same extent as postnatal feeding up-regulate tissue markers of NEC such as HSPs and HIF-1α, as well as oxidative stress and tissue perfusion and homeostasis (arginine, citrulline, and NO).

In conclusion, the introduction of enteral formula feeding into the fetal intestine induced a remarkable growth and adaptation response, which in many aspects were similar to that induced by feeding of the preterm newborn after birth. The proteomic analysis of the intestinal proteins before and after fetal enteral feeding showed marked effects on pathways related to energy and nutrient metabolism, probably required to facilitate rapid changes in intestinal growth and function. However, we also identified a number of intestinal proteins that differed in response to feeding between the present fetal pigs and the earlier studies on preterm newborn pigs. Specifically in the fetal pigs, differential regulation of proteins related to bacterial colonization and NEC development was either absent or showed a direction of change not consistent with tissue inflammatory lesions.