After birth the gastrointestinal tract of the neonate is exposed to a vast array of food, bacterial, and environmental antigens. A properly functioning mucosal barrier and subsequent appropriate function of the mucosal immune system is essential in controlling responses to these antigens (1). In neonates, intestinal permeability is transiently increased (2,3). In the rat, gut closure (a defined stage in development when intestinal permeability decreases to adult levels) occurs at d 21, due to active maturational events, which include activation of the mucosal immune system (36).

TGF-β is a multifunctional polypeptide growth factor. Five isoforms of TGF-β (711), are synthesized as a larger protein precursor consisting of the mature TGF-β dimer and latency associated protein. Besides transcription/translation regulation, the expression of active TGF-β is dependent on the release of the biologically active dimer (12). TGF-β1 is the most abundant isoform in the mucosa of the adult gut (11). TGF-β is synthesized by epithelial cells, macrophages, and lymphocytes, and has a central role in modulating gut mucosal cell growth (inhibiting gut epithelial cell growth), differentiation, migration, gut immune responses, extracellular matrix synthesis, and wound repair (13,14).

Early weaning or lack of breast feeding in infants has been associated with the development of autoimmune diseases, such as celiac disease and type 1 diabetes mellitus (15). This sensitization to self or food antigens may result from premature exposure to antigen in the periphery (16). In this study, we determine the appearance of TGF-β1-secreting cells in the suckling rat pup proximal and distal small intestine and assess maternal milk concentrations of TGF-β in a time course from birth to post weaning.


Animals. Pregnant Hooded Wistar rats were obtained from the University of Adelaide breeding colony (Adelaide, South Australia). The study was carried out with the approval of the Women's and Children's Hospital Animal Ethics Committee, Adelaide, Australia.

Rat pups (n = 8 per time point) were killed by decapitation at 3, 10, 19, 22, 29, and 84 d after birth. The gastrointestinal tract was rapidly excised and placed onto an ice-cold glass slab. The small intestine was isolated and two 1-cm segments were collected from the proximal duodenum starting at 1 cm caudal to the pyloric sphincter. These 1-cm segments were also collected from the ileum, ensuring that Peyer's patches were present. Tissue was either embedded and cooled by iospentane surrounded by liquid nitrogen or fixed in 4% buffered formaldehyde (pH 7.0) for 24 h and embedded in paraffin.

Maternal rats were milked at various days after birth. The animals were given 5 IU/mL oxytocin intramuscularly (Syntocinon, Sandoz-Pharma, Basle, Switzerland) to induce lactation, 60 min before milking. At the same time, the rat pups were also removed from the maternal rats. Milk was collected from the lactating nipple by using silastic tubing connected to a gentle suction. After collection, aliquots of the whole milk were centrifuged at 1700 × g to remove cells and lipid to obtain the aqueous phase. The lipid layer was obtained from a fresh aliquot by centrifuging the milk at 680 × g for 10 min, the aqueous and lipid fractions were then stored at -70°C until assayed. The pellet from the 680 × g centrifugation was washed three times in PBS. Aliquots of cells in the pellet were sonicated for 5 min using a Soniclean (Lab Supply, Australia), at room temperature. The sonicated cell pellets were centrifuged at 1000 × g for 5 min, and supernatants were collected and stored at -70°C until assayed.

Stomach contents from 6 pups per time point, with full stomachs, at d 6, 10, and 19 of age, respectively, were collected, diluted in PBS, and centrifuged at 1000 × g for 5 min. The supernatants were collected and stored at -70°C until assayed.

Immunohistochemical analysis. Immunohistochemical analyses was carried out on segments of the duodenum and ileum. Three-micrometer sections were cut from paraffin-embedded tissue and placed on gelatin-coated slides. Sections were deparaffinized with xylene and rehydrated in graded ethanol in water. Sections were incubated in 1% (vol/vol) hydrogen peroxide for 30 min at room temperature to quench endogenous peroxidase activity. Tissue sections were then incubated with 1% normal chicken IgY (R&D Systems, Minneapolis, MN) for 30 min at room temperature to block non-specific binding of primary antibody. The blocking antibody was decanted, and 100 µL of a 1/500 dilution of chicken anti-TGF-β1 (R&D Systems) were added, and the sections were incubated overnight at 4°C. The sections were then washed twice in Tris-buffered saline containing 0.01% Tween 20 for 5 min each wash. Sections were blocked with 100 µL of 1% normal rabbit serum (Sigma Chemical Co., St. Louis, MO) for 30 min. After removal of the blocking serum, biotinylated rabbit anti-chicken IgG (H+L) (Zymed, San Francisco, CA) was added as the secondary antibody, and the sections were incubated for 60 min at room temperature. The sections were then washed twice in Tris-buffered saline, as above. An avidinbiotin peroxidase complex (Dako, Carpinteria, CA) was then applied to the sections for 30 min, after which sections were washed twice in Tris-buffered saline, as above. Substrate, 1 mg/mL diaminobenzidine (Dako) and 0.02% hydrogen peroxide, was then added to the sections, and the tissue was incubated for 10 min. After washing in distilled water, the sections were counterstained with methyl green or hematoxylin (Sigma), dehydrated with graded ethanol, and mounted in Ultramount mounting medium (Histolabs, Riverstone, NSW, Australia).

All antibody dilutions were made in 0.2% BSA (Sigma) in Tris-buffered saline. Control samples included: 1) sections incubated with normal chicken IgY (R&D Systems) added as the primary antibody in place of the chicken anti-TGF-β1; 2) sections incubated only with secondary antibody before avidin biotin peroxidase labeling; and 3) sections incubated with primary antibody, chicken anti-TGF-β1, which had been pre-incubated with 10 ng/mL recombinant TGF-β1 (R&D Systems) for 60 min at room temperature to absorb out anti-TGF-β1 activity.

Analysis of TGF-β1 expression. Computer-aided video image analysis was used to examine TGF-β1 expression during the development of the rat intestine. Sections were examined with an Olympus BH-2 LM microscope at a 10× magnification. The image was captured by a monochrome high resolution CCD camera (TM-7 camera, Pulnix Industrial Products Division, Claytion, Vic, Australia) and analyzed using an image analysis software program "Video Pro" (Leading Edge, Adelaide, SA, Australia) (17). Villi, crypts, and lamina propria were analyzed, only intact whole villi were assessed. For each specimen, at least 15 microscopic fields were analyzed. On each microscopic field, the total area was measured by drawing a 1-pixel thick line around the villi and crypt area. This was followed by a measurement of the total number of stained cells within this area, and their intensity, density, and IOD. Intensity of labeling was estimated by the ratio of IOD of the color reaction over the total area measured, i.e. OD equals IOD/total area of the field.

TGF-β mRNA in situ hybridization. The in situ hybridization kit for human TGF-β1 (R&D Systems, Oxon, UK) contained a cocktail of seven digoxigenin-labeled oligonucleotide probes, complementary to a total of 210 nonoverlapping coding sequence bases of human TGF-β1 mRNA. The TGF-β1 probe does not hybridize with TGF-β2 and β3 mRNA (R&D Systems). The kit also contained a TGF-β1 sense probe, including the same regions of sequences as the antisense probe cocktail. Following a protocol modified from Wang et al. (18), in situ hybridization was performed on 8-µm cryostat sections of the rat duodenum which were mounted on 3-aminopropyl-triethoxysilane-coated glass slides and fixed for 5 min in 4% paraformaldehyde in 10 mM PBS (pH 7.4). Sections were first acetylated for 10 min in 0.25% acetic anhydride in 0.1 M triethanolamine, washed in water, and dehydrated by graded ethanol. Sections were then treated with 25 µL of hybridization mix (preheated for 5 min at 85°C), containing 0.75 µg/mL sense or antisense probe, 50% formamide, 10% dextran sulfate, 0.05% Triton X-100, 500 µg/mL herring sperm DNA (Boehringer Mannheim), 0.05% polyvinylpyrrolidone, and 5× SSC (750 mM NaCl and 75 mM sodium citrate, pH 7.0). Sections were covered with glass coverslips and incubated in a humidified chamber for 18 h at 37°C. Post-hybridization washes were undertaken at 34°C with gentle shaking as follows: once in 2× SSC solution for 5 min to remove coverslips; twice in 2× SSC for 20 min; once in 1 × SSC for 10 min; and once in 0.1 × SSC for 10 min. After blocking of nonspecific binding sites with 10% normal sheep serum for 30 min, the sections were incubated for 30 min at 37°C with an alkaline phosphatase-coupled sheep anti-digoxigenin IgG (1:450, Boehringer Mannheim), and developed by incubating in NBT/X-phosphate substrate for 18 h in the dark according to the supplier (Boehringer Mannheim), in the presence of 2.5 mM levamisole (Sigma) to inhibit endogenous alkaline phosphatase activity. Color reaction was stopped by incubating sections with 0.01 M Tris containing 1 mM EDTA (pH 8) for 10 min, before the sections were lightly counterstained with methyl green and mounted in glycerol gel-mounting medium (Dako).

Quantitative determination of TGF-β concentration in milk. Sandwich type ELISA kits (Quantikine, R&D Systems) were used for the quantification of active TGF-β1 and TGF-β2. Milk (aqueous, lipid, and pellet phase) from rats with 6-, 10-, and 19-d-old pups was assayed with and without acid activation (incubation of milk in 1 M (final) acetic acid for 60 min at room temperature). All measurements were done in triplicate. Total milk protein levels were determined by the Lowry (19) method. In addition the gastric contents of suckling rat pups were assayed by the above method.

Statistical analysis. Values were expressed as mean ± SD. Statistical analysis was carried out by the nonparametric method of Mann-Whitney (p < 0.05). TGF-β levels in milk (mean ± SD) was assessed by the t test (p < 0.05).


TGF-β in maternal milk. The change with time of TGF-β2 levels in maternal milk is shown in Table 1. Acid activation of the milk samples resulted in significant concentrations of TGF-β2 being detected in the aqueous phase, variable concentrations in the lipid phase, and low levels in the sonicated cell pellet. TGF-β2 levels were highest at the earliest time point measured (d 6), and progressively decreased over the suckling period in the lipid and aqueous phase. On d 19 postpartum, no TGF-β2 was detected in the lipid phase. The mean concentrations of TGF-β2 in the aqueous phase, expressed as a proportion of the total protein in milk were 7.7 ng of TGF-β2/mg of total milk protein, 4.6 ng of TGF-β2/mg of total protein, and 1.5 ng of TGF-β2/mg of total protein on d 6, 10, and 19, respectively. Total milk protein levels were 43 ± 16, 34 ± 12, and 41 ± 12 mg/mL on d 6, 10, and 19, respectively. The TGF-β2 present in milk was predominantly in the latent form. Milk without acid activation had active TGF-β2 at concentrations of 6 ± 0.5, 6 ± 0.9, and 4 ± 1 ng/mL in the aqueous phase on d 6, 10, and 19, respectively. In the lipid phase without acid activation, TGF-β was detected in the d-6 samples at only 34 ± 6 ng/mL. The levels of TGF-β1 (with or without acid activation) were below the level of detection of the ELISA assay (<31.2 pg/mL TGF-β1).

Table 1 TGF-β concentration (ng/mL ± SD) in maternal rat milk

The TGF-β2 concentration in gastric washouts of suckling rat pups was 8.5 ± 6 ng/mL at d 6, 7 ± 5 ng/mL at d 10, and 5 ± 9 ng/mL at d 19.

Mucosal TGF-β expression. To determine when endogenous TGF-β1 expression occurred in the rat pup proximal and distal small intestine, sections of duodenum and ileum were assessed in a time course from birth to postweaning by immunohistochemistry. TGF-β1-labeled cells in the duodenum increased over time during the weaning period (Figs. 1 and 2, A-D). Only a few TGF-β1-positive cells were detected in 3-d-old rat pup intestine; goblet cells were stained, as seen in Figure 2A. In 10-d-old pups the number of TGF-β1 positive cells detected increased, with strong staining of goblet cells, as well as some faint labeling of enterocytes in the villi (Fig. 2B). No TGF-β1-positive cells were detected in the lamina propria; at d 3 or 10. At d 19, a few TGF-β1-positive cells were detected in the lamina propria of the duodenum, and enterocytes were strongly labeled along the villi in duodenum (Fig. 2, C and E). At d 29 the labeling pattern in the duodenum (Fig. 2D) was similar to that seen at d 19. Similar staining patterns to those seen in the duodenum were observed in the ileum at d 3 and 10. In the ileum at d 3 only goblet cells labeled (data not shown). At d 10 faint labeling of enterocytes and goblet cells was observed (Fig. 2F). In the immature ileum on d 19 (Fig. 2G) enterocytes are characterized by supranuclear vacuoles. TGF-β labeling was detected in the cytoplasm of these enterocytes all along the villi. In the ileum at d 29 (Fig. 2H) the supranuclear vacuoles are absent and staining can be seen on epithelial cells along the villi as well as on goblet cells.

Figure 1
figure 1

Immunohistochemical labeling of TGF-β1-positive cells in the duodenum over a time course from birth to post-weaning expressed as number (mean ± SD) of labeled cells/mm2 of rat mucosa. Statistics: d 3 vs 10, 19, 22, 29, and 84, p < 0.05 significant; d 10 vs 19, 22, 29, and 84, p < 0.05 significant; d 19 vs 22, 29, and 84, p < 0.05 significant; and d 22 vs 29 and 84, and d 29 vs 84, p > 0.05 not significant.

Figure 2
figure 2

Immunohistochemical labeling of cells expressing TGF-β1 peptide in infant rat duodenum and ileum. Immunohistochemical labeling of (A) TGF-β1-positive cells in paraffin sections of rat duodenum at d 3 (400×), (B) d 10 (400×), (C) d 19 (400×), (D) d 29 (400×), (E) d 19 (1000×), and (F) ileum at d 10 (400×), (G) ileum d 19 (400×), and (H) ileum d 29 (400×). Negative control (I) ileum; anti-TGF-β1 preabsorbed with recombinant TGF-β1 (400×).

The immunostaining could be abolished by preincubation of the primary antibody with recombinant TGF-β1 (Fig. 2I, ileum) (duodenum, data not shown). Other negative controls (normal chicken IgY, as well as absence of the secondary antibody) also consistently gave negative staining (data not shown). Quantification of labeling of cells in the duodenum by video image analysis showed that there was a significant increase in the total number of TGF-β1-positive cells detected at d 19 (1914 ± 307 cell/mm2 when compared with d 10 (801 ± 205 cells/mm2) or d 3 (327 ± 172 cells/mm2) post-partum (Fig. 1). After d 22, there was no significant difference in the numbers of TGF-β1-positive cells detected (1384 ± 223, 1262 ± 195, and 1312 ± 164 cells/mm2) on d 22, 29, and 84, respectively. IOD measurements showed that TGF-β1 staining increased from d 3 to 19 before stabilizing thereafter (Fig. 3). There was no significant difference in the intensity of TGF-β1 staining (OD) of cells over the time course from d 3 to 84.

Figure 3
figure 3

Labeling intensity (mean ± SD) of TGF-β1-labeled cells in rat duodenum, as estimated by the ratio of integrated OD of the color reaction over the total area of mucosa measured by image densitometry. Statistics: d 3 vs 19, 22, 29, and 84, p < 0.05; d 10 vs 19, 29, and 84, p < 0.05; d 19 vs 29 and 84 p > 0.05 not significant; d 3 vs 10, d 19 vs 22, and d 22 vs 29 and 84, p > 0.05 not significant.

TGF-β-positive cells were also detected in the colon. TGF-β peptide was located in the upper crypt and surface epithelium in 19-, 22-, and 84-d-old animals (data not shown).

TGF-β-positive cells in the lamina propria. TGF-β1-positive cells were also present in low numbers in the duodenal lamina propria in older suckling rats and adults. No TGF-β1-labeled cells were detected in 3- or 10-d-old rat pups in the lamina propria (Fig. 4). A few cells (31 ± 13 cells/mm2) were detected on d 19. The number of TGF-β1-positive cells progressively increased over the time course (Fig. 4).

Figure 4
figure 4

Immunohistochemical labeling of TGF-β1-positive cells in the lamina propria of the duodenum over a time course from birth to post weaning, expressed as number (mean ± SD) of labeled cells/mm2 of rat lamina propria. Statistics: d 3, 10, 19, 22, and 29, vs 84, p < 0.05 significant.

TGF-β1-labeled cells in Peyer's patches in the ileum were counted on d 22, 29, and 84. Low numbers of TGF-β1-positive cells were detected throughout the follicular and parafollicular zones of the Peyer's patches at these times. No significant increase in number (i.e. 16.5 ± 9.1, 17.2 ± 9.62, and 18.2 ± 8.5 cells/mm2) was observed over the days assessed.

TGF-β mRNA by in situ hybridization. In situ hybridization was performed on frozen tissues of the neonatal rat small intestine. No TGF-β1 mRNA was detected in the duodenum of 3-d-old (data not shown) and 10-d-old rats (Fig. 5A). At d 19 (Fig. 5B) and 22 (data not shown), the TGF-β1 mRNA signal was localized in the villi, with staining detected on enterocytes and goblet cells. By d 29 (Fig. 5C), hybridization was intense, with a similar intensity as observed in d-84 rats (data not shown), distributing mainly on the villous epithelium, particularly at the upper villi, with other regions being negative, although very faint labeling was seen in the crypt area in some sections. No labeling was seen with three negative controls (sections pretreated with RNase A to degrade mRNA, in the absence of the antisense probe, or in the presence of a sense probe) (Fig. 5D).

Figure 5
figure 5

In situ hybridization in frozen sections of rat duodenum. (A) TGF-β1 mRNA at d 10 (250 ×), (B) d 19 (250 ×), (C) d 29 (250 ×), and (D) negative control with a sense probe on d-29 tissue (250 ×).


We have shown that TGF-β2 levels are higher in early milk, but reduce toward weaning. On the other hand, endogenous TGF-β1 expression increases over the weaning period as the maternal source of TGF-β2 decreases. Although this decline might be compensated for by increased volume intake in pups at d 10, the effect of TGF-β2 in milk by d 19 would be small as consumption of milk has reduced by mid-weaning. TGF-β2 but not TGF-β1 is present in maternal rat milk. TGF-β isoforms have been shown to be equivalent in function in vitro in terms of biologic activity, although differences in potency on different cell types do occur (20).

Feeding antigens to newborn animals may result in immune priming for humoral as well as cell-mediated immune responses, whereas tolerance occurs when the same antigen is fed to adult animals (21). Because oral tolerance is not fully developed during the suckling period, modulation of the immune response of the infant may be mediated by milk factors. The regulation of the suckling infant's immune response could be due, at least in part, to TGF-β present in milk. Human colostrum has high levels of TGF-β2 (1365 ± 242 ng/mL). Lower levels (925 ± 212 ng/mL) are detected in late milk (22). Importantly, maternal TGF-β2 in breast milk is essential for the normal function and survival of newborn mice. TGF-β1 null mice progress toward a wasting disease with systemic inflammation, and die after weaning (23). TGF-β1 null newborn pups born to TGF-β heterozygotes have detectable levels of TGF-β in tissues, whereas those born to null females were negative for detectable TGF-β using histochemical analysis. Labeled TGF-β given orally was shown to cross the intestine and could be recovered intact (active form) in various tissues, implying a role for maternal rescue of TGF-β in the neonate (23). We have detected nanogram levels of active TGF-β in the gastric contents of suckling rat pups. Further activation of milk-borne TGF-β may occur in the in the small intestine as a consequence of decreasing pH due to lactose fermentation or binding of latent peptide to receptors in the intestine (23). The autoimmunity associated with TGF-β1 deficiency was reported to result from an increase in MHC class II molecule expression on antigen presenting cells, leading to active proinflamatory cytokine expression (24).

Weaning in the rat is a gradual process from d 15 to 28 (25). We observed an increase in the number of TGF-β1-producing cells over the weaning period; the number peaked at d 19 and then leveled off. In neonates, intestinal barrier function is immature, allowing passage of antigens across the epithelium (2,3,16). Production of endogenous TGF-β1 in the intestine may play a role in maintaining the mucosal barrier and integrity, and therefore limit antigen exposure in the periphery (26). TGF-β has been shown to reduce the capacity of interferon-γ or the human pathogen Cryptosporidium parvum to disrupt epithelial cell barrier function. Moreover, this effect was prolonged over a number of days after initial exposure to TGF-β (26). Maternal milk also promotes maturation and development of barrier function (27). Breast-fed rabbits when fed BSA have lower plasma levels of this antigen than do formula-fed rabbits of an equivalent age. The passage of antigen decreases with increasing age of the infant (16). We have shown that TGF-β1 production in the intestine is increased at the time that antigen passage is reported to be decreasing.

TGF-β has an effect on the growth and differentiation of the epithelium (9,26,28,29). TGF-β1-positive cells in our study on infant intestine were detected along the villi where well differentiated and mature cells are present, and to a lesser extent in the crypt area where the undifferentiated immature cells are located. This localization of TGF-β1 may function to arrest cell growth as the enterocytes leave the proliferative zone and may help to maintain intestinal integrity along the villus at the epithelial surface. TGF-β is expressed in the neonatal gut during mouse embryogenesis (30). In adult intestine, the distribution of TGF-β1 has been controversial. However, Northern blot analysis and histochemical analysis has shown that TGF-β1 expression is predominantly on cells located on the villus tip with no TGF-β1 in the crypts, suggesting that TGF-β1 may function to arrest growth of cells emerging from the crypts. Only low level expression was detected in cells in the lamina propria of the adult gut (11).

Active TGF-β expression is highly regulated. Many cell types have been reported to be able to synthesize it and have cell surface receptors for binding it, including early goblet cells (29,31). We observed that the immunohistochemical distribution of TGF-β1-labeled cells predominantly correlated with the distribution of the corresponding TGF-β1 mRNA with both peptide and mRNA expression localizing in the villus. This would imply that protein accumulates close to the site of synthesis, and that TGF-β1 mRNA undergoes translation throughout the time course studied. Weak peptide expression can be seen at later time points in the crypts; mRNA labeling in the crypts was very weak in comparison to the epithelium at d 29 (Fig. 5C). It has been previously shown that immunohistochemical distribution of TGF-β does not always correlate with the distribution of its mRNA peptide that may accumulate away from its site of synthesis (20).

At weaning, the intestine is exposed to a large influx of antigens, which stimulate the gut-associated lymphoid tissue (46). The gut-associated lymphoid tissue favors induction of immune cells that secrete cytokines including IL-4, IL-5, IL-10, and TGF-β (32). TGF-β is a potent immunosuppresive cytokine. We observed TGF-β1- positive cells in the epithelium, as well as in low numbers in the lamina propria around mid-weaning when maternal TGF-β levels were decreasing, and antigenic stimulation by food antigens is increasing. Only low levels of TGF-β-producing cells in the lamina propria have also been reported by Barnard et al. (11). Tolerance develops only around mid-weaning in the rodent (21). During inflammation in the intestine, TGF-β-positive cells have been shown to increase in number in the lamina propria (33). Interaction of lipopolysaccharide, a major enteric antigen in the intestine, with T helper cells has been reported to increase mRNA for TGF-β, as well as expression of TGF-β receptors and production of active TGF-β in lymphocytes (20). Priming of CD8 T cells to Staphylococcus B in the absence of TGF-β results in production of lymphocytes which produce interferon-γ, a proinflamatory cytokine, whereas priming of CD8 cells in the presence of TGF-β results in generation of cells which secrete IL-10 as well as TGF-β and act to down-regulate Staphylococcus B T cell proliferation of CD4 and CD8 cells (34). In the neonatal intestine CD8-positive cells predominate in the lamina propria (35).

The epithelial cell and the cytokine profile secreted by the epithelial enterocyte is now being recognized as a primary player in determining the type of immune response generated by the intestine (36,37). The enterocyte expresses classical and nonclassical MHC class antigens and is capable of presenting antigen to the mucosal immune system (38). TGF-β is known to regulate MHC antigen expression as outlined by studies on TGF-β null mice and down-regulate inflammation (24). The role of maternal TGF-β in milk may be to regulate immune responses in the neonate until weaning (23,39,40).

We have shown that maternal milk TGF-β2 levels are highest in early milk and decrease toward weaning. In contrast, expression of TGF-β1 in the intestine increased with age, with a significant increase in expression after mid-weaning.