The small intestine is able to compensate for the loss of functional surface area. This adaptive response is characterized by distinctive morphological changes that are associated with enhanced absorptive capacity. Enhanced crypt cell proliferation leads to crypt hyperplasia and increased crypt depth [reviewed in Hanson (1982)]. Villus lengthening also occurs and, as we have recently shown, villus enterocytes directly contribute to the early adaptive response by augmenting synthesis of a variety of cellular mRNA and proteins (Rubin et al. 1996). Although enteral nutrients are necessary for these adaptive changes to occur, little is known about the specific extracellular mediators and intracellular mechanisms required for full expression of the adaptive response [reviewed in Saxena et al. (1993)]. Identification of these should suggest strategies for augmenting the adaptive response that would benefit patients dependent on parenteral nutrition such as those with short bowel syndrome resulting from Crohn's disease.

As an approach to defining the molecular mechanisms underlying the adaptive response, we have used subtractive hybridization cloning techniques to identify genes that are differentially regulated in the adapting rat intestine 48 h after 70% resection (Dodson et al. 1996). One group of cDNA clones induced during intestinal adaptation includes cellular retinol binding protein II (CRBP II),⁵ liver fatty acid binding protein, ileal lipid binding protein and apolipoprotein (apo) AIV. All of these encode proteins involved in the absorption, metabolism and trafficking of fatty acids, lipids or retinoids. Induction of these genes in the early adaptive period may occur to augment assimilation of these nutrients specifically or to regulate or modulate the adaptive response.

The latter possibility is most likely for CRBP II. This abundant small intestinal retinol and retinal binding protein plays an essential role in intestinal vitamin A trafficking [reviewed in Levin (1994)]. For example, the magnitude of retinol absorption, esterification with long-chain fatty acids and the secretion of retinyl esters correlate with cellular levels of CRBP II (Levin 1993, Lissoos et al. 1995). Retinol and its metabolites, including retinoic acid, are essential for normal embryogenesis, cell growth and differentiation, and maintenance of epithelial tissues. These functions are thought to be mediated through binding of retinoids to retinoid nuclear receptors. Homo- and heterodimeric receptor complexes modulate the transcription of target genes containing retinoic acid receptor response elements [reviewed in Chambon (1996)]. The retinoid receptors that have been identified include the retinoic acid receptors and retinoid X receptors (RXR). Retinoic acid receptors are activated by all-trans- and 9-cis-retinoic acid, whereas RXR bind exclusively to 9-cis-retinoic acid.

Retinoid receptor response elements have been identified in the CRBP II promoter (Mangelsdorf et al. 1991, Nakshatri and Chambon 1994), and we have demonstrated that retinoic acid regulates CRBP II expression in the human intestinal cell line, Caco-2 (Levin and Davis 1997). Thus, the increased expression of CRBP II in the remnant ileum postresection suggests a regulatory role for retinoids in the adaptive process. Furthermore, it is possible that changes in CRBP II levels after partial small intestinal resection could directly affect intestinal gene expression (and thus the adaptive process) by modulating the synthesis and nuclear delivery of all-trans- and 9-cis-retinoic acid.

To address these issues, we compared the expression patterns of CRBP II and apo A-I, an enterocytic gene that also has an RXR- response element in its promoter [reviewed in Kardassis et al. (1996)]. We observed that the temporal pattern of CRBP II expression in the remnant intestine after partial resection was distinct from that of apo A-I and the fatty acid binding proteins. These results are consistent with the hypothesis that retinoids may play a unique role in the initiation and/or maintenance of the adaptive response. Furthermore, our analysis of the effects of in vivo retinoic acid administration on crypt cell proliferation in the remnant ileum after 70% small bowel resection, indicates that retinoids may modulate the early intestinal adaptive response by stimulating crypt cell proliferation.

MATERIALS AND METHODS

Tissues (n = 3-10 rats per group per time point) were harvested at 2, 4, 8, 16, 24 and 48 h and at 1 wk following surgery for the time course studies. For all studies, rats were killed by sodium pentobarbital overdose (150 mg/kg, intraperitoneal). The intestines were divided into duodenal and jejunal segments (i.e., intestine proximal to the anastomosis) and three 5-cm ileal segments (proximal, mid- and distal) starting 15 cm proximal to the ileocecal valve. The protocols for all animal studies were reviewed and approved by the Washington University Animal Studies Committee.

Northern blots were hybridized to assess the effect of retinoic acid administration on CRBP II, apo A-I and RXR- mRNA levels. RNA samples from the combined duodenal-jejunal segment and from the ileal segments were analyzed for each experimental and control animal.

Rat CRBP II (Li et al. 1986), rat apo A-I (Elshourbagy et al. 1985) and RXR- (Gearing et al. 1993, courtesy of E. Widmark and J. Gustaffsson, Karolinska Institute, Huddinge, Sweden) cDNAs were labeled with [³²P]dCTP to high specific activity and the blots were hybridized at 42°C for 20 h (CRBP II and apo A-I) or 48 h (RXR-) and then washed as described (Levin et al. 1987). The stringency of these hybridization and washing conditions was confirmed to be specific for each of the study genes by Northern blot hybridization. For each cDNA, a single band of appropriate size was noted on Northern blots.

Densitometric quantitative analysis was used to compare expression in experimental resected rats and control sham-resected rats. Dot blots were quantitated by scanning laser densitometry (Levin et al. 1987, Rubin et al. 1989) and Northern blots by NIH Image 1.55 (W. Rasband, National Institute of Mental Health) analysis of digitized images obtained with a UMAX PS-2400X scanner using UMAX Magicscan V1.2 (UMAX Technologies, Fremont, CA). The relative magnitude of ileal expression of the study genes was the same in the proximal, mid- and distal segments. Thus, for statistical analyses of changes in ileal gene expression in the remnant ileum, data obtained from each of these segments were treated as a representative sample of total ileal RNA. All Northern blots were stripped and rehybridized with radiolabeled human glyceraldehyde 3-phosphate dehydrogenase (Tso et al. 1985) to correct for differences in RNA loading and transfer efficiency. This human cDNA hybridizes with rat RNA, producing a single band on Northern blot (data not shown). Levels of intestinal glyceraldehyde 3-phosphate dehydrogenase expression were not modified by intestinal resection or by retinoic acid or vehicle administration.

RESULTS

The expression of the CRBP II and apo A-I genes was two- to fivefold higher in the proximal small bowel (duodenal/jejunal segment) than in the ileum, consistent with maintenance of the normal cephalo-caudad gradient in expression of these genes (Rubin et al. 1989). However, CRBP II and apo A-I mRNA levels were not significantly increased in this segment by partial intestinal resection (data not shown).

The ability of the small intestinal epithelium to rapidly acquire and metabolize plasma-derived all-trans-retinoic acid is well established (Cullum and Zile 1985, Zile et al. 1982). Thus, to minimize dosage effects resulting from differences in absorptive capacity or efficiency, retinoic acid was administered intravenously upon completion of resection surgery. Two doses of retinoic acid were studied because plasma clearance deviates from first-order kinetics dose dependently and relative plasma concentrations of different retinoic acid metabolites vary dose dependently (Swanson et al. 1981). The 18-µg dose is thought to be "physiologic" and the 100-µg dose is well below "pharmacological" amounts (2-3 mg) used in many studies of retinoic acid metabolism [for examples see Cullum and Zile (1985), Swanson et al. (1981), Zile et al. (1982)].

To assess the ability of retinoic acid to stimulate ileal crypt cell proliferation postresection, 5-BrdU incorporation into DNA was quantitated. Representative tissue sections demonstrating 5-BrdU immunostaining in ileal segments of control (vehicle-treated) and retinoic acid-treated resected rats are shown in Figure 3. In the remnant adaptive gut, the number of 5-BrdU-labeled cells per crypt was significantly increased by retinoic acid (100 µg but not 18 µg). At 6 h postresection, the ileal crypts of treated rats contained 53% more labeled cells compared with vehicle-treated controls (Fig. 4; retinoic acid mean cells per crypt = 15 vs. vehicle mean = 9.8, P < 0.001). At 6 and 16 h after resection, on the basis of the total number of cells per crypt, expansion of the total crypt cell population had not yet occurred. Thus, 6 h after resection, retinoic acid increased the labeling index (i.e., 5-BrdU-stained crypt cells/total crypt cells) by 50% (from 20 to 30%; P = 0.002 compared with vehicle-treated rats). In contrast to resected rats, retinoic acid did not stimulate intestinal proliferation in rats that did not undergo surgery. In fact, by 6 h after either dose of retinoic acid, the crypt labeling index was reduced by ~6% (data not shown).

In Figure 4, the number of labeled crypt cells at time 0 reflects the basal proliferation rate before surgery. The increase in the number of 5-BrdU-labeled crypt cells in the resected vehicle control group reflects changes induced by intestinal adaptation alone. By 16 h, there were no significant differences in 5-BrdU labeling between the retinoic acid-treated group and the vehicle-treated group. However, as expected, the labeling index was greater in both groups of resected rats compared with nonresected rats.

In addition, at 6 h after surgery and retinoic acid administration, there were no differences in steady-state RXR- mRNA levels, whereas, at 16 h, RXR- mRNA levels were decreased 40% (P < 0.05; 100 µg dosage; data not shown). The regulation of RXR- by retinoids has been demonstrated only in F9 teratocarcinoma cells and in the vitamin A-deficient rat testis after readministration of retinyl acetate (van Pelt et al. 1992, Wan et al. 1994). These observations suggest that retinoic acid receptors (RAR) and RXR may help to regulate RXR- expression and underscore the possibility that changes in intestinal gene expression may be influenced by changes in the quantity and type of retinoic acid receptors. In addition, changes in the concentrations and binding of receptor ligands and interactions with other trans acting factors are also important.

DISCUSSION

As discussed above, crypt cell proliferation is enhanced in the adapting remnant intestine following partial resection. The direct effects of vitamin A and retinoic acid on intestinal epithelial proliferation and differentiation have not been reported. Studies using different models for inducing vitamin A deficiency in rats have produced conflicting results. For example, when rats with absent vitamin A stores were maintained on retinoic acid, there were no effects on the rate of division or migration up the villus of intestinal epithelial cells (Rojanapo et al. 1980a). However, using rats in the early stages of vitamin A deficiency (i.e., rats whose weight had plateaued after weaning to a vitamin A- and retinoic acid-deficient diet), there was significant prolongation of the S2-phase of the cell cycle in the jejunal crypts (Zile et al. 1977). In these vitamin A-deficiency models, the morphological appearance of the small intestine was nearly normal except for a quantitative decrease in goblet cells. Thus, these studies have provided indirect evidence supporting a role for retinoids in maintenance of goblet cells (Rojanapo et al. 1980b) but not enterocytes, enteroendocrine cells or Paneth cells.

To assess directly the ability of retinoic acid to modulate crypt cell proliferation in the postresection adapting intestine, crypt cell incorporation of 5-BrdU was assessed. Administration of 100 µg of retinoic acid resulted in a 53% increase in the number of proliferating ileal crypt cells at 6 h after small bowel resection. Thus retinoic acid may have a role in enhancing the intestinal adaptive response. Although our data are compatible with a stimulatory role for retinoic acid in small intestinal epithelial renewal after partial resection, administration of retinoic acid to normal (i.e., nonoperated) rats resulted in a 6% decrease in the crypt labeling index. Thus the stimulatory effects of retinoic acid on crypt cell proliferation may be specific to the adapting intestine or may vary depending on the physiological context.

The intestinal effects of retinoic acid are consistent with those in other epithelial tissues in which it has been shown to modulate cell proliferation and/or differentiation. In vivo demonstrations of retinoid inhibition of cell proliferation and stimulation of differentiation include the use of postnatal retinoic acid to increase the number of pulmonary alveoli in rats (Massaro and Massaro 1996) and the use of retinoids to treat and prevent premalignant and malignant skin lesions [reviewed in Lotan (1996)]. In vitro, retinoids can also have the opposite effect as illustrated by the suppression of terminal differentiation and stimulation of proliferation in cultured keratinocytes [reviewed in Fisher and Voorhees (1996)]. Retinoids are also active in nonepithelial cell types as indicated by differentiative effects on HL60 promyelocytic cells and F9 teratocarcinoma cells and by the successful use of retinoic acid as differentiation therapy in promyelocytic leukemia [reviewed in Chomienne et al. (1996) and Gudas et al. (1994)].

As shown in Figure 4, by 16 h after resection, the number of cells incorporating 5-BrdU was the same whether retinoic acid was or was not administered at the time of surgery. Thus, these studies indicate that the usual increase in ileal crypt cell proliferation after massive small bowel resection can be induced to occur earlier by treatment with retinoic acid. The equalization of the number of labeled cells by 16 h after retinoic acid administration suggests that the acute effect is limited to a discrete population of G1 cells already primed to initiate DNA synthesis. Although these data might suggest that retinoid effects are limited to the immediate postresection period, the quantity and time course of retinoic acid delivery to the intestinal mucosa were not addressed. Pharmacokinetic studies indicate that circulating retinoic acid metabolites but not retinoic acid would still be present by 16 h after either dosage (Swanson et al. 1981). Thus, it will be worthwhile to evaluate whether regular dosing with retinoic acid can further expand the proliferating crypt cell population.

The temporal, spatial and cellular patterns of CRBP II expression after resection were determined as a basis for addressing a potential role for CRBP II in early adaptation. CRBP II mRNA levels in the remnant ileum were increased by 8 h and remained elevated for at least 1 wk postoperatively. In contrast, apo A-I mRNA levels are not increased at 8 h, are maximally increased at 16 h, are back to base line by 24 h and, as previously demonstrated, are still at base line for at least 1 wk after small bowel resection (Rubin et al. 1996). Thus, despite the presence of a common RXR- response element in the promoter of each of these genes, the temporal regulation of apo A-I and CRBP II expression during adaptation is different. This observation underscores the potential importance of other transcription factors. For example, the apo A-I promoter also contains binding sites for the orphan receptors ARP-1, HNF-4, Ear3-COUP-TF and a response element for a peroxisome proliferator-activated receptor [reviewed in Kardassis et al. (1996)]. The CRBP II promoter contains an element that efficiently binds HNF-4 and ARP-1 in vitro (Nakshatri and Chambon 1994). Although the functions of these receptors are not known, all but HNF-4 can form heterodimers with RXR and thus are potentially regulated by retinoids. In addition, peroxisome proliferator-activated receptors are implicated in fatty acid signaling pathways, which could also be important in the adapting intestine. Despite the regulation of CRBP II expression by retinoic acid in Caco-2 cells (Levin and Davis 1997), the potential importance of other regulators of CRBP II expression postresection is underscored by the inability of exogenous retinoic acid to consistently increase CRBP II mRNA above the levels resulting from surgical resection alone. In addition, the fact that retinoic acid administration downregulated RXR- expression by 16 h postresection underscores the possibility that changes in intestinal gene expression may be influenced by variation in the number and type of retinoic acid receptors, along with interactions with other trans acting factors and changes in the concentrations and binding of their ligands.

The sustained increase in CRBP II expression at 1 wk after resection may reflect an adaptive mechanism to increase retinol absorption per unit of small bowel surface area. Such a response would be consistent with our demonstration that CRBP II levels are determinants of vitamin A absorption in Caco-2 cells (Levin 1993, Levin and Davis 1997, Lissoos et al. 1995) and evidence that intestinal CRBP II levels are regulated by physiological stimuli such as lactation (Levin et al. 1987) and vitamin A deficiency (Rajan et al. 1990).

Conversely, the induction of CRBP II expression before 8 h postresection may reflect a regulatory role for retinoids in the early stages of the adaptive response. The normal intestine and the adapting intestinal remnant are likely target tissues for retinoic acid because they are actively proliferating and differentiating epithelial tissues. Furthermore, the small intestine is able to absorb, metabolize and synthesize retinoic acid. Because CRBP II clearly plays an important role in intestinal vitamin A absorption and metabolism [reviewed in Levin (1994)], increases in CRBP II levels in the adapting remnant intestine might directly affect cellular levels of retinoic acid, 9-cis-retinoic acid and other putative ligands of nuclear retinoid receptors. Nevertheless, CRBP II expression remains confined to villus-associated enterocytes postresection, and thus, any regulatory effects deriving from the observed increase in CRBP II expression are probably limited to this cell population.

In conclusion, these studies have shown that retinoic acid can stimulate crypt cell proliferation in the adapting remnant intestine by 6 h after partial small bowel resection. The early and sustained increase in CRBP II expression suggests that CRBP II may have a specific role in mediating the effects of retinoids on the adapting intestine. These observations indicate that retinoids are likely modulators of intestinal adaptation after loss of functional small bowel surface area.

ACKNOWLEDGMENTS