The exocrine pancreas of the cystic fibrosis (CF) mouse (cftrm1UNC ) is only mildly affected compared with the human disease, providing a useful model to study alterations in exocrine function. The CF mouse pancreas has ∼50% of normal amylase levels and ∼200% normal Muclin levels, the major sulfated glycoprotein of the pancreas. Protein biosynthetic rates and mRNA levels for amylase were not altered in CF compared with normal mice, and increases in Muclin biosynthesis and mRNA paralleled the increased protein content. Stimulated pancreatic amylase secretion in vitro and in vivo tended to be increased in CF mice but was not statistically significant compared with normal mice. We show for the first time that the CF mouse duodenum is abnormally acidic (normal intestinal pH = 6.47 ± 0.05; CF intestinal pH = 6.15 ± 0.07) and hypothesize that this may result in increased signaling to the exocrine pancreas. There were significant increases in CF intestinal mRNA levels for secretin (310% of normal,P < 0.001) and vasoactive intestinal peptide (148% of normal, P < 0.05). Furthermore, CF pancreatic cAMP levels were 147% of normal (P < 0.01). These data suggest that the CF pancreas may be chronically stimulated by cAMP-mediated signals, which in turn may exacerbate protein plugging in the acinar/ductal lumen, believed to be the primary cause of destruction of the pancreas in CF.
- bicarbonate ion
- adenosine 3′,5′-cyclic monophosphate
- cystic fibrosis
- vasoactive intestinal peptide
in patients with“severe” mutations in cystic fibrosis transmembrane conductance regulator (CFTR), the pancreas is seriously affected with progressive and eventual widespread destruction of the exocrine portion (28). By contrast, in the mouse model of cystic fibrosis (CF), where the CFTR gene has been disrupted, the exocrine pancreas exhibits only mild CF pathologies (10, 12, 13, 19, 25). The milder pathology in the mouse may be due to an alternative Cl− channel that is able to compensate for loss of CFTR in the mouse pancreas (5).
Although the CF mouse pancreas is mildly affected, it does exhibit some CF-like pathologies, the explanation of which may yield important insights into CF pathogenesis. We reported that the CF mouse pancreas has fewer and smaller zymogen granules (10), and others found that amylase levels are decreased (25) compared with normal. In addition, the luminal plasma membranes of acini tend to be dilated and filled with protein aggregates (12, 13, 19).
It is controversial whether the pancreatic acinar cell is a site of expression of CFTR. The strongest support for CFTR expression in acinar cells comes from immunocytochemical labeling for CFTR in mouse pancreas (41), although CFTR Cl− channel activity was not demonstrated to be present in acini. Therefore, it is not certain whether the effects observed in CF mouse pancreatic acini are due to loss of CFTR in acinar cells, to duct cell dysfunction, or to indirect effects of CF. Because endocrine cells in the gastrointestinal tract regulate the pancreas, changes in the pancreas may be due to alterations in these distal sites. Indeed, the most striking phenotypic changes in the CF mouse are its poor nutritional status and obstructive pathologies in the small intestine (34). In this report, we investigated changes in the CF mouse pancreas and provide evidence that such changes are mediated by alterations in the lumen of the small intestine due to loss of functional CFTR.
MATERIALS AND METHODS
Mice heterozygous for deletion of the CFTR gene (CFTR knockout mouse;cftrm1UNC ) were obtained from Jackson Labs (Bar Harbor, ME) and bred to obtain mice with normal or deleted CFTR alleles (34). Except for body weight determinations, where heterozygous animals were included, all experiments used homozygous normal and CFTR knockout mice only. Our CFTR knockout colony has been bred into the C57BJ/6 background, in which others have shown results in a more severe and consistent CF phenotype (33). Mice were genotyped as described by PCR analysis of tail-snip DNA (13). All mice were maintained from postnatal age of 10 days on Peptamen (Nestle, Deerfield, IL), a complete liquid diet, to prevent intestinal obstruction that occurs in CF mice on solid mouse chow (18). Mice were used after postnatal day 40 to avoid confounding morphological changes that occur transiently during normal pancreatic development (13).
Mice were anesthetized with pentobarbital sodium and perfuse fixed through the aorta with 4% paraformaldehyde in phosphate-buffered saline. Cryosections (2.5 μm) were prepared and immunolabeled for amylase and Muclin with rabbit primary antibodies and donkey anti-rabbit-FITC secondary antibody (Jackson ImmunoResearch, West Grove, PA) as previously described (10).
Stimulated pancreatic secretion in vitro and in vivo.
Pancreatic acini were prepared from normal and CF mice by collagenase digestion (27), and equal aliquots of cells were used on the basis of DNA content (3). Cells were incubated at 37°C for 30 min in the absence (basal) or presence of the indicated stimuli for 30 min. The cells were pelleted, and amylase in the supernatant was measured as described (26). In vivo pancreatic secretion was stimulated by a single intraperitoneal injection of the cholinergic agonist pilocarpine (10 μg/g body wt) (14). At the specified time points, the mice were killed, the pancreas was removed and homogenized, and DNA, amylase, and Muclin content were measured. Muclin was measured by immuno-dot-blot using a rabbit antibody (8, 9) or by Western blot using comparison to a standard curve of known Muclin amounts (12) (seeresults).
In vitro protein synthesis.
Isolated pancreatic acini prepared from normal and CF mice were preincubated for 1 h in Met/Cys-free medium (Sigma, St. Louis, MO). Equal cell aliquots of 50 μg of DNA were labeled for 30 min with 0.5 mCi/ml of 35S-labeled Met/Cys (TranSLabel; ICN, Costa Mesa, CA). Cell pellets were analyzed by SDS-PAGE and phosphorimaging (Cyclone; Packard Instrument, Meriden, CT). Because Muclin is of a unique size in the acinar cell (12), its incorporated radioactivity was measured directly from the total sample; amylase was immunoprecipitated with a rabbit anti-human amylase antiserum (Sigma), run on SDS-PAGE, and quantified by phosphorimaging.
mRNA quantitation by Northern blot analysis and quantitative real-time RT-PCR.
Total RNA was prepared from the pancreas or from the entire small intestine using the TRIzol reagent (Life Technologies, Grand Island, NY). Northern blot analysis of tissue mRNA was performed with the use of total RNA with probes for amylase, Muclin, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as previously described (12, 21). For quantitation of mRNAs for secretin, vasoactive intestinal peptide (VIP), chromogranin A (CgA), and GAPDH (GenBank accession nos. U07568, X74297, NM_007693, andM32599, respectively), a real-time RT-PCR approach was used with a LightCycler and an RNA amplification kit (Roche, Indianapolis, IN). Primers were designed to cross RNA splice junctions when possible to minimize potential amplification from any contaminating genomic DNA. Control PCR reactions performed omitting the RT step resulted in a signal of <0.1% of the complete RT-PCR reaction (not shown). The following primers were used: secretin, plus = GCACCTCCCAGGACCCCAAG, minus = TTCTGTGTCCTGCTCGCTGC, product = 129 bp; VIP, plus = GTCAGCTTGGACAGCAGAGC, minus = TGGTCCAAAGAGAGGCCAGG, product = 263 bp; chromogranin A, plus = CTGTGCGCCGGGCAAGTTTT, minus = TGCTGCGTCTCTGTGCTTGG, product = 393 bp; and GAPDH, plus = TTCACCACCATGGAGAAGGC, minus = GGCATGGACTGTGGTCATGA, product = 238 bp. The RT-PCR products were recovered from the LightCycler capillaries, reamplified by conventional PCR, cloned into a T-vector (pCR2.1-TOPO; Invitrogen, Carlsbad, CA), and submitted for DNA sequencing to verify that the products were as expected.
cAMP enzyme immunoassay.
For cAMP determinations, pancreatic tissue was taken and snap frozen in liquid nitrogen. The frozen tissue samples were then lyophilized to dryness (72 h). The dry samples were homogenized in 100 μl of 0.1 N HCl/mg dry wt with a Tekmar homogenizer (Cincinnati, OH). The homogenized samples were centrifuged at 600 g × 10 min, and the supernatants were used for cAMP quantitation by use of an enzyme immunoassay (Assay Designs, Ann Arbor, MI).
In situ duodenal pH measurement.
Mice were fasted overnight with free access to water. Mice were anesthetized with an intramuscular injection of ketamine and placed on a heat pad (Deltapase; Braintree Scientific, Braintree, MA), and the abdomen was opened. A small incision was made in the duodenum just below the stomach, and a needle pH probe (Orion, Beverly, MA) was inserted into the lumen. The probe was equilibrated for 1 min, and a millivolt reading was taken. pH values were determined from a standard curve of millivolt vs. pH standard buffers.
Paired sample data were analyzed using Systat software (Chicago, IL) by t-test and multiple group data by ANOVA with post hoc Tukey's test; P < 0.05 was considered significant. Data are presented as means ± SE.
CF mice have obvious gastrointestinal problems and fail to gain body weight compared with normal mice, even when maintained on a complete liquid diet. As shown in Fig. 1, CF mice are significantly smaller than normals as early as 16 days of age. There is no difference in body weight at any age between homozygous normal and heterozygous CFTR(+/−) mice, and these values are combined in the data shown in Fig. 1. For all other experiments, CFTR(+/+) mice were used as normals and CFTR(−/−) mice as CF. The reason for the smaller size of the CF mice has not been explained; it is reminiscent of the “failure to thrive” observed in CF children (39).
The exocrine pancreas of the CF mouse is morphologically altered compared with normal, with fewer zymogen granules and frequent dilations of the acinar lumens (Fig. 2). The dilated lumens are filled with amylase-positive aggregates and are lined by high amounts of the sulfated glycoprotein Muclin (Fig. 2). The decrease in number of zymogen granules is reflected in the statistically significant difference in the amount of amylase in the CF pancreas, which was ∼50% of normal (P < 0.05; see Fig. 3 C). The decrease in amylase is not simply a downregulation of the secretory pathway. In contrast to amylase, Muclin, the major sulfated glycoprotein of the pancreas and the most abundant component of the zymogen granule membrane (9), is upregulated to ∼200% of normal (P < 0.05; Fig.3, A and C).
To determine the mechanism for decreased amylase in the CF pancreas and deposition of Muclin on the apical plasma membrane, we performed a series of experiments investigating the secretory pathway starting from mRNA expression through stimulated protein secretion. By Northern blot analysis, pancreatic mRNA for amylase was not significantly different in CF tissue compared with normal (Fig.4, A and C) and could not explain the difference in tissue protein levels. At the same time, Muclin mRNA was significantly elevated to ∼300% of normal in the CF pancreas (Fig. 4, B and C), in parallel with the elevated tissue content of Muclin.
We next measured the rates of amylase and Muclin biosynthesis using isolated acini to determine whether the changes in amylase levels were due to translational regulation of gene expression. Isolated acini were labeled with 35S-Met/Cys, and the amounts of radioactivity incorporated into amylase and the biosynthetic precursor to Muclin (pro-Muclin) were quantified. (With the short labeling time, only pro-Muclin is labeled; Ref. 16). As shown in Fig.5, there was no difference in amylase biosynthesis comparing normal and CF acini, excluding translational control of amylase expression. By contrast, the CF acini synthesized ∼230% pro-Muclin compared with normal cells (Fig. 5), again paralleling the observed change in tissue Muclin.
We next measured basal and stimulated amylase secretion from normal and CF pancreas and normalized the levels of secretion to the initial cell content of amylase. As shown in Fig. 6, there was a tendency for increased amylase release from CF acini when stimulated individually by carbachol (CCh) or cAMP (8-bromoadenosine 3′,5′-cyclic monophosphate; 8-BrcAMP). However, the effect was not statistically significant. Consistent with previous reports (2), CCh was found to be a strong secretagogue (230–240% of basal), whereas cAMP was weak (107–126% of basal), and the combination of the two showed strong potentiation (437–511% of basal). With both stimuli, there was no difference in the percentage of amylase release when comparing CF to normal. The tendency for increased release suggested that the CF acini may be more activated in some fashion. Because it takes 1.5–2 h to prepare acini, it is possible that the difference may be stronger in vivo.
To test this, we examined amylase release in vivo, using a single intraperitoneal administration of the cholinergic agonist pilocarpine (14, 38). In both normal and CF mice, there was a robust amylase loss from the pancreas over the first 1–2 h after stimulation (Fig. 7). In CF mice, amylase levels were decreased to their lowest after 1 h (29% of initial), whereas in the normal mice, the level was decreased to 46% of initial at 1 h. The normal mice reached their lowest value at 2 h poststimulation (39% of initial). Although the differences between normal and CF mice were greater in vivo than in vitro, because of high variability, the values at any one time point were not statistically different (P > 0.1). In the subsequent 3 h, there was a partial recovery of amylase content in both normal and CF mice to ∼50% of initial values by 4 h after the stimulus.
We also measured in vivo turnover of Muclin in these animals. Muclin is a poorly secreted peripheral membrane component on the inner face of the granule, and it reaches the apical plasma membrane by exocytosis of the zymogen granule. Previous work with normal mouse acini showed that in vitro, little Muclin is released from cholecystokinin octapeptide (CCK-8)-stimulated acini (9) or CCh plus 8-BrcAMP-stimulated acini (11). To determine whether Muclin is also poorly secreted in vivo, we measured release of Muclin in normal mice into pancreatic juice stimulated by pilocarpine, as previously described (14). Muclin in pancreatic juice was 6 μg/mg protein, whereas in the unstimulated pancreas it was 20 μg/mg protein (12); amylase activity was greater in the juice (0.19 U/mg protein) compared with the whole unstimulated pancreas (0.10 U/mg protein), similar to previous reports (15, 29). The ratio of Muclin to amylase in pilocarpine-stimulated juice was 15.4 ± 0.10% of that in unstimulated pancreatic tissue (n = 3 pancreata and 5 juice samples, P= 0.003), demonstrating that Muclin is poorly secreted compared with amylase. Despite the fact that Muclin is not released, it is still lost from the pancreas after stimulation, presumably by endocytosis and degradation in lysosomes. Therefore, loss of Muclin from the pancreas after stimulation can be used as an indicator of the endocytic process that follows exocytosis.
As shown in Fig. 7, Muclin loss from the pancreas is slower than amylase release. In the normal pancreas, Muclin content is at its minimum (28% of initial) 2 h after stimulation. In CF mice, the decrease of Muclin is as fast or faster during the first 0.5 h after stimulation but then appears blunted compared with normal. The CF mice reached their lowest level of Muclin content by 0.5 h after stimulation (50% of initial) and remained relatively unchanged over the subsequent time periods. The differences in the CF mice were not statistically different compared with normal. By 4 h, both normal and CF values were similar and had recovered to ∼50–60% of the starting level.
Although we could not document a difference in amylase release, a small but chronic increase in amylase release conceivably could be enough to result in the significant decrease observed in tissue amylase levels. A possible explanation is that signaling from the intestine to the pancreas may be altered in CF because of CF-related changes in the intestine. To test this, we measured the luminal pH of normal and CF mice duodena. As predicted, the CF intestine was statistically more acidic, averaging 0.32 pH units more acid than normal (Fig.8).
We then tested whether signaling molecules that would stimulate bicarbonate secretion in response to an acidic intestine were upregulated in the CF mouse. We measured secretin mRNA in the intestine and VIP mRNA in the intestine and pancreas. In the CF mouse, the intestinal mucosa is hypertrophied: the wet weight of the duodenum (first 5 cm of the small intestine) is ∼140% of normal (CF = 0.52 ± 0.05 g vs. normal = 0.37 ± 0.05 g;P < 0.05, n = 3 each). Although it has been reported that there are increased numbers of certain cell types such as goblet cells in the CF mouse intestine (33), it is unknown whether this hypertrophy also includes the minor population of endocrine cells. Therefore, for normalization of endocrine mRNAs, we used CgA mRNA, which is ubiquitously expressed in endocrine cells (17, 23). mRNA for the neurotransmitter VIP was normalized to GAPDH.
We used real-time RT-PCR to quantify mRNAs for secretin and CgA. As shown in Fig. 9 A, the RT-PCR reactions using the primer pairs given in materials and methods produced the expected product sizes. The PCR products were cloned and sequenced to confirm that they matched the sequences in GenBank. At most, there were two bases different in any one RT-PCR product compared with the database sequences, likely due to PCR artifacts. As presented in Fig. 9 B, CF intestinal secretin mRNA was 310% of normal, VIP mRNA was 148% of normal, and pancreatic VIP mRNA was 270% of normal.
We then measured whether there was a change in pancreatic cAMP levels as a result of the increased secretin and VIP expression. Pancreatic tissue was obtained from mice fed ad libitum, and cAMP levels were determined by enzyme immunoassay. As shown in Fig. 9 C, the CF pancreas had 147% of the normal cAMP levels.
In the human disease CF, the exocrine pancreas has the strongest phenotype-genotype relationship: the greater effect a mutation in the CFTR gene has on the CFTR Cl− channel activity, the more severe the effect on the pancreas will be (7). It is widely believed that the pancreas is destroyed in CF because of protein plugging of the acinar/ductal lumens. It has been proposed that loss of fluid volume and alkaline bicarbonate secretion cause secreted digestive enzymes to be poorly soluble, resulting in their aggregation in the lumen with subsequent damage to the acinar tissue.
The CFTR knockout mouse model of CF offers a unique opportunity to study the effects of CF on the exocrine pancreas, since the pancreas in this mouse is only mildly affected (12, 22). We started with the observation that the amylase content was decreased in the CF mouse pancreas and worked to explain the relationship between loss of CFTR and this phenotype. Our results lead us to the deduction that amylase release may be elevated in the CF mouse. Although the difference in the amylase secretory rate was too slight to be measured at a level of statistical significance, it can be argued that a small chronic increase in secretory rate may be sufficient to result in the decrease observed in steady-state amylase stores.
To gain evidence for an elevated secretory activity, we focused on a signaling pathway that could be expected to be different in CF. CFTR is a cAMP-activated Cl− channel, and its proper functioning is required for bicarbonate ion secretion in the gastrointestinal organs. Bicarbonate is secreted by several sources: crypt epithelial cells, submucosal glands, and the ductal trees of the pancreas and biliary tract (6, 24, 35). The function of this bicarbonate is to neutralize the acid delivered into the duodenum from the stomach, and a neutral pH is required for proper pancreatic digestive enzyme function and nutrient absorption.
Bicarbonate ion secretion likely involves multiple mechanisms, including CFTR-mediated Cl− secretion, which is then exchanged for bicarbonate; direct permeation of bicarbonate ion through CFTR; and other less-well-defined pathways (35). Interestingly, it recently has been shown that CFTR mutations that affect regulation of bicarbonate and Cl− exchange but are normal for Cl− secretion are also the mutations correlated with pancreatic insufficiency in CF patients (4). That study emphasizes the importance of bicarbonate ion secretion in addition to normal fluid secretion driven by CFTR-mediated Cl− transport.
Bicarbonate secretion is stimulated by the gastrointestinal hormone secretin and by the neurotransmitter VIP, the releases of which are mediated by acid entering the duodenum (24). We document here for the first time that the duodenal pH of the CF mouse is significantly more acidic than normal, most likely because of the impaired ability to secrete bicarbonate in the absence of CFTR. Furthermore, we show that intestinal secretin and VIP mRNA levels are elevated in the CF mouse, probably in response to the abnormal acidity. In addition, pancreatic VIP mRNA is also elevated in the CF mouse. Finally, we also show that steady-state pancreatic cAMP levels are elevated in CF mice, consistent with increased secretin and VIP signaling to the pancreas.
Thus our data suggest that the changes in the exocrine pancreas of the CF mouse may be due to the acid-base imbalance in the duodenum. Loss of CFTR function is the primary CF defect, resulting in an acid-base imbalance in the intestine. In an attempt to neutralize duodenal pH, cAMP-mediated signaling may be upregulated with the consequence that zymogen secretion by the acinar cell is potentiated. Tang et al. (37) have presented evidence for an increased secretory response in CF mouse acini compared with normal, and they concluded that CF acini show a greater degree of potentiation by cAMP of Ca2+-mediated secretion. Our data differ from theirs, and we found that there is no difference in the potentiated maximal secretion under combined Ca2+ and cAMP stimulation (Fig.6). Our data support the hypothesis that there may be a chronically higher protein secretory rate due to elevated cAMP in the CF mouse pancreas.
The observation of dilated acinar luminal membranes in the CF mouse pancreas can also be explained by a chronic elevated secretory state. Under these conditions, exocytosis may exceed the compensatory endocytosis of granule membrane back into the acinar cell, resulting in dilation of the luminal membrane as happens during acute stimulation (32). At the same time, the presence of amylase-positive aggregates in the lumen suggests that the ability of the duct system to transport the secreted protein to the intestine has been exceeded. We have previously documented that such morphological changes occur transiently in the normal mouse pancreas during postnatal development as the mice are weaned (13). We interpreted this as a situation where the acinar protein secretory capacity may exceed the ability of the duct to secrete enough alkaline fluid to solubilize and transport these proteins. A similar condition may exist in the CF mouse pancreas as a consequence of chronic secretin signaling and thus is an indirect effect of CF. Others have suggested that the morphological changes in the CF pancreas are more directly related to loss of CFTR and are a consequence of perturbed endocytosis due to abnormal acidity in the acinar lumen (20). However, there is no direct evidence for this. In addition, the fact that the CF mouse pancreas is not progressively destroyed argues against any significant ductal obstruction as occurs in CF humans.
Although exocrine function appears normal in the CF mouse pancreas, there is a clear increase in Muclin expression that remains to be explained. It is not yet known what regulates Muclin gene expression, whether it is by some signaling from the apical pole of the acinar cell, or if it is responsive to the elevated cAMP levels in the CF pancreas. Because of its abundance in zymogen granules, we have proposed Muclin to function in the formation of the granules from thetrans-Golgi network (16). We have also proposed a protective role for Muclin on the apical plasma membrane of epithelial cells in the gastrointestinal system, based on its mucin-like biochemical composition (12). In addition, the human ortholog of the gene that encodes Muclin, dmbt1/gp-340, has been suggested to be involved in inflammatory responses (30). RNAs transcribed from this gene are alternatively spliced to give rise to tissue-specific forms of the protein, the overall structure of which is largely the same but differs in the presence or absence of a COOH-terminal transmembrane domain and short cytosolic tail, which appear to be expressed in exocrine cells only (36). The upregulation of Muclin expression in the CF pancreas may be a protective response to the chronic secretory state. Like other inflammatory responses in CF, upregulation of Muclin may actually be detrimental. The presence of high levels of Muclin on the luminal membrane may hinder release of the exocytosed zymogens and may also interfere with endocytosis. These issues will require further work to resolve.
How do these data relate to the human disease? It is known that duodenal pH in CF patients is more acidic and remains so longer after a meal than in normal individuals (1). The resultant acid environment of the duodenum in CF can inhibit and inactivate pancreatic digestive enzymes, especially lipase (1, 31), and this is true whether the enzymes are endogenous in pancreas-sufficient patients or oral supplements in pancreas-insufficient patients. It has also been documented that plasma secretin levels are elevated in CF patients, being 230% of normal (40), similar to the increases we measured in CF mouse intestinal secretin and VIP mRNAs. Notably, we also found an increase in pancreatic VIP mRNA. Thus, in CF individuals, the exocrine pancreas is likely chronically stimulated by these signaling molecules that act through cAMP. Such increased stimulation would be expected to exacerbate the problems in the pancreas and promote the progressive destruction of the exocrine tissue.
It is suggested by these data that there may be a beneficial effect of inhibiting gastric acid secretion on the pathogenesis of CF in the gastrointestinal system. In fact, a study using the highly selective H+-K+-ATPase inhibitor omeprazole documented improvement of nutritional status in CF patients (1). By reduction of stomach acid production, the acid load delivered to the duodenum would be decreased, which in turn would decrease the stimulus for secretin and VIP production. This could relieve the chronic stimulation to the pancreas and, in CF patients with some pancreatic function, may help preserve the health of the pancreas. In patients that are already pancreatically insufficient, it still may be beneficial to inhibit gastric acid secretion to promote better activity of exogenous enzymes and help nutrient absorption.
This work was supported by a Cystic Fibrosis Foundation grant and National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-56791.
Address for reprint requests and other correspondence: R. C. De Lisle, Dept. of Anatomy and Cell Biology, Univ. of Kansas School of Medicine, Kansas City, KS 66160 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2001 the American Physiological Society