Main

Since the identification of the reg gene by the group of Okamoto in rat(1), in human(2), and more recently in mouse(3) pancreas, the role of the expressed protein has been widely debated. The first experiments performed on 90% depancreatized-nicotinamide treated rats have shown that the regeneration of pancreatic islets was associated with the expression of the reg gene. This gene is expressed in regenerating islets and not in normal islets, suggesting that the reg protein could be involved in β cell regeneration or growth(1). This association between reg gene expression and islet cell replication was supported by experiments on rat isolated islets(4) and in vivo 90% depancreatized rats(5). Opinions on the location of the major site of reg expression remain controversial(6, 7). In humans,in situ hybridization and immunohistochemistry experiments have revealed the reg protein in acinar cells of the exocrine pancreas(8–10), indicating that this protein is a physiologic secretory product expressed in the exocrine pancreas, whereas it is never expressed in normal endocrine tissue. This is consistent with the cDNA sequence of the human reg gene(2) which encodes the sequence of a secretory glycoprotein of 19 kD, named P19(11), precursor of the protein X of 14 kD devoid of carbohydrates, isolated from slightly activated pancreatic juice(12), and successively named PTP (pancreatic thread protein)(13) and PSP (pancreatic stone protein)(14).

In a recent study on the expression levels of human genes encoding islet hormones and other pancreas-specific proteins, Mally et al.(9) stated that the reg gene exhibits a 20-fold increase in mRNA levels at the 16th wk of gestation and that its expression remains high until birth with levels comparable to those in the adult pancreas. These data are in contrast with our previous immunohistochemical studies of secretory proteins in the developing human pancreas(15) which have shown that the ontogenesis of P19/reg was parallel to that of trypsinogen and chymotrypsinogen, with weak immunoreaction intensities until the 24-27th wk and a marked increase at birth.

To further clarify the role of the reg gene in human pancreatic development, we have studied the level of reg gene expression from the 17th to the 29th wk of gestation. In an attempt to elucidate the possible function of the reg gene in β cell and/or acinar cell growth we have compared its expression with that of insulin and chymotrypsinogen genes, two endocrine and exocrine products, respectively.

METHODS

Materials . Pancreatic tissues. Pancreatic tissues were collected from 32 fetuses (17-29 wk of gestation) after spontaneous or induced terminations. Macroscopic and microscopic examinations of pancreases were normal and the fetuses were devoid of digestive diseases. Pancreases were snap-frozen in liquid nitrogen and stored at -80°C. This study was performed as part of the teratovigilance network, the latter being approved by the ethical committee. Gestational ages were determined by pathologists according to classical criteria including biparietal diameter and foot length. Seven adult pancreases were obtained from transplant donors and/or from surgery and immediately frozen. None was a consequence of chronic pancreatitis or removed postmortem. The pancreases appeared macroscopically normal but no histologic study was performed.

cDNA probes. Four human cDNA probes were used to determine the levels of specific mRNAs. Human reg and insulin probes were kindly provided by Dr. H. Okamoto (Japan) and Dr. M. D. Walker (Israël), respectively; both were full-length cDNA. Reg cDNA and insulin cDNA were subcloned in pBS and pBR322 vectors, respectively. A 28S cDNA probe was from ATCC (American Type Culture Collection, Rockville, MD). The chymotrypsinogen probe was synthesized by reverse transcription and polymerase chain reaction (RT-PCR).

Synthesis of chymotrypsinogen cDNA probe by RT-PCR. The chymotrypsinogen cDNA probe was synthesized by RT-PCR with 1 μg of total adult pancreatic RNA according to the instructions provided with the GeneAmp RT-PCR kit (Perkin-Elmer, Norwalk, CT). The primers used in the PCR reaction were deduced from the cDNA sequence of human chymotrypsinogen(16). They were as follows: sense primer 5′-CGGCGCTGGGCAGGCACACGGCGG-3′ (nucleotides 410-433) and antisense primer 5′-GACGCCGTCCCCGGCTCCTGGCCC-3′ (nucleotides 115-138). The thermal cycle profile used for amplification consisted of: 1) denaturation for 1 min at 95°C, then 2) primer annealing for 1 min at 71°C, and 3) extension for 5 min at 72°C. The PCR was stopped after 35 cycles. After amplification, the RT-PCR product was electrophoresed on a 10% polyacrylamide gel. The correct size of the fragment (319 bp) was controlled by ethidium bromide visualization, and its sequence was confirmed by the dideoxy chain termination method using a Sequenase sequencing kit (U. S. Biochemical Corp., Cleveland, OH).

Isolation of total pancreatic RNA and Northern-blot analysis. The isolation of total pancreatic RNA from the fetal pancreas was performed using the RNAzol method (Bioprobe Systems, Montreuil, France) which is particularly suitable for small samples (weight < 300 mg). Adult pancreatic RNA was prepared using the procedure described by Chirgwin with slight modifications as previously explained(17). The RNA concentration was monitored spectrophotometrically (absorbance at 260 nm with 1 OD = 40 μg/mL), and the integrity of RNAs was confirmed by 1% formaldehyde-agarose gel electrophoresis.

Northern-blot analysis of electrophoretically separated total RNA from adult pancreases was performed to verify the hybridization specificity of each probe and to assess the length of each mRNA. In brief, RNA, first denatured by formaldehyde before electrophoresis through 1% agarose in the presence of formaldehyde, was transferred to nitrocellulose according to the procedure described by Maniatis et al.(18) and hybridized with the cDNA probes labeled with [α-32P]dCTP (3000 Ci/mmol, Amersham, Les Ulis, France) using for chymotrypsinogen a random primer kit (Boehringer-Mannheim, Meylan, France) and for insulin and reg the nick-translation system (Life Technologies, Inc., Eragny, France). The specific activities of the probe used were routinely 109 and 108 for random priming and nick-translation labeling, respectively. Results were visualized by autoradiography.

Quantitative analysis of mRNA by dot-blot hybridization. Quantification of each specific mRNA was determined by dot-blot analysis in duplicate on each pancreas as previously described(17). In brief, sequential dilutions of RNA were dotted onto a nitrocellulose filter and then hybridized with the radiolabeled cDNA probes. A 28 S cDNA probe was used to correct the differences in sample RNA loading. After film development, the filter-bound radioactivity was determined by scanning the autoradiograph densitometrically at 490 nm using an automatic microplate reader (MR 5000 Dynatech, Saint-Cloud, France). The scanned readings of each autoradiographic film were corrected for background. The estimates for mRNA contents were made from the linear portion of the curves obtained by linear regression and expressed in arbitrary units. The results were subsequently normalized with the 28 S cDNA probe. Statistical significance of results was achieved using paired two-tailed t test.

RESULTS

Qualitative analysis of mRNA by Northern blot. The results of Northern blot analysis are shown in Figure 1. Each mRNA was identified as a single band, and as expected the transcript sizes were 0.90 kb for reg, 0.95 kb for chymotrypsinogen, and 0.55 kb for insulin.

Figure 1
figure 1

Northern blot analysis of pancreatic RNA samples. Total RNA (2 μg/lane) isolated from adult human pancreases, first denatured, was submitted to electrophoresis on a 1% agarose gel containing 2.2 M formaldehyde, then transferred to a nitrocellulose filter and hybridized with nick-translated (insulin, reg) or random-primed (chymotrypsinogen) cDNA probes. CHY, chymotrypsinogen; INS, insulin;REG, reg.

Full size image

Quantitative gene expression of reg, chymotrypsinogen, and insulin in the fetal pancreas during development. Comparison with the adult pancreas. Quantitative changes in the mRNA levels of reg, insulin, and chymotrypsinogen were determined in human pancreases from 32 different fetuses. The low amount of total RNA did not permit the quantitation of all mRNAs on each pancreas but, at least two probes in addition to the 28 S probe were used. Because of the long period of time necessary to collect the pancreases and to complete this study, comparison of fetal mRNA levels in each experiment was performed using the same references (total RNA from two adult pancreases) loaded onto each dot-blot. This procedure was carried out throughout the entire study, and the results are expressed in arbitrary units. For the seven adult pancreases, a similar approach was used, and the results are presented as a mean ± SEM.

As shown in Figure 2, the patterns of gene expression of the three proteins during fetal development displayed a large dispersion of values. We did not find any correlation with gestational ages; no progressive increase of gene expression was apparent during the period studied (17-29 wk of gestation). No peculiar difference in gene expression was noticed for fetal pancreases of different origin (induced or spontaneous abortions) (data not shown). However, a noticeable difference exists between the levels of reg and chymotrypsinogen gene expression on one hand and insulin gene expression on the other. In 26 cases out of 30 for reg and 17 out of 21 for chymotrypsinogen, the fetal values remained below the mean adult values, whereas in 18 out of 22 cases, the levels of insulin gene expression were very high, well above the mean adult level. Statistical analysis indicates that the difference between the levels of insulin mRNA in fetus and adults is highly significant (p ≤ 0.001). In addition, it is worthwhile to notice the extremely low level of reg mRNA in the fetal pancreas compared with that in the adult in half of the samples studied, whatever the gestational age of the fetus.

Figure 2
figure 2

Relative levels of reg, chymotrypsinogen, and insulin mRNA during fetal development. Insulin (•), reg (▴), and chymotrypsinogen (♦) mRNA levels are expressed in arbitrary units (A.U.) as the values of the densitometric scans of autoradiographs from dot-blot experiments after normalization to the 28 S signals. Adult results are given as a mean ± SEM: insulin (○) mRNA = 0.75 ± 0.23, reg(▵) mRNA = 0.67 ± 0.22, and chymotrypsinogen (⋄) mRNA = 0.78± 0.08.

Full size image

Correlation between insulin and reg gene expression. We were looking for a relationship between the gene expression of insulin and reg because an association between reg gene expression and islets growth has been found in animals. Figure 3 shows that no correlation exists between the two human gene expressions during fetal development. In contrast, in the adult pancreas, reg gene expression is correlated with insulin gene expression (r = 0.92; p ≤ 0.01)(Fig. 4). It is interesting to notice that no relation was found either between reg and chymotrypsinogen expression or between insulin and chymotrypsinogen expression (data not shown) in fetal as well as adult pancreases.

Figure 3
figure 3

Relation between reg and insulin gene expression during fetal development. mRNA levels for 22 pancreases are expressed in arbitrary units (A.U.) as explained in the legend of Figure 2.

Full size image
Figure 4
figure 4

Correlation between reg and insulin gene expression in the adult pancreas. mRNA levels are expressed in arbitrary units (A.U.) as explained in the legend of Figure 2. Correlation coefficient, r = 0.92.

Full size image

DISCUSSION

The levels of gene expression of reg and chymotrypsinogen during the 17th to 29th wk of fetal development show that the two genes are expressed as early as the 17th wk, and that, despite the scattered values probably due to the differences in the degree of maturity of human pancreas before birth and perhaps to natural variations between individuals, no evident increase of expression was observed during this fetal period. These results correlate with the pattern of the appearance of corresponding proteins revealed by immunohistochemistry of fetal pancreases during the same period of gestation. A weak immunoreactivity of chymotrypsinogen and P19/reg was detected in acinar cells at the 16th wk, and a low increase of labeling was observed around the 27th wk(15). In the human pancreas, the absence of increased gene expression during the studied period (17-29-wk gestation) may indicate a period of fetal development without any important regulatory changes as reported by Track et al.(19) who found a dramatic increase in enzyme activities only after the 32nd wk of pregnancy.

The high insulin gene expression in fetal pancreases as early as the 17th wk of gestation concurs with the appearance of β cells in human pancreases at approximately 8.0-8.5 wk of gestation(20), and the insulin immunoreactivity characterized at the 8th gestational wk by immunohistochemistry(21) and by immunofluorescence(22). Our data are in agreement with the results of Miettinen and Heikinheimo(23) who detected insulin mRNA by PCR, Northern blot, and in situ hybridization from the 15th wk of gestation. The significant difference between fetal and adult mRNA levels (up to 3-4 times higher in the fetus) can be explained by the high relative volume of endocrine tissue in fetuses compared with that in the adult. As reported by Rahier et al.(24) this relative volume of endocrine tissue was found to be particularly high from the 17th wk on, decreasing to 50% in the youngest neonate, but nevertheless remaining 6 times more abundant than that in adults. In contrast, using RNase protection assays, Mally et al.(9) found fetal insulin mRNA levels 2-6 times lower than that in adult. The reasons for this striking discrepancy are not clear, but could be ascribed in part to the different probes or methods used. The elevated mRNA level of insulin that we found would support the already recognized role of insulin as a growth factor during fetal development(25, 26). As discussed by Francis et al.(4) several mechanisms have been proposed to account for the apparent growth-promoting effect of insulin, with insulin potentiating the action of growth factors including the IGF.

The level of reg gene expression was extensively studied in 30 fetal pancreases, and interestingly, it was found to be very low in most pancreases. Compared with the mean adult level (0.65), 20 out of 30 fetal mRNA levels remained ≤0.3, without any relation with the gestational age. These very low reg mRNA levels are not due to a lack of sensitivity of the hybridization technique, because we were able to measure by this same method the mRNA levels of the cystic fibrosis transmembrane conductance regulator gene in cells containing less than 10 copies/cell(27). Our data on the reg gene expression throughout the fetal period studied (17-29 wk) differ from the observations of Mally's(9) group. These authors have quantified the reg gene expression in 13 pancreases from 13-24-wk-old fetuses. They described a very low mRNA level before the 16th wk and noticed a dramatic increase of reg gene expression between the 14th and 16th wk. However, it is worthwhile to notice that the reg mRNA levels they found are in a wide range after the 16th wk, as shown in this study. From these dispersed values it is difficult to believe that the levels remained high throughout gestation and comparable to those found in the adult pancreas. Our data show that, even though the fetal values are dispersed, they tend to be lower than adult.

The low level of reg expression and the absence of correlation with insulin and chymotrypsinogen gene expression suggest that the reg gene does not play a direct role in β or acinar cell growth during development. The difference observed between the mRNA levels of reg and insulin cannot be due to a difference in the sensitivity of the two probes to detect its complementary mRNA, because they have approximately the same specific activity. However, possible differences in the rate of degradation between the two mRNAs, with insulin mRNA half-life longer than that for reg, as well as differences in the rate of mRNA transcription cannot be eliminated. The absence of correlation that we found with insulin gene expression (r = 0.32) illustrates that the reg gene might not be involved in the expansion of the islet mass during human fetal development, at least after the 17th wk. The same lack of correlation between reg mRNA level and β cell growth was reported by Smith et al.(28) who studied two models of rat pancreatic growth and reported that it is unlikely that reg is aβ-cell specific growth factor.

In the adult pancreas, the mean levels of expression for the reg, chymotrypsinogen, and insulin genes are very similar and do not have a wide range, probably because of the maturity of the adult gland. The correlation that we have found between reg and insulin expression is consistent with the hypothesis that the reg gene would be associated with β cell dysfunction in the adult as detected in aging mice by Perfetti et al.(29) or that the reg gene would be necessary to maintain the β cell mass in the normal adult pancreas as suggested by Watanabeet al.(5).

In conclusion our data suggest that the reg gene might not be involved in human pancreatic β cell growth during fetal development, and they support the known role of insulin as a growth factor.