Des-(27-31)C-peptide. A novel secretory product of the rat pancreatic beta cell produced by truncation of proinsulin connecting peptide in secretory granules.

Insulin and connecting peptide (C-peptide) are produced in equimolar amounts during proinsulin conversion in the pancreatic beta cell secretory granule. To determine whether insulin and C-peptide are equally stable in beta cell granules (and thus secreted in equimolar amounts), neonatal and adult rat beta cells were pulse-chased, and radiolabeled insulin and C-peptide analyzed by high performance liquid chromatography. A novel truncated C-peptide was identified and shown by mass spectrometry to be des-(27-31)C-peptide (loss of 5 C-terminal amino acids). Des-(27-31)C-peptide is a major beta cell secretory product, accounting for 37.4 ± 1.6% (neonatal) and 8.5 ± 0.6% (adult) of total labeled C-peptide in secretory granules after 10 h of chase. Des-(27-31)C-peptide is also secreted in a glucose-sensitive manner from the perfused adult rat pancreas, accounting for ∼10% of total C-peptide immunoreactivity secreted. Human C-peptide is also a substrate for truncation in granules. Thus, when human proinsulin was expressed (infection with recombinant adenovirus) in transformed (INS) rat beta cells, human des-(27-31)C-peptide was secreted along with the intact human peptide and both intact and truncated rat C-peptide. In addition to truncation, 33.1 ± 1.2% of C-peptide in neonatal but not adult rat beta cell granules was further degraded. Such degradation was completely inhibited by ammonium chloride (known to neutralize intra-granular pH), whereas truncation was only partially inhibited by ∼50%. In conclusion, a novel beta cell secretory product, des-(27-31)C-peptide, has been identified and should be considered as a potential bioactive peptide. Both truncation and degradation of C-peptide are responsible for non-equimolar secretion of insulin and C-peptide in rat beta cells.

Proinsulin conversion occurs within immature secretory granules (1,2) resulting in the production of equimolar amounts of insulin and C-peptide (3). These molecules are stored in mature secretory granules until they are released from the beta cell. It has thus been widely assumed that insulin and C-peptide must always be released in equimolar amounts (4 -6). However, this assumption is based upon the tenet that insulin and C-peptide are equally stable within beta cell secretory granules and thus their relative proportion does not change with time. Such is not always the case.
First, C-peptide can be subject to endoproteolytic cleavage. If such cleavage is restricted to one or only few sites in the molecule, it leads to the production of discrete truncated forms of the peptide (7,8). If such truncation occurs in the beta cell secretory granule (rather than lysosomes), it will lead to the secretion of these truncated forms. More extensive proteolysis will result in the degradation of C-peptide without any detectable truncated forms, and this appears to be the case in transformed rat beta cells (INS cells), but not in isolated adult rat islets (9). Both limited and extensive proteolysis will elevate the ratio of insulin to intact (non-truncated) C-peptide to values greater than one. Second, a novel post-granular, constitutivelike secretory pathway has been proposed in which vesicles bud from granules carrying C-peptide and not insulin leading to selective release of C-peptide from the beta cell (10). When this occurs, the secretory granules will be left with an insulin to C-peptide ratio greater than unity.
Our previous data suggested the presence of a discrete truncated form of C-peptide in secretory granules of INS cells, as well as more extensive degradation (9). In the present work, we identified this truncated peptide by sequence analysis and mass spectrometry as a novel form of C-peptide lacking the 5 C-terminal residues (des-(27-31)C-peptide). We further show that this truncated C-peptide is a major secretory product of both neonatal and adult rat beta cells and as such not a peculiarity of insulinoma cells. Both truncation and complete degradation appear to contribute to non-equimolar secretion of insulin and C-peptide from beta cells of newborn rats, whereas only truncation is apparent in granules of beta cells from adult rats.

EXPERIMENTAL PROCEDURES
Cell Preparation and Culture-Monolayer cultures were prepared from the pancreases of neonatal (3-to 5-day-old) Sprague-Dawley rats as described previously (11). Studies were performed following 3 days of growth in culture medium consisting of 45% NCTC 135, 45% medium 199 (v/v; Life Technologies, Inc., Grand Island NY) and 10% fetal calf serum supplemented with 16.7 mM glucose and 50 g/ml gentamicin.
Adult rat islets were isolated from male Sprague-Dawley rats weighing 200 -220 g by collagenase digestion as described previously (12). Prior to experiments the islets were cultured overnight in Dulbecco's modified Eagle's medium containing 10% fetal calf serum and 8.3 mM glucose.
Note that in separate experiments 1 we have confirmed that neither the culture medium (Dulbecco's modified Eagle's medium used for adult islets versus the mixture of 45% NCTC 135, 45% medium 199 used for the newborn islet monolayers) nor the glucose concentration (8.3 versus 16.7 mM) influence the stability of C-peptide within adult islet cells. The choice of Dulbecco's modified Eagle's medium with 8.3 mM glucose for the adult islets was based upon previous studies demonstrating that under these conditions islets remain the most viable with well maintained responses toward secretagogues and an appropriate balance between secretion and intracellular degradation (crinophagy) (13).
Expression of Human Proinsulin in INS Cells Using Recombinant Adenovirus-Recombinant adenovirus for expression of human proinsulin was prepared by homologous recombination of the plasmid pJM17 (14), which consists of a modified Ad5 genome, and pACCMV-hins, which was constructed by inserting human proinsulin cDNA into the SmaI site of the polylinker of pACCMV.pLpA (15). To this end, semiconfluent 293 cells were co-transfected with the two plasmids using a modified calcium phosphate precipitation method (16). The 293 cells were seen to lyse some 2-3 weeks post-transfection and clonal recombinant virus was then prepared by reinfection of 293 cells followed by overlay with 0.7% agar. Plaques were visible 7-10 days later and were taken to correspond to viral clones. INS cells were maintained in tissue culture as described in detail previously (17), and were seeded in 35-mm diameter Petri dishes at a density of 10 6 cells/dish. Four days after seeding the dishes, the cells were infected with recombinant adenovirus for 1 h at a multiplicity of infection of 13. After washing, the cells were kept in culture for 24 h before being used for the pulse-chase experiments.
Pulse-Chase Methodology-Following culture, neonatal rat beta cells were washed 3 times in 1.0 ml of prewarmed Krebs-Ringer bicarbonate buffer (KRB) 2 containing 0.1% bovine serum albumin, 20 mM Hepes, 1.67 mM glucose, and no calcium. The cells were preincubated in 1.0 ml of this medium for 15 min at 37°C and then pulse labeled for 10 min at 37°C in 0.30 ml of KRB containing 400 Ci/ml [ 3 H]leucine (Amersham Corp.) and 16.7 mM glucose. The labeled cells were washed 3 times with 1.0 ml of ice-cold KRB containing 1.67 glucose and 1 mM unlabeled leucine, then chased for 2 or 10 h at 37°C in medium composed of 45% NCTC 135, 45% medium 199 with 20 mM Hepes, 5.5 mM sodium bicarbonate, 10% fetal calf serum, and 8.3 mM glucose. Human C-peptide (10 g/ml; Peninsula Laboratories, Belmont, CA) and human insulin (100 g/ml; Eli Lilly and Co., Indianapolis, IN) were also added to prevent degradation of labeled C-peptide and insulin in the media. In some wells, NH 4 Cl was added after 2 h of chase to obtain a final concentration in the medium of 25 mM. Following the chase period, the medium was harvested and the cells were extracted in 0.5 ml of 1 M acetic acid with 0.1% bovine serum albumin, freeze-thawed, and centrifuged 5 min at 10,000 rpm to remove insoluble material. Chase media and cell extracts were stored at Ϫ20°C prior to analysis.
Adult rat islets were pulse-labeled and chased using a similar protocol with the following modifications. The 10-min pulse label was in a volume of 0. Stimulation of Regulated Granule Exocytosis-Following chase periods of either 2 or 10 h, exocytosis of beta cell secretory granules was stimulated to allow assessment of labeled peptides in secretory granules. To this end, at the end of the 2-or 10-h chase, cells were washed 3 times in ice-cold KRB containing 1.67 mM glucose, and thereafter 500 l of a pre-warmed stimulation buffer containing a mixture of beta cell secretagogues was added (KRB supplemented with 16.7 mM glucose, 0.1 mM isobutylmethylxanthine (Sigma, St. Louis, MO), 5 M carbachol (Sigma), and 10 mM arginine). After 1 h of incubation at 37°C, the stimulation medium was collected and the cells extracted in acid as above. All samples were stored at Ϫ20°C prior to analysis.
Perfused Pancreas-The technique for surgical isolation and vascular perfusion of the pancreas has been previously described (18). Male Wistar rats weighing 300 -350 g were used. The perfusate was a Krebsbicarbonate buffer containing 0.2% bovine serum albumin, 3% dextran (Clinical grade; Sigma) and either 4.4 or 16.7 mM glucose. It was gassed with water-saturated 95% O 2 , 5% CO 2 to maintain a physiological pH, heated to 37°C and pumped through the pancreas via the abdominal aorta at a flow rate of 3 ml/min. Glucose was administered as follows: min 0 -20, 4.4 mM; min 20 -55, 16.7 mM; min 55-80, 4.4 mM. Fiveminute fractions (15 ml) of portal venous effluent were collected in chilled plastic vials containing 500 l of acetic acid (10 M) to inhibit protease activity.
Collected fractions were pooled to represent beta cell secretion under four conditions: 1) pre-stimulus basal (min 0 -20); 2) first phase glucosestimulated (min 20 -25); 3) second phase glucose-stimulated (min 45-55); and 4) post-stimulus basal (min 60 -80). Pooled fractions were extracted on C18/OH cartridges (Bond Elut, Varian, Harbor City, CA) using a vacuum manifold system. A maximum of 8 ml of perfusate was extracted on any individual cartridge. The loaded cartridges were washed with 5 ml of 0.1% trifluoroacetic acid followed by 5 ml of 10% acetonitrile plus 0.1% trifluoroacetic acid. The peptides were then eluted with 4 ml of 60% acetonitrile plus 0.1% trifluoroacetic acid. Material eluted from the cartridges was vacuum-dried on a Speed-Vac Plus (Savant, Farmingdale, NY) and reconstituted in 0.1% trifluoroacetic acid plus 0.1% bovine serum albumin. The dried eluates from each of the four conditions described above were reconstituted in a total volume of 0.3 ml and stored at Ϫ20°C for subsequent HPLC analysis and radioimmunoassay (RIA). Recovery of rat C-peptide II and rat insulin using this procedure was 70 Ϯ 4 and 68 Ϯ 5%, respectively (both n ϭ 3).
HPLC Separation of Insulin and C-peptide-Media samples were acidified by the addition of acetic acid to a final concentration of 1 M prior to injection. The HPLC system used was a Beckman System Gold pump module 126 and detector module 166 equipped with a LiChrospher 100 RP-18 column from Merck (Darmstadt, Germany). Separation of rat insulin I and II and C-peptide I and II was achieved using a previously described method (9,19). Scintillation mixture (3.5 ml; LumaFlow II; Lumac, Olen, Belgium) was added to 1-ml fractions and the radioactivity measured in a Beckman LS 7500 liquid scintillation counter.
We have previously shown that the recovery of proinsulin, insulin, and C-peptide by this method is quantitative and reproducible (9,20). To correct for the number of leucine residues in insulin and C-peptide (6 leucines in insulin, 5 leucines in C-peptide and des- (27)(28)(29)(30)(31)C-peptide), radioactivity obtained for C-peptide I and II and their truncated forms was multiplied by 1.2. Only results for insulin I and C-peptide I (and its truncated form) are presented since in the neonatal cells, in which the ratio of insulin I:II is approximately 4, the radioactivity in C-peptide II and its truncated form was on occasion too low to be useful for precise quantification. Note, however, that this was not the case for all of the samples, and when C-peptide II and its truncated form could be measured with confidence, the results for insulin:C-peptide ratios and the relative amounts of truncated C-peptide were not significantly different from those obtained using insulin and C-peptide I (data not shown). For the adult islets, in which the ratio of insulin I:II is 1.5, radioactivity in C-peptide II and its truncated form was always high enough for precise quantification and here again no significant differences were observed between values obtained using these data or those for insulin I and C-peptide I.
Degradation of Insulin and C-peptide in Chase Media-To prevent possible degradation of labeled insulin and/or C-peptide in the chase and stimulation media leading to anomalous ratios of insulin to Cpeptide, 100 g/ml human insulin and 10 g/ml human C-peptide were added to media prior to incubation. To assess degradation under these conditions, 125 I-labeled human insulin (Amersham) and 125 I-labeled human C-peptide (provided by the Diabetes Endocrinology Research Center Radioimmunoassay Core, University of Washington) were added to the chase media in the presence or absence of the excess unlabeled peptides and exposed to cells at 37°C for 2 or 10 h (to mimic the chase conditions) or for just 1 h in the medium used for stimulating exocytosis. The amount of 125 I-labeled peptide was determined in appropriate fractions following HPLC separation and recovery of each was calculated. The addition of unlabeled insulin and C-peptide essentially prevented any degradation of labeled peptide as recovery of both 125 Iinsulin and 125 I-C-peptide following the 1-h incubation was Ͼ98% under all conditions tested.
Radioimmunoassay of Rat Insulin and C-Peptide II-To determine the percent of total C-peptide secreted from the perfused adult rat pancreas that is comprised of des-(27-31)C-peptide, C-peptide II (intact and truncated) levels were determined in appropriate fractions of HPLC eluate by a previously described RIA for rat C-peptide (21). The rat C-peptide assay employs synthetic rat C-peptide II as standard and reliably detects only C-peptide II but cross-reacts to a limited extent and in a non-linear fashion with rat C-peptide I (data not shown). It is therefore necessary to separate C-peptides I and II by HPLC before RIA. The antiserum used in this RIA detects both intact and des-(27-31)C-peptide II equally well (data not shown). Des-(27-31)C-peptide II was not reliably detectable in the pre-stimulus basal (4.4 mM glucose) perfusion samples. Therefore, to calculate the incremental increase in the rate of truncated C-peptide II release over basal, the minimum detectable level of truncated C-peptide II (2 fmol/min) was taken as basal.
Identification of Truncated Rat C-Peptide-Based on preliminary data, the peptides responsible for the early eluting peaks in Fig. 1 were thought to represent truncated forms of rat C-peptides I and II. To identify these peptides unequivocally, large quantities of the peptides were purified from acid extracts of adult rat islets by HPLC and subject to several analyses.
Amino Acid Analysis, N-terminal Microsequencing, and Determination of C-terminal Residues-Amino acid analysis after gas-phase hydrolysis or time scale carboxypeptidase Y digestion was performed as described (22). N-terminal sequence determination was carried out with a pulsed liquid phase microsequencer, model 477 from Applied Biosystems (Foster City, CA).
Mass Spectrometry-Truncated C-peptide I and II and their intact counterparts were purified by HPLC and then analyzed by mass spectrometry according to established procedures (23). In brief, samples were dissolved in 10 ml of water/acetonitrile/glacial acetic acid (50:50:1, v/v) and infused at 2 ml/min into the source of a Trio 2000 machine (Fisons, United Kingdom) equipped for electrospray ionization. The machine was calibrated externally using a solution of horse myoglobin.
Data Analysis-Data are presented as mean Ϯ S.E. Statistical comparisons were performed by two-tailed unpaired Student's t test or by one-way analysis of variance followed by Dunnett's test, where appropriate. p Ͻ 0.05 was considered statistically significant.

RESULTS
Characterization of Rat Truncated C-peptide-HPLC analysis of rat cell extracts and media allowed for the separation of rat insulin I and II as well as rat C-peptide I and II. HPLC profiles of media from neonatal rat beta cells revealed the presence of a pair of radioactive peaks which eluted approximately 20 min earlier than the two intact C-peptides (Fig. 1). Preliminary studies on these peptides had suggested that they were truncated C-peptide molecules (9) but they had yet to be characterized definitively.
In order to identify the precise site of truncation of rat C-peptide, four approaches were used: amino acid composition, partial N-terminal sequencing, C-terminal amino acid analysis, and mass spectrometry. The amino acid composition of the truncated form of both C-peptide I and II suggested that the last five residues of C-peptide had been lost (data not shown). This hypothesis was reinforced by sequencing the N-terminal region of the truncated molecules which showed that this region was unaltered compared to intact C-peptide I or II (sequence obtained: EVEDPQVP for truncated C-peptide I and EVEDPQVA for truncated C-peptide II). The C-terminal region of the truncated C-peptide I was analyzed by digestion with carboxypeptidase. The results were again in keeping with the loss of the last five residues (Leu, 190 pmol; Ala, 110 pmol; Thr, 52 pmol; Gln, 44.5 pmol; Asp, 41.7 pmol; Gly, 60.5 pmol). Definitive identification of the truncated rat C-peptides was based upon their molecular mass determined by mass spectrometry. The intact C-peptides were included in this study as a control. As detailed in Table I, the predicted and observed masses for the intact C-peptides were close to identical. The observed mass for the truncated forms of both C-peptides confirms unequivocally that the truncation results in the loss of five residues from the C terminus leading to the generation of des- (27)(28)(29)(30)(31)C-peptides I and II (Fig. 2).
Neonatal Rat Islet Cell Monolayers-As discussed under "Experimental Procedures," only results for insulin I and C-peptide I (and its truncated form) are presented. Radioactive insulin, C-peptide, and des- (27)(28)(29)(30)(31)C-peptide were quantified by HPLC in cell extracts after 2 and 10 h of chase, and in the medium following 1-h stimulation of the cells with a secretory mixture containing isobutylmethylxanthine, carbachol, arginine, and glucose. Such stimulation resulted in the release of greater than 50% of cellular labeled insulin after both 2 and 10 h of chase. Products released during the 1-h stimulatory period are taken to reflect the contents of secretory granules at the two times of chase.
Significant amounts of label were recovered in the form of des-(27-31)C-peptide in both cells and granules. After 2 h of chase, 11.9 Ϯ 0.2% of granular labeled C-peptide was truncated FIG. 1. Representative HPLC profile of radioactive insulinrelated peptides secreted from neonatal rat beta cells. Cells were pulse-labeled for 10 min with [ 3 H]leucine and chased for 10 h. Exocytosis of secretory granules was then stimulated for 1 h with a beta cell secretagogue mixture (as described under "Experimental Procedures") and the 1-h stimulation medium was injected onto HPLC. Peaks were identified from known elution times of standards. Note the presence of two peaks representing truncated C-peptides I and II secreted into medium from neonatal rat beta cells.

TABLE I Predicted and measured molecular mass of intact and truncated rat C-peptides I and II as determined by mass spectrometry
Intact and truncated C-peptides from adult rat islets were purified by reversed-phase HPLC and their molecular mass measured by electrospray mass spectrometry. The amino acid sequences of the two rat C-peptides and predicted site of truncation are shown in Fig. 2. The last 5 residues of the rat C-peptides which are removed in the truncated forms are identical for both molecules and the difference in molecular mass between the intact and truncated rat C-peptides I and II is thus predicted to be the same. The measured values given were precise (based on observed variation) to Ϯ1 mass unit and accurate (as determined by running standards) to Ϯ3 mass units.  Fig. 3). At all times, it is apparent that there was less labeled Cpeptide than insulin. In extracts of the neonatal cells, the ratio of insulin to intact C-peptide (I/CP) was greater than one at both chase times, being 1.53 Ϯ 0.08 after 2 h and increased further to 2.68 Ϯ 0.07 by 10 h (Fig. 4, upper panel, open bars). After 2 h of chase, I/CP in the 1-h stimulated medium ("granules": Fig. 4, lower panel, open bars) was already greater than 1 (1.29 Ϯ 0.03), and by 10 h of chase, it was markedly elevated (2.39 Ϯ 0.1; p Ͻ 0.001 versus 2 h). Even the ratio of labeled insulin to the sum of labeled intact and truncated (des-(27-31)) C-peptide (I/(CPϩtCP)) was greater than unity (Fig. 4, hatched  bars). The selective loss (considered to be degradation) of Cpeptide relative to insulin can be readily calculated from the insulin to C-peptide ratios. On this basis, selective loss of C-peptide from granules, aside from that due to formation of des-(27-31)C-peptide (calculation based upon I/(CPϩtCP) in the 1-h stimulated medium), was 12.2 Ϯ 1.3 and 33.1 Ϯ 1.2% following 2 and 10 h of chase, respectively. The net loss of intact C-peptide (calculation based upon I/CP in the 1-h stimulated medium) was 22.6 Ϯ 1.6% and 58.1 Ϯ 1.7% at these same two times. I/(CPϩtCP) in the 1-h stimulated medium was lower than in the cell extracts, indicating that some selective degradation of C-peptide relative to insulin must have also occurred in a non-secretory granule compartment, presumably multigranular bodies.
Since selective loss of C-peptide from granules might occur by constitutive-like secretion of vesicles enriched in C-peptide relative to insulin (10), we assessed the activity of this pathway in these cells by measuring I/(CPϩtCP) in the 2-and 10-h chase medium, prior to the 1-h stimulation period. Release of Cpeptide via this pathway is predicted to result in an I/CP ratio less than one in the medium (10). Surprisingly, I/(CPϩtCP) in the pre-stimulation chase medium was close to 1 at both time points (2 h, 0.96 Ϯ 0.04; 10 h, 1.02 Ϯ 0.02). Since no extracellular degradation of either insulin or C-peptide was observed (see "Experimental Procedures") and I/(CPϩtCP) in secretory granules was greater than 1 (see above), the equimolar release of labeled insulin and C-peptide observed during the pre-stimulation chase period must be due to the net contributions of both constitutive-like secretion (I/(CPϩtCP) Ͻ 1) and granular secretion (I/(CPϩtCP) Ͼ 1). Thus, C-peptide loss via constitutive-like secretion occurs in neonatal rat beta cells and contributes to the I/(CPϩtCP) Ͼ 1 present in secretory granules. However, it can be calculated that this pathway only accounted for 1.7 and 9.1% of the C-peptide selectively lost from granules during the 2-and 10-h chase periods, respectively. Thus, the majority of C-peptide selectively lost from granules as reflected in the elevated I/(CPϩtCP) in the 1-h stimulated media was due to degradation rather than constitutive-like secretion.
Insulin secretory granules are known to be acidic (1,24). We reasoned that the selective loss of C-peptide from granules and/or truncation may be dependent on this acidic environment. Therefore, we examined the impact of NH 4 Cl on Cpeptide, since this agent has been shown to effectively restore the pH of the intragranular milieu to neutrality (24) and we have shown it to inhibit loss of C-peptide from INS cell secretory granules. 3 However, since proinsulin to insulin conversion is dependent on granular acidification (24 -26), we added NH 4 Cl only after the first 2 h of chase, in order to allow extensive conversion of newly synthesized (labeled) proinsulin to have occurred. Addition of NH 4 Cl for the last eight of the 10 h of chase was followed by 1 h stimulation with the secretory mixture (without NH 4 Cl). I/(CPϩtCP) in the 1-h stimulated medium of NH 4 Cl-treated cells was significantly lower than that of cells that were chased for 10 h in the absence of NH 4 Cl (1.14 Ϯ 0.04 versus 1.50 Ϯ 0.03; p Ͻ 0.001; Fig. 4, lower panel, hatched bars). In fact, I/(CPϩtCP) in the NH 4 Cl-treated cells 3 M. Neerman-Arbez and P. A. Halban, unpublished data.

FIG. 2. Amino acid sequence of C-peptides from rat and human. Truncated rat C-peptides I and II (des-(27-31)C-peptides I and II)
are produced by truncation of rat C-peptides I and II at Leu 26 -Glu 27 (underlined), which will also liberate the C-terminal pentapeptide. Note that the Leu 26 -Glu 27 cleavage site is also present in human C-peptide. Differences in amino acid sequence from rat C-peptide I are shown in bold.

FIG. 3. Release of truncated C-peptide I (des-(27-31)C-peptide I) from neonatal rat islet cells expressed as a percent of total C-peptide I (intact plus des-(27-31)C-peptide I).
Neonatal cells in monolayer were pulse-labeled (10 min; [ 3 H]leucine) and then chased for 2 or 10 h as indicated. In one set of cells NH 4 Cl was added for the last 8 of the 10-h chase ("10 h ϩ NH 4 Cl"). At the end of these periods, cells were incubated for a further 1 h in the presence of a mixture of beta cell secretagogues to stimulate exocytosis (as described under "Experimental Procedures"). C-peptide truncation was measured by HPLC analysis of products in the 1-h stimulated media which are taken to reflect the contents of secretory granules. Data are presented as mean Ϯ S.E., n ϭ 3. was identical to that observed following only 2 h of chase. Thus, since NH 4 Cl had only been present in the chase medium as from the 2-h time point, it had totally inhibited the selective loss of C-peptide from secretory granules during the last eight of the 10-h chase. Although I/(CPϩtCP) was lower in cell extracts after 10 h of chase with NH 4 Cl than without (1.58 Ϯ 0.03 versus 1.86 Ϯ 0.11; p Ͻ 0.05), it was still higher than that in cell extracts following the 2-h chase (1.58 Ϯ 0.03 versus 1.37 Ϯ 0.06; p Ͻ 0.05; Fig. 4, upper panel, hatched bars), suggesting that loss of C-peptide in non-granular compartments is less dependent on an acidic environment or that NH 4 Cl is less effective at neutralizing such compartments.
Interestingly, although addition of NH 4 Cl for the last eight of the 10-h chase did reduce the proportion of labeled C-peptide that was truncated to 23.6 Ϯ 0.2%, this value was still significantly greater than that observed after 2 h of chase (11.9 Ϯ 0.2%; p Ͻ 0.001; Fig. 3).
Isolated Adult Rat Islets-C-peptide in granules of adult islet beta cells was considerably more stable than in the newborn cells. I/CP in extracts of isolated adult rat islets was 1.21 Ϯ 0.03 and 1.71 Ϯ 0.05 after 2 and 10 h of chase, respectively (Fig. 5,  upper panel, open bars). I/(CPϩtCP) was only slightly lower than I/CP (Fig. 5, compare open and hatched bars) reflecting the relatively small, albeit significant, amounts of des- (27)(28)(29)(30)(31)C-peptide in extracts of adult beta cells (4.7 Ϯ 0.5 and 10.3 Ϯ 0.5% of total C-peptide at 2 and 10 h of chase, respectively). However, unlike the neonatal cells, I/(CPϩtCP) in the granules (1 h stimulated medium) of adult islets was close to unity following both the 2-h (0.95 Ϯ 0.03) and the 10-h (1.06 Ϯ 0.03) chase periods (Fig. 5, lower panel, hatched bars). The present study thus confirmed our previous observation (9) that C-peptide is indeed not selectively lost from granules of adult islets aside from truncation. The proportion of des- (27)(28)(29)(30)(31)Cpeptide in granules of adult rat beta cells was lower than that of neonatal rat beta cells, being 3.8 Ϯ 1.6% of total (intact plus truncated) C-peptide following 2 h of chase and increasing to 8.5 Ϯ 0.6% after 10 h of chase.
Perfused Adult Rat Pancreas-In order to confirm that the generation of des-(27-31)C-peptide in isolated adult islets was not an artifact of the culture system or the experimental approach used, we next sought to confirm, by an independent approach, that des- (27)(28)(29)(30)(31)C-peptide is indeed secreted from normal adult rat pancreas. Therefore, we perfused the pancreas of three adult rats, used HPLC to separate C-peptides I and II (and their des- (27)(28)(29)(30)(31) truncated forms) in the venous effluent and quantified their amount by RIA. Since the Cpeptide RIA detects C-peptide II but not C-peptide I (see "Experimental Procedures"), we used C-peptide II and des-(27-31)C-peptide II to estimate the percent of C-peptide released from the adult rat pancreas in this truncated form.
Expression of Human Proinsulin in INS Cells-The comparison of the sequence of rat C-peptides I and II with that of human C-peptide (Fig. 2) revealed conservation at the site cleaved in the generation of des-(27-31)C-peptide. This led us to suspect that human C-peptide may also be susceptible to truncation in the same way as the rat peptides. In order to monitor the truncation of rat and human C-peptide in the same cellular setting we took advantage of recombinant adenovirus to express human proinsulin at high levels in transformed INS cells. We have shown previously that truncation of rat C-peptides is an active process in these cells (9). Following a pulse label and a 2-h chase, the INS cells were stimulated for 15 min and products released during this period analyzed by HPLC. The infected cells released two radioactive products which co- eluted from HPLC with internal standards of human C-peptide and des-(27-31)C-peptide (Fig. 7). No such products were released from control (non-infected) cells (Fig. 7), confirming their identity as products derived from human proinsulin. Of the total human C-peptide (intact plus truncated) released from the infected cells, 31% was in the form of des-(27-31)Cpeptide, compared with 37% truncation for rat C-peptide I (not shown). Human C-peptide is thus truncated in transformed rat beta cells, and the degree of truncation is comparable to that seen for rat C-peptide in the same cells.

DISCUSSION
The pancreatic beta cell not only synthesizes and stores insulin and C-peptide, it is also responsible for degrading these products if they are not released (13,27). Such degradation arises by crinophagy (28). This process, common to all secretory cells, involves fusion of secretory granules with lysosomes resulting in the formation of multigranular bodies (29), the classic degradative compartment for granular products. Once insulin and C-peptide have been targeted to this degradative compartment, they are no longer subject to regulated secretion. Insulin from most animal species exists in crystal form within granules and the relative stability of insulin within multigranular bodies has been attributed to its crystal state (28,30). Such is not the case for C-peptide, which is located in the halo of the secretory granule (28,31), is soluble, and is thus degraded very rapidly once introduced into the lysosome (28,30). The result of this notable difference in stability of these two molecules formed as a result of proinsulin processing is an elevated ratio of insulin to C-peptide in extracts of beta cells (19). Measuring such ratios in extracts of whole cells does not, however, provide an accurate indication of the situation within functional secretory granules and thus cannot provide an indication of the relative proportion of insulin and C-peptide secreted from the beta cell.
It is commonly assumed that both insulin and C-peptide are equally stable within beta cell secretory granules (in contrast to multigranular bodies) and that they will thus be secreted in equimolar amounts. However, recent work from our group has shown that this is not always the case. In transformed beta cells (INS cells), we have found that C-peptide but not insulin is lost from secretory granules (9). Such selective loss results in the disproportionate release of insulin relative to C-peptide. In the present study, we demonstrate that the selective loss of granular C-peptide is not unique to transformed cells but that this process is also operative in primary neonatal (but not adult) rat beta cells. In addition to the selective loss of granular C-peptide, we have determined that limited proteolysis of Cpeptide also occurs in beta cell secretory granules of both neonatal and adult rats, resulting in the production and secretion of a truncated form of C-peptide, des- (27)(28)(29)(30)(31)C-peptide.
Truncation of C-peptide was first shown in 1973 to occur within beta cells (8) but has been largely ignored since that time. In the initial description of this occurrence in rat islets, it was suggested that cleavage of C-peptide arose prior to proinsulin conversion (8). The conversion process would then liberate a truncated form of C-peptide. This truncated form (Cpeptide 1-22 or des- (23)(24)(25)(26)(27)(28)(29)(30)(31)C-peptide to use currently accepted nomenclature) was lacking its nine C-terminal amino acids (8). The truncated C-peptide described in the present study differs both in its mode of generation and its composition. We have found that the proportion of this particular truncated form increases between 2 and 10 h of chase. Given that proinsulin conversion is complete within granules of newborn rat beta cells within 2 h (32), this increase thus suggests that the truncated form studied here is generated from free C-peptide as well as possibly from proinsulin. Furthermore, molecular characterization of the truncated forms of both C-peptide I and II show clearly the loss of just the last five, not nine, residues. Note, however, that in the present study we have not attempted to identify other truncated forms of C-peptide by HPLC and our data thus do not exclude the existence of the previously identified truncated form of C-peptide (8) or, indeed, yet smaller forms.
The detailed molecular characterization of truncated C-peptide I and II was performed on peptides from adult rat islets. It is assumed that the same truncation is also occurring in the newborn cells based upon the elution times from HPLC. Furthermore, based on the known amino acid sequences of rodent and human C-peptide, we predicted that this truncation event could occur in human C-peptide given that the same Leu 26 -Glu 27 cleavage site is present in the human molecule (Fig. 2) and indeed in other higher species (33). To test this hypothesis, human proinsulin was expressed in transformed rat beta cells, and the human C-peptide stored in granules of these cells was indeed found to be truncated to roughly the same extent as its rat counterpart. It should be noted in this context that a cascade of different truncated C-peptides has recently been identified in extracts of a human insulinoma (7). As discussed above, the study of truncated peptides in tissue extracts does not provide any indication of the situation prevailing in secretory granules since the extracts will reflect the combined contents of granules and multigranular bodies. Thus, the finding of such truncated peptides in extracts does not necessarily imply that they are secreted by the beta cell. Moreover, it is also well established that truncation can arise as an artifact of extraction, and such has indeed been found for both dog (removal of the N-terminal octapeptide) (34) and rhinoceros (removal of the C-terminal octapeptide) (35) C-peptides. Our demonstration that des- (27)(28)(29)(30)(31)C-peptide is not only present in cell extracts, but is also secreted in a regulated manner from several different preparations (primary cells from newborn and adult rats, the perfused rat pancreas and transformed rat beta cells), tends to rule out that this truncation event is an artifact and rather suggests that it is a normal post-translational modification of C-peptide in the beta cell secretory granule.
Two mechanisms have been proposed for the selective loss of C-peptide from granules (9). First, C-peptide could be degraded directly within secretory granules. Second, C-peptide could be selectively removed from secretory granules in microvesicles. Such vesicles could in turn have two fates. They could discharge their contents by exocytosis representing the post-granular constitutive secretory pathway suggested by some to be operative in secretory cells including the pancreatic beta cell (10,36,37). The data from this study suggest that the constitutive-like pathway makes a significant, albeit small contribution to the loss of C-peptide from secretory granules of neonatal rat beta cells. In this respect, neonatal rat beta cells differ from transformed beta cells, in which constitutive-like secretion makes a negligible contribution to the loss of C-peptide from secretory granules (9). A second fate of these microvesicles is that they could be targeted to lysosomes where the C-peptide would be degraded. Our data do not allow us to distinguish whether C-peptide is being degraded directly within secretory granules or being selectively transferred to lysosomes. Although neutralization with NH 4 Cl indicates clearly that loss of C-peptide from within secretory granules is dependent upon compartmental acidification, this observation is compatible with either mechanism as both could be dependent on an acidic environment.
Truncation of C-peptide was much more evident in immature than in mature beta cells. However, even in the adult beta cell secretion of des- (27)(28)(29)(30)(31)C-peptide was not trivial, comprising approximately 10% of total C-peptide immunoreactivity released from the perfused adult rat pancreas under both basal and glucose-stimulated conditions. Interestingly, unlike the process responsible for selective C-peptide loss from the secretory granule, the truncation process is not totally acid-dependent since it was not fully inhibited by addition of NH 4 Cl. This suggests that these are separate processes although it is possible that truncation may be an early step in a more extensive degradative process. Alternatively, it is possible that the truncated form of C-peptide may represent a biologically important peptide or may be an intermediate in the formation of a yet smaller product of biological importance that we could not identify in our HPLC analysis.
The findings from the present study may have implications for clinical pathophysiology, both in terms of measurement and potential biological activity of C-peptide related molecules. If disease states in which beta cell function is perturbed (such as insulinoma or non-insulin-dependent diabetes mellitus) are associated with selective granular loss of C-peptide and consequently non-equimolar release of insulin and C-peptide, then use of C-peptide as a marker of functional status of the beta cell may underestimate true function. Furthermore, should truncated forms of human C-peptide escape detection by other analytical techniques, such as radioimmunoassay, then beta cell secretory function may similarly be underestimated. Finally, the identification and characterization of a novel truncated form of C-peptide, des- (27)(28)(29)(30)(31)C-peptide, also raises interesting questions regarding the potential biological activity of C-peptide. Presently, a definite biological role of intact C-peptide has not been demonstrated (see Ref. 38, for review). If des- (27)(28)(29)(30)(31)C-peptide is indeed released in humans, it may serve a biological function. It appears less likely that the Cterminal pentapeptide liberated by the truncation would have biological activity since this region of C-peptide is not well conserved ( Fig. 2 and Ref. 33). The possibility that des- (27)(28)(29)(30)(31)C-peptide is biologically active has not been considered in studies in which intact C-peptide has been administered in search of its biological function. Indeed, our data raise the possibility that intact C-peptide may itself not be a bioactive peptide but merely the biosynthetic precursor of such a peptide.