Lithostathine, the Presumed Pancreatic Stone Inhibitor, Does Not Interact Specifically with Calcium Carbonate Crystals*

Lithostathine protein) is, in addition to albumin, the major nonenzymatic protein of the pancreatic juice. It has been assumed to inhibit calcium carbonate precipitation and therefore to prevent stone formation in the pancreatic ducts. This function is, however, debatable. The assumption is based on the inhibition of in vitro crystal nucleation and growth by lithostathine. Considering that these phe-nomena occur only under certain critical conditions, we re-examined the question using a protein preparation where the purity and folding have been tested by mass spectroscopy and NMR in the absence of nonprotein contaminants. Under these conditions, we showed con-clusively that lithostathine does not inhibit calcium carbonate nucleation and crystal growth. We demonstrated that previous findings on the alleged inhibition can be attributed to the uncontrolled presence of salts in the protein preparation used. Moreover, the affinity of lithostathine to calcite crystals, expressed as the half-life of bound iodinated protein in the presence of unlabeled competitor, was significantly lower than that of bovine serum albumin (8.8 and 11.2 h, respectively). Using glass microspheres instead of crystals did not significantly change the half-life of bound lithostathine (8.0 h). These findings are incompatible with the hypothesis of a specific interaction of lithostathine with calcium carbonate crystals. In conclusion, considering that components of pancreatic

Lithostathine (pancreatic stone protein, PSP) or Reg protein is a 144-residue glycoprotein synthesized by the exocrine pancreas (1,2). It is also expressed in the endocrine compartment of the regenerating pancreas (3,4), as well as in pancreatic ductal cell carcinoma and in colon and rectal tumors (5). The function of the protein is uncertain. It has been shown to stimulate islet regeneration in partially depancreatized rats (6), and it is regarded by several authors as a ␤-cell growth factor implicated in the maintenance of ␤-cell mass and func-tion (7)(8)(9)(10)(11). Reg protein may be involved also in cell proliferation and differentiation in nonpancreatic structures (12).
The occurrence of lithostathine in the pancreatic juice has led to the suggestion that it may prevent stone formation by controlling nucleation, growth, and aggregation of calcite crystals. The two main arguments that have been developed to support this assumption are as follows. First, lithostathine has been shown to delay crystal nucleation and to inhibit growth of preformed CaCO 3 from supersaturated solutions (13), and second, it acts as a calcite crystal habit modifier, leading to the formation of smaller crystals (14). This presumed physiological function has been challenged recently (15). At least three conditions must be met to demonstrate the presumed role of lithostathine: 1) Crystal nucleation and growth occur only under certain critical conditions; thus any interference, such as for example the uncontrolled presence of certain salts, will lead to false positive results. 2) A partially denatured protein may give rise to artifacts; therefore, the sample of lithostathine employed must have the correct three-dimensional structure. 3) It is essential that a "standard" protein, such as BSA, 1 be employed as control. Because these conditions were not all met in the experiments cited earlier, we decided to reinvestigate the role of lithostathine using the approach of Ref. 13, with rigorous purification of the protein and with the checking of its folding (16). We found that under these conditions lithostathine does not inhibit calcium carbonate precipitation and crystal growth. Moreover, we demonstrated that a slight contamination of the protein solution by the buffer used for immunopurification mimics the reported inhibitory effects of lithostathine. Because the inhibitory activity had been attributed to the N-terminal undecapeptide, we determined also the effects of this peptide, in both its glycosylated and nonglycosylated forms, as well as the effects of synthetic peptides of related sequence. We showed also that inorganic compounds in the pancreatic juice, such as NaCl and above all phosphate, are powerful inhibitors of calcium carbonate crystal growth. Lastly, we performed competitive adsorption experiments to determine the affinity of lithostathine to calcite crystals, as compared with an amorphous phase (glass), using BSA as a control. We concluded that lithostathine cannot be a specific inhibitor of pancreatic stone formation. In a critical examination of this phenomenon, lithostathine would instead appear as a promoter of stone formation through its 133-residue isoform (PSP-S 1 ).

Protein Purification and N-terminal Peptide Isolation
Human pancreatic juice was collected by endoscopic cannulation of the main pancreatic duct of patients suffering from various pancreatic diseases. Collection of the pancreatic juice was part of the clinical * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
routine. Lithostathine was immunopurified as described previously (2). It was then concentrated on a cation exchange cartridge (MemSep SP, Millipore), which was eluted with pH 9.0, 50 mM borate buffer containing 0.15 M NaCl. The latter step had the advantage of completely removing the salts used in immunopurification, namely glycine/HCl buffer neutralized with Tris. For the isolation of the N-terminal peptide, the Arg 11 -Ile 12 bond was cleaved by trypsin as described in Ref. 16. The glycosylated N-terminal peptide was isolated from the resulting supernatant. The dry material was dissolved in 0.5% trifluoroacetic acid, and the solution was then loaded onto a 4.6 ϫ 150 mm C18 column (Altex) equilibrated in 0.05% trifluoroacetic acid and eluted with a linear gradient of 0 -60% methanol for 60 min at a flow rate of 1 ml/min. The peptide eluted at 18 min. The eluate was partly evaporated under nitrogen and lyophilized.
In addition, several peptides were synthesized (Neosystem). The natural and the synthetic peptides were quantified by amino acid analysis.

In Vitro CaCO 3 Nucleation and Crystal Growth
Effect of Lithostathine and Derived Peptides-The initiation of CaCO 3 precipitation in vitro was examined as described in Ref. 17. To 0.2 ml of a solution of lithostathine (10 mg/ml in 0.15 M NaCl), 10 ml of 25 mM sodium bicarbonate, pH 8.8, and 10 ml of 16 mM calcium chloride were added with gentle stirring at 25°C. Calcium carbonate precipitation was monitored by the decrease in pH; 0.2 ml of 0.15 M NaCl served as control.
CaCO 3 crystal growth was investigated according to Ref. 13. A metastable supersaturated CaCO 3 solution containing 0.27 mM CaCl 2 and 4.8 mM NaHCO 3 , adjusted to pH 8.8, was seeded with calcite crystals in the proportion of 1 mg of crystals for 4 ml of metastable solution. Seed crystals were prepared as described in Ref. 13; their mean linear size was 0.5 mm, with a specific surface area of 0.5 m 2 /g. The mixture was maintained at 25°C with gentle stirring. The free calcium concentration decreased as Ca 2ϩ ions were incorporated into the crystals. Ca 2ϩ concentration changes were determined by colorimetric assay (Boehringer Mannheim) performed on aliquots, which were collected at 1, 2, 4, and 6 h after seeding (beyond that, the concentration remained stable for more than 24 h). The compounds to be tested were diluted in the metastable solution before crystal seeding. The concentrations mentioned in this report refer to the latter metastable solution. Lithostathine and its N-terminal undecapeptide were used at the highest concentration described in Ref. 13, i.e. 6 M. BSA and synthetic peptides were at the same concentration by weight (100 g/ml).
The following peptides were used: The peptide lith-I had the same sequence as the N-terminal undecapeptide of human lithostathine, i.e. pEEAQTELPQAR. To determine the effect of the glutamic residues, we used the peptide lith-Is, which had the same sequence as lith-I, except that Glu was substituted by Gln, i.e. pEQAQTQLPQAR. The peptide lith-II had the same sequence as the N-terminal undecapeptide of human lithostathine II, product of the regl gene (18), i.e. QESQ-TELPNPR. The peptide rat had the same sequence as the N-terminal undecapeptide of rat lithostathine, i.e. pEEAEEDLPSAR (19). All synthetic peptides were provided by Neosystem (Strasbourg, France) Statistical Analysis-In all experiments described in the present report, assays were done at least in quadruplicate. The inhibitory activity of the various substances tested, in the experiments described above and in the following, was analyzed as follows: the Ca 2ϩ concentrations reached at the end of the assays, i.e. after 6 h, were compared using Student's t test. For the study of lithostathine binding to calcite crystals, regression analysis was performed according to Ref. 21.

Inhibitory Effects Due to Contaminants Potentially Present in the Lithostathine Solution and Effects of Components of Pancreatic Juice-
The protein used in Ref. 13 was prepared according to Ref. 20. The protocol includes an elution step using 0.2 M glycine buffer, pH 2.8. The solution was then neutralized with 10% of 1 M Tris, pH 8.9, and dialyzed overnight against phosphate-buffered saline. Here is a potential source of artifact, because the phosphate buffer used contains 10 mM PO 4 2Ϫ i.e. 10 times the concentration required for the complete inhibition of Ca 2ϩ incorporation into crystals (see "Results"). In addition, because dialysis is unlikely to be 100% effective, the final lithostathine solution would contain residual glycine/Tris buffer. The critical aspect of this experiment is that the amount of phosphate and glycine/Tris introduced into the inhibition assay would parallel the amount of protein (increasing lithostathine concentrations were obtained by adding to a given volume of metastable solution increasing volumes of a stock solution of lithostathine). For this reason, we checked the possible inhibition of calcium crystal growth by glycine/Tris at a final concentration ranging from 0.125 to 1 mM, i.e. 0.5% or less of the buffer used in the immunopurification process. Moreover, we determined the effects of some components of the pancreatic juice, either inorganic compounds or proteins, namely 0.15 M NaCl, 50 M sodium phosphate, 6 M trypsinogen, chymotrypsinogen. To determine the relevancy of inorganic phosphate and 6 M as a potential inhibitor of pancreatic stone formation, we measured the physiological phosphate levels in 10 samples of human pancreatic juice (basal secretion) using the kit provided by Boehringer Mannheim.

Kinetic Parameters of Lithostathine Binding to CaCO 3 Crystals
Iodination-Lithostathine and BSA were equilibrated in metastable CaCO 3 solution (pH 8.8, 0.27 M CaCl 2 and 4.8 mM NaHCO 3 ) at a concentration of 1 mg/ml. 500 g of protein were iodinated with 2 MBq of [ 125 I]iodine in the presence of 10 g of Iodogen (Pierce). Proteins were then desalted on PD10 columns (Pharmacia Biotech Inc.). The specific radioactivity of lithostathine and BSA were 3.5 and 3.1 KBq/g, respectively.
Solid Phase Adsorption and Release-500 l of a metastable solution containing 62.5 g of iodinated lithostathine or BSA, respectively, were incubated for 2 h with 5 mg of calcite crystals. Iodinated lithostathine was also incubated with an equivalent amount of glass microspheres, i.e. an amount that bound the same quantity of radioactivity. The solution was then centrifuged (2,000 ϫ g for 1 min), and the pelleted crystals or microspheres were rinsed twice with 2.5 ml of metastable solution. They were then incubated with 2.5 ml of unlabeled protein solution equilibrated in metastable buffer, according to the data in Table I. The solution was periodically renewed, and the radioactivity of the collected supernatants was measured. All incubations were performed with gentle shaking. The concentration of unlabeled lithostathine was equal to 6 M; unlabeled BSA was used at the same concentration by weight (i.e. 100 g/ml or 1.5 M). Fig. 1 shows the decrease in free Ca 2ϩ ions, resulting from Ca 2ϩ incorporation into calcite crystals. Ca 2ϩ concentration dropped from 270 nM to about 70 nM in 2 h, and then it stabilized at about 50 nM after 6 h. Adding carefully purified native human lithostathine did not significantly modify the phenomenon: the Ca 2ϩ concentration reached at the end of the assay did not differ significantly from one experiment to another (p Ͼ Ͼ 0.05). A similar result was obtained when the isolated N-terminal undecapeptide of lithostathine was added to the metastable solution; the same was true with BSA. All compounds were at 6 M in the final metastable solution, i.e. the concentration used by previous workers (13). The figure shows also a nucleation experiment, where the pH of the solution decreases as a consequence of CaCO 3 precipitation. Again, unlike in previous reports, lithostathine had no effect on the reaction.

Lithostathine Has No Effect on Either CaCO 3 Nucleation or Crystal Growth-
Inhibitory Effects of Synthetic Peptides Related to Lithostathine N-terminal Undecapeptide-Following Ref. 13, we checked the effect of the synthetic peptide named lith-I, which reproduces the N-terminal sequence of lithostathine. It did not reduce the rate of Ca 2ϩ incorporation into calcite crystals when used at 6 M (Fig. 2, f compared with g). This observation is consistent with the fact that considering its amino acid chain, one can predict that the peptide is unlikely to display an organized three-dimensional structure. However, the peptide showed a marked effect, provided it is used at a much higher concentration, i.e. 80 M (about 100 g/ml) (Fig. 2b). Other synthetic peptides with related sequences were also checked for their potential inhibitory effectiveness. The peptide rat, derived from the N-terminal region of rat lithostathine, had the same effect as peptide lith-I when used at the same concentration (Fig. 2a). Glutamic residues were involved in this activity because free glutamic acid, when used at a concentration equivalent to its occurrence in peptide lith-I (160 M), displayed some inhibitory effect (Fig. 2e). However, glutamic acid itself is not sufficient, because the reduction of the Ca 2ϩ incorporation into crystals was partially, but not totally, abolished when the glutamic residues of peptide lith-I were substituted by glutamine (peptide lith-Is) (Fig. 2, d compared with b). Moreover peptide lith-II, derived from human lithostathine II, with the same glutamic acid equivalent as lith-I but with a slightly different sequence, had a significantly reduced inhibitory activity (Fig. 2, c compared with b). These findings indicate that a number of different compounds can reduce the rate of Ca 2ϩ incorporation into calcium carbonate crystals. Electrostatic in-teractions via glutamic acid are presumably involved, and synergy of action with the peptide fragment is probable. Fig. 3 shows that the inhibitory activity attributed to immunopurified lithostathine might reflect the effect of traces of the glycine/Tris buffer used in the purification procedure. Minute amounts of the buffer added to the test solution (final concentrations as low as 0.05-0.6% of that of the initial buffer) led to an inhibition that mimicked that previously found with lithostathine. The Ca 2ϩ concentration reached at the end of the assay (after 6 h) differed significantly from one experiment to another, i.e. with increasing glycine/Tris concentration. The figure shows also that NaCl at a physiological level had a strong inhibitory effect, and 50 M phosphate totally inhibited the growth of calcium crystals (higher phosphate concentrations gradually disaggregated the crystals; data not shown). To determine the physiological relevance of the latter finding, we measured the phosphate levels in 10 human pancreatic juice samples (basal secretion). The concentration was 95.8 Ϯ 18.6 M, with a 77-150 M range; one sample, however, markedly differed from the others, displaying only 14 M, and it was excluded from the calculation of the mean. Thus, according to our analysis, phosphate levels in pancreatic juice are usually higher than 50 M, a concentration that has been shown above to totally inhibit calcium crystal growth in vitro. We determined also the inhibitory activity of 6 M tryspsinogen and chymotrypsinogen. The inhibition, expressed as the difference in free Ca 2ϩ concentration in test experiments versus the control shown in Fig. 3, was 30.8 and 18.8%, respectively.

Artifact Introduced by Possible Residual Glycine/Tris in Lithostathine Solution and Inhibitory Activity of Physiological Components of Pancreatic Juice-
Lithostathine Binding to Solid Phases-To gain further insight into the interaction of lithostathine with calcite crystals, we determined the kinetics of the binding of lithostathine to the crystals, using iodinated protein and unlabeled competitors. As control, we used on the one hand a protein frequently employed for this purpose, BSA, and on the other hand glass micro-  (Table I). Kinetic analysis showed that bound iodinated proteins were released by their unlabeled counterparts. The rate of release fitted a linear regression on a semi-logarithmic scale, as shown by the coefficient of determination: the nul hypothesis H 0 ␤ ϭ 0 was rejected for all the experiments. The comparison of the slope of the regression lines at the p ϭ 0.05 threshold showed that three experiments did not differ significantly from each other (Table I and Fig. 4). We can deduce that the half-life on the calcite crystals was significantly greater for BSA than for lithostathine (11.2 and 8.8 h, respectively); therefore we can assume that BSA has a higher affinity for the crystals (Table I, experiments a and b). In agreement, free BSA released bound lithostathine, with the half-life of lithostathine being reduced significantly to 6.4 h (p ϭ 0.001) ( Table I, experiments b and e). Another important aspect was the finding that the half-life of lithostathine on glass microspheres (8.0 h) was not significantly different from that on crystals; this means that lithostathine binds to calcite crystals in the same way as to glass (Table I, experiments b and c). DISCUSSION Our investigations showed that lithostathine does not inhibit the in vitro incorporation of Ca 2ϩ into CaCO 3 crystals, even at 6 M (about 100 g/ml), the highest concentration employed by other workers (13). The inhibitory effect described by Bernard et al. (13) was probably due to the presence, in the lithostathine preparation employed, of contaminants such as PO 4 2Ϫ and Trisneutralized glycine buffer. Phosphate was introduced by dialysis of the protein solution against phosphate-buffered saline. We stress that according to the method used (20), the PO 4 2Ϫ concentration in the stock solution of lithostathine was 10 times the level required for complete inhibition of Ca 2ϩ incorporation into calcite crystals. Glycine/Tris was used to immunopurify the protein, and it was hardly totally removed by the dialysis procedure: we showed that increasingly minute amounts (less than 0.5% of the initial concentration) of Tris/ glycine can reproduce the inhibition graphs attributed to lithostathine. We stress that specific protein-crystal interactions require a correct folding of the protein; improper folding may lead to nonspecific interactions, with potentially irrelevant effects. For this reason, the three-dimensional structure of our lithostathine preparation, isolated from human pancreatic juice, has been investigated by high resolution NMR spectros-copy (16). The protein appeared to be properly folded, because the secondary structure elements assigned by NMR were in agreement with those present in the three-dimensional organization obtained by x-ray diffraction (22).
In apparent contradiction with some of our findings, "measurable calcite crystal inhibitory activity" was observed recently with a rat lithostathine produced in a baculovirus expression system (15). We have no explanation for the discrepancy. However, the latter protein preparation was stabilized by protease inhibitors. This indicates that contaminants, such as proteases, may be present, because we observed that lithostathine is stable when completely purified (our preparation did not need any stabilizing agent). Knowing the inhibitory effect of trypsinogen, no definitive conclusion should be drawn before the protein produced by genetic engineering has been demonstrated to be devoid of trace contaminants, especially proteases, and its three-dimensional structure has been checked.
We showed also that the undecapeptide derived from the N-terminal region of lithostathine inhibits Ca 2ϩ incorporation into CaCO 3 crystals, however only at a concentration equivalent to about 1 mg/ml of lithostathine, i.e. at a physiologically irrelevant level. Similar results were obtained with the Nterminal undecapeptide from rat lithostathine; substituted peptides had intermediate effects, indicating that no change in the peptide sequence has a determinant effect on the rate of Ca 2ϩ incorporation into calcium crystals. A number of pancreatic juice constituents can also prevent Ca 2ϩ binding, especially NaCl at a physiological level and, among the proteins, trypsinogen. Moreover, special attention should be drawn to phosphate, because of its well known potency as an inhibitor of calcite precipitation (23,24). Our assays show that generally the phosphate level in normal pancreatic juice may suffice per se to inhibit calcium carbonate precipitation. According to Ref. 23, calcium carbonate precipitation cannot occur in biological fluids, unless the phosphate to carbonate ratio decreases because of physiological or pathological changes in cellular activity. This aspect of lithogenesis control has not yet received the attention it deserves.
In addition to this, our findings show that the adsorption of lithostathine to calcite crystals was similar to that of albumin. Under the conditions of our test, the concentration of both proteins at the crystal surface was about 200 ng⅐cm Ϫ2 . This means that binding is a function of total protein mass rather than molecular concentration. Lithostathine bound to calcite crystals was released more effectively by BSA than by lithostathine itself. This means that BSA adheres to crystal surface with higher affinity than lithostathine. Thus, lithostathine is unlikely to act on crystal growth in a way different from albumin, which is present in the pancreatic juice at a mean concentration of 600 g/ml (25). As a comparison, lithostathine occurs at 25-60 g/ml (26); according to Ref. 27, the highest value attained is 137 g/ml. Moreover, the binding of lithostathine to an amorphous phase such as glass microspheres and its rate of release therefrom is comparable with what is observed with calcite crystals. It has been shown that lithostathine alters crystal growth by interfering with the apposition of new layers of calcite (14). However, the same general loss of well defined crystal edges and shapes was observed with plasma fibronectin, which shows high nonspecific affinity for the {104} faces of calcite; again, the result is the reduction of average crystal size, even at a much lower protein concentration (28). Hanein et al. (28) state that they have reason to believe that this result may be applicable to a variety of globular proteins. One may expect, from our adsorption experiments, that the reduction in crystal size by globular proteins may also apply to albumin.
In conclusion, not only does lithostathine not inhibit Ca 2ϩ incorporation into calcium carbonate crystals, but also no relevant specific function can be attributed to it with respect to the control of lithogenesis. In the light of these findings, the role postulated for lithostathine in the development of stones during the course of chronic calcifying pancreatitis should be re-evaluated. We underline that the onset of chronic pancreatitis is marked by precipitation of protein plugs in the pancreatic ducts; proteinaceous calculi later calcify (29), as confirmed by successive examination of patients (30). These calculi are made mainly of PSP-S 1 , the 133-residue isoform of lithostathine. PSP-S 1 , which is generally present in the pancreatic juice (2), tends to precipitate because it is insoluble at the physiological pH. As a result, PSP-S 1 is the major component of the organic matrix of pancreatic stones and was first isolated from these structures (31). In this sense, lithostathine can actually promote stone formation. In apparent contradiction with this assumption, it has been shown that patients at risk with chronic calcifying pancreatitis display reduced levels of pancreatic lithostathine (26,27). We should consider, however, that these patients may have higher levels of altered lithostathine isoforms, which are likely to be ignored by the assays used, based on chromatographic methods. In addition, it should be underlined that patients at risk with chronic calcifying pancreatitis include users of alcohol and that the calcification of proteinaceous calculi is known to be stimulated by alcohol abuse (30). In conclusion, in agreement with Ref. 15, the name "lithostathine" is inappropriate, and the function(s) of the protein remain(s) to be elucidated.