Differential expression of proline-rich proteins in rabbit salivary glands.

Salivary glands synthesize and secrete an unusual family of proline-rich proteins (PRPs) that can be broadly divided into acidic and basic PRPs. We studied the tissue-specific expression of these proteins in rabbits, using antibodies to rabbit acidic and basic PRPs as well as antibodies and cDNA probes to human PRPs. By immunoblotting, in vitro translation, and Northern blotting, basic PRPs could be readily detected in the parotid gland but were absent in other salivary glands. In contrast, synthesis in vitro of acidic PRPs was detected in parotid, sublingual, and submandibular glands. Ultrastructural localization with immunogold showed heavy labeling with antibodies to acidic PRPs of secretory granules of parotid acinar cells and sublingual serous demilune cells. Less intense labeling occurred in the seromucous acinar cells of the submandibular gland. With antibodies to basic PRPs, the labeling of the parotid gland was similar to that observed with antibodies to acidic PRPs, but there was only weak labeling of granules of a few sublingual demilune cells, and no labeling of the submandibular gland. These results demonstrate a variable pattern of distribution of acidic and basic PRPs in rabbit salivary glands. These animals are therefore well suited for study of differential tissue expression of PRPs.

Aggregated, conformationally changed fibrinogen exposes the stimulating sites for t-PA-catalysed plasminogen activation.

The present paper shows that conformationally changed fibrinogen can expose the sites A alpha-(148-160) and gamma-(312-324) involved in stimulation of the tissue-type plasminogen activator (t-PA)-catalysed plasminogen activation. The exposure of the stimulating sites was determined by ELISA using mABs directed to these sites, and was shown to coincide with stimulation of t-PA-catalysed plasminogen activation as assessed in an assay using a chromogenic substrate for plasmin. Gel permeation chromatography of fibrinogen conformationally changed by heat (46.5 degrees C for 25 min) demonstrated the presence of both aggregated and monomeric fibrinogen. The aggregated fibrinogen, but not the monomeric fibrinogen, has exposed the epitopes A alpha-(148-160) and gamma-(312-324) involved in t-PA-stimulation. Fibrinogen subjected to heat in the presence of 3 mM of the tetrapeptide GPRP neither aggregates nor exposes the rate-enhancing sites. Thus, aggregation and exposure of t-PA-stimulating sites in fibrinogen seem to be related phenomena, and it is tempting to believe that the exposure of stimulating sites is a consequence of the conformational changes that occur during aggregation, or self-association. Fibrin monomers kept in a monomeric state by a final GPRP concentration of 3 mM do not expose the epitopes A alpha-(148-160) and gamma-(312-324) involved in t-PA-stimulation, whereas dilution of GPRP to a concentration that is not longer anti-polymerizing, results in exposure of these sites. Consequently, the exposure of t-PA-stimulating sites in fibrin as well is due to the conformational changes that occur during self-association.

Stimulating effect on tissue-type plasminogen activator--a new and sensitive indicator of denatured fibrinogen.

A high clottability and a short thrombin clotting time have routinely been considered as evidence of genuineness of the fibrinogen molecule. Since denatured fibrinogen stimulates the t-PA-catalysed conversion of plasminogen to plasmin, it was of interest to study the sensitivity of t-PA-stimulation as evidence of fibrinogen denaturation. Therefore, fibrinogen was intentionally exposed to various denaturating conditions (freeze-drying, heating, EDTA, alkali), and the clottability, the thrombin clotting time and the t-PA-stimulating effect were recorded. We found that the clottability was a poor indicator of fibrinogen denaturation, whereas the t-PA-stimulating effect could detect even mild fibrinogen denaturation. The thrombin clotting time was shortened after freeze-drying or heating at 47 degrees C, in spite of what might have been expected. Thus, denaturation is not necessarily accompanied by a prolonged clotting time. In some instances therefore, the t-PA-stimulation is an even more sensitive and reliable indicator of fibrinogen denaturation than is the thrombin clotting time. Consequently, this parameter should be combined with the thrombin clotting time to characterise preparations of fibrinogen.

D-dimer specific monoclonal antibodies react with fibrinogen aggregates.

Human fibrinogen exposed to 46.5 degrees C was subjected to gel permeation chromatography. The protein eluted in two distinct peaks. The first peak appeared in the void volume containing soluble fibrinogen aggregates, while the other peak represented monomeric fibrinogen. In contrast to the monomeric peak material, the aggregate fraction reacted with a panel of monoclonal antibodies specific for fragment D-dimer using an ELISA system. Edman degradation showed that both the aggregate and the monomeric fractions were devoid of soluble fibrin, and immunoblots of SDS-PAG electrophoretic profiles disclosed no sign of stabilized high molecular weight derivatives. We have previously shown that the aggregate fraction of similarly treated fibrinogen, in contrast to the monomeric fraction, stimulates the t-PA catalyzed conversion of plasminogen to plasmin and concomitantly exposes the sequences Aalpha-(148-160) and gamma-(312-324) involved in t-PA stimulation. Our present and previous findings suggest that soluble fibrinogen aggregates possess a fibrin-like structure, and that fibrin or fibrinogen polymer formation is a prerequisite for the enhancing effect on t-PA-mediated plasminogen to plasmin conversion which is seen even with the polymers in the soluble state.

Normal and fibrinaemic patient plasma contain high-molecular weight crosslinked fibrin(ogen) derivatives with intact fibrinopeptide A.

Freshly drawn plasma samples from healthy subjects and from fibrinaemic patients were subjected to electrophoresis on SDS-agarose (unreduced material) or on SDS-PAG (1D and 2D, reduced material) and Westernblotted. The blots were immunovisualized using either polyclonal anti-fibrinogen or a monoclonal antibody (Y18) to fibrinopeptide A-containing molecules. The following results were obtained: 1. Normal plasma as well as plasma from patients with fibrinaemia contained FXIII-crosslinked HMW oligomers, stabilized through gamma gamma-dimerization as well as alpha-polymer formation and these oligomers contained molecules with intact A alpha-chain N-termini. 2. Cross-linked material amounted to less than 0.1% of the fibrinogen pool regardless of the sample studied, and apparently less in fibrinaemic patient plasma than in normal plasma. Thus, since the ratio of crosslinked fibrin(ogen) derivatives to that of fibrinogen was lower for fibrinaemic plasma than for normal plasma, it is suggested that the type of soluble fibrin which gives rise to a positive EGT in fibrinaemia patients is not crosslinked.

Immunovisualization of fibrinogen A alpha-chain heterogeneity in normal plasma and plasma from patients with DIC or on streptokinase therapy.

Purified fibrinogen as well as normal plasma, or plasma from patients with DIC or undergoing streptokinase(SK)-therapy was subjected to 1D- and 2D SDS-electrophoresis under reducing conditions. The pattern was revealed either by Coomassie-staining or immunostaining after Western-blotting and then compared. The use of polyclonal antibodies to fibrinogen as well as two monoclonal anti-bodies reacting with FPA and C-terminal part of the A alpha-chain confirmed immunologically the previously reported molecular weight heterogeneity of the A alpha-chain of the fibrinogen molecule as being a constituent of normal plasma, and lead to the following conclusions: 1. The MW-heterogeneity is observed in the fibrinogen pool of normal plasma as well as in DIC-plasma, SK-plasma and in purified fibrinogen, being the least noticeable in normal plasma and most advanced in SK-plasma. Patterns obtained using immunostaining with monoclonal anti-FPA confirm that the MW-heterogeneity of fibrinogen is mainly due to C-terminal degradation of the A alpha-chain. 2. Numerous A alpha-chain remnants (at least 9), with intact N-terminal ends, are found to be present in normal plasma, with a MW range from 66,200 to 36,000 Da, demonstrating that each of the "classical" HMW, LMW, LMW' subgroups consist of fibrinogen molecules which are very heterogeneous. 3. Two populations of A alpha-chains in purified fibrinogen and in fibrinogen in plasma react with the C-terminal specific Mab G-8. This is in contrast to the findings in plasma from streptokinase-treated patients, where several bands of lower molecular weights than the gamma-chain can be seen, suggesting the presence of free, circulating A-alpha chains split in the N-terminal half of the chain beyond the last inter-chain disulphide bond. 4. 2D electrophoresis disclosed substantial deviations in the patterns obtained with DIC-plasma, SK-plasma and with fibrinogen purified by beta-alanine-precipitation from that observed with normal plasma. The present technique allows selective characterization of fibrinogen independently of the other proteins present in plasma and offers extreme sensitivity.

Soluble, cross-linked fibrin(ogen) hybrid oligomers do not stimulate t-PA conversion of plasminogen.

Cross-linked hybrid oligomers of fibrinogen and fibrin are found in plasma from fibrinaemic patients and normal individuals as well as in preparations of purified human fibrinogen. The present study was undertaken to see if such hybrid oligomers have the same stimulatory effect on tissue plasminogen activator (t-PA) conversion of plasminogen as do polymeric and monomeric fibrin. Hybrid oligomeric fibrin(ogen) material was provided by subjecting purified human fibrinogen to gel filtration in urea-containing buffer at pH 5.6. Well separated fractions of hybrid oligomeric material and monomeric fibrinogen were thus obtained. Some of this material was converted to soluble polymeric or monomeric fibrin using insolubilized thrombin. Hybrid polymeric fibrin, polymeric fibrin or monomeric fibrin were then added to citrated, normal plasma to 2.5 or 5 per cent of the plasma fibrinogen concentration. The added material was kept in solution by plasma fibrinogen. The "COA-SET Fibrin Monomer Test" (Kabi,Stocholm,Sweden), based on the ability of fibrin monomers to enhance t-PA mediated plasminogen-plasmin conversion, was used to compare the potential stimulatory effect of the preparations above. The results led to the following conclusions: 1) Cross-linked, soluble fibrin(ogen) hybrid polymers in a concentration of 5 per cent of plasma fibrinogen concentration (w/w) do not stimulate t-PA. 2) Thrombin conversion of the fibrin-fibrinogen hybrid material resulted in an increase in the rate of t-PA mediated plasminogen conversion, corresponding to the one observed with equivalent (w/w) amounts of fibrin monomers. Compared on a mole to mole basis, fibrin oligomers are more powerful than fibrin monomers as stimulators of t-PA activity.

Structural and genetic aspects of proline-rich proteins.

Considerable advances have been made in the genetics of salivary proline-rich proteins (PRP). The genes for acidic, basic, and glycosylated PRP have been cloned. They code for precursor proteins that all have an acidic N-terminal followed by proline-rich repeat sequences. Structural studies on secreted proteins have demonstrated that not only acidic but also some basic PRPs have this general structure. It is possible that mRNA for different PRP may have originated from a single gene by differential mRNA splicing, but post-translational cleavages of the primary translation product apparently also occur. In vitro translation of salivary gland mRNA results in a single precursor protein for acidic PRP. Such in vitro translated protein can be cleaved by salivary kallikrein, giving rise to two commonly secreted acidic PRPs, and kallikrein or kallikrein-like enzymes may be responsible for other post-translational cleavages of PRPs. Acidic as well as some basic PRPs are phosphorylated. A protein kinase has been demonstrated in salivary glands which phosphorylates the PRPs and other secreted salivary proteins in a cAMP and Ca2+-calmodulin-independent manner. Knowledge of the conformation of PRPs is limited. There is no conclusive evidence of polyproline-like structure in the proline-rich part of PRPs. Ca2+ binding studies on acidic PRPs indicate that there is interaction between the Ca2+ binding N-terminal end and the proline-rich C-terminal part. This interaction is relieved by modification of arginine side-chains. 1H, 32P, and 43Ca NMR studies have further elucidated the conformation of acidic PRPs in solution.(ABSTRACT TRUNCATED AT 250 WORDS)

Phosphorylation of salivary proteins by salivary gland protein kinase.

Human saliva contains a number of phosphorylated acidic proline-rich proteins (APRP). Monkey parotid saliva contains a similar protein with the same phosphorylated sequences as the human proteins. A crude protein kinase was prepared from Macaca fascicularis parotid glands which phosphorylated human APRP. The enzyme was activated by Mg2+, it had a pH optimum between pH 7.0 and 7.5, the Km for ATP was 78 mumol/L, and for APRP it was 85 mumol/L. Phosphorylation of APRP was independent of cAMP and calmodulin. Phosphate was incorporated as phosphoserine, and the kinase phosphorylated the same residues in dephosphorylated APRP which are phosphorylated in the secreted protein. In addition, the enzyme preparation also phosphorylated dephosphorylated and native APRP in a region which is not phosphorylated in the secreted protein. There was no difference in the rate of phosphorylation of APRPs and their tryptic peptides. The kinase also phosphorylated other dephosphorylated salivary phosphoproteins. An enzyme was demonstrated in the human salivary gland which gave the same pattern of phosphorylation of APRP as did the simian kinase. More than one kinase may be necessary for the observed phosphorylation.

Quantitation of human salivary acidic proline-rich proteins in oral diseases.

Acidic proline-rich proteins (APRP) were quantitated immunochemically in salivary secretions from groups of: caries-resistant (CR) and caries-susceptible (CS) subjects; heavy- and light-calculus-formers; and patients with Sjögren's Syndrome, drug-induced xerostomia, and recurrent parotitis. In all groups except the parotitis patients, there were comparable levels of APRP, about 40-50 mg%, with similar values in parotid and submandibular saliva. In chronic recurrent parotitis, the values were somewhat higher (about 60 mg%). There were no differences in the proportion of APRP-A to C in a subset of CR and CS. Taken as a whole, the data support the view that the secretion of APRP is stable and that caries status and propensity to calculus formation are not associated with abnormal levels of these phosphoproteins.

Fibrinogen present in EDTA--anticoagulated plasma stimulates the tissue-type plasminogen activator-catalysed conversion of plasminogen to plasmin.

The presence of soluble fibrin in plasma is an early and sensitive indicator of activation of the coagulation system. Quantitative spectrophotometric assays for soluble fibrin can be based on the principle that soluble fibrin stimulates the tissue-type plasminogen activator-catalysed conversion of plasminogen to plasmin. It was previously shown that treatment of purified fibrinogen by EDTA, which removes the three tightly bound Ca2+ ions, results in exposure of tissue-type plasminogen activator-catalytic sites similar to those unveiled by thrombin. Since EDTA is a common anticoagulant, it was of interest to study the effect of EDTA on a test based on plasminogen activation. It is concluded that the determination of soluble fibrin in EDTA-anticoagulated plasma from healthy individuals gives a false positive indication of the presence of soluble fibrin. This was true irrespective of whether the test was performed at pH 7.4, 7.8 or 8.5. The most probable explanation is that tissue-type plasminogen activator-stimulating sites are exposed in fibrinogen by EDTA. Therefore, EDTA-plasma is unsuitable for assaying soluble fibrin with tests based on the tissue-type plasminogen activator-mediated conversion of plasminogen to plasmin.

Freeze-dried fibrinogen or fibrinogen in EDTA stimulate the tissue-type plasminogen activator-catalysed conversion of plasminogen to plasmin.

Both soluble and insoluble fibrin stimulate the tissue-type plasminogen activator-catalysed conversion of plasminogen to plasmin. Whether fibrinogen can exert a similar effect has been a controversial issue. The present investigation shows that while fibrinogen purified by beta-alanine precipitation does not stimulate the tissue-type plasminogen activator-catalysed plasminogen activation, fibrinogen which has been either lyophilized or stripped of bound Ca2+ ions by EDTA chelation, stimulates this reaction. The data indicate that such procedures alter the molecular conformation of fibrinogen, and expose stimulatory sites which are hidden in the native fibrinogen molecule. These results may explain previous findings concerning the capacity of fibrinogen as a stimulator of the tissue-type plasminogen activator-catalysed plasminogen activation. Since even slight alteration of the molecular structure of fibrinogen leads to an increase in the tissue-type plasminogen activator stimulation, the authors suggest that this can be used to test if the fibrinogen is in a native state.

Characterization of fibrinogen/fibrin derivatives isolated from normal and fibrinaemic plasma and serum using paramagnetic particles coated with a monoclonal antibody (mAb) to D-dimer.

Paramagnetic particles coated with a monoclonal antibody to D-dimer (mAb S4) were used to isolate and concentrate D-dimer and D-dimer-containing complexes in plasma and serum. Antibody-captured material was eluted with SDS-urea buffer and examined by either SDS-polyacrylamide or submerged SDS-agarose gel electrophoresis followed by Western blotting. The protein pattern was visualized by either polyclonal antibodies to human fibrinogen or monoclonal antibodies specific for fibrinogen derivatives containing fibrinopeptide A (FpA; mAb Y18), the N-terminus of the beta-chain in fibrin (mAb 59D8) or the gamma-chains (mAb J88B). The results obtained show that paramagnetic particles coated with mAb S4 catch intact D-dimer as well as a variety of cross-linked fibrin molecules of high-molecular-weight (HMW) in plasma. The existence of HMW derivatives in fibrinaemic serum indicates that some of the HMW fibrin related material in such plasma is not clottable. Fibrinogen/fibrin monomers and some of the fibrinogen/fibrin related derivatives found in the eluates were probably non-covalently bound to the complexes caught by mAb S4 coated particles. The present technique combines the selective concentrating power of immunoparticles and the sensitivity of immunovisualization and allows rapid and direct identification of minute amounts of circulating immunoreactive fibrinogen/fibrin derivatives.

Early cross-linked fibrin in human plasma contains alpha-polymers with intact fibrinopeptide A.

alpha-polymer formation, as opposed to gamma-chain dimerization has been considered a relatively late event in factor XIII-induced fibrin stabilization. Recently it has been shown, however, that plasma from healthy individuals and from patients with fibrinaemia contains small amounts of soluble fibrin/fibrinogen oligomers interlinked through dimerized gamma-chains as well as cross-linked alpha-chains. The present work was carried out to see if these early alpha-chain polymers also arise during coagulation of plasma in vitro. Plasma samples from healthy individuals, prepared by immediate centrifugation of blood collected without anticoagulant, were allowed to clot spontaneously for varying periods. The plasma clots were solubilized in SDS-urea-mercaptoethanol and samples were subjected to SDS-PAGE and Western blotting using polyclonal antibodies to human fibrinogen, or monoclonal antibodies specific either for A alpha/alpha-chains, for fibrinopeptide A-containing chains, for the N-terminus of the fibrin beta-chain or for the gamma-chains. Fibrin/fibrinogen oligomers were seen to form long before visible gelation of plasma. These oligomers were cross-linked through gamma-chain dimerization, but also through A alpha- or alpha-chain polymerization. The number and amount of alpha-polymers containing A alpha-chains increased immediately after clot formation, but these disappeared about 20 min later, due to complete removal of fibrinopeptide A (FPA) by thrombin. It is concluded that alpha-polymer formation is a very early event during plasma coagulation in vitro, and that both A alpha- and alpha-chains are involved.

Immuno-radiometric assays for human salivary acidic proline-rich proteins and their N- and C-terminal fragments.

An immuno-radiometric assay was developed for acidic proline-rich proteins from human saliva, and assays designed which specifically detect the N- or C-terminal parts of the proteins. The immuno-radiometric assay depends on the binding of antigen to paper discs coated with specific antibodies. Subsequently, the discs are incubated with radioactive specific antibodies. The amount of radioactive antibodies bound to the discs depends on the amount of antigen already adhering to the discs.

Interaction of tannin with human salivary proline-rich proteins.

Tannins are polyphenolic compounds, widely distributed in plant-based foods, which have harmful effects on animals including humans. Salivary proline-rich proteins (PRPs) may act as a defence against tannins by forming complexes with them and thereby preventing their interaction with other biological compounds and absorption from the intestinal canal. The aim here was to compare the ability of members of the family of human PRPs to form insoluble complexes with tannin and to assess the stability of such complexes under conditions similar to those in the alimentary tract. Basic PRPs (BPRPs), which have no other known biological functions, were very effective in forming insoluble complexes with both condensed tannin and tannic acid. Practically no tannin bound to acidic PRPs (APRPs) and glycosylated PRPs (GPRPs), suggesting that tannin in the diet would not affect their biological activities. There were only small differences in the tannin-precipitating ability of various BPRPs of different sizes or sequences, indicating that, although there is considerable phenotypic variation of PRPs, it is not likely to cause marked individual variation in tannin-binding ability. Tryptic digestion of an APRP led to a marked increase in tannin binding to the resulting proline-rich peptides, supporting observations in other studies that there may be an interaction between the proline-poor N-terminal and the proline-rich C-terminal regions in native APRPs, which inhibits the biological activities of the proteins. Deglycosylation of a GPRP also led to a dramatic increase in tannin-binding ability, showing that the carbohydrate side-chains prevent binding of tannin. Most of the condensed tannin-PRP complexes remained insoluble under conditions similar to those in the stomach and small intestine, supporting the proposal that PRPs act as a defence against tannin.

Isolation and characterization of six proteins from rabbit parotid saliva belonging to a unique family of proline-rich proteins.

Proline-rich proteins are major components of salivary secretion from humans non-human primates, rats, hamsters and rabbits. They are also synthesized in mice in response to chronic stimulation by beta agonists. This study to provide an understanding of the structural and genetic relationships within these families of proteins to determine the possible function of the proline-rich proteins. Rabbit parotid saliva was collected and proline-rich proteins were affinity purified using goat antibodies to human proline-rich proteins. Purification was achieved by repeated cation exchange chromatography on a Mono S column a Fast Protein Liquid Chromatography system. Six basic proline-rich proteins were purified. The apparent molecular weights were between 75,000 and 125,000, based on their mobilities in sodium dodecyl sulphate-polyacrylamide gel electrophoresis. Glycine, glutamine (and glutamate) and proline accounted for 79-87% of total amino acids in all proteins, but proline was present in smaller amounts (17-21%) than in proline-rich proteins from other species. All proteins were glycosylated but not phosphorylated. Circular dichroism of two proline-rich proteins, MS7A and MS5B, indicated the absence of secondary structure. The N-terminal sequences of three proteins electro-eluted after preparative gel electrophoresis were determined. A high degree of similarity was found in various regions of mouse, rat, monkey and human proline-rich proteins. Rabbits thus synthesize constitutively a family of proteins that are immununologically and structurally related to proline-rich proteins other species.

Demonstration of proline-rich proteins in rabbit parotid saliva and partial characterization of some of the proteins.

Rabbit parotid saliva was collected by cannulation of the secretory duct in anaesthetized animals. Proteins which cross-react with antibodies to human acidic proline-rich proteins were demonstrated in the secretion and fractionated by chromatography on an immunosorbent and CM32-cellulose. At least 5 proline-rich proteins were identified with molecular weights ranging from 19,000 to 61,000 as determined by sodium dodecylsulphate acrylamide gel electrophoresis. The major immunoreactive components were basic proteins with similar size and charge properties. In two of the proteins, proline, glycine and glutamic acid or glutamine accounted for 84 or 99 per cent of all residues. In contrast to proline-rich proteins from other species, proline constituted only 13 or 17 per cent of total amino acids.

Characterization of human sublingual-gland protein kinase by phosphorylation of a peptide related to secreted proteins.

Phosphoproteins in human saliva include proline-rich proteins, statherins, histatin 1 and cystatin SA-III. The presence of phosphate in these proteins is necessary for various functions in the mouth including calcium binding, inhibition of precipitation of calcium phosphate, inhibition of growth of hydroxyapatite crystals and adherence to hydroxyapatite. To elucidate the process of phosphorylation of these proteins, the phosphorylation of a peptide (APRP8) with an amino acid sequence identical to one of the phosphorylated sites in acidic proline-rich proteins by a kinase from the human sublingual gland was investigated. The kinase, which was highly labile, was purified 58-fold by fractionation of sublingual gland homogenate and gel filtration, but the enzyme was inactivated when further purification by chromatographic techniques commonly used for protein kinases was attempted. To compare the enzyme with other kinases, and to obtain information that could be used in its further purification, a characterization was undertaken. The enzyme required 10 mM Mg2+ for optimum activity, it had a KM of 0.09 mM for ATP and the KM for the peptide substrate APRP8 was 0.42 mM. It was not activated by cAMP or calmodulin, characteristics that are shared with casein kinases and mammary gland kinase. The sublingual kinase as well as casein kinase 2 were inhibited by heparin, but in other respects the two kinases had different properties. While casein kinase 2 is activated by polylysine and has optimal activity in 150 mM KCl, sublingual kinase was inhibited by polylysine and the addition of KCl. Moreover, casein kinase 2 can utilize both ATP and GTP as phosphoryl donors, but GTP was not a substrate for sublingual kinase. The sublingual kinase shared a substrate recognition sequence with mammary gland kinase, but, unlike that kinase, it could not utilize Ca2+ instead of Mg2+. While the sublingual kinase thus shared some properties with both casein kinase 2 and mammary gland kinase, distinct differences were also seen and the relationship to these enzymes remains to be determined. The characterization of the sublingual kinase will be useful in its further purification.

Interaction of tannin with human salivary histatins.

The ability of all major human salivary histatins to precipitate condensed tannin was demonstrated, and it was found that histatins 3 and 5 share the same condensed tannin-binding region but less tannin bound to histatin 1. The condensed tannin-binding region of histatin 5 includes both the N- and the C-terminal parts, although more tannin binding occurs in the C-terminal region. Epigallocatechin gallate (EGCG) showed similar binding characteristics as condensed tannin, but much less EGCG was precipitated. Pentagalloyl glucose (PGG) was precipitated equally well by histatins 1, 3, and 5 and bound equally well to the N- and C-terminal regions of histatin 5. In contrast to condensed tannin, cleaving histatin 5 into N- and C-terminal fragments increased their ability to precipitate PGG. Together, these results show a number of differences in the nature of interaction of histatins with condensed tannin, EGCG, and PGG. Most of the condensed tannin-protein complexes remained insoluble under conditions similar to those in the stomach and the small intestine, suggesting that histatins may act as a defense against dietary tannin in humans.