Induction of Proline-rich Proteins in Hamster Salivary Glands

Treatment of hamsters with the fl-agonist isoproterenol caused a dramatic increase in a series of unusual proteins in the parotid and submandibular glands. These proteins are acid soluble and they contain high amounts (mol %) of glutamate plus glutamine (30-35), proline (23-30), and glycine (12-25). Three proteins (HP.15, HlP.3a, and HP,13b) were isolated from trichloroacetic acid extracts of parotid glands of isoproterenol-treated hamsters. The basic protein (HP,15) was not retained by l)EAE-cellulose and did not contain phosphate or carbohydrate. Two acidic proteins (llP13a and ltP,13b) had the same apparent molecular weight on sodium dodecyl sulfate-polyacrylamide gel electrophoresis, but these separated by were DEAEcellulose chromatography. HP,13a and HP43b contained -1.3 and 5.7 phosphate residues/mol of protein, respectively. Levels of mRNAs encoding this series of proteins showed striking increases following isoproterenol treatment as determined by cell-free translations and Northern analysis. Feeding tannins to rats and mice mimicks the effects of isoproterenol treatment on the parotid gland (Mehansho, H., Hagerman, A., Clements, S., Butler, L., Rogler, J, and Carlson, 1). M. (1983) Proc. Natl. Acad. Sci. U. S. A. 80, 394P3952; Mehansho, H., Clements, S., Sheares, B. T., Smith, '., and Carlson, D. M. (1985) J. Biol. Chem. 260, 4118-4-123)). However, hamsters on a high tannin diet (2%) did not respond like rats and mice and instead displayed an unusual growth inhibition. %Veanlinghamsters maintained on a 2% tannin diet initially lost weight for 3 days and then failed to gain weight for up to 6 months when kept on this diet. Essentially a normal growth rate was observed when the tannin-fed hamsters were switched to a normal diet.


Induction of Proline-rich Proteins in Hamster Salivary Glands by Isoproterenol Treatment and an Unusual Growth Inhibition by Tannins*
Treatment of hamsters with the fl-agonist isoproterenol caused a dramatic increase in a series of unusual proteins in the parotid and submandibular glands. These proteins are acid soluble and they contain high amounts (mol %) of glutamate plus glutamine (30-35), proline (23-30), and glycine (12-25). Three proteins (HP.15, HlP.3a, and HP,13b) were isolated from trichloroacetic acid extracts of parotid glands of isoproterenol-treated hamsters. The basic protein (HP, 15) was not retained by l)EAE-cellulose and did not contain phosphate or carbohydrate. Two acidic proteins (llP13a and ltP,13b) had the same apparent molecular weight on sodium dodecyl sulfate-polyacrylamide gel electrophoresis, but these separated by were DEAEcellulose chromatography. HP,13a and HP43b contained -1.3 and 5.7 phosphate residues/mol of protein, respectively. Levels of mRNAs encoding this series of proteins showed striking increases following isoproterenol treatment as determined by cell-free translations and Northern analysis. Feeding tannins to rats and mice mimicks the effects of isoproterenol treatment on the parotid gland (Mehansho, H., Hagerman, A., Clements, S., Butler, L., Rogler, J, and Carlson, 260, 4118-4-123)). However, hamsters on a high tannin diet (2%) did not respond like rats and mice and instead displayed an unusual growth inhibition. %Veanlinghamsters maintained on a 2% tannin diet initially lost weight for 3 days and then failed to gain weight for up to 6 months when kept on this diet. Essentially a normal growth rate was observed when the tannin-fed hamsters were switched to a normal diet.
Salivary glands of' various animals can synthesize, or can be induced to synthesize, a group of l)roteins which are unit sially high in proline, tile so-called proline-rich proteins 'This research was supp,)rted in part by Public Health Service Grants I)K-36812 and I-I L-;60:31. This is .Jurnal Paler 11 .156 froim (PI1s)' (1). These proteins collectively constitute about 70% of the proteins in human salivary secretions (2). The PRPs are encoded by multigene families (1) and can undergo various post-translational modifications including proteolysis, phos l)horylation, and glycosylation. These unusual proteins are presumfably constitutive in human saliva (2), but families of similar proteins are dramatically increased or induced in parotid and submandibular glands of rats (3, 4), mice (5,6), and hamsters (this report) by isoproterenol treatment. The nucleotide sequences of several PRImRNAs from rat (7,8), mouse (8), and human (9) and the structures and organiza tions of' complete genes of' PIP inultigene families from the mouse (10), hamster (11), and human (12) have been reported. Proteins derived from the nucleotide sequences are all char acterized by four general regions: a putative signal peptide, a transition region, and the repetitive region, a carboxyl-ter minal region (1).
Previously we found that feeding tannins mimicked the effects of' isoproterenol on parotid glands, causing glandular hypertrophy and induction of l'RPs in rats (5) and mice (13). The apparent tissue-specific synthesis and the appearance of PRPs in saliva suggest a biological function in the oral cavity and gastrointestinal tract. Evidence has been presented that these proteins have high affinities for tannins and that they can reverse the detrimental effects of tannins in the diets of rats (13) and mice (5). However, hamsters do not respond to dietary tannins by inducing the synthesis of' PRPs and as a result tannins show an unusual ability to inhibit growth. Tannins are also unusually toxic to hamsters. We have iso lated and partially characterized three proteins from hamster l)arotid glands which are high in glutamate (or glutanine), l)roline, and glycine and which are dramatically induced by isoproterenol treatment. In an attempt to determine the rea sons for the different responses to tannins, assays for 0 adrenergic receptors and adenylate cyclase veie performed on membrane preparations from parotid glands of rats and hamsters. Cell-free translations and Northern analysis were carried out on control and isoproterenol-treated animals to determine the extent of induction of PRI mRNAs.

EXPERIMENTAl IROCEItJRES
Materials-All materials were of highest purity available and were purchased from commercial sources unless otherwise indicated. The following were purchased from resl)e(tive -,apanics: I. - F'eding Trials---Male (;olden Syrian hamsters (50-60 g) were maintained on Purina lab ('how for 5-6( days before starting the feeding experiments. Sorghum diets were prepared as described elsewhert, I:0. Feed and water were provided ad libiturn. Tannin contents were ineasured by a tompetitive binding assay (14). The condensed htnnin contents of Savanna sorghuai grain and Quebracho extract were 2 and 5")., respect ively. Gielatin, Quebracho extract, or an amino acid mix equivalent to gelatin in amino acid composition was added it sirghium diets as indicated at -' of the dry weight.
Isolation and Analysis of mRNA-Total RNA was prepared by using the guanidine thiocyanate-cesium chloride procedure described in Chirgwin etal. (21). RNA was translated in a rabbit reticulocyte lysate system, and the translation products were analyzed as described previously (6). Northern blot analysis was performed as described by Thomas (22) using electrophoresis in 1.5% agarose gel containing 2.2 mM formaldehyde (23). : 0 P-Labeled exon IllI of hamster PRP gene H29 (11) was prepared by nick translation and was used as a probe for hybridizations. The filters were treated as described previously (6).

RESULTS
Effects of Isoproterenol Treatment-Isoproterenol treat ment has a dramatic hypertrophic effect on the parotid and submandibular glands of rats (3, 4) and mice (5) with the sizes of these glands increasing by about 10and 5-fold, resI)ectively, after 10 days of' treatment. Increases in sizes of the parotid and submandibular glands of' hamsters were es sentially negligible (<0.5-fold), but isoproterenol treatment of' Lamsters (lid induce the svnthesis or accumulation of acid soluble proteins abott 6-8-fold (Table I). As found with rats (:1) and mice (5), these )roteins were high in proline but were aIlso tnusually high in glutamate (or glutamine) ( Table II).
Elect rophoret ic patterns of' acid-soluble proteins and gly coproteins induced in salivary glands of hamsters treated with "Amino acid composition derived fro the nucleic acid sequence of hamister Pil' gene H29 (11) its given in Fig. 9 aid assuming "SAla is the N-terminal amino acid of' the mature protein.
'Asx = Asp + Asn. GIx Gil+ Gin. ' ND, not detected. varying amounts of isoproterenol are shown in Fig. 1. The profiles and levels of most proteins were essentially unchanged when the amount ofisoproterenol was increased from 0.5 to 3.5 mag/animal. Several proteins are glycosylated, especially in isoproterenol-treated submandibular glands (Fig.  1B). A glycoprotein of M H,, of 89,000 (GP89) showed an unusual regulatory pattern; it was induced dramatically in parotid glands of hamsters treated with 2.5 mg of isoproterenol/day.
lProtein Isolations--The acid-soluble proteins extracted with l0"i, trichloroacetic acid from parotid glands of isoproterenol-treated hamsters were partially resolved by DEAEcellulose chromatography (Fig. 2). Fractions in peak-1, 111, IV, and V lacked absorhance at 280 tm, which is consistent with results obtained witi other PRlPs (3). To check for homogeneity, each fraction from )eaks I through V was assayed by S1)S-PAGE (data not shown). As illustrated in Fig.  2 (inset), peak I which was not retained by the DEAEcellulose C0lumn11 appears as one protein in SDS-PAGE. and this protein has a M,,,,,, of'-15,000 (HP-5). Peak III includes at least 3 proteins. Peaks IV and V contain glycoproteins HP43a and H1431), each with a M,,,,,, of' 43,000. A minor contaminant of -P43h was removed by repeating the DEAEcellulose chromatography wit h a lower salt gradient.
(ompositional Analysis-The amino acid compositions of HP45, HP43a, and HP43h are presented in Table I. These proteins are all high in glutamic acid (or glutaminet, proline, glycine, and aspartic acid (or asparagine). They either lack or contain very low amounts of aromatic and sulfur-containing amino acids. HP-n3a and HP-131) are different, but they do have similar amino acid compositions. HP-13a and HP-,3b contain -1.3 and 5.7 meol of phosphate/meol of protein, respeclively (Table It. The amino acid composition of the protein encoded by hamster PRP gene H29 ( 11) is similar to, but not identical with, Ather HP43a or HP43b.
Partial ,in:no Acid Seq ences Jf HPI5, ttP43a, and Hl'.13h--Sequences of the amino-terminal regions of HP,15, HP43a, and HP-1b are compared together wit h,sequence data from proline-rich proteins of rat, mouse, and human (Fig. 3). There is about 73C' homology in the first I1 amino acids of HP-l4a and HP43h. When the sequence of H P43a is compared with the amino acid sequence derived from PRP gene H29 (residues 15-25) (11), there is a single substitution at position 8; aspartic acid HP-13a) (codon GA'/,.) for isoleucine (H29) (codon ATA). \Vhen sequences of HP43a and HP43h were aligned with peptides encoded by PRP cDNAs of mouse

A. C00massie
(pMP125) and rat (pRP33), there were two regions of homol ogy for all four peptides. The basic PRP, HP45, had no homolog, with either HP43a or HP43b (or with other basic PRPs from rat (8), mouse (8), and human (24)). Both HP43a and HP43b have relatively high amouros of serine and thre onine which are potential sites for phosphorylation and gly cosylation. For example, residue 9 in HP43a and HP43b is missing from sequence analysis as would be expected if it was either glycosylated or phosphorylated. Treatment with Proteases and ('NBr-HP45 and HP43a were treated with various proteases and with CNBr. A char acteristic difference was observed between HP15 and HP43a (Fig. 4). HP45, the basic PRP, was resistant to S. aureus protease VS whereas HP43a was an excellent substrate. Both behaved similarly toward clostripain and trypsin. Chymotryp sin hydrolyzed HP45 extensively, but HP43 appeared to have a single cleavage site. As expected, both HP45 and HP43a were resistant to CNBr.
Binding AssaYs of Hamnster I'roline-rich Proteins to Tan nins-A high affinity for tannins is one of the characteristic properties of PRPs (13). The relative affinities of tannins for four PRPs from hamster parotid glands were measured by a binding assay using '"C labeled bovine serum albumin (14). These proteins have about an 8-fold greater affinity for tannin than does bovine serum alhunin (Fig. 5). Results of these binding assays are comparable to those obtained fo rat (13) and mouse '5) PRPs.
Cell-free Translations-Total RNAs prepared from parotid glands of isol)roterenol-treated and control hamsters were translated in vitro by the reticulocyte lysate system (Fig. 6) in the presence of [HIH-proline. Isoproterenol treatment in creased demonstrably the in vitro synthesis of several pro teins. Many proteins induced by isoproterenol treatment were apparently present normally in low aniounis. Two polypep. tides (Al, 58,000 and a low molecular weight protein) disap peared with isoproterenol treat ment. This observation is con sistent with the reduction or disappearance of a-amylase and parotid-sl)ecific )rotein in cell-free translations of RNAs from isoproterenol-treated rats (7) and mice (6). Cell-free transla tions with ["Sjmethionine (not shown) clearly demonstrated the dramatic decrease in n-amnyiase.
Northern Hvbridization-ReIati,,e changes of PP mRNAs after isoproterenol treatment were rdetermined by Northern analysis using exon Ill of hamster 1PRP gene H29 (1H) as the probe (Fig. 7). Isoproterenol treatment caused about a 10-fold induction in PRP i1RNAs. 'his increase is considerably lower than the 70-fold increase observed in mice (6). Parotid glands lost about 20( of their weight and within about 20 days after of hamsters, however, have higher basal levels of PRP mRNAs the diets were switched, both groups were close to the same than did the mouse. As observed in rat (8) and mouse (6), two weight. Growth curves for rats on high and low tannin diets major size classes of RNAs were detected, are shown in Fig. 8 (inset). The addition of either gelatin or Effects of Feedinga Diet High in Tannin-Unlike with rats the amino acid mix to the low tannin diet had little effect on (13) and mice (5), feeding high tannin sorghum (Savanna) to growth rate (Table IV) In ccntrast, when gelatin, which also hamsters for 3 days failed to cause either glandular hypertrohas a high affinity for tannins, was added to the high tannin phy or an increase in the levels of PRPs (Table I1). Upon diet, the weight gains of the hamsters improved dramatically prolonged feeding of high tannin sorghum, an unusual inhi- (Table IV) and were essentially the same as hamsters main bit ion of growth w ,s observed (Fig. 8). In fact, after 6 months tained on low tannin sorghum. Savanna plus an amino acid on Savanna sorghum, hamsters were essentially (±5 g) the mix equivalent to the composition of gelatin, on the other same weight as at 3 days (data not shown). At this point (6 hand, had no effect on the tannin-mediated growth inhibition. months), diets were switched and hamsters previously on The lack of effect of the crystalline amino acid mix eliminates Savanna grew at close to the normal rate observed for the the possibility that the beneficial response to gelatin in the younger aninmals. Hamsters switched to the high tannin diet high tannin diet involves the amino acids or total nitrogen Tissue-specificInducible Gene Expression       (aspartate instead of' isoleucine) (Fig. 3). The amino acid sequence of HP43a is blank in position 9, and the derived sequence from H29 has a serine in this position. Therefore, this serine is likely phosphorylated or glycosylated. This same reading frame also encodes a polypeptide of 5 tandemly re peated peptides of amino acid sequence Pro-Pro-Gin-Gin Glu-Gly-Gln-Gln-Gln-Asn-Arg-Pro-P ro-Lys-Pro-Gly-Asn-Gln-Glu-Gly ( Fig. 9 (11)). If HP43a is encoded by H29, it would contain 19 glutamates in 169 amino acids, which would explain why it is an excellent substrate for S. aureus V8 (Fig.   4, lane 2). From residues 53-63 (Fig. 9, Asp-Glu-Glu-Gly-Asp-Asp-Asp-Gly-Glu-Glu-Asp), nine out of eleven amino acids are acidic, identifying the polypeptide encode by H29 as an acidic PRP. Based on the derived amino acid sequence, there is only one chymotrypsin site among 169 amino acids (F-45) which is consistent with the protease studies of HP43a (Fig.   4, lane 5). From the results obtained so thr (composition analysis, acidic nature, N-terminal sequence analysis, and sensitivity to proteases), we concluded that H29 codes for HP43a or a very closely related protein.
Northern analysis (Fig. 7) showed two size classes of mes sages, 1100 and 850 bases, which is common to rat and mouse p (6 extract was extremely toxic to hamsters (Table IV), and within 3 days halt'of t oxhe animals died. withi 3 days lat~yclaThe halffth anda died. Assays forp-ReceptorsandAden*vlat C*vclase Activity-The tannin-niediated induction of' PllPs inrats and mice was completely blocked by the s3-agonist propranolol (25). Since hamsters failed to respond to tannins, it was possible that hamster parotid glands have fewer il-receptors and less adenlate tyclase activity. Results obtained from t'l-hydroxybenzylpindolol binding studies clearly showed that membrane preparations from both rat and hamster parotid glands contained essentially the same nut.ber ofl-receptors (Table V). Moreover, membrane preparations from rat and hamster parotid glands formed 18.5 and 13.5 pmol of cyclic AMP, respectively, per ing of protein (Table V). Thus, the failure of hamster parotid glands to respond to tannins is not due to hamster basic PRP HP45 was not phosphorylated (Table 1I), which is different from human basic PRP LB-1 (24). Also there was no N-terminal homology of HP45 with other basic PRPs from rat, mouse, and human. On the other hand, N-terminal sequences among human, rat, and mouse acidic PUPs, and hamster acidic PRPs HP43a and HP43b are quite conserved (Fig. 3). As we reported previously (11), the evolution of the N-terminal region of acidic PRPs is apparently under negative selection against replacement sub stitution. In other words, the N-terminal regions of basic PRPs are under less stringent functional constraint than those of acidic PRPs. Recently, Braunlin et al. (28) reported that the N-terminal fragment of a human acidic PRP is involved in calcium binding, and this might explain the func either a decrease in adenylate cyclase activity or to fewer /I-tional constraint. receptors. DISCUJSSION Proline-rich proteins and glycoproteins are specifically synthesized and secreted by salivary glands of various animals. In addition to the hamster PRPs reported here, these unusual proteins have been isolated and characterized from human (see Ref. 2 for a review), rat (3, 4), mouse (5,8), monkey (26), and rabbit (27). Mainly as a result of work from our laboratory it is now known that PRPs of mice, rats, and hamsters are tissue-specific products of multigene tamilies which are dramatically induced by isoproterenol. These unusual proteins pose a number of interesting problems concerninggene structure and evolution, regulation of gene expression, post-translational niodification,, and protein conformation.
Ann et al. H11) reported the complete sequence of hamster lPRI gene H29. The amino acid composition of the derived polypeptide encoded by H29 (Fig. 9) is compared to those of HP43a and -1P.13b (Table I1). The derived sequence of' one reading frame from exon 11 of H29 is Ala-Thr-Ile-Tyr-Glu-As)-Ser-lle-Ser-(iln-Leu-Ser (1 1) which is identical to the Nterminal sequence ol' HP43a (Fig. 3) except in position 8 With regard to regulation of PRP gene expressions, a different picture is presented by hamsters: (i) hamsters re spond to isoproterenol treatment by an increased synthesis of PRPs (Figs. 1, 6, and 7), but there is little, if any, hypertrophic response (Table I) when compared to mice and rats; (ii) feeding tannins has essentially no effect on hamster salivary glands (Table III); (iii) hamsters fed a diet containing 2% tannin lose weight for about 3 days, as do rat. (13) and mice (5), but then an unusual continued growth inhibition is ob- 27C a a o6 Q a 0P6 0 t11 Q a P a 0FPP 48 0 p S A S D I L 6 0 D D t 1 E D 6 N A PI G 70pp 0 Q 6Gf I Q K P R P P K P 6 N a 6 REPEAT REGION 1PPa a t a a a N RP P KP 6 N 0 F 6 1i1 pa a a a a N RPP K P I 0 13P P a a a a a Q p KP GN 0 C6 11 P 0 F6Q0a NRP P K P G N a F G lilP P a a S t tI S t 18