THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 266, No. 18, Iasue of June 25, pp. 11718-11725.1991 Printed in U.S.A.

Recombinant Rhizopuspepsinogen EXPRESSION, PURIFICATION, AND ACTIVATION PROPERTIES OF RECOMBINANT RHIZOPUSPEPSINOGENS*

(Received for publication, January 16, 1991)

Zhong ChenS, Gerald Koelsch, He-ping Han, Xin-Juan Wangp, Xin-li Lin, Jean A. Hartsuck, and Jordan Tang From the Protein Studies Program, Oklahoma Medical Research Foundation and Department of Biochemistry and Molecular Biology, Uniuersity of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma 73104

A cDNA clone, which contained the complete rhizo- puspepsin structure and the putative proregion, was placed in three different Escherichia coli expression vectors for the synthesis of rhizopuspepsinogen (Rpg). Recombinant Rpgs which were expressed in the cytosol of E. coli as inclusion bodies (cRpg and tRpg) were not active. After solubilization in 6 M urea and refolding by rapid dilution, both of these Rpgs were purified to homogeneity. The third zymogen, pRpg, which was secreted to the periplasmic space of E. coli with an omp leader, was fully active and also was purified. The expression level of pRpg was higher (over 40 mg/liter culture) than that of cRpg (about 1.5 mg/liter culture). Amino-terminal sequence analysis of the zymogens re- vealed that cRpg and pRpg contain 40 and 51 residues of prosequence, respectively. tRpg, which was ex- pressed under the control of T7 promoter, was synthe- sized at 500 mg/liter culture and was purified at 50 mg/liter culture. This zymogen contained, in addition to 51 residues of proregion, 16 residues inherited from the expression vector construction. All of these Rpgs spontaneously converted to rhizopuspepsin in solutions of pH less than 5. Each of the conversions was associ- ated with a change of molecular weight as monitored in sodium dodecyl sulfate-polyacrylamide electropho- resis. At least one intermediate of conversion was ob- served in the pH range of 2 to 3 for both the cRpg and pRpg zymogens. For pRpg and tRpg, kinetic data dem- onstrated that the Rpg to rhizopuspepsin conversion was accomplished by a first order, unimolecular reac- tion at pH 2. The first order kinetic constants in this pH at 15 OC were 1.1 and 2.4 min” for pRpg and tRpg, respectively. The activation rate decreased as pH was raised above pH 2. At pH greater than 3.0, rhizopus- pepsin-catalyzed, second-order activation also takes place. Consequently, the recombinant Rpgs are acti- vated by either of two cleavage mechanisms as is the case for pepsinogen. These results also support the hypothesis that Rpg is synthesized in Rhizopus chinen- sis as a zymogen. Rpg in the host fungus is probably

* This study was supported by Research Grant AM-01107 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) M63451.

$Present address: The Hematology Section, Dept. of Medicine, University of Oklahoma Health Sciences Center, Oklahoma City, OK 73190.

Present address: Beijing Medical University, Building 8, Group 12, No. 225, Beijing, People’s Republic of China.

activated by an acid environment of pH less than 5 in the secretory granules to become rhizopuspepsin be- fore secretion.

Rhizopuspepsin (EC 3.4.23.6) is an aspartic protease pro- duced by the fungus Rhizopus chinensis for the digestion of protein substrates in the growth medium. The specificity of this enzyme, which is similar to many other general purpose aspartic proteases such as mammalian pepsins, and the ki- netics of this enzyme are well documented (1, 2). The amino acid sequence (3, 4) and high resolution crystal structure (5) of rhizopuspepsin have been determined. Additionally, rhizo- puspepsin is closely related in tertiary structure and in active center structure to other aspartic proteases, including renin (6) and aspartic proteases from Rous sarcoma virus (7) and human immunodeficiency virus (8). These structural similar- ities imply that all these enzymes are related in evolutionary origin and in catalytic mechanism. All these factors make rhizopuspepsin an important model for studies of structure and function relationships of aspartic proteases.

The zymogens of the aspartic proteases are of considerable interest because the biological functions of these enzymes are intimately related to their zymogen structures and their mech- anisms of activation. I t is well known that most aspartic proteases are derived from their precursors by an acid-trig- gered, intramolecular activation mechanism (9). The only known exception is the activation of prorenin, which is ap- parently accomplished via proteolysis by a yet unidentified protease (10). Fungi, however, secrete mature aspartic pro- teases such as rhizopuspepsin. It has not been established whether the aspartic proteases of fungi are expressed as zymogens, and, if they are, what would be their activation mechanism.

During the cloning and sequencing of rhizopuspepsin cDNA, we observed a stretch of sequence upstream from the amino-terminal position of the mature enzyme (4).’ The length of this region (68 amino acid residues) and its moderate sequence homology to the “pro” region of pepsinogen sug- gested that this sequence may be that of the “prepro” sequence of a rhizopuspepsin precursor, rhizopuspepsinogen (Rpg).’ We, therefore, decided to express cloned cDNA in Escherichia

R. Delaney, R. N. S. Wong, and J. Tang, unpublished results. * The abbreviations used are: Rpg, rhizopuspepsinogen; Rp, rhizo-

puspepsin, cRpg, rhizopuspepsinogen expressed from vector pGBT- Rpg into the cytosol of E. coli; pRpg, rhizopuspepsinogen expressed from plasmid PIN-111-ompA3-Rpg into the periplasmic space of E. coli; tRpg, rhizopuspepsinogen expressed from vector PET-3a-Rpg; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electropho- resis.

11718

Recombinant Rhizopuspepsinogen 11719

coli in order to test whether an active Hpg could be obtained for activation studies. In this paper we report the results of these experiments.

EXPEKIMENTAL PKOCEDUHES:'

RESULTS

('onstruction of Rhizopuspepsinogen cDNA and Expression Vectors-The Rpg gene was constructed from two clones. Clone 33332 (4) contained the entire protease region and part of the putative prosequence. Another clone' 36C6 contained par t of the protease region, the putative proregion and part of the "pre" region (by sequence homology to the pepsinogen proregion). A ScaI site was utilized to ligate the 5"fragment to t he 3"fragment of 33E2 to form a rhizopuspepsinogen gene. The nucleotide and amino acid sequences of the putative prepro region of this clone are shown in Fig. 1. There are 68 amino acid residues in front of the amino-terminal position of rhizopuspepsin (Fig. lA, residues -1 to -68). Hased on the genomic structure of Rpg (isozyme PI 5);' residue -68 is part of the initiation codon (Met) with the adenosine base missing in the cDNA clone. For the construction of expression vectors in E. coli, we estimated the approximate position of the Hpg amino terminus based on the homology of its proregion to tha t of pepsinogen (see "Discussion"). Hpg cDNA was sub- cloned into M13 so that modification by nucleotide-directed mutagenesis would be possible. In this process, the putative presequence was removed, and an b h R I site and an initiation codon were installed in front of the proregion (see "Experi- mental Procedures" and Fig. 1H) . T h e I.:coRI/l-'stI fragment was cloned into vector pGBT-T19 to form pGRT-Rpg, which contained a tac promoter and provided a Shine-Dalgarno site. Although, in pGBT-Rpg vector, the Rpg gene is not in phase with the initiation codon, it directed the synthesis of pRpg at a level of about 1.5 mg/liter culture, as monitored by Western blots. No proteolytic activity of rhizopuspepsin was detected for the expressed cRpg (results not shown). Another expres- sion vector, pIN-III-ompA3-Rpg, which contains an ump leader sequence to secrete Rpg to the periplasmic space of E. coli, was constructed by inserting the EcoRI/HindIII fragment from pGRT-Rpg into plasmid PIN-111-ompA3. In this con- struction (Fig. lC), the Rpg cDNA fragment is connected in phase after the omp leader with the addition of three extra amino acids (Gly-Ile-His) between the omp leader and the zymogen. The PIN-111-ompA3-Rpg was transformed into E . coli JM109 cells so that fully active pRpg was synthesized at a level of over 40 mg/liter culture. A third expression vector, PET-3a-Rpg, controlled by a T 7 bacteriophage, was con- structed (Fig. 1D). The EcoRI/HindIII fragment from vector pIN-111-ompA3-Rpg, containing the entire pepsinogen-coding region, was cloned into vector PET-3a (11) and expressed in the host strain BL21(I>E3). tRpg was synthesized as inclusion bodies a t a level of about 500 mg/liter of culture.

Purification of Recombinant Hhizc~ppuspep,sino~en,s-Rhizo- puspepsinogens expressed from all vectors were purified to homogeneity. Insoluble cRpg was dissolved in urea, refolded by dilution, and purified (Table I ) by a series of chromato- graphic separations, on DEAF,-Sephacel (Fig. 2) , Sephadex ( i - 7 5 (Fig. 3 ) , and Mono Q (Fig. 4) columns. The purified cRpg migrated as a single band on SDS-polyacrylamide gel

' Portions o f this paper (including "E:xperinlental I'rocedures," 'I'ahles 1-111, and Figs. 2-4. (i-8) are presented in miniprint a t the end o f t h i s paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microl'ilm edition o f the Journal that is available from Waverly Press.

~ ~~

' I<. 1)elaney. personal communication.

2 Si '"- . : :E ". 1" :::E Eli?,:SS::!; :'CT:3 CF t.i?;

5 2 I i l C !;2r I "

?;, , . . h : l l . T ~ T h . i: AT; C.:: A:C AT: A:: J;: ;;A Ch: t M AT; ?,+.I *:s i,.: Y,.: TL: c:p " . y ;:n z:?. Y I I

1 5 -.- _". LC.: G C A TtA AT7 C A T i:A G T i M i . 6:)r ~ r c , - ! y Srr : : r H;r A:. Y a I Arn . .

>I

" , t : - ,> L,..!,,: I I L%!.:,,; .....;,_ -..;,,.-- , : !.*

FIG. 1. The cDNA and protein sequences of the putative preproregion of rhizopuspepsinogen (Rpg) and the relevant structures in the constructions of expression vectors. A. coding sequence o f preproregion of Hpg. 'The cDN.4 sequence is that o f the clone' : M X and the nunlhcrirlg is on the right margin. The amino acid residue numbering for the preproregion starts from the amino terminus o f rhizopuspepsin and is in the reverse negative numhers. liesidue -68 is the initiation Met 1)ased on the gene sequence o f Hpg? 'I'he douhlr. and singlc triangIcs indicate the amino termini o f cHpg anti pl ipg respectively. 13. sequence from expression vector pGBT- Hpg which produced cHpg. The parent vector. pGHT-'TlS. contains a lac promotor (1', ,2,), a Shine-Dalgarno ( S / I l ) site. and a cloning I.:c.r,HI site. The putative presequence (hases 1 t o 51. see port .-I) u'as removed, and the b h R 1 and A'TG initiation codons were added to the cl)NA by primer-directed mutagenesis (see "Experimental Pro- cedures"). 'I'he nucleotide sequence which follows the initiation hlet is that o l l i p g cl)NA. 'I'he amino terminus o f expressed cKpg is shou.n by 11 L'orticd orrow. ('. sequence from expression vector PIS-111- ompA:%-lipg which produced pRpg. The cDKA fragment resulting I'rom the I.:coKI hydrolysis of vector pGHT-Rpg is connected t o the parent vector plX-lIl-ompA:i to form this expression vector. The vector contains a lipoprotein promoter a Inc promoter. l',,,,. a Shine-l);llg;lrno site, and an omp leader sequence. The amino termi- nus o f ' expressed pHpg is marked with a rcJrticnI arrow. The dot.\ indicate sequences which are not shown. I ) , sequence from expression vector pF,'I'-:%:>-l<pg which produced tIipg. 'The E,'coHI lragment from vector p(;H'I'-Rpg and HnrnHI-digested vector pET-:la were treated with Klenow fill-in and blunt end-ligated. The sequence shwvs Shine- 1)alg:arno (kS/[)) site, initiation codon Met (residue 1 ). I(< extra amino- terminal rehidues. and the sequence from native rhizol)us~)el)~inoge~l starting at Ala-\'al-Asn.

11720 Recombinant Rhizopuspepsinogen

electrophoresis corresponding to a molecular mass of 41 kDa (Fig. 5). The automated Edman degradation produced an amino-terminal sequence of Ser-Ile-Pro. pRpg was fully active and was purified in three steps. After the osmotic shock extraction, pRpg was chromatographed on a DEAE-Sephacel column (Fig. 6) and a Superose 12 column (Fig. 7). pRpg produced a single band of about 41 kDa in SDS-gel electro- phoresis. The amino terminus of the pRpg was found to be Ala-Val-Asn. tRpg was recovered from the culture as insoluble inclusion bodies, washed, refolded and chromatographed on a column of Sephacryl S-300 (Fig. 8). The active zymogen from the column appeared as single band on SDS-PAGE which corresponded to the expected size of 41 kDa. The finai yield of tRpg was about 50 mg from 1 liter of the original culture (Table 111).

Activation of Rhizopuspepsinogem-Recombinant Rpgs spontaneously generated proteolytic activity in acid solutions. This was accompanied by the removal of the propeptide to form rhizopuspepsin. Fig. 5 shows the polyacrylamide electro- phoresis patterns of cRpg and pRpg, which had been incu- bated in acidic buffers ranging from pH 2 to 6 for 10 min a t 37 "C. Incubations at pH 4 or below resulted in the conversion of both these Rpgs to rhizopuspepsin bands, which were about 6 kDa smaller. At least one intermediate of the activation can be seen in the gel. The amino-terminal sequences of the activated products were then examined. The results showed that the amino terminus of the larger species was Thr-Ser- Thr- and the smaller one, mature Rp, was Ala-Gly-Val-.

Rhizopuspepsinogen Actiuation Kinetics-The activation kinetics of Rpg was studied at pH 2 and 3.5 using both pRpg and tRpg rhizopuspepsinogen preparations. The semilogarith- mic plots of zymogen remaining uersus time for two typical experiments at different zymogen concentrations are shown in Fig. 9. Although the semilogarithmic plots produced linear relationships, suggesting the predominant first-order activa- tion reactions, the data were fitted to a mixed first- and second-order activation model (12) in order to detect the possible presence of the two reactions. The first-order acti- vation rate constants resulting from experiments at pH 2 are presented in Table IV. The value of k, for the rhizopuspepsi- nogen activation is concentration independent, and the sec- ond-order constant produced from the model is essentially zero. There is a 2-fold difference in the kl for pRpg and tRpg. For comparison, the activation of porcine pepsinogen was carried out under the same conditions, producing a first-order rate constant almost identical to that of pRpg. Based upon

Activation of Rhizopuspepsinogen

at Different pH Values

koa A B koa

66. - 68

45 .

36 -

. 4 3

. Rpg R W .

RP.

29 . . 2 9

2 3 4 5 6 2 3 4 5 6 5

FIG. 5. SDS-PAGE of activation of both cRpg and pRpg and their activation products generated at different pHs. The cRpg and pRpg were the final purified products as described in Tables I and 11. For the activation experiments, the zymogens were treated with citrate buffers from pH 2.0 to 6.0 at 37 "C for 10 min, respec- tively, then subjected to SDS-PAGE (10% gel) and stained with Coomassie Blue. The results indicated that both forms of Rpg can be activated at pH 4.0 or below.

I O 1 20

1.0 mg/mL

0.5 mg/mL

A I

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

Time (min)

FIG. 9. Semilogarithmic plot of pRpg activation at 15 "C, pH 2, for two different protein concentrations.

TABLE IV First-order activation rate constants for Rpg at 15 "C, p H 2

Activation of porcine pepsinogen (PPgn) was carried out with the same conditions and is shown for direct comDarison.

[Protein] kt m d ml min"

tRpg 0.1-1.0 2.4 (0.8)o n = 6 PRPg 1.0 1.0 (0.2) n = 7 PRPg 0.5 1.1 (0.2) n = 10 PPgn 0.75 1.2 (0.1) n = 4

a Standard errors are shown in parentheses.

the linearity of the semilogarithmic activation plot, a concen- tration-independent first-order rate constant, and a negligible second-order constant produced by the regression analysis, it is concluded that the activation of Rpgs occurs intramolecu- larly at pH 2.

The pH dependence of Rpg activation was demonstrated by observing the first minute of activation of the zymogen a t various pH values. Experimental conditions were chosen to favor a unimolecular activation mechanism with low protein concentration (0.5 mg/ml) and short time of activation (1 min). As shown in Fig. 10, the maximum activation rate is near pH 2. At pH 3.5 and higher, the activation in 1 min is immeasurably small.

In order to demonstrate clearly the presence of a rhizopus- pepsin-catalyzed bimolecular mechanism, activation experi- ments were done in the presence of preformed rhizopuspepsin a t higher pH values where such a mechanism is favored. Excluding the activity due to the presence of preformed enzyme, an increase in the percentage activation of rhizopus- pepsinogen beyond that caused by intramolecular activation is observed (Fig. 10). The data at pH 3.5 demonstrate that the activation rate rises with an increase of concentration of initial rhizopuspepsin, which confirms that rhizopuspepsino- gen may be activated by a second-order, rhizopuspepsin- catalyzed reaction.

Activation of tRpg was also studied in order to compare the activation properties of different recombinant Rpgs. Fig. 11 presents a typical activation experiment carried out at pH 3.5 and fitted to the mixed kinetics model as described above for experiments at pH 2. This experiment alone does not verify the mixed kinetics model. However, work described above does substantiate simultaneous first- and second-order acti- vation processes.

Results are presented in Table V. In contrast to data at pH

Recombinant Rhizopuspepsinogen 11721

100-

80. I

0 1 1 I 1.5 2.0 2.5 3.0 3.5 4.0

PH FIG. 10. pH profile of intramolecular pRpg activation at

15 O C , 0.4 mg/ml. Data are expressed as percentage of activation in 1 min versus pH. Circles represent data from activation experi- ments in the presence of 0.4 mg/ml preformed rhizopuspepsin. The triangle represents data from experiments with 0.8 mg/ml preformed rhizopuspepsin. In all cases the error bars show the positions of two entirely independent measurements.

" - , , , , , , , . ~ 0 5 10 15 20 2 5 30 35 40

Time (min)

FIG. 11. Activation time course of tRpg at 15 "C, pH 3.5. The curve is generated from a mixed kinetic model involving both k , and k , activation rate constants determined by a non-linear regression to the experimental data.

TABLE V Activation rate constants for tRpg at 15 "C, p H 3.5

Experiments were done at various enzyme concentrations. Acti- vation of porcine pepsinogen (PPgn) was carried out with the same conditions and is shown for a direct comDarison.

k. k, kJk, ~ ~ ~~ ~~~

min" rnin X mglml"

tRpg 0.015 (0.005)" 0.035 (0.006) n = 4 2.3 PPgn 0.07 (0.05) 0.87 (0.06) n = 3 12 Standard errors are shown in parentheses.

2, the observed rate constants kl and k p for Rpg are signifi- cantly less than those for porcine pepsinogen. Also, the ratio of k p to kl is smaller for Rpg than that of porcine pepsinogen. The lower activation constants of tRpg, together with a smaller second-order contribution to activation relative to first-order result in a noticeably slower activation of tRpg at pH 3.5 compared with porcine pepsinogen.

Modeling of Rhizopuspepsinogen-Features of the porcine pepsinogen activation peptide which participate in interac- tions with the mature portion of porcine pepsin were used to create a sequence alignment (Fig. 12). Using the first align- ment of Fig. 12, an activation peptide was built into the

10P

PRPg

PPqn

pRpg

20P

30P 40P

1 10 1 .

"

FIG. 12. Two possible alignments of the putative activation peptide of pRpg with that of porcine pepsinogen (PPgn). Boxed regions indicate positions of identity with porcine pepsinogen; dashes represent positions of insertions or deletions. Numbering is according to porcine pepsinogen. Position 1 indicates the amino terminus of mature porcine pepsin; the position with the "*" above indicates the amino terminus of mature rhizopuspepsin.

structure of mature rhizopuspepsin (5). The activation peptide structure was based upon the porcine pepsinogen crystal structure (13) whose coordinates are available from the Brookhaven Protein Databank, Chemistry Department, Brookhaven National Laboratory, Upton, NY 11973. A struc- tural model of rhizopuspepsinogen then was built using com- puter graphics and molecular dynamics (see "Experimental Procedures" in Miniprint) and is presented in Figs. 13 and 14. A principal feature of conservation in the alignment and the model involves residues Lys-36P and Tyr-37P (pepsin numbering). In porcine pepsinogen, Lys-36P lies medially between the two active-site aspartic acid residues 32 and 215, and the adjacent Tyr-37P occupies the P1'-binding pocket and hydrogen bonds to Asp-215 (Fig. 14). Residues of the modeled activation peptide which directly interact with ac- tive-site elements were constrained in the molecular dynam- ics, as these were expected to be structurally conserved in rhizopuspepsinogen.

The amino-terminal residues 2P-8P (pepsin numbering) of the activation peptide participate in a P-sheet (Fig. 13), as dictated by the porcine pepsinogen-staking template. The residues of this strand interact directly with residues of the pepsin portion and are in the position which is to be occupied by the amino terminus of the mature enzyme. This set of interactions is expected to have implications on the activation process (13).

In the rhizopuspepsinogen model, five ion pairs involving residues of the activation peptide are present, compared with eight for porcine pepsinogen. Of the eight ion pairs involving the activation peptide residues in porcine pepsinogen, five cannot possibly be conserved in rhizopuspepsinogen, due to lack of the acidic or basic side chains in the sequence. The three which it is possible to conserve in rhizopuspepsinogen are indeed conserved in the model; one involves the amino- terminal p-structure, and the other two are at the active site. Two additional ion pairs, excluded from pepsinogen because of sequence differences, occur in the model. They are Asp- 51P to Lys-4OP, and Asp-10 to Lys-35P.

11722 Recombinant Rhizopuspepsinogen

FIG. 13. Stereo view of the struc- tural model of rhizopuspepsinogen. The heauy line is the a-carbon backbone of Rpg activation peptide residues OP- 52P; the thin line is the a-carbon back- bone of Rpg residues 1-328 (pepsin num- bering).

FIG. 14. Stereo view of an overlay of Rpg activation peptide (solid line) with the porcine pepsinogen acti- vation peptide (dotted line) residues 1P-44P. The orientation in this view is different from that in Fig. 13. Active-site residues of the Rpg model (solid line) which may exert an effect upon activa- tion are pictured overlayed with their homologous counterparts in porcine pep- sinogen (dotted line): Lys-36P, Tyr-37P, and Tyr-9, all of which interact with the active-site aspartic acids 32 and 215 (pepsin numbering).

The presence of 3 proline residues between 1OP and 30P hinders the continuity of regular secondary structure in this region where two helices occur in porcine pepsinogen. The residues of those helices in porcine pepsinogen have limited involvement with the mature enzyme, as the consecutive hydrogen bonding of the a-helix implies (13), and are not expected to have significant influence on the activation of pepsinogen.

DISCUSSION

Prior to this work, fungal aspartic proteases, including rhizopuspepsin, have been observed only as mature enzymes and no zymogen has been found during the extraction and purification of these enzymes (14, 15). To investigate the possible zymogen of rhizopuspepsin, we rationalized that ac- tive recombinant zymogen might be synthesized in E. coli for the study of its properties. Expression vectors, which were constructed to include in rhizopuspepsin cDNA a putative proregion, directed the synthesis in E. coli of three structural variants. These putative Rpgs have all the expected properties of an aspartic protease zymogen. They are stable in neutral and mildly alkaline solutions and are spontaneously converted to active proteases upon acidification (Table I11 and Fig. 5). Thus, this fungal aspartic protease must be synthesized in viuo as a zymogen and, possibly, is activated shortly afterward in the secretory vesicles; consequently, no zymogen can be observed from the extracts of the cells. Since several putative proregions have been observed in the genomic structure of other fungal aspartic proteases (16-18) and since the absence of zymogen from the fungal cells is a general phenomenon, rapid intragranular conversion of aspartic protease zymogens to mature proteases may also be a general pattern occurring in fungi. The pH of maximal activation of Rpg is not neces-

sarily the same as that of a secretory granule. Since activation is catalyzed by the Rp active site, the characteristics of that active site will determine the pH dependence of activation. However, the function of the active enzyme may have exerted more influence on the evolution of that active site than has the activation process. I t is sufficient with respect to survival of the organism that activation of the zymogen occurs.

One of the problems in the expression of Rpg was that we did not have a complete cDNA from a single isozyme. The cDNA from the rhizopuspepsin region is that of isozyme pI6 (4) while the putative proregion is from isozyme pI5. The cDNA used to express Rpg was constructed by fusing the cDNA fragments from these two isozymes. The use of a chimeric cDNA is justifiable because the two isozymes, $5 and pI6, differ only at eight positions with rather conservative replacements. The fact that a 3-A resolution crystal structure of Rp was determined from the cocrystallized isozymes testi- fies to the high conformational similarity of the isozymes (19). Finally, the cDNA overlapping region of the two isozyme clones were identical in sequence. All these facts predicted that the two isozymes are extremely similar in structures and properties, and a chimeric zymogen would be representative of the two zymogens.

The second problem was that since Rpg had not been observed previously, the position of the pre/pro junction was not known. Comparing the sequences of the putative prose- quence of Rpg and the corresponding region of other aspartic protease zymogens provides some assistance. Fig. 12 shows that the putative proregion of Rpg is moderately related to the proregion of porcine pepsinogen. This alignment (Fig. 12) emphasizes the overall charge distributions and the position of the Lys-Tyr (residues 36P-37P in pepsinogen), which are known to be important in the crystal structure of pepsinogen

Recombinant Rhizopuspepsinogen 11723

(13, 20). Based on this homology alignment, pRpg produced from expression vector PIN-111-ompA3-Rpg is only a single residue longer at the amino terminus than pepsinogen in the alignment (Fig. 12). However, these sequence alignments are not conclusive. If the amino-terminal position of pRpg is near that of the native zymogen, then, interestingly, cRpg is 11 residues shorter at the amino terminus than is pRpg (Fig. 1). In pepsinogen, residues 2P through 9P take part in p structure and a stable cRpg zymogen structure is difficult to imagine without completion of these elements of secondary structure. Moreover, the successful refolding of synthetic cRpg is per- plexing if elements of the secondary structure are missing.

It should be noted also that in the design of vector PIN-111- ompA3 the expressed pRpg should have 4 additional amino acid residues (Ala-Gly-Ile-His-) at the amino terminus. These 4 residues are not found at the amino terminus of pRpg. It is not clear, a t present, whether this is due to a change of cleavage site by the signal peptidase or whether these 4 residues are removed by other proteases in the periplasmic space of the host bacterium.

To our surprise, vector pGBT-Rpg expressed cRpg out of the coding phase (Fig. 1B). The cRpg obtained was 11 residues shorter at the amino-terminal end than pRpg. This provided an opportunity to compare the activation properties of the two zymogens. The mechanism by which pGBT-Rpg directed the synthesis of cRpg in E. coli is not clear. Most probably, there is an unknown alternative initiation site.

The activation kinetics of pRpg and tRpg showed that they are activated by an intramolecular mechanism similar to that of porcine pepsinogen (12, 21, 22). The first-order activation constant of tRpg is about twice that of pRpg. The faster constant for tRpg may be related to the extra stretch of 16 residues at the amino terminus of this zymogen (Fig. 1D). However, in spite of the difference between the activation constants for pRpg and tRpg, the values of the activation rate constants for the recombinant Rpgs are very similar to that for pepsinogen activation (23). The pH profile for fractional activation revealed that the most rapid activation is a t pH 2 as is the case with pepsinogen. Moreover, bimolecular, rhizo- puspepsin-catalyzed activation becomes important at pH val- ues above 3.5, and the value of the second-order activation rate constant is quite similar to that of pepsinogen. The strong similarity of the activation properties of pRpg and tRpg to those of pepsinogen is uncanny considering the difference in kinetic properties of the two enzymes. (For example, pepsin has a specific activity six times greater than Rp in the milk clotting assay employed in this study.) Consequently, the essential elements of the zymogen structures must be very similar. The results of activation experiments of the Rpgs indicated that all three zymogens are activated in acidic solutions. This indicates that the first 11 residues in pRpg are not essential for the conformation of Rpg and also are not essential for zymogen activation. Since the amino-terminal position of native Rpg is not known, the possibility has been considered whether these 11 residues could belong to the preregion and thus not be important for activity. Although

this possibility cannot be completely excluded, the first align- ment in Fig. 12 places the amino terminus of pRpg near the native zymogen and was used for computer modeling.

The structure resulting from the modeling procedure sug- gests that the alignment of Fig. 12 results in a structure for the activation peptide which is sterically plausible and com- parable to that of porcine pepsinogen. Structural elements of porcine pepsinogen most likely related to activation (amino- terminal p structure, ion pairs, active-site elements) are con- served in the model. Structural features which were not re- tained by the molecular dynamics simulation are those which probably have a limited effect upon the activation process.

Acknowledgments-We wish to thank Dr. Ricky N. S. Wong for contributions during the early phase of this work and Dr. Robert Delaney for making the rhizopuspepsinogen gene sequence available to us prior to its publication.

REFERENCES 1. Oka, T., and Morihara, K. (1973) Arch. Biochem. Biophys. 156,543-551 2. Oka, T., and Morihara, K. (1974) Arch. Biochem. Biophys. 165,65-71

4. Delaney, R., Wong, R. N. S., Meng, G., Wu, N., and Tang, J. (1987) J. Biol. 3. Takahashi, K. (1987) J. Bid. Chem. 262,1468-1478

Chem. 262,1461-1467 5. Suguna, K., Bott, R. R., Padlan, E. A,, Suhramanian, E., Sheriff, S., Cohen,

G. H., and Davies, D. R. (1987) J. Mol. Bid. 196,877-900 6. Sielecki, A. R., Hayakawa, K., Fujinaga, M., Murphy, M. E. P., Fraser, M.,

Muir, A. K., Carilli, C. T., Lewicki, J. A,, Baxter, J. D., and James, M. N. G. (1989) Science 243,1346-1351

7. Miller, M., Jaskolski, M., Rao, J. K. M., Leis, J., and Wlodawer, A. (1989)

8. Wlodawer, A,, Miller, M., Jaskolski, M., Sathyanarayana, B. K., Baldwin, Nature 337,576-579

E., Weber, I. T., Selk, L. M., Clawson, L., Schneider, J., and Kent, S. B. H. (1989) Science 245,616-621

9. Tang, J., and Wong, R. N. S. (1987) J. Cell. Biochem. 33,53-63 10. Heinrikson, R. L., Hui, J., Zurcher-Neely, J., and Poorman, R. A. (1989)

11. Studier, W. F., Rosenherg, A. H., Dunn, J. J., and Dubendorff, J. W. (1990) Am. J . Hypertens. 2 , 367-368

12. AI-Janabi, J., Hartsuck, J. A., and Tang, J. (1972) J. Biol. Chem. 247 , Methods Enzymol. 185,60-89

13. Hartsuck, J. A,, and Remington, J. (1988) The 18th Linderstrom-Lung 4628-4632

14. Otsuru, M., Tang, J., and Delaney, R. (1982) Int. J . Biochem. 14,925-932 Conference, Abstract 28, July 4-8, 1988, Elsinore, Denmark

15. Sodek, J., and Hofrnann, T. (1970) Can. J . Biochem. 4 8 , 1014-1017 16. Gray, G. L., Hayenga, K., Cullen, D., Wilson, L. J., and Norton, S. (1986)

18. Tonouchi, N., Shoun, H., Uozumi, T., and Beppu, T. (1986) Nucleic Acids 17. Esumi, H., Sato, S., and Sugimura, T. (1978) FEES Lett. 86,33-36

Res. 14,7557-7568 19. Suhramanian, E., Swan, I. D. A,, Liu, M., Davies, D. R., Jenkins, J. A,,

Tickle, I. J., and Blundell, T. L. (1977) Proc. Natl. Acad. Sci. U. S. A. 74,556-559

Gene (Amst.) 48,41-53

20. James, M. N. G., and Sielecki, A. R. (1986) Nature 319,33-38 21. Bustin, M., and Conway-Jacobs, A. (1971) J. Biol. Chem. 246,615-620 22. McPhie, P. (1972) J. Biol. Chem. 247,4277-4281 23. Marciniszyn, J., Jr., Huang, J. S., Hartsuck, J. A., and Tang, J. (1976) J.

Biol. Chem. 251 , 7095-7102 24. Lin, X., Wong, R. N. S., and Tang, J. (1989) J. Biol. Chem. 264 , 4482-

4489 25. Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982) Molecular Cloning: A

Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

26. Ausuhel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A., and Struhl, K. (1987) Current Protocols in Molecular Biology, Greene Publishing Associates, Toronto

27. Zoller, M. J., and Smith, M. (1983) Methods Enzymol. 100 , 468-500 28. Kunkel, T. A. (1985) Proc. Natl. Acad. Sci. U. S . A. 8 2 , 488-492 29. Bedouelle, H., and Duplay, P. (1988) Eur. J. Brochem. 171,541-549 30. Laemmli, U. K. (1970) Nature 227,680-685 31. McPhie, P. (1976) Anal. Biochem. 7 3 , 258-261 32. SAS Institute, Inc. (1985) SAS User's Guide: Statistics, SAS Institute, Inc.,

33. Jones, T. A. (1978) J. Appl. Crystallogr. 11 , 268-272 34. Singh, U. C., Weiner, P. K., Caldwell, J. W., and Kollrnan, P. A. (1986)

Cary, North Carolina

AMBER 3. 0, University of California, San Francisco

Continued on next page.

11724 Recombinant Rhizopuspepsinogen

Recombinant Rhizopuspepsinogen 11725

9

0 10 20 30 40 50 60

G-75 Ssphodex

o.800 - 0.600 1

Mono 0

0 600 1 6000

DM-SEPHACEL

Sephacryl 5-300

1.200

1'200 i 1 0.900 0.900 :

0

0 0.600 0.600 ; 10

4 E B 0.300 0.300 ;

0.000 9

I - 0.000

15 25 35 45 55

Rapid Dilution

OEAE

2 2 b 2 4 . 2 100

13 8.G 35

c-7s 2 2 3 . 8 1 6

Mono 9 1.6 1.5 I