PROTEINS: Structure, Function, and Genetics 7:280-290 (1990)

The Characterization of Recombinant Mouse Glandular Kallikreins From E . coli Michael Blaber,’ Paul J. Isackson,’,’ and Ralph A. Bradshaw’ Departments of ‘Biological Chemistry and ‘Anatomy and Neurobiology, College of Medicine, University of California, Irvine, California 9271 7 A system has been developed ABSTRACT for the expression in E . coli of 12 of the 14 expressed mouse submandibular gland kallikreins as cassettes subcloned directly from cDNA. Using the epidermal growth factor binding protein (mGK-9)and the y-subunit of nerve growth factor (mGK-3), as test cases, mature processed forms, obtained as functionally active proteins, as well as various precursor forms, were isolated. The expression system described allows rapid isolation of kallikrein protein from corresponding cDNA with yields of approximately 1.0 mg of purified protein from 10 g of initial cell paste. This expression system will facilitate structurelfunction studies of the mouse glandular kallikrein gene family and help elucidate the regions of the mature proteins responsible for the diverse catalytic behavior and growth factor interactions observed in this family of proteins. Key words: proteases, protein turnover, posttranslational processing INTRODUCTION The mouse glandular kallikrein gene family is a multigene family of closely related serine type proteases with differing substrate specificities and functi~nalities.l-~ Members of the mouse glandular kallikreins have been implicated in the processing of the precursor forms of various bioactive peptides including epidermal growth factor (EGF) and nerve growth factor (NGF).7-10 The diverse activities of the mouse glandular kallikreins include the ability to proteolytically process bioactive pep tide^,^,^ process and subsequently form a stable complex with bioactive peptides,8.10,11and in at least one case (aNGF; mGK-4), form a stable complex with P-NGF (even though it has lost all catalytic ability4). Only a few of the 14 putatively expressed mouse kallikreins’ have been characterized with regard to substrate specificity. The bonds known to be cleaved are diverse and occur after a variety of residues including lysine, arginine, leucine, and histidine. 5,6,12-14 However, despite these differences, the mouse kallikrein family shares a high degree of amino acid sequence identity (73-83%),l thus rendering this family an ideal system for elucidating 0 1990 WILEY-LISS, INC

structureifunction relationships for serine type pr0tea~es.l~ In this report we describe an E. coli expression system for the mouse glandular kallikrein gene family. These vectors take advantage of conserved restriction endonuclease sites among the various glandular kallikrein genes, allowing virtually any member of this family isolated via cDNA techniques to be inserted directly as a cassette into an efficient expression vector. This expression system is evaluated using two mouse kallikrein genes, y-NGF (mGK-3) and EGF-BP (mGK-9).

MATERIAL AND METHODS Unless otherwise indicated all chemicals and reagents were purchased from Sigma Chemical Company. Restriction endonucleases were purchased from New England Biolabs. Vector construction. The expression vectors utilized are shown in Figure 1 and were constructed from fragments of pBR322,16 pUC19,I7 M13mp18,l” pPL-lambda,” PING1 (Ingene, Inc.), and mGK-9 cDNA.~ Vector MB4-29 differs from MB5-36 by having an additional EcoRI endonuclease restriction site in the region of the lac promoter, which does not direct the transcription of any gene. A cDNA-derived mouse kallikrein gene can be inserted as an EcoRIiPstI fragment 3‘ to the Salmonella typhimur i u m A r a B genezo(derived from the PING1 plasmid) and expressed as a fusion protein with a 151 amino acid fragment of the A r a B gene. An arginine residue has been introduced between the AraB sequence and the mature kallikrein sequence to facilitate separation of the fusion protein from the kallikrein by lim-

Received November 27, 1989; accepted December 27, 1989. Address reprint requests to Dr. Ralph A. Bradshaw, Department of Biological Chemistry, College of Medicine, University of California, Irvine CA 92717. Abbreviations used: NGF, nerve growth factor; EGF, epidermal growth factor; EGF-BP, epidermal growth factor binding protein; a-NGF, NGF alpha subunit; 6-NGF, NGF beta subunit; y-NGF, NGF gamma subunit, r-, recombinant; Tes, 2- tL2-hydroxy-l,1-bis(hydroxymethyl)-ethy1 1amino)ethanesulfonic acid; Ris-tris, 2-bis[2-hydroxyethyll-amino-2-~hydroxymethyll-1,3-propanediol); Tween 80, polyoxyethylenesorbitan monooleate; STI, soybean trypsin inhibitor, PAGE, polyacrylamide gel electrophoresis; NAGE, native acrylamide gel electrophoresis, I.-BAPNA, L-benzoyl arginine paranitroanilide.

EXPRESSION OF MOUSE KALLIKREINS IN E. COLI

Fig. 1. Mouse glandular kallikrein/araB (from Salmonella typhimunurn) fusion protein expression vectors. The mouse kallikrein gene is inserted as a cassette with a 5' EcoRl site and 3' Pstl site. The araB promoter is regulated by the bifunctional araC gene product, which is carried on the plasmid, and induced by the addition of L-arabinose. An arginine residue preceeds the mature kallikrein protein sequence to faciliate removal of the fusion peptide by bovine pancreatic trypsin. The transcription terminator T, , derived from lambda phage, is 3' to the kallikreiniaraB fusion. Also included in the vector is an ampicillin resistance gene. A lac promoter is also present but does not direct the transcription of any gene (see text). Fector MB4-29 differs from MB5-36 by an additional EcoRl restriction endonuclease site in the region of the lac promoter.

ited proteolysis by trypsin. The EcoRI site is located 10 amino acids into the mature sequence of the mouse kallikrein gene. The fragment for this region was derived from mGK-9 cDNA3 and is identical in mGKs 1, 3, 9, and 16.' The 3' insertion site for the mouse kallikrein gene is a PstI site which may be generated during construction of a cDNA library from the commonly utilized step of dC-tailing cDNA prior to annealing with dG-tailed PstI restricted pBR322,'l Thus any mouse glandular kallikrein present in a cDNA library constructed using the dCtailing methodology may be directly inserted into this expression vector by restriction with EcoRI and PstI endonucleases. Expression of the AraBi kallikrein fusion protein is under control of the AraB promoter, which is regulated by the bifunctional regulatory AraC gene product, also carried on the plasmid, and induced by addition of L-arabinose. The inculsion of the T,, transcription terminator (derived from pPL-lambda) allows for a stable, wellregulated expression system. To test the utility of the vector and the functionality of glandular kallikreins expressed in E. coli, cDNA clones of the mouse kallikreins y N G F (mGK3)'" and EGF-BP (mGK-9I3 were restricted as described above and the EcoRIIPstI kallikrein fragment inserted into the vector. Growth of cells. DH5a cells (Bethesda Research Laboratories) were utilized as the host cells for vector transformation and were plated on LBiampicillin

281

(100 mg/liter) agar plates.21A 10 m l LBiampicillin (100 mgiliter) culture of the transformed E . coli was grown overnight a t 37°C and used to innoculate a 1.0 1 LBiampicillin (100 mgiliter) culture. The culture was grown for 3 hours a t 37°C followed by induction of expression from the AraB promoter induced by the addition of 20 ml of 20% L-arabinose. Growth was allowed to continue overnight and the cells were harvested by centrifugation for 5 minutes a t 5000 g in a Sorvall GSA rotor, a t 4°C. Approximately 2.5 g cell paste per 1.0 liter culture was produced. The cell pellet was stored frozed at -20°C prior to use. Initial solubilizationicleavage of the fusion peptide. The cell paste from four 1 liter fermentation cultures was thawed a t room temperature and combined. The cell paste was suspended in 200 ml of 50 mM Tris, 1 mM EDTA, 1.0 mg/ml lysozyme by homogenization for 5 minutes. The sample was then divided into two 100 ml aliquots and sonicated for 3 minutes using a Branson model 200 Sonifier microtip cell disruptor a t 50% duty cycle, setting #3. The samples were combined and centrifuged for 20 minutes a t 10,OOOg in a Sorvall GSA rotor a t 4°C Western blot analysisz3 of the supernatant and pellet indicated that virtually all of the kallikrein is present as an insoluble disulfide-cross-linked aggregate (Fig. 2). Consequently the supernatant was discarded and the pellet utilized for further purification. The pellet was solubilized by addition of 60 ml of 7.0 M unbuffered guanidine hydrochloride followed by stirring overnight at 4°C. The sample was then diluted into 1 liter of 1.0 M guanidine hydrochloride, 50 mM Tris, 5 mM EDTA, 0.005% Tween 80, 1.25 mM reduced glutathione, 0.25 mM oxidized glutathione, and allowed to incubate with stirring for 24 hours at 4°C. The sample was then dialyzed, using Spectropore #3 dialysis tubing (Spectrum Medical Industries), verses two 20 x volumes of 50 mM Tris, 0.005% Tween 80, pH 8.5, for 16 hours with each buffer change, at 4°C. The sample was then centrifuged for 30 minutes, 10,OOOg in a Sorvall GSA rotor a t 4°C. The supernatant was pooled and the temperature raised to 37°C. Bovine pancreatic trypsin (10 pl of a 40 mgiml solution) was added to the sample to cleave the AraB fusion protein from the mature kallikrein. The cleavage of the fusion protein was allowed to proceed for 1 hour and resulted in the appearance, to near asymptotic levels, of kallikrein enzymatic activity as judged by hydrolysis of the kallikrein substrate D-Val-Leu-Arg-p-nitroanilide (Helena Labs). The sample was then adjusted to pH 7.5 by the addition of 0.5 M HC1 and the kallikrein protein was subjected to further purification utilizing column chromatography. Purification of recombinant kallikreins. The following steps were carried out at 4"C, and the kallikrein activity was followed using the substrate D-

282

M. BLABER ET AL

Fig. 2. SDS-PAGE and Western blot analyses of extracts from E. colicarrying the MB4-29(mGK-9) expression vector. Lanes 1-3

are whole cell, cell lysis supernatant,and cell lysis pellet, respectively, from L-arabinose induced cultures. Lanes 4-6 are corresponding samples from noninduced cultures. All samples have been reduced with 1 % 2-mercaptoethanol. Lane 7 contains mo-

Val-Leu-Arg-p-nitroanilide. After cleavage of the fusion protein the sample was applied to a DEAE cellulose column (2.5 cm x 15.0 cm) equilibrated to 50 mM Tris, 0.005% Tween 80, pH 7.5. After the sample was loaded the column was washed with the equilibration buffer. The kallikrein activity flowed through the DEAE cellulose column under these conditions. The pooled sample was adjusted to pH 5.0 with 0.5 M acetic acid and the sample was loaded onto a CM cellulose column (2.5 x 7.0 cm) equilibrated to 50 mM sodium acetate, 0.005% Tween 80, pH 5.0. Under these conditions the kallikrein activity bound to the column matrix. The column was eluted with a linear gradient of 0 to 0.4 M NaCl in the equilibration buffer over a total of 10 column volumes, and fractions which contained kallikrein activity were pooled. The NaCl concentration of the sample was adjusted to 0.15 M (either by dilution with the equilibration buffer or by addition of 5.0 M NaCl) and adjusted to pH 7.5 by the addition of 1.0 M Tris pH 9.5. The sample was then passed over a (Bio-Rad Corpora1.0 ml column of STI-Affigel EZ4 tion) to remove any residual trypsin. STI efficiently removes any trypsin in the sample but EGF-BP and y-NGF do not bind significantly to this inhibitor. The sample was concentrated using an Amicon pres-

lecular mass standards (values reported in kDa). The arrow next to lane 1 identifies the expressed araB/kallikrein fusion protein. Lanes 8 and 9 are a Western blot analysis of whole cell extracts from induced cultures. The sample in lane 8 was not reduced with 2-mercaptoethanol whereas the sample in lane 9 was reduced prior to analysis.

sure cell with a YMlO membrane to a volume of 2.0 ml and applied to a Sephadex G-75 column (1.5 x 50 cm) equilibrated to 50 mM sodium acetate, 0.1 M NaC1, 0.005% Tween 80, pH 5.5. Fractions of 2.0 ml were collected and the peak of kallikrein activity was pooled. The sample was concentrated to a n absorbance a t 280 nm of 2.0 using a n Amicon YMlO Centricon concentrator with a Sorvall SS34 rotor a t 5,OOOg. This final sample was then aliquoted and kept frozen a t -70°C. The yield of kallikrein for several purifications varied between 0.50 and 1.2 mg per 10 g cell paste. A m i n o acid composition and sequence analyses. Amino acid composition was determined using the Picotag (phenylthiocarbamyl derivative) method (Waters Gorp.) and a Hewlett Packard model 1090 liquid chromatograph. Duplicate samples were hydrolyzed for both 20 and 96 hours. Values for threonine and serine were determined from the 20 hour hydrolyses, values for isoleucine and valine were determined from the 96 hour hydrolyses, and all other amino acid values were an average of both time points. Half-cystine, methionine, tyrosine, and tryptophan were excessively degraded and were not included in the calculations. Amino acid sequences were determined using au-

EXPRESSION OF MOUSE KALLIKREINS IN E . COLI

tomated Edman degradation chemistry on an Applied biosystems model 470A protein sequenator. Ten cycles were determined for each kallikrein. SDS PAGE analysis. Either 5.0 or 10.0 pg of recombinant kallikrein was analyzed using 20% PAGE.z5 The subchain composition of the samples was determined by addition to the samples of 1% 2-mercaptoethanol prior to analysis. Kinetic analysis. The kinetic constants of the recombinant kallikreins were determined for the Nacetylated tripeptide nitroanilide substrates AlaThr-Arg and Glu-Leu-Arg. Kinetic constants for these substrates with the natural kallikreins EGFBP and y-NGF, as well as the methodology for their determination, has been previously reported6 except that a nonlinear curve fit to the Michaelis-Menten equation was utilized instead of a linear fit to the direct linear plot. Growth factor complexation analysis. The ability of the recombinant kallikreins to complex with the associated gromth factors of the natural kallikreins was determined using a NAGE systemz6 for the NGF complex and gel exclusion chromatography for the EGF complex. Formation of the 7 S NGF complex was evaluated following addition of 20 pg of either natural or recombinant y-NGF, 20 pg natural a-NGF, and 20 pg natural p-NGF in 0.05 M TES, pH 6.8. The mixture was then resolved on a 10% BisTris NAGE system. The gel was visualized by staining with 0.25% Coomassie brilliant blue. Formation of the HMW EGF complex was evaluated following incubation of 16 kg recombinant kallikrein and 2 kg natural EGF (Gibco Corp.) in a final volume of 20 pl of 20 mM Tes, 0.1 M NaCl, pH 7.2. The sample was incubated overnight at 37°C and chromatographed on a Sephadex G-75 column (2 mm x 80 cm). The fractions were blotted on nitrocellulose and developed using a Western blot methodz3 with anti-EGF antisera. Processing of recombinant kallikreins to mature forms. Samples of the recombinant kallikreins, approximately 1.0 mgiml, were incubated for 6 hours a t 37°C in 20 mM Tris, 0.1 M NaC1, pH 7.5. The samples were flash frozen and stored a t -70°C prior to kinetic and SDS-PAGE analyses.

RESULTS Expression of Recombinant Kallikreins SDS-PAGE analysis of crude cell extracts indicates that the AraB/kallikrein fusion protein is expressed as the major cellular protein in the transformed DH5-a E . coli host (Fig. 2). Microscopic examination of the induced host indicated the presence of refractile bodies within the cells and Western blot analysis confirmed the expressed fusion protein existed in a variety of disulfide-linked aggregates (Fig. 2). The level of expression of the fusion protein from SDS-PAGE analysis is estimated a t 10% of the total cell protein. This corresponds to

283

approximately 100 mg of fusion protein or 62 mg of kallikrein in 10 g of cell paste. The lac promoter in the expression vector arose from a vestige of an intermediate plasmid during the construction of the final vectors. It was noted that the levels of plasmid DNA for the vectors in Figure 1 were approximately 5-fold higher than the pUC family of vectors. The presence of the lac promoter near the origin of replication may be responsible for this phenomenon. While vector MB5-36 readily allows insertion of an EcoRIiPstI DNA fragment, vector MB4-36, with two EcoRI restriction endonuclease sites, requires a more tediously obtained partially EcoRI digested vector. However, it was observed that vector MB429 gave consistently higher yields of expressed protein with certain glandular kallikrein genes. The reason for this behavior is not clear.

Purification of Recombinant Kallikreins A summary of the purification steps for recombinant kallikreins (EGF-BP (mGK-9) and y-NGF (mGK-3) is presented in Table 1. Prior to solubilization in the guanidineiglutathione buffer, no kallikrein activity is present with or without trypsin cleavage of the AraB fusion protein. The yield of activatable kallikrein following solubilization is estimated to be between 2 and 3% of the total kallikrein protein in the sample. Thus, the major loss in the purification is the ability to solubilize the AraB/ kallikrein fusion product, the yields for each of the remaining purification steps being substantially higher. The two kallikreins evaluated (mGK-3 and mGK-9) behave identically in the purification scheme except for the CM cellulose chromatography step; r-EGF-BP elutes with approximately 0.225 M NaCl and r-y-NGF elutes with 0.125 M NaC1. Under the conditions utilized approximately 75% of the trypsin activity, added to cleave the kallikrein from the fusion protein, and monitored using the substrate L-BAPNA, binds to the initial DEAE cellulose column. The remaining trypsin in the DEAE cellulose flow-through binds to the CM cellulose resin and elutes at approximately 0.25 M NaCl, well separated from r-y-NGF but close to r-EGF-BP. The subsequent STI-Affigel column efficiently removes the remaining trypsin. In a test of the binding affinity of the STI Affigel resin for trypsin, 400 pg of trypsin was loaded directly onto the STI-Affigel column. No activity was detected in the column flowthrough. The assay conditions utilized in this evaluation would have detected greater than 0.4 pg of trypsin. Thus the STI Affigel bound at least 99.9% of the trypsin loaded onto the column. Furthermore, in a control purification, with no functional kallikrein expressed, no trypsin activity was demonstrable in the final sample. In this control the pooled fractions from the CM cellulose column covered 0.125 to 0.25 M NaCl.

284

M. BLABER ET AL.

TABLE 1. The Purification of Recombinant Kallikreins EGF-BP (mGK-9) and y N G F (mGK-3) From 4.01 Cultures* Total Sample Whole cell lysate Cell pellet Pre-fusion cleavage Post-fusion cleavage DEAE flow-through CM elution STI-Affgel flow-through YM 10 concentration G-75 pool YM 10 concentration

Vol (ml) 200 NIA 1000 1000 1000 50 50 2.0 15 0.5

A280

A280

12.5 NIA 0.86 0.86 0.08 0.11 0.11 1.6 0.11 2.6

2500 NIA 860 860 80 5.7 5.3 3.3 1.67 1.28

Kallikrein (mg) (62) (61) (1.50) 1.31 1.3 1.1 0.92 0.90 0.76 0.75

Yield (%) -

98 2.4 90 95 85 85 98 85 98

*Prior to solubilization and subsequent cleavage of the fusion protein by trypsin, no kallikrein activity is present; thus the values for kallikrein prior to this step are presented within parentheses as estimates from SDS-PAGE of kallikrein protein. The total activatable kallikrein in the sample after guanidineiglutathione solubilization is given in parentheses. The amount of kallikrein reported for the other steps of the purification is based on the amino acid composition data for the final sample and the proportional enzymatic activities of the various stages of the purification towards the substrate o-Val-Leu-Arg-p-nitroanilide. The yields follow the enzymatic activity of the samples.

TABLE 11. Amino Acid Compositions of Recombinant EGF-BP (mGK-9) and y-NGF (mGK-3)* Amino acid Lysine Histidine Arginine Aspartic acid Threonine Serine Glutamic acid Proline Glycine Alanine Valine 112-Cystine Methionine Isoleucine Leucine Phenylalanine Tyrosine Tryptophan Total

r-EGF-BP 19.9 7.3 5.8 26.0 14.2 13.8 21.0 14.2 22.0 12.8 13.8 NID NID 9.9 22.2 6.0 NID NID

r-v-NGF 16.1 5.4 7.1 25.5 16.4 12.0 18.5 14.2 19.9 15.2 12.3 NID NID 11.6 23.7 7.1 NID NID

(10) (7)

-

"Experimental and theoretical yields for mature nonprocessed forms of the kallikreins are presented. Values for l/a-cystine, tryptophan, methionine, and tyrosine were not determined (NiD). Residue values derived from primary sequence data are given in parentheses.

Characterization of the Recombinant Kallikreins The amino acid compositions of r-EGF-BP and ry-NGF are shown in Table 11. The values for r-EGFBP are in excellent agreement with theory suggesting high purity, while the values for r-y-NGF suggest detectable levels of other protein. Amino acid sequence analyses for r-EGF-BP and r-y-NGF, shown in Table 111, indicate that initially, isolated recombinnnt kallikreins are only partially cleaved a t internal sites in comparison to the natural mature form^^,"^,^^ (Fig. 3 ) . Sequence analysis of the recombinant kallikreins after an 18 hour incubation at 37"C, pH 7.5, indicates that the levels of the internal sequences His-Pro-Glu-Tyr- and Phe-Gln-

Asn-Ala- for EGF-BP, and Phe-Leu-Glu-Tyr- and Phe-Gln-Phe-Thr- for y-NGF, respectively, become equimolar with the amino terminal sequence IleVal-Gly-Gly (data not shown). During this additional limited proteolytic processing to the mature forms, a tetrapeptide may be excised in both kallikreins (Fig. 3).3,2"However, from the sequence analysis of the initial and the 18 hour autodigests, it may be deduced that a t least one-third of the EGFBP molecules have this tetrapeptide excised by the longer treatment. SDS-PAGE analyses of the final r-EGF-BP and r-y-NGF samples, shown in Figure 4, are consistent with the amino acid composition and amino acid sequence data. The r-EGF-BP sample is of a

285

EXPRESSION OF MOUSE KALLIKREINS IN E. COLI

TABLE 111. Amino Acid Sequence Analysis of r-EGF-BP and r-y-NGF as Initially Isolated From E. coZi.* r-EGF-BP cycle Amino acid Amino acid Amino acid Amino acid r-y-NGF cycle Amino acid Amino acid Amino acid

1 Ile 377 His 22 Asn 189 Phe 114

2 Val 369 Pro 71 His N/D Gln 69

Ile 526 Phe N/D Phe NID

Val 699 Leu 797 Gln 41

3 Gly 319 Glu 64 Ile

5 Phe 27 1 Asp 54 His 26 Lys 72

6 Lys 252 Tyr 50 Pro 67 Asp 80

7 Cys NID Ser 16 Glu 68 Leu 80

8 Glu 206 Asn 30 Tyr

Asn 75

4 Gly 241 Tyr 44 Arg 46 Ala 65

Gly 555 Glu 474 Phe 105

Gly 488 Tyr 41 1 Thr 71

Phe 433 Asp NID Asp NID

Lys 550 Tyr 40 1 Asp 167

NID Ser 447 Leu 119

Cys

138

Gln 44

9 Lys 239 Asp NID Asp NID Cys NID

Asn 144 Leu 50 Tyr 82 Val 51

Glu NID Asn NID Tyr NID

Lys 475 Asp 268 Cys NID

Asn 116 Leu 332 Val 84

61

10

*Approximately 750 pmol of r-EGF-BP and 1 nmolf of r-y-NGF were analyzed. Values (pmol) at each cycle are given beneath each residue.

higher purity than the r-y-NGF sample, and the initially isolated recombinant kallikreins are only partially processed. The mature natural kallikreins in both cases consist of three disulfide-linked polypeptide chains of approximately 9.1 kDa, 5.8 kDa, and 10.7 kDa (Fig. 3). The reduced forms of the recombinant kallikreins display significant levels of a 17 kDa band and virtually no 6 kDa band, indicating their partially processed states. A time course of autodigestion for r-EGF-BP indicates that after approximately 6 hours the majority of the molecules are processed to the mature three chain form (Fig. 5). A similar incubation of r-y-NGF indicates that it is processed to the mature three-chain form in an analogous manner (data not shown). The conversion to the mature three chain form coincides with the disappearance of the 17 kDa band and the appearance of a 6 kDa band and increased intensity in the region of the 10 kDa band. Additionally, conversion to the mature three chain form coincides with slight retardation of migration in SDS-PAGE and yields a migration value similar to the natural mature three chain form. Kinetic analyses of both the partially processed forms of the recombinant kallikreins and the fully processed mature forms (after 6 hours a t 37"C, pH 7.5) are summarized in Table IV. Also included in this table for purposes of comparison are kinetic constants for natural EGF-BP and y-NGF for these substrates determined previously.6 The data indicate that the recombinant kallikreins are kinetically less active than their natural counterparts. Also, differentially processed forms of the recombinant kallikreins do not display significantly altered kinetic behavior for the substrates tested. Growth F a c t o r Complexation Analysis The reassociation of the 7s-NGF complex with recombinant and natural y-NGF, as demonstrated by NAGE, shows that r-y-NGF is able to participate

in formation of the high molecular weight NGF complex (Fig. 6). Additionally, in this electrophoretic system r-y-NGF resolves as a doublet with a retardation in migration distance in comparison with the natural form. This is a n indication of the existence of differentially processed forms in the recombinant sample. The formation of the high molecular EGF complex with recombinant and natural EGF-BP, as demonstrated by chromatography on Sephadex G-75, is also shown in Figure 6. The data show that r-EGFBP is able to substitute for the naturally isolated protein in formation of the HMW EGF complex. However, the EGF complex with r-EGF-BP does not appear to be as stable as with natural EGF-BP, as indicated by the presence of a greater amount of noncomplexed EGF in the recombinant sample (Fig. 6b, lane 4). DISCUSSION A generic expression system for the mouse glandular kallikreins would allow the facile production of this varied family and provide the opportunity t o probe structure and function relationships. Although the complete nucleotide sequences for the entire set of mouse kallikreins has yet to be reported, restriction digest analysis of genomic DNA suggests that a 5' EcoRI site is a unique site within the region of the mature protein and is shared by most of the kallikreins.' From the complete nucleotide sequences reported for mGKs-1, 3 , 4,9, 13, 16, and 22, no PstI restriction endonuclease sites exist internal to the region of the mature p r ~ t e i n . ~ ~ ~ " , " ~ 3 3 Thus, a 3' PstI site generated via cDNA construction is also a unique restriction endonuclease site shared by most, if not all of the glandular kallikeins. This has allowed the production of expression vectors that treat the various kallikrein genes as cassettes with 5' EcoRI and 3' PstI restriction endonuclease sites. A potential drawback of this approach is

M. BLABER ET AL.

286

r -EGF-BP 37X

I VGG.

25t.

142:

. . . . .SLHRNHIRHPEY. . . . . . TPFKFONA . . .

NH 2

COOH -9

I

KDa --3+-6

KDa-f-10

r

-+

-8-NGF

I OOX

1

7 KDa

I VGG, . . . . . SLMRKH IRFLEY .

I6X

. . . . . TPTKFOFT . . .

COOH

NH 2 -9

1

KDa ----++-6

KDo -3-

10 7 KDO+

Fig. 3. A schematic representation of the extent of internal processing of recombinant kallikreins EGF-BP (mGK-9) and y-NGF (mGK-3) as initially isolated from E. coli. The sites of internal processing indicated by vertical arrows, of the recombinant kallikreins correspond to regions of the natural protein which are subject to posttranslational hydrolysis.

the limited heterogeneity which exists in the 10 amino acid region 5' to the sequence coded in the cassette.' It remains to be seen if variations in this region affect functionality. The levels of expression of the AraB/kallikrein fusion protein constitute a major protein product of the host (Fig. 2). The main inefficiency in the purification of the recombinant kallikreins is the solubilization step (Table I). A point or concern regarding the insolubility of the fusion protein is that the AraB fragment utilized contains an odd number (three) of half-cystine residues." This allows interchain disulfide bonds, and subsequent aggregation, to occur during isolation. Removal of one of these residues may decrease the amount of aggregation. Also, conditions for growth of the host have not been optimized, and the yield of cells could probably be substantially improved. The overall yield of the recombinant kallikreins is approximately 1.0 mg of protein from 10 g of initial cell paste. It was observed that the yield of mGK-9 was consistently about 50% higher than that of mGK-3. The reason for this not clear, but mGK-3 does appear to be more susceptible to internal processing than mGK-9 (Fig. 3). If processing occurs prior to the solubilization step by E. coli proteases, these peptide chains may separate due to the reduc-

ing effects of the glutathione in the solubilization buffer. The isolation of intermediate proteolyzed forms of the glandular kallikreins suggests that some adjustment in the activation step by trypsin, or perhaps introduction of a more specific cleavage site between the fusion peptide/kallikrein, may allow isolation of nonproteolyzed single chain kallikreins. The ability of the recombinant proteins to be processed to the mature three chain form after incubation for several hours a t 37°C suggests that the kallikreins may be undergoing autolysis a t these sites. Based on tripeptide substrate specificities both mGK-3 and mGK-9 have been postulated to be able to cleave at least one internal site.6 Alternatively, trace amounts of an E. coli protease or trypsin causing these processing steps cannot be ruled out. The initially isolated form of y-NGF is similar to isolated subforms of the natural p r ~ t e i n . No ~ ~forms . ~ ~ of natural EGF-BP other than the fully processed three chain form have been reported. Thus the intermediate form(s) isolated from this expression system provides the first opportunity to study the effects of processing upon functionality of this enzyme. EGF-BP and y N G F have two general activities; they are active proteases with distinct substrate specificities6 and they have the ability to form complexes with specific growth Thus, the

EXPRESSION OF MOUSE KALLIKREINS IN E . COLI

1 2

3 4 5

Fig 4 SDS-PAGE analysis of recombinant kallikreins EGFBP (mGK-9) and y-NGF (mGK-3) as initially purified from E coh Lanes 1 and 2 are unreduced samples of recombinant EGF-BP and y-NGF, respectively Lanes 4 and 5 are samples of recombinant EGF-BP and y-NGF, respectively, which have been re-

6

287

7 8

duced with 1% 2-mercaptoethanol prior to analysis. Lane 6 and 8 are samples of naturally isolated EGF-BP; the sample in lane 8 has been reduced with 1% 2-mercaptoethanol, the sample in lane 6 has not. Lanes 3 and 7 contains molecular mass standards with the values reported next to lane 3 in kDa.

functionality of the recombinant proteins was evalnot quite as pronounced, the substrate which gave uated by both kinetic analyses and by the ability to the highest k,,, value with r-y-NGF is Ac-Ala-Thrform a stable growth factor complex. The kinetic Arg-p-nitroanilide (representing the carboxyl termianalyses indicate that when comparing the mature nal sequence of P-NGF). The kinetic analyses also forms of the recombinant kallikreins to the partially indicate that for the substrates tested the recombiprocessed forms there are only minor alterations in nant enzymes are approximately 25-50% less active specificity or catalytic efficiency for the tripeptide than the naturally isolated forms and the K, values substrates tested. The substrates Ac-Glu-Leu-Argfor the different enzymehbstrate combinations are p-nitroanilide and Ac-Ala-Thr-Arg-p-nitroanilide between 1.5 and 4-fold higher for the recombinant represent the carboxyl terminal sequences of the enzymes as compared to their naturally isolated growth factors EGF and p-NGF, r e ~ p e c t i v e l y .A~ ~ , ~ ~counterparts.6 These results suggest the presence of variety of experimental data suggest that the interdenatured forms of the kallikrein andlor the presaction of the growth factors with their associated ence of contaminating proteins which competitively kallikreins involves these carboxyl terminal seinhibit the enzyme. Alternatively, since the natural quences and the substrate binding sites of the kallikreins are g l y ~ o s y l a t e d ,the ~ ' ~lack ~ of carbohykallikreins12,14,38,39as well as specific interactions drate on the recombinant kallikreins may alter the external to these regions.6 Subsequent kinetic charkinetic parameters. Active site titrations with natacterization of the natural kallikreins with tripepurally isolated EGF-BP with both p-nitrophenyltide substrates representing the carboxyl terminal p-guanidinobenzoate and 4-methylumbelliferyl-psequence of EGF and 6-NGF indicated comparaguanidinobenzoate proved unsuccessful due t o the tively higher kcat values for the kallikreinhbstrate efficient hydrolysis of these substrates. Additionally, treatment of natural EGF-BP with endoglycombinations representing the associated growth cosidases D, F, and H did not alter its kinetic pafactor.6 The recombinant kallikreins also display rameters toward Ac-Glu-Leu-Arg-p-nitroanilide this k,,, characteristic (Table IV). The substrate which yields the highest k,,, value for r-EGF-BP is (data not shown). Thus, further purification may be Ac-Glu-Leu-Arg-p-nitroanilide(representing the required to obtain kinetic results identical with the natural counterparts. carboxyl terminal sequence of EGF), and although

M. BLABER ET AL.

288

1 2 3 4 5

6 7 8 9 10 1112

Fig. 5. SDS-PAGE analysis of an autodigestion time course of recombinan! EGF-BP and a comparison with naturally isolated EGF-BP. Lanes 1-5 are unreduced samples 0, 0.5, 2.0, 6.0, and 18 hours incubation at 37"C, pH 7.5. Lanes 7-1 1 are the same time point samples that have been reduced with 1 % 2-mercap-

13 14

toethanol prior to analysis. Lanes 12 and 14 are samples of naturally isolated EGF-BP; the sample in lane 14 was reduced with 1% 2-mercaptoethanol prior to analysis. Lanes 6 and 13 are rnolecular mass standards with the values reported next to lane 6 as kDa.

TABLE IV. Kinetic Constants for the Partially and Fully Processed Forms of r-EGF-BP and r-y-NGF With the Substrate Glu-Leu-Arg-p-Nitroanilide (AcELRpNa)and Ac-Ala-Thr-Arg-p-Nitroanilide (AcATRpNa)* Enzyme Substrate Partially processed kallikreins r-EGF-BP AcELRpNa r-y-NGF AcELRpNa r-EGF-BP AcATRpNa r-y-NGF AcATRpNa Fully processed kallikreins r-EGF-BP AcELRpNa r-y-NGF AcELRpNa EGF-BP AcELRpNa y-NGF AcELRpNa r-EGF-BP AcATRpNa r-y-NGF AcATRpNa EGF-BP AcATRpNa vNGF AcATRDNa

Km (mM)

(sec)

0.442 0.038 1.18 0.271

35.6 10.5 5.70 12.9

0.386 0.083 0.074 0.024

35.2 8.75 51.5 14.3 7.93 8.25 17.8 29.0

1.14

0.308 0.506 0.094

kcat

*Values for nature BP and y N G F a r e from Ref. 6 .

Recombinant y-NGF is able to substitute for natural y-NGF in t h e formation of t h e stable 7 s complex (Fig. 6). Although t h e NAGE 7s-NGF reassociation assay is r u n for approximately 6 hours at

room temperature there is minimal preincubation time of t h e sample. Thus, t h e form of r-y-NGF in the 7s complex may be a two chain precursor form. Various subforms of natural 7-NGF have been

EXPRESSION OF MOUSE KALLIKREINS IN E. COLI

1

2

3

4

5

6

289

shown to form stable NGF complexes including two chain form^.^^-^' The formation of the high molecular EGF complex with natural EGF-BP is a very slow process and involves a preincubation step of at least 4 hours,' preferably overnight (M. Blaber, unpublished observations). In the course of such a lengthy incubation, processing of the r-EGF-BP to the mature three chain form is possible. The formation of the HMW EGF complex with r-EGF-BP is less pronounced than with the natural protein (Fig. 6). This lowered affinity for EGF is most likely a reflection of the higher K , value observed for the Ac-Glu-Leu-Arg-p-nitroanilide substrate. Thus, from both the kinetic analysis and HMW-EGF reassociation analysis the interaction of the carboxyl terminal of EGF within the substrate binding site of r-EGF-BP displays less affinity than the natural kallikrein. The basis for this difference may also be due to the presence of contaminants or the lack of glycosylation.

ACKNOWLEGMENTS The authors are indebted to Drs. John Burnier and James Marsters of Genentech, Inc. for providing the tripeptide substrates, Dr. Harry Haigler of the Department of Physiology, U.C. Irvine for the EGF antiserum, Dr. Jar How Lee of Ingene, Inc. for making available the PING1 plasmid, and Messrs. Steven Disper and Kevin Burke of the ProteiniNucleic Acid Analysis Laboratory of the Department of Biological Chemistry, U.C. Irvine for the amino acid composition and protein sequence analyses. Ms. Kisma Stepanich provided expert assistance in the preparation of this manuscript. This work was supported by U.S.P.H.S. research grant NS19964 and American Cancer Society research grant BC273. REFERENCES

.Fig. 6. Top: Growth factor reassociation assays of the 7 SNGF complex and the HMW EGF complex with recombinant and natural kallikreins. Formation of the 7 S-NGF complex as determined by NAGE analysis. Lane 1 is a mixture of natural y- and p-NGF with recombinant y-NGF; lane 2 is IT-y-NGF;lane 3 is a mixture of natural y-, f3-, and c-NGF. Lanes 4-6 are samples of natural c-, b-, and a-NGF, respectively. The arrow next to lane 1 identifies the 7 S-NGF complex. Bottom: Formation of the HMW EGF complex as demonstrated by chromatography on Sephadex G-75 resin followed by slot blot analysis using anti-EGF antisera. Column fractions # I 6 4 6 were analyzed, representing the range of the excluded and included column volumes. Lane 1 is a native EGF-BP; lane 2 is a sample of EGF; lane 3 is a mixture of EGF with native EGF-BP; lane 4 is a mixture of EGF with recombinant EGF-BP. EGF standards of 1.O, 0.1, or 0.01 bg are shown at the bottom of the blot.

1. Evans, B.A., Drinkwater, C.C., Richards, R.I. Mouse glandular kallikrein genes: Structure and partial sequence analysis of the kallikrein gene locus. J. Biol. Chem. 262: 8027-8034, 1987. 2. Thomas, K.A., Baglan, N.C., Bradshaw, R.A. The amino acid sequence of the y-subunit of mouse submaxillary gland 7 s nerve growth factor. J. Biol. Chem. 256:91569166, 1981. 3. Blaber, M., Isackson, P.J., Bradshaw, R.A. A complete cDNA for the major epidermal growth factor binding protein in the male mouse submandibular gland. Biochemistry 26:6742-6749, 1987. 4. Isackson, P.J., Bradshaw, R.A. The alpha subunit of mouse 7 s nerve growth factor is an inactive serine protease. J. Biol. Chem. 2595380-5383, 1984. 5. Poe, M., Wu, J.K., Florance, J.R., Rodkey, J.A., Bennett, C.D., Hoogsteen, K . Purification and properties of renin and y-renin from the mouse submaxillary gland. J. Biol. Chem. 258:2209-2216, 1983. 6. Blaber, M., Isackson, P.J., Marsters, J.C., Jr., Burnier, J.P., Bradshaw, R.A., Substrate specificites of growth factor associated kallikreins of the mouse submandibular gland. Biochemistry 28:7813-7819, 1989. 7. Woo, J.E., Fahnestock, M., Snow, J., Walz, D., Mobley, W.C. 6-NGF endopeptidase is a mouse submaxillary kallikrein (mGK-22). Abst. SOC. Neurosci. 14:685, 1988. 8. Taylor, J.M., Cohen, S., Mitchell, W.M. Epidermal growth

290

M. BLABER ET AL

factor: High and low molecular weight forms. Proc. Natl. Acad. Sci. U.S.A. 67:164-171, 1970. 9. Frey, P., Forand, R., Maciag, T., Shooter, E.M. The biosynthetic precursor of epidermal growth factor and the mechanism of its processing. Proc. Natl. Acad. Sci. U.S.A. 76: 6294-6298, 1979. 10. Edwards, R.H., Selby, M.J., Garcia, P.D., Rutter, W.J. Processing of the native nerve growth factor precursor to form biologically active nerve growth factor. J . Biol. Chem. 263: 6810-6815, 1988. 11. Varon, S., Nomura, J . , Shooter, E.M. The isolation of the mouse nerve growth factor protein in a high molecular weight form. Biochemistry 6:2202-2209, 1967. 12. Moore, J.B., Jr., Mobley, W.C., Shooter, E.M. Proteolytic modification of the beta nerve growth factor protein. Biochemistry 132333-840, 1974. 13. Tavlor.,~ .J.M.. Mitchell. W.M.. Cohen. S. Characterization of t h e binding protein for epidermal growth factor. J. Biol. Chem. 249:2188-2194, 1974. 14. Greene, L.A., Shooter, E.M., Varon, S. Subunit interactions and enzymatic activity of mouse 75 nerve growth factor. Biochemistry 8:3735-3741, 1969. 15. Blaber, M., Blevins, R.A., Thomas, K.A., Isackson, P.J., Bradshaw, R.A. Epidermal growth factor binding protein: A model serine protease for enzyme engineering. In: “Protein Structure, Folding, and Design,” Vol. 2. Oxender, D. (ed.). New York: Liss, 1987: 131-142. 16. Bolivar, F., Rodriguez, R.L., Greene, P.J., Betlach, M.C., Heyneker, H.L., Boyer, H.W. Construction and characterization of new cloning vehicles. 11. A multipurpose cloning system. Gene 2:95-113, 1975. 17. Yanisch-Perron, C., Vieira, J., Messing, J. Improved M13 phage cloning vectors and host strains: Nucleotide sequences of the Ml3mp18 and pUC19 vectors. Gene 33: 103-119, 1985. 18. Messing, J., Gronenborn, B., Muller-Hill, B., Hofschneider, P.H. Filamentous coli phage M-13 as a cloning vehicle: insertion of a HinDII fraement of the lac reeulatorv region in phage M-13 rep1icati;e form zn uztro. Pqoc. Nacl. AYad. SCI.U S A . 74:3642-3646, 1977. 19. Drahos, D., Szybalski, W. Antitermination and termination functions of the cloned nutL, N and tL, modules of coliphage lambda. Gene 16:261-274, 1981. 20. Lin, H.-C., Lei, S.-P., Wilcox, G. The araBAD operon of Salmonella typhimurium LT2I. Nucleotide sequence of araB and primary structure of its product, ribulokinase. Gene 34:111-122, 1985. 21. Maniatis, T., Fritsch, E.F., Sambrook, J . In: “Molecular Cloning: A Laboratory Manual.” Cold Spring Harbor, NY: Cold spring Harbor Laboratory Press, 1982: 218-219. 22. Ullrich, A., Gray, A,, Wood, W., Hayflick, J., Seeberg, P.H. Isolation of a complementary DNA clone coding for the gamma-subunit of mouse nerve growth factor using a high-stringency selection procedure. DNA 3:387-392, 1984. 23. Burnette, W.N. “Western blotting”: Electrophoretic transfer of proteins from sodium dodecyl sulfate-polyacrylamide gels to unmodified nitrocellulose and radiographic detection with antibody and radioiodinated protein A. Anal. Biochem. 112:195-203, 1981. 24. Bouma, B.N.,Miles, L.A., Beretta, G., Griffin, J.H. Human i--

i

~

25 26.

27. 28.

~

29.

30.

31. 32. 33.

34. 35.

36. 37. 38.

39. 40.

pla3ma prekallikrein. Studicx of its acti\atioii bv activated lactor X11 and Of its iriwrt~vntitinby diisopropyl phosphofluoridate. Riuchemixtry 19:1151-1160, 1980. 1,aemrnli. U.K. C‘leavnuc of structuml Uroteins during the assembly of the head Gf bacteriophage‘ T-4. Nature (YLondon) 227:680-685, 1970. Chrambach, A,, Jovin, T.M., Svendsen, P.J., Rodbard, D. Analytical and preparative polyacrylamide gel electrophoresis. An objectively defined fractionation route, appratus, and procedures. In: “Methods of protein Separation.” Catsimpoolas, N. ed. New York: Plenum Press, 1976: 129. Silverman, R.E. Interactions within the mouse nerve growth factor complex. Ph.D. Thesis, Washington University, St. Louis, MO. 1977. Isackson, P.J., Silverman, R.E., Blaber, M., Server, A.C., Nichols, R.A., Shooter, E.M., Bradshaw, R.A. Epidermal growth factor binding protein: Identification of a different protein. Biochemistry 262082-2085, 1987. Mason, A.J., Evans, B.A., Cox, D.R., Shine, J., Richards, R.I. Structure of mouse kallikrein gene family suggests a role in specific processing of biologically active peptides. Nature (London) 303:300-307, 1983. Isackson, P.J., Ullrich, A,, Bradshaw, R.A. Mouse 75 nerve growth factor: complete sequence of a complementary DNA coding for the alpha-subunit precursor and its relationship to serine proteases. Biochemistry 23:5997-6002, 1984. Lundgren, S., Ronne, H., Rask, L., Peterson, P.A. Sequence of an epidermal growth factor-binding protein. J . Biol. Chem. 2593780-7784, 1984. Drinkwater, C.C., Evans, B.A., Richards, R.I. Sequence and expression of mouse y-renin. J. Biol. Chem. 263:85658568,1988. Drinkwater, C.C., Evans, B.A., Richards, R.I. Mouse glandular kallikrein genes: Identification and characterization of the genes encoding the epidermal growth factor binding proteins. Biochemistry 26:6750-6756, 1987. Burton, L.E., Shooter, E.M. The molecular basis of the heterogeneity ofthe y-subunit of 7 s nerve growth factor. J. Biol. Chem. 256:llOll-11017, 1981. Stach, R.W., Server, A.C., Pignatti, P.-F., Piltch, A,, Shooter, E.M. Characterization of the y-subunit of the 7 s nerve growth factor complex. Biochemistry 15:1455-1461, 1976. Angeletti, R.H., Bradshaw, R.A. Nerve growth factor from the mouse submaxillary gland: amino acid sequence. Proc. Natl. Acad. Sci. U.S.A. 68:2417-2420, 1971. Savage, C.R. Jr., Inagami, T., Cohen, S. The primary structure of epidermal growth factor. J. Biol. Chem. 247:76127621, 1972. Server, A.C., Sutter, A,, Shooter, E.M. Modification of the epidermal growth factor affecting the stability of its high molecular weight complex. J . Biol. Chem. 251:1188-1196, 1976. Bothwell, M.A., Shooter, E.M. Thermodynamics of interaction of the subunits of 7 s nerve growth factor. J. Biol. Chem. 25323458-8464, 1978. Nichols, R.A., Shooter, E.M. Characterization of the differential interaction of the microheterogenous forms of the y-subunit of 7 s nerve growth factor with natural and synthetic ligands. J. Biol. Chem. 258:10296-10303, 1983.