Avian glutamine phosphoribosylpyrophosphate amidotransferase propeptide processing and activity are dependent upon essential cysteine residues

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1992 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 267, No. 11, Issue of April 15,pp. 7936-7942 1992 Printed in ~ . S . A .

Avian Glutamine PhosphoribosylpyrophosphateAmidotransferase Propeptide Processing and Activity Are Dependentupon Essential Cysteine Residues* (Received for publication, December 12, 1991)

Gaochao Zhou,Steven S. Broyles, Jack E. DixonS, and Howard Zalkin From the Department of Biochemistry, Purdue University, West Lafayette, Indiana 47907

Avian glutamine phosphoribosylpyrophosphateamidotransferase contains an NHz-terminal propeptidelike sequence. NH2-terminal sequence analysis of immunoaffinity purified enzyme from chicken liver indicates that the propeptide is processed and the mature enzyme starts with Cys’. Propeptide processing was investigated by site-directed mutagenesis using a system for expression in HeLa cells. Glutamine-dependent activity and processing were abolished by replacement of the conserved cysteine at position 1, whereas NH3dependent activity was retained. Cys’ is thus inferred to have a role in glutamine-dependent activity and in propeptide processing. Inactive, insoluble enzymes in which the propeptide was not processed were obtained as a result of replacements of cysteines 415 and 488. Cysteine residues at positions 416 and 488 are inferred to be ligands to an Fe-S cluster on the basis of sequence similarity to the enzyme from Bacillus subtilis. Mutation of CyszB9and Cysze6led to loss of enzyme activity and propeptide processing, although solubility was unchanged. The results suggest that incorporation of an Fe-Scluster is needed for native structure, resultant propeptide processing, and glutamine-dependent activity.

Glutamine PRPP‘ amidotransferase catalyzes thefirst committed step in the pathway for de novo purine nucleotide synthesis: glutamine PRPP +phosphoribosylamine glutamate + PPi. Similar to other glutamine amidotransferases, NH, can replace glutamine as a substrate. Inprokaryotes and mammals, this enzyme is subject to end product inhibition by purine nucleotides and inhibition is assumed to have a regulatory function (1-4). Genes encoding the amidotransferase have been cloned and sequenced from Escherichia coli ( 5 ) and Bacillus subtilis (6) andthe enzymes characterized (1, 2). The amidotransferases from E. coEi and B. subtilis are 40% iden-

+

+

tical in primary sequence (6) and have similar catalytic and regulatory properties (7). However, the B. subtilis enzyme undergoes two unique posttranslational modifications. (i) An 11 amino acid propeptide is cleaved from the NH2 terminus to generate a mature enzyme having an NH2-terminal active site cysteine residue (6, 8, 9) and (ii) a [4Fe-4S] cluster is assembled into each subunit (10,ll). Therole of the propeptide is not understood, but processing is obligatory for the glutamine-dependent activity (8, 9). The 4 cysteinyl ligands to the Fe-S cluster have been identified (6, 12). One role of the [4Fe-4S] cluster is to serve as a site for the oxidative inactivation of the enzyme that occurs during nutrient starvation preceding sporulation (11).Oxidative inactivation appears to be the rate-limiting step in enzyme turnover (13). Glutamine PRPP amidotransferase has been highly purified from avian sources (14, 15) and partially purified from human placenta (16). These enzymes are very oxygen labile and this lability has precluded all subsequent attempts to isolate purified enzyme. There are two paramount questions about the structure of the amidotransferase from animals. (i) Are the avian andmammalian enzymes synthesized as proenzymes? (ii) What is the structure of the earlier noted iron component inthe avian enzyme (14, 15) and theFe-S cluster in the human enzyme (16, 17)? cDNA encoding chicken glutamine PRPP amidotransferase was recently isolated (18). The primary sequence of the chicken amidotransferase is 40% identical to that of the B. subtilis enzyme, and there is correspondence of the sequence at the NH2 terminus with the B. subtilis amidotransferase propeptide (see Fig. 2). There is, however, no direct evidence that the propeptide-like sequence is actually processed. In this paper we report for the first time the use of an antibody affinity column which can be used to obtain the amidotransferase in a nearly pure form. Using this column we have determined the NH2-terminal amino acid sequence of the enzyme from chicken liver. These data establish that the propeptide-like sequence is processed and themature enzyme starts with Cys’. A system for expression in HeLa cells was used to investigate the roles of several cysteine residues. Overall, the cysteine replacements indicate that propeptide processing is extremely sensitive to structural perturbations.

* This work wassupported by United States Public Health Service Grant GM25658 (to H.Z.), National Institute of Diabetes and Digestive and Kidney Diseases 18849 (to J.E.D.), and NationalInstitute of Allergy and Infectious Diseases 28432 (to S.S.B.). This is journal paper number no. 13290 from the Purdue University Agricultural Experiment Station. The costs of publication of this article were EXPERIMENTALPROCEDURES defrayed in part by the payment of page charges. This article must Plasmids-pTM1, obtained from Bernard Moss, National Institherefore be hereby marked “advertisement” in accordance with 18 tutes of Health, was used as thevector for expression of heterologous U.S.C. Section 1734 solely to indicate this fact. $ Present address: Dept. of Biological Chemistry, University of cDNA in HeLa cells. This plasmid contains aphage T7 promoter and transcription terminator, an encephalomyocarditis virus sequence to Michigan Medical School, Ann Arbor, MI 48109-0606. The abbreviations used are: PRPP, 5-phosphoribosylpyropbos- enhance cap-independent mRNA translation (19), thymidine kinase phate; GAR, phosphoribosylglycinamide; SDS, sodium dodecyl sul- left and right arms, and sequences for propagation in E. coli. E. coli fate; DMEM, Dulbecco’s modified Eagle’smedium; PBS, phosphate- lac2 from pLT61T (20) was cloned by polymerase chain reaction into buffered saline; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesul- pTMl NcoI-BamHI polylinker sites to yield pTMl-lacZ (Fig. 1). Chicken glutamine PRPP amidotransferase cDNA encoding Met-” fonic acid.

’

7936

Avian Glutamine PRPP Amidotransferase

col

BamHl

Y.

\

EcoRl FIG. 1. Plasmids for expression of chicken glutamine PRPP amidotransferase cDNA and E. coli lac2 in HeLa cells. The abbreviations used are: AT, amidotransferase cDNA; Ap, ampicillin resistance; EMC, encephalomyocarditis virus untranslated region; TKI, TKr, left and right segments of thymidine kinase, respectively; promoter and terminator, respectively, for phage T7 polymPt7, Tt7, erase. The black box in pTM-AT is DNA for the amidotransferase propeptide. It has been deleted in pTM2-ATm. Vector-derived sequences are indicated by a thin line. TT7

to Trp4$' proenzyme was cloned into pTMl NcoI-BamHI polylinker sites to yield pTM1-AT. Met' in the amidotransferase cDNA (18) has been renumbered Met-". Amidotransferase cDNA in which codons -10 to -1 were deleted was cloned into pTMl toyield pTM2ATm which encodes Cys' to Trp4* mature enzyme after removal of the initiator Met. For synthesis of chicken glutamine PRPP amidotransferase in E. coli, cDNAs encoding Met" to Trp4= proenzyme and an enzyme containing residues Met" to Trp4%were ligated into vector pT7-7 (21) and expression was in E. coli strain BL21DE3pLysS. Expression in HeLa Cells-HeLa cells grown in DulBecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum to 80% confluence in 25-cm2flasks were infected with recombinant vaccinia virus vTF7-3 (22), containing the phage T7 RNA polymerase gene, at a multiplicity of infection of 30. After incubation a t 37 "C for 30 min, infected cells were washed three times with

7937

DMEM medium prior to transfection by the lipofectin procedure (23). For transfection 50 pg of plasmid DNA and 30 pg of lipofectin were added to each flask containing 3 ml of DMEM medium. After incubation at 37 "C for 7 h, 3ml of additional DMEM plus 20% fetal calf serum were added, and incubation was continued for 20 h. Lysates were made by four freeze-thaw cycles from 8 X lo6 cells in 400 pl of buffer A containing 50 mM HEPES, pH 7.0, 10 mMMgC12, 0.30 M sucrose, and 1 mM phenylmethanesulfonyl fluoride. The lysate was separated into soluble and particulate fractions by centrifugation for 4 min in a microcentrifuge. The particulate fraction was resuspended in 400 pl of the same buffer solution to restore its original volume. For metabolic labeling 100 pCi of [2-3H]glycinewas added to a dish of transfected cells 4 h prior to harvest. Serum used for transfection was dialyzed against phosphate-buffered saline (PBS) todecrease the glycine pool. In addition, all four ribonucleotides and deoxynucleotides were added to minimize the conversion of serine to glycine. Antisera-Three polyclonal antibodies were raised against chicken glutamine PRPP amidotransferase. The antigens were (i) enzyme fragment Ser'L2-Trp499, which was produced as a glutathione S-transferase fusion from pGEX-2T (24). The fusion protein was synthesized in strain JM101, purified by glutathione-agarose affinity chromatography, cleaved with thrombin (24), and theenzyme fragment isolated by electroelution from an SDS-polyacrylamide gel. (ii) propeptide, NH2-MELEELGIREEC was coupled through Cys+' to maleimideactivated bovine serum albumin (Pierce Chemical Co.). (iii) COZHterminal peptide, CLTGDYPVELEW-C02H was coupled to bovine serum albumin through Cys'% using m-maleimidobenzoyl-N-hydroxysuccinamide (25). Peptides were synthesized by the Biochemistry Biotechnology Facility, Indiana University School of Medicine. Antibodies were raised in mature New Zealand White rabbits by injection of 500 pg of protein in Freund's complete adjuvant followed by boosts at 2, 8, and 14 weeks with 250 pg of protein in Freund's incomplete adjuvant. Blood was collected at 10 and 17 days after the third and fourth injections and IgG partially purified by ammonium sulfate precipitation followed by dialysis against PBS. Antibodies raised against enzyme fragment Ser"' to Trp4*, propeptide, and CO,H-terminal peptide are designated AT-Ab, N-Ab, and C-Ab, respectively. Antibody Affinity Column-AntibodyC-Abwas attached covalently to protein A-Sepharose CL-4B (Sigma) by cross-linking with dimethyl suberimidate (25). Protein Blots and Immunoprecipitation-Western blots (26) from SDS-polyacrylamide gels were carried out using goat anti-rabbit IgG coupled with alkaline phosphatase. For protein blots approximately 60 pg of soluble protein from about 6X lo5 cells or particulate protein from 1.2 X lo5cells was used in each lane of the SDS-polyacrylamide gel. Immunoprecipitation of amidotransferase from [2-'H]glycine-labeled HeLa cells was first carried outin a 0.5-ml mixture containing cell lysate (-5 X lo6 cells), NET buffer solution (50 mM Tris-HC1, pH 7.5, 150 mM NaC1,5 mM EDTA, 0.05% Nonidet P-40), and 20 pl of preimmune serum. After incubation for 1 h at 0 "C, nonspecific IgG-antigen complexes were precipitated by addition of 200 p1of Staphylococcus aureus cells containing protein A (Pansorbin, Calbiochem). Chicken glutamine PRPP amidotransferase was then specifically immunoprecipitated by addition of 8 p1 of C-Ab antiserum, incubation, and treatment with S. aureus protein A as above. The resultant pellet was dissolved in 50 p1 of 10 mM Tris-HC1, pH 8.0,l.O mM EDTA and samples of 10 and 40 pl used for SDS-polyacrylamide gel electrophoresis (25) and radiosequencing, respectively. Autoradiography of gels treated with EN3HANCE was at -70 "C for 8 days. The sample for radiosequencing was dialyzed against 70% formic acid. Samples from radiosequencing were dried under nitrogen, taken up in 50 pl of methanol, and counted for radioactivity. Immunoaffinity Purification of Chicken Liver AmidotransferaseFrozen chicken liver was homogenized in 2 ml/g tissue of buffer A. The homogenate was centrifuged at 15,000 X g for 30min. The soluble extract (4 ml) was dialyzed against PBS for 4 b at 4 "C, then passed through a 4-ml Sepharose CL-4B-200 column. The treated extract was incubated overnight a t 4 "Cwith 0.5 ml of packed immunoaffinity C-Ab-Sepharose resin using gentle mixing. The resin was washed twice in a batchmode with PBS and thenpacked into asmall column and washed further with 20 ml of PBS. Amidotransferase was eluted with 1.5 ml of PBS containing 60 p M COzH-terminal peptide by batch mode using gentle agitation. The immunoaffinity-purified chicken amidotransferase was dialyzed against 100 volume of 20% formic acid for 2 days with four changes, and lyophilized. After SDS-polyacrylamide gel electrophoresis, the protein was transferred to a polyviny-

Avian Glutamine PRPP Amidotransferase

7938

lidine difluoride membrane and sequenced by Edman degradation using an Applied Biosystems model 470A gas-phase protein/peptide sequencer in the Purdue Laboratoryfor Macromolecular Structure. Enzyme Assay-Glutamine- and NH3-dependent amidotransferase activities were assayed by coupling product formation to the synthesis of ["CIGAR using [14C]glycineand GAR synthetase (27). Purified E. coli GAR synthetase (28) was a generous giftof J. Stubbe, Massachusetts Institute of Technology, Cambridge, MA. @Galactosidase was assayed according to Miller (29).

(A) 1 I

AT-Ab

2

3

4

M "

RESULTS

Avian AmidotransferasePropeptide-like Sequence Is Not (B) N-Ab Processed in E. coli-The recombinantchickenglutamine 1 2 3 4 P R P P amidotransferase is not functional in E. coli, whereas a n enzyme with a n engineered deletion of the 11 NHz-terminal amino acids restores limited function (18).This finding suggests that cleavage of the avian11amino acid propeptidelike sequence (Fig. 2) is essential for activity and cannot take " p place in E.coli. Protein blots withspecific antisera were used t o characterize the chicken amidotransferase synthesized in E.coli. Recombinant avian enzyme synthesized inE. coli was size fractionated by SDS-polyacrylamide gel electrophoresis and detected by Western blotting using antiserum AT-Ab raised againstenzyme fragment Ser"2-Trp499.The blot inFig. FIG. 3. Western blots of amidotransferase from chicken 3A shows that the recombinant avian amidotransferase (lane 3 ) is larger than the enzyme with an engineered propeptide liver and E. coli. Each lane contained60 pg of soluble protein. The deletion (lane 2 ) . Lane 1 in Fig. 3A shows that this antibody lanes are: I , E. coli transformed with vector; 2, E. coli with pT7-ATm; 3, E. coli with pT7-AT, 4 , chicken liver. Blot A wasprobedwith E. coli proteins. Lower molecular antiserum did not cross-react with AT-Ab andblot B with antiserum N-Ab. P, proenzyme; M, weight chicken amidotransferase fragments from expression mature enzyme. in E. coli (lanes 2 and 3 ) may result fromproteolysis or aberrant translation initiation from internal AUG codons and M 1 can be ignoredfor this analysis. T o prove that this size difference is the result of NH2-terminal processing, a duplicate blot was probedwith antiserum N-Ab raisedagainstthe the propeptide. Fig. 3B shows that this antibody detects larger 200 amidotransferasein lane 3 but not the enzyme in lane 2 97having the NH2-terminaldeletion. The blots inFig. 3, A and 68 B,lanes 2 and 3, thus establish the fidelity of antiserum NAb and provide the basis to distinguish whether or not the propeptide-like sequence is processed from the native avian 43 amidotransferase. 29 Amidotransferase Propeptide Is Processed in Chicken Liver-The Western blot in Fig. 3A, lane 4 , shows that the enzyme fromchicken liver is similar in size to the amidotrans18ferase having an engineered propeptide deletion (lane 2 ) but is smaller than the primary translation product (lane 3 ) .The duplicate blot inFig. 3B, shows that antiserum N-Ab did not react with amidotransferase from chicken liver(lane4 ) . Thus, FIG. 4. Immunopurified chicken liver amidotransferase. Ceither all or part of the 11amino acid propeptide-like sequence Ab immunopurified amidotransferase (3 fig) was electrophoresed on has been removed from the chickenliver amidotransferase. In order to isolate amidotransferase from chicken liver and a 10% SDS-polyacrylamide gel and stained with Coomassie Blue. Lane M,molecular weight markers; lane 1, amidotransferase. The establish the site of propeptide cleavage, a n affinity column arrow points to theenzyme. was prepared using antibody directed against a CO2H-terminal peptide. The column allows for specific adsorption of the amidotransferase fromchickenliver extracts and selective E. coli NI~CIVC---~---H---PUVYGID---QQFEDSVF-elution with a synthetic C02H-terminalpeptide. Under these B.S&n'lir NLAEIKGLEEEI~CVFC---~---L---P~FYCID---~GQ~LA~F-non-denaturing conditions,active enzyme was recovered that Chiclrm H E L E E L C I R E E I ~ C V F C - - - ~ - - - ~ - - - P ~ Y H G I D - - - I G ~ ~ T A ~ Lis- -approximately 90% pure upon analysis by SDS-polyacrylC' c269 (w c!yc*'3 e cc Cd(C48) amide gel electrophoresis (Fig. 4, lane 1). Transfer of the FIG. 2. Aligned regions of glutamine PRPP amidotransfer- purified protein to a polyvinylidine difluoride membrane and ase. Regions containing the NH2-terminal propeptide andcysteinyl subsequent Edman degradation yielded the amino acid seligands to the [4Fe-4S] cluster in the B. subtilissequence (6) are quence X-G-V-F-G.This sequence confirmsthe identification aligned with corresponding regions in the enzymes from Saccharoof the affinity-purified protein as avian glutamine P R P P myces cereutsiae (35), E. coli (5), and chicken (18).Cysteine residues amidotransferase anddefines the NH2 terminus of the mature are highlighted and numbered according to the avian sequence. The enzyme. The results of Edman degradation indicate that the 4 cysteinyl ligands to the Fe-S cluster in B. subtilis amidotransferase shown in Fig. 2 was cleaved are designated by a, b, c, and d. Vertical lines mark the processing primarytranslationproduct site. between Glu a t position 11 and Cys at position 12. Cysteine S. cenvisiae

~ImCILC---D---N---EHIYGID---TKFEDCVF--

Avian GlutaminePRPP Amidotransferase (A)

I

A

AT-Ab Soluble

1

2

3

7939 1

2

Particulate

6

4 " 5

7 ' 8

Cycle

6 1

(B)

2

5

4

5

6

7

8

9

1

0

C-Ab Soluble

1 I

3

2

3

Particulate 4115

6

7 ' 8

" C G V F G C I A A G P M E L E E L G I R E

M

4

(C)

N-Ab Soluble

1

' 2

3

Paniculate 4115

6

7 I8

FIG. 6. Immunoprecipitation and radiosequencing of chicken amidotransferase synthesized in HeLa cells. Top, HeLa cells were infected with arecombinant vaccinia viruslT7 polymerase, transfected with pTM1-AT andlabeled for 4 h with [3H]glycine.The (lane 2 ) lysate (lane 3 ) was firsttreated withpreimmuneserum followed by antiserum C-Ab (lane 1). Autoradiography was for 8 days. Bottom, radiosequencing of protein shown in lane 1 of top panel. Fractions from Edman degradation were counted for radioactivity. A background value of 45 cpm was subtracted from each cycle. The amino acid sequence in one-letter code is shown for the mature ( M ) and pro ( P )enzymes.

TABLE I Activity of chicken glutaminePRPP amidotransferase i n transfected HeLa cells Plasmid

FIG.5. Expression of chicken amidotransferase cDNA in HeLa cells. Western blots of triplicate SDS-polyacrylamide gels were probed with antisera AT-Ab ( A ) , C-Ab ( B ) , and N-Ab ( C ) . HeLa extract soluble fraction (60 pg of protein, 6 X lo5cells) is in lanes 2-4; particulate fraction (1.2 X lo5 cells) in lanes 5-7. The lanes are: I , chicken liver; 2 and7, HeLa/pTMl vector; 3 and 5, HeLa/ pTM1-AT; 4 and 6, HeLa/pTMZ-ATm; 8, fetal calf serum. Arrows indicate the positions of bovine serum albumin ( B S A ) ,proenzyme ( P ) ,and mature enzyme ( M ) .

was not identified in cycle 1because sulfhydryl groups were not alkylated. Accordingly, Met' in the primary translation product is numbered Met-" and residues -11 to -1 correspond to a cleavable propeptide. Fig. 2 shows the similarity of the avian and B. subtilis amidotransferasepropeptide sequences and the conservation at the NH, terminus of the mature enzyme. Expression of Chicken Amidotransferase in HeLa CellsSynthesis of enzymatically active amidotransferase is needed to examine function by mutagenesis. Since the recombinant avian enzyme produced in E. coli is not active (18), we turned to expression of cloned cDNA in HeLa cells. Phage T7 RNA polymerase from a recombinant vaccinia virus was used to transcribe cloned cDNA under the control of a T7 promoter.

Enzyme

Enzyme activitf Glutamine NHI nmollminlmg

1.87 1.08 None pTMl pTM1-AT Wild type 17.5 10.4 A(Met"'-Glu") 17.1 10.7 pTM2-ATm 4.40 0.90 C1F pTM1-AT(C1F) 1.74 1.03 C269S pTM1-AT(C269S) pTM1-AT(C295S) C295S 2.62 2.01 0.90 1.70 C415S pTM1-AT(C415S) pTM1-AT(C488S) C488S 1.40 0.80 299 NDb E. coli pTM2-EcAT Values are theaverage of two to five determinations with deviations of less than 20%. Not determined.

To test thisexpression system in HeLa cells, several control experiments were conducted. Plasmids pTM1-lac2 encoding E. coli P-galactosidase (Fig. 1)and pTM2-EcAT encoding E. coli amidotransferase (not shown) were expressed in HeLa cells to yield approximately 10 and 2%, respectively, of the soluble protein as estimated by enzyme activity (not shown). This system is therefore potentially suitable for synthesis of active chicken glutamine PRPP amidotransferase directed from plasmid pTM1-AT encoding the intactproenzyme, residues Met-" to Trp4", and pTM2-ATm encoding amidotransferase, residues Met' to Trp4$' (Fig. 1). The protein blot in Fig. 5 A shows that the soluble amido-

Avian GlutaminePRPP Amidotransferase

7940

BSA

(A)

C-Ab

(6)

N-Ab

-

MP -

Soluble

' 2

3

Particulate 4'

' 5

6

7l

BSA --C M

P

t

FIG. 7. Synthesis of wild type and C1F mutant amidotransferase i n HeLa cells. DuplicateSDS-polyacrylamide gels were probed with antisera C-Ab ( A ) and N-Ah ( B ) . HeLa cell extracts were fractionated into soluble (lunes 2-4) and particulate (lanes 5-7) preparations. The lunes are: I , chicken liver extract; 2 and 5, wild type; 3 and 6, C1F mutant; 4 and 7, vector control. Arrows indicate and the position of bovine serum albumin ( B S A ) , proenzyme (P), mature enzyme (M). 1

2

3

4

FIG. 8. Protein blots of chicken amidotransferase mutants C269S and C295S. Soluble preparations of HeLa cell extracts were fractionated with SDS-polyacrylamide gel electrophoresis, followed by probing with antisera AT-Ab. The lunes are: I , pTM1-AT (wild type) plus pTM1-ATClF); 2, pTM1-ATC(C269S); 3, pTM1AT(C295S); 4, pTM1-AT(C1F).

tide processing. This enzyme was likewise detected by antiserum C-Ab but not by N-Ab (Fig. 5, B and c , lane 4). AntiseraN-Aband C-Ab reactedwith two proteinsin addition to chicken glutamine P R P P amidotransferase. (i) Antibody C-Ab cross-reacted with the endogenous HeLa cell amidotransferase because of COzH-terminal sequence identity' (Fig. 5B, lane 2 ) . (ii) Both antisera reactedwith traces of serum albumin that originated from the culture medium (Fig. 5, B and C, lanes 2-4). This was a result of raising antibodies to peptide antigens coupled to bovine serum albumin.These two spuriousreactions did not, however, obscure the detection of pro- and mature avian amidotransferase in HeLa cells. NHz-terminal radiosequencing was performedto verify the conclusion that propeptide was processed from the soluble chicken amidotransferase in HeLa cells. Transfected HeLa cells were labeled with radioactive glycine and the amidotransferase isolated by immunoprecipitation with C-Ab antiserum. A single radioactive amidotransferase band was detected by SDS-polyacrylamide gel electrophoresis (Fig. 6A, top, lane 1). The results of radiosequencing are shown in Fig. 6, bottom. Peaks of radioactive glycine were obtained a t cycles 2, 5, and 10, consistent with the NHz-terminal sequence of the mature enzyme. Two distinct forms of chicken amidotransferase, soluble and particulate,were synthesized in HeLacells (Fig. 5 ) . Based on protein blotsderived from known amounts of cell extract, we estimated thatapproximately 10% of the amidotransferase encoded by plasmid pTM1-AT and pTM2-AT was soluble and 90% insoluble. The insoluble avian amidotransferasewas not processed based on size (Fig. 5A, lane 5 ) and reaction with antiserum N-Ab (Fig. 5C, lane 5). Insolubility, however, was not a consequence of the propeptide processing defect since the engineered mature enzyme generated a comparable amount of both soluble and insoluble products (Fig. 5, A and B). The activityof chicken glutamineP R P P amidotransferase in transfectedcells was9-10-fold higher than thebackground activity in HeLa cells (Table I), The ratio of NH3-dependent to glutamine-dependent activity was 1.7, a value identical to the HeLaenzyme. Fractionation of cell extracts demonstrated that all of the amidotransferase activity was in the soluble fraction. The unprocessed particulate enzyme was not active. Virtually identical activities were obtained for enzymes encoded byplasmids pTM1-AT and pTM2-ATm, indicating that propeptideprocessing is not rate-limiting. Effect of Cysteine Replacements on Propeptide Processing and Enzyme Activity-There are 12 cysteines in the avian glutamine P R P P amidotransferase, 4 in the NHp-terminal glutamine amide transfer domain and 8 in the distal synthase domain (18). Of these only Cys' is conserved in all of the glutamine P R P P amidotransferases from bacteria, yeast, and avian and human cells. Cys' is notonly essential for glutamine amide transfer function but been has implicatedin propeptide processing in the B. subtilisenzyme (8, 9). To determine whether Cys' has a comparable role in the chicken enzyme, a C1F mutation was constructed in plasmid pTM1-AT(C1F). The analysisof the C1Fenzyme is shown in Fig. 7 and Table I. Protein blots in Fig. 7, lanes 2 and 3 and 5 and 6, indicate that the C1Fenzyme was distributed in soluble and particulate fractions. The soluble C1F enzyme was not processed (Fig. 7, A and B, lane 3 ) and had undetectable glutaminedependent enzyme activity (Table I). The basal level of glutamine-dependent activity in cells transfected with pTM1-

transferase encoded by plasmids pTM1-AT and pTM2-ATm (lanes 3 and 4, respectively) was identical insize to theenzyme from chicken liver (lane 1). Thus, chicken proamidotransferase was processed in HeLa cells. For confirmation, duplicate protein blots were probed with antisera N-Ab and C-Ab to the NH, and COzH terminii, respectively. Amidotransferase synthesized from pMT-At (Met"'-Trp'gg) was detected by antiserum C-Ab (Fig. 5B, lane 3 ) but not by N-Ab (Fig. 5C, lane 3 ) confirming that the soluble chicken proenzyme was processed in HeLa cells. The engineered mature enzyme from '2. Chen, G. Zhou, A. Gavalas, J. E. Dixon, and H. Zalkin, plasmid pTM2-ATm served as a positive control for propep- unpublished results.

Avian Glutamine PRPP Amidotransferase

7941

the chicken amidotransferase cDNA was expressed in HeLa cells using a vaccinia virus/phage T7 system. Synthesis of active chicken glutamine PRPP amidotransferase differed in two important respects from synthesis of the @-galactosidase control. (i) The level of chicken amidotransferase, although not quantitated, was considerably lower than thatof 8-galactosidase. (ii) Approximately 90% of the wild type avian glutamine PRPP amidotransferase was insoluble and inactive, whereas the 0-galactosidase was entirely soluble. Insolubility was not due to a rate-limitingcapacity for propeptide cleavage since the engineered mature enzyme was comparably insoluble. Insoluble glutamine PRPP amidotransferase was not detected for the native enzyme in chicken liver or in HeLa cells (Fig. 5B, lane 7 and Fig. 7A, lane 7). Insolubility does not resultsimply from high level synthesis of the amidotransferase in HeLa cells. High level synthesis of E. coli amidotransferase was obtained in HeLa cells. The E. coli enzyme, which does not contain an Fe-S cluster, was synthesized at 30-fold higher levels than the avian amidotransferase, and this enzyme was entirely soluble (not shown). We therefore conclude that the rateof chicken amidotransferase synthesis exceeded the capacity of transfected HeLa cells for incorporation of an iron component and correct folding and oligomerization. Expression of chicken amidotransferase cDNA in HeLa cells permits a functional analysis by site-directed mutagenesis. Five of the 12 cysteines in the amidotransferase were replaced. Of the cysteines only Cys' is invariant in all of the available sequences. A C1S replacement abolished the glutamine-dependent enzyme activity while some activity with NH3 was retained (Table I). Thisresult supports the conclusion that Cys' in glutamine PRPP amidotransferase (30, 31) and in related amidotransferases (32, 33) is an active site residue that functions in glutamine amide transfer. Propeptide processing was abolished in the avian amidotransferase C1S mutantashas also been observed for a comparable mutation in the B. subtilis enzyme (8). It has been suggested (8, 9) that this defect results from an altered processing site DISCUSSION or is a reflection of autocatalytic processing, as occurs in Chicken liver glutamine PRPP amidotransferase is synthe- prohistidine decarboxylase (34). Replacements of Cys415 and Cys4= were designedto invessized with an 11amino acid propeptide-like sequence preceding theinferred active site cysteine (18).In thiswork, Western tigate the possibility that these residues are ligands to anFeS cluster since they align with cysteines that serve this funcblot analysis with antibodies preparedagainstNH2-and C02H-terminal peptides, as well as direct sequence analysis tion in theB. subtilis enzyme. Our working hypothesis is that of immunoaffinity-purified enzyme has established that the the chicken enzyme contains an Fe-Scluster. Evidence for an propeptide-like sequence is removed and the mature amido- Fe-S cluster in the avian amidotransferase is inconclusive. By transferase in chicken liver starts with an NH2-terminal cys- direct measurement, iron (14, 15) but not sulfide (15) has teine residue. Accordingly, methionine at position 1 in the heen detected. However, other properties such as release of amino acid sequence (18) has now been renumbered Met-" tightly bound Fe by treatment with mercaptoethanol and long to reflect processing of the propeptide. This propeptide, wavelength absorption arecharacteristic of Fe-Sproteins Met-" to Glu' is identical in5 of 11 positions with the (15). Furthermore, Fe and Sz-were detected in the homologous corresponding B. subtilis glutamine PRPP amidotransferase human enzyme (17). The C415S and C488S replacements propeptide (Fig. 2). Recent evidence indicates that thehuman were unique. All of the glutamine PRPP amidotransferase enzyme contains a propeptide sequence identical to that of was insoluble in thesemutants. The propeptide was not the chicken enzyme.' Notwithstanding the similarity of the processed in the insoluble enzyme (Fig. 7). Mutations of the avian and B. subtilis propeptides, processing does not take cysteinyl ligands that interfere with assembly of the Fe-S place in heterologous cells. The propeptide in the avian en- cluster in B. subtilis glutamine PRPP amidotransferase also zyme is not processed in bacteria (Fig. 3) and the bacterial lead to loss of solubility and prevent propeptide processing propeptide is not processed in animal cells.3 The discussion (13, 25). We therefore infer that Cys415and Cys4= may be that follows leads to two conclusions. (i) An intact Fe-Scluster ligands to an Fe-S cluster that is structurally distinct from is not assembled in heterologous cells. (ii) An Fe-S cluster is that in the B. subtilis enzyme. The clusters must be different required for native structure and this structure is essential for because cysteinyl ligand Cb is not conserved in the avian propeptide processing. sequence (Fig. 2). Lack of normal amidotransferase function In order to examine the effect of mutations on function, in heterologous cells (18),3is thus suggested to result from limitations in assembly of the Fe-S cluster which is needed G. Zhou and H. Zalkin, unpublishedresults. for native structure. The complete insolubility of the c415S AT(C1F) is attributed to the mammalian enzyme. The C1F mutant enzyme, however, retained NHa-dependent activity. Western blots using comparable amounts of extract protein suggest a 2-3-fold reduction in soluble C1F protein(not shown). From assays in cell extracts it is not possible to distinguish whether the decreased NHs-dependent activity relative to the wild type results from partial mutational inactivation in addition to a decreased content of soluble enzyme. Of the 8 cysteines in the synthase domain, those at positions 415,485, and 488 align with cysteinyl ligands to the[4Fe-4S] cluster in the B. subtilis enzyme (Fig. 2). To investigate whetherthese residues may function as ligands to an Fe cluster inthe avian enzyme, we replaced cysteines at positions 415 and 488 with serines. Data in Table I indicate that there was undetectable glutamine and NHs-dependent activity in (2415s and C488S enzymes. Only background activity from HeLa cells was detected. Western blot analyses with antisera N-Ab and C-Ab indicated that these enzymes were localized entirely in the particulate fraction and contained the propeptide (not shown). In order to determine whether the phenotype of the cysteine replacements at positions 415 and 488 is unique, two additional cysteines in thesynthase domain, residues 269 and 295 were mutated. CysZ6'is conserved in thebacterial and human enzymes, whereas CysZg5is only conserved in the human amidotransferase (Fig. 2).' The content of soluble enzyme in the C269S and C295S mutants was comparable to the wild type (not shown). The Western blot in Fig. 8 shows that the soluble C269S enzyme was unprocessed whereas the (2295s enzyme was partially processed. Lane 1 in Fig. 8 shows the resolution of the two forms from a mixture of wild type and C1F. The results of enzyme assays, given in Table I, indicate that the unprocessed C269S mutant enzyme is without glutamine- and NH3-dependent activity whereas the partially processed C295S enzyme has low residual activity with both substrates.

7942

Glutamine Avian

PRPP Amidotransferase

9. Souciet, J.-L., IL.modson, M. A., and Zalkin, H. (1988) J. Biol. Chem. 263,3323-3327 10. Wong, J. Y., Meyer, E., and Switzer, R. L. (1977) J. Biol. Chem. 252,7424-7426 11. Switzer, R. L. (1989) BioFactors 2 , 77-86 12. Makaroff, C. A., Paluh, J. L., and Zalkin, H. (1986) J. Biol. Chem. 261, 11416-11423 13. Grandoni, J. A., Switzer, R. L., Makaroff, C. A., and Zalkin, H. (1989) J. Biol. Chem. 264, 6058-6064 14. Hartman, S. C. (1963) J. Biol. Chem. 238,3024-3035 15. Rowe, P. B., and Wyngaarden, J. B. (1968) J. Biol. Chem. 243, 6373-6383 16. Itakura, M., and Holmes, E. W. (1979) J. Biol. Chem. 254,333338 17. Leff, R. L., Itakura, M., Udom, A., and Holmes, E. W. (1984) Adu. Enzyme Regul. 22,403-411 18. Zhou, G., Dixon, J. E., and Zalkin, H. (1990) J. Biol. Chem. 265, 21152-21159 19. Elroy-Stein, O.,Fuerst, T. R., and Moss, B. (1989) Proc. Natl. Acad. Sci. U. S. A. 86,6126-6130 20. Linn, T., and Pierre, R. S. (1990) J. Bacteriot. 172, 1077-1084 21. Tabor, S., and Richardson, C.C. (1985) Proc. Natl. Acad. Sci. U. S. A. 82,1074-1078 22. Fuerst, T. R., Niles, E. G., Studier, F. W., and Moss, B. (1986) Proc. Natl. Acad. Sci. U. S. A. 83,8122-8126 23. Felgner, P. L., Gadek, T. R., Holm, M., Roman, R., Chan, H. W., Wenz, M., Northrop, J. P., Ringold, G. M., and Danielson, M. (1987) Proc. Natl. Acad. Sci. U. S. A. 8 4 , 7413-7417 24. Smith, D. B., and Johnson, K. S. (1988) Gene (Amst.) 6 7 , 31-40 25. Harolow, E., and Lane, D.(1988) Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 26. Towbin, H., Staehelin, T., and Gordon, J. (1979) Proc. Natl. Acad. Sci. U. S. A. 76,4350-4354 27. Schendel, F. J., Cheng, Y. S., Otvos, J. D., Wehrli, S., and Stubbe, REFERENCES ' J. (1988) Biochemistry 27,2614-2623 Messenger, L. J., and Zalkin, H. (1979) J. Bwl. Chem. 254,338228. Shen, Y., Rudolph, J., Stubbe, J., Flannigan, K. A., and Smith, 3392 J. M. (1990) Biochemistry 29,218-227 Meyer, E., and Switzer, R. L. (1979) J. Biol. Chem. 2 5 4 , 5397- 29. Miller, J. H. (1972) Experiments in Molecular Genetics, p. 403, 5402 Cold Spring Harbor Laboratory, Cold Spring Harbor, NY Becker, M. A., and Kim, M. (1987) J. Biol. Chem. 2 6 2 , 14531- 30. Tso, J. Y., Hermodson, M. A,, and Zalkin, H. (1982) J. Biol. 14537 Chem. 257,3532-3536 Holmes, E. W., McDonald, J. A., McCord, J. M., Wyngaarden, J. 31. Vollmer, S. J., Switzer, R. L., Hermodson, M. A., Bower, S. G., B., and Kelley, W. N. (1973) J. Biol. Chem. 2 4 8 , 144-150 and Zalkin, H. (1983) J. Biol. Chem. 2 5 8 , 10582-10585 (1983) Tso, J. Y., Zalkin, H., vancleemput, M., Yanofsky, C., and Smith, 32, Badet, B., Vermoote, P., Haumont, P-Y. (1987) Biochemistry 2 6 , J. M. (1982) J. Biol. Chem. 257,3525-3531 1940-1948 Makaroff, C. A., Zalkin, H., Switzer, R.L., and Vollmer, S. J. 33. VanHeeke, G., and Schuster, S. M. (1989) J. Biol. Chem. 264, (1983) J. Biol. Chem. 258, 10586-10593 19475-19477 Switzer, R. L. (1989) in Allosteric Enzymes (Herve, G., ed) pp. 34. Recsei, P. A,, Huynh, Q. K., and Snell, E. E. (1983) Proe. Natl. 129-151, CRC Press, Boca Raton, FL Acad. Sci. U. S. A. 80,973-977 Mantsala, P., and Zalkin, H. (1984) J. Biol. Chem. 2 5 9 , 14230- 35. Mantsala, P., and Zalkin, H. (1984) J. Biol. Chem. 2 5 9 , 84788484 14236

and C488S mutant enzymes supports the view that the putative Fe-S cluster is essential for native structure and propeptide processing. The final two mutant enzymes, C269S and C295S, are distinguished from the preceding in two respects. (i) Their solubility is similar to the wild type. (ii) There is a trace of propeptide processing in C295S and a trace of activity with glutamine and NH,. The simplest interpretation is that these cysteine replacements in thesynthase domain result in subtle structural perturbations and loss of NH3-dependent activity. The structural changes may prevent propeptide processing which is required for glutamine-dependent activity. The forgoing interpretations lead to a working model for amidotransferase function. Concomitant with or following synthesis an Fe-S cluster isincorporated into the avian amidotransferase in animal cells. This allows the enzyme to acquire its native structure and to undergo propeptide processing. The mature enzyme has glutamine-dependent activity. When enzyme synthesis in HeLa cells exceeds the capacity forFe-S incorporation, much of the enzyme cannot fold properly and is insoluble. Likewise, an Fe-S cluster is not assembled into theavian enzyme in bacterial cells; the native structure is not attained and the propeptide is not processed. Structural perturbations resulting from lack of an intact FeS cluster, as well as from mutations in cysteine residues at positions 1, 269, and 295, preclude propeptide processing and glutamine-dependent enzyme activity. 1.

2. 3. 4. 5. 6. 7. 8.