J. Nol. Biol. (1981) 146, 169-192

Genetic and Biochemical Characterization of the birA Gene and Its Product: Evidence for a Direct Role of

Biotin Holoenzyme Synthetase in Repression of the Biotin Operon in Escherichiu coli

DAVID F. BARKER? AND ALLAN izI. CAMPBELL

Lkpartmrnt of Biological Scimces Stanford I’t~iwrsity, Stanford, CA 94305, I’.S.L4

1,esions at the hir.4 locus of Escherichia coli produce, in varying degrees, derepression of the biotin operon and an increased minimum biotin growth requirement (Barker & Campbell, 1980) as well as diminished biotin uptake and defective biotin holoenzyme synthetase activity (Campbell et al., 1972,198O). In the accompanying paper, we showed that three bird mutants produce biotin holoenzyme synthetase with altered in vitro properties and that they carry lesions in the structural gene for this enzyme. The pleiotropic hiril defect was attributed to structural interactions between a protein domain which includes the holoenzymr synthetase active site and a second protein domain, possibly part of the same polypeptide, which functions as the hio repressor.

To determine if one or more genes reside at bir.4, we tested pairwise combinations of nine mutations with representative phenotypes for their ability to establish repression of hio expression. The mutations define a single complementation group. Instances of partial complementation appear to be intracistronic, suggesting that the bird product, forms a multimer active as both biotin holoenzyme synthetasc and repressor.

DNA segments that include and express the hirrl gene have been cloned into multicopy plasmids. Plasmid-mediated expression of hir=i can produce a state of superrepression of the hio operon and a concomitant increase in holoenzymr svnthetase specific activity. The complementation properties of derivative plasmids. dith insertions of Tn5 or small deletions in the bacterial DNA se I.6 x IO’ base region that includes the hir.4 gene and a @9 x IO Q

ment, define a base segment

essential to biotin holoenzyme synthetase and repressor function. The region is flanked by the thrT and tufB genes in a previously unassigned region of th(t bacterial DNA carried by Mrifd18.

A preparation of holoenzyme synthetase, purified nearly lO.MK-fold, contains a protein that binds specifically to biotin operator DNA as determined by its abilit,y to protect a Tag1 endonuclease site that borders the imperfect inverted repeat where the hio repressor is presumed to bind. Biotinyl .5’-adenylate or biotin plus ATI’ are more effective corepressors than biotin alone, suggesting that biotinyl 5’. adenylate. a presumed intermediate in the holornzyme synthetase reaction. is the true corepressor.

470 I). F. BARKER AND A. M. CAMPBELL

1. Introduction Th h e p ysiological alterations caused by mutations at the hirtl locus of Eschrrich%n co/i include derepression of the bio operon. defective biotin uptake and increased minimum biotin growth requirement, indicating that an element common to biotin operon repression, biotin uptake and utilization is encoded there (Campbell it al., 1972,398O; Barker & Campbell, 1980). We showed in the accompanying paper (Barker & Campbell, 1981) that some t&A mutants produce biotin holoenzyme synthetase with demonstrably altered properties in vitro and concluded that bir=l is the structural gene for this enzyme. We also showed that certain hir.4 mutations, belonging to a class previously designated bioR and presumably affecting the bio

repressor. are defective in the synthetase activity. We have also observed that, in our collection of birrl mutants, there is no simple relationship between the minimum biotin growth requirement and the concentration of biotin required to effect repression, supporting the idea that there are structural interactions between a protein domain which includes the holoenzyme synthetase active site and a second protein domain which functions as the hio repressor. Further evidence of the intimate association of BHSt and repressor is provided by the discovery by Yrakash & Eisenberg (1979), which is confirmed here. that biotin,vl 5,‘.adenylate. a presumed intermediate in the biotin holoenzyme synt,hetase reaction, is apparently the true corepressor.

In principle, the Oir=l locus could encode one or more polypeptides. If one polypeptide is produced, it could include interacting domains directly involved in BHS and repressor activity. Alternatively. the hirS product could be part, of a multienzyme complex with both of these activities. Either hypothesis could accommodate the properties of those mutations that apparently cause structural changes in both the BHS a,nd repressor functional units. as well as those whose principal effect is on only one of these units. If two or more polypeptides were affected by the hir=l mutations, one of these could be multifunctional or at least one would have to be involved in a multienzytne complex in order to explain the mutant phenotypes.

We report here that nine bird mutations, with phenotypes representative of the entire spectrum of alterations observed, fall into a single complementation group as judged by their inability to complement, to establish normal repression of hio operon expression in complementation tests. Some of the mutations sbudied probably affect the hio repressor since repression of operon expression is completely eliminated while biotin uptake and utilization are only moderately altered. Other mutations severely affect biotin uptake, cause a dramatic increase in minimum biotin growt,h requirement and drastically reduce the biotin holoenzyme synthetase activity measured in rdro. The three mutations with demonstrably altered BHS activity in vitro. as described in the accompanying paper, were included in this analysis. The failure of these mutant’s to complement indicates that some component of the hio repressor and BHS are encoded by the same gene.

We hare also cloned and characterized a 2.1 kb segment of bacterial USA that

t Abbreviations used BHS. biotin holoenzyme synthetawz; kb. base-pairs x IO’; bp. base-pair: SDS. sodium dodecyl &fate: biotinyl-AMP. biotinyl ii’-denylate: bio. biotin.

ROLE OF birit IS bio I<EGlTLA’I’IOK 471

includes and expresses the bir.-l gene. An elevated level of the hr.3 product, is ljrescnt in strains bearing certain plasmids that include the gene, resulting in a &ate of superrepression of the hio operon and a correspondingly increased BHS specitit acativity. Deletion and insertion mutations within a 0.9 kh region of this cloned I)K’A eliminate plasmid-mediated complementation of the entire spectrum of hir.4 mutations.

I+~ally, an extensively purified preparation of BHS contains a protein that) sp~ificallp hinds to hiotin operator DNA when appropriate corepressor (biotin. biotin plus ATI’ or biotinyl 5’-adenylate) is present. We conclude that the bird product is a multifunctional protein that possesses active sites directly involved in

Ijiotin holoenzyme synthetase activity and binding to hio operator DSA.

2. Materials and Methods

I’ropagation of phage and bacteria and transduction with Plvir have been described (Barkthr & (‘ampbell, 1980,1981). Some of the recombination deficient strains used in this work were c*onstructed by transducing parental strains with Plvir grown on BM7096. which carries an insertion of TnlO adjacent to the ~c.456 allele. MacConkey indicator medium (Difco) was used in some experiments to assess the level of expression of lac genes fused to hio. Strains wit.h t’hr standard fusion @(hioFC-ZaciQWI and normal bio regulation form red c*olonirs on unsupplemented MacConkey plates and white colonies when supplemented wit,h aI n&I or more biotin.

(I)) Phage and bactwia

Strains used in this work are listed in Table 1. The derivative of hTR16ir.l with the imm434 immunit,y region was constructed by growing hbir.4c’ on S396, which carries a Xdg imm434 prophagra. Plaque-forming imm434 recombinants were recovered by plating the lysate on Bsl2661, which carries an immX prophage, and Iysates of purified recombinants tested for Hir+ t,ransduction. as described below.

To construct A781 bir.4 phages bearing bir=l alleles of interest, the birJ mutations were first transduced with Pleir from the bio-lac fusion strain in which they were isolated (Barker & (‘ampbell. 1980), to BM6026, a strain with the (gal-bio)&l fusion described by Ketner & (‘ampbell (1975). In this strain, expression of galK is under bio control, and the ga/E gene is illactivated by the deletion. Derepression of the biotin promoter under low biotin growth c.onditions causes these cells to become sensitive to the presence of galactose. presumably dut> to the accumulation of galactose-1-phosphatfx.

Xrg+ transductants were selected, and cotransduction of the bird allele was scored either I)y the increased biotin requirement conferred or loss of repression of the biotin operon. c,ridenced by sensitivity to galactose in the presence of biotin. Drops of a h781birJ imm434 Iysate were spotted on lawns of the hir=l -, (gal-bio)dfil strains on plates of minimal agar (which contains the indicator dye triphenyltetrazolium chloride), supplemented with 0.4”,, galactose. (P40/6 glycerol (v/v). 0.04% casein amino acids (Difco) and 41 nM-biotin. (:alact,oscs- resistant, ((iaIR) lysogens were detected among the red papillae appearing after 2 days in th(s area of phage growth. These arose, in the majority of cases, by integration of the h781bir.J genomfa ~icr homologous recombination with the bacterial chromosome segment which contains bir.4 Lysates produced by induct,ion of t,he GalR lysogens usually contained bir.-l + and birrl - phages.

Plaque-purified progeny from inductions of presumed heterogenotes were checked for their bir.-l allele by 3 spot tests. Portions of lysates were placed on BM2661 on minimal lactose plus 41 nxr-biotin plates: on B&l6048 on the GalR selective medium described above

472 I). F. BARKER AND A. M. CAMPBELL

TABLE I

Strains of bacteria curd hnctrriophage

Bactei-ial strain Genotype Source. origin or reference

(‘BKOl7 Ml82 MC4100 S396 BM2661 BM4046 BM4050 BM4O(iz, HM40ti4 HM407:! HM4084 BM4086 BM4110 HM4112 BM4134 BM4140 HM6010 BM6020 HM6026 BM6038 BM604X KMTO!%

Pha,qe strains hhir.4 c+ hbirA intnr4.34 hbirA cl857 Smti hNK467

Plasmids pBR3d2 pAWC184 pHI’H4 pKK%48clls pBA2 pBA 1 I pBAI:! pHA% pHA36 pBA40 pHAd03 pHAdZl pBAi37 pHA14:! pm2

CV3110 thy-. qEOli:;Tn5 F-AktcX71. g&P. grrlK, &A. thi F-ara1)139, AlarlJ169, strd . thi LV3350 (hdyimm434) .G MIc4100, @(bioFC-lacZ)SOl As BM2661. birA206 As BMZ661, birA215 As BM2661. birA85 As BM1661. birA91 As BM’J661, b&A 707 As BM2661. birASO1 AS BM2661. birA303 As BM2661. rpoR4R3, birA35Z As BM2661. rpoR423. birA.361 As BMg661. birA879 As BM2661, birA85, recA1 As MC4100. A(yal-bio)61 As Ml%%, @(bioFC-lacZ)501 As BM6010. nrgEOl7::Tn5 As BM6010. birARS As BM6010. birA301 Hfi .url: :‘l’nlO rwAc56

(AbZ21 cl857 res::Tns’ Oar&Y I’nmRO)

Derivative of pBR32:! Derivative of pACYC184 Derivative of pHUB4 Derivative of pHlTB4 Derivative of pBASZ Derivative of pBA22 Derivative of pBA11 Derivative of pBAl1 Derivative of pBAl1 Derivative of pBAl1 Derivative of pACYC184

K. Shaw 8r C. Berg M. (‘asadaban M. Casadaban Our c~ollection Barker & Campbell (1980) Barker & Campbell (1980) Barker & Campbell (1980) Barker & Campbell (1980) Barker & Campbell (1980) Barker (t Campbell (1980) Barker 8r Campbell (1980) Barker & (‘ampbell (1980) Barker & (‘ampbell (1980) Barker Rr Campbell (1980) Barker B Campbell (1980) This work This work This work This work This work This work K. McEnter via M. Casatlaban

This work This work This work N. Klec-kner

Holivar el nl. (1977) (‘hang & Cohen (1978) I’lric,h-Hernard ef ul. (1979) LTlrich-Bernard d al. (1979) This work This work This work This work This work This work This work This work ‘l’his work This work 1). Barker. .I. Kuhn and

A. Campbell

and also on BM4062 on glucose, 41 nM-hiotin medium. In the first test, infection of &.-I+ cells by a birA - phage apparently allows rare recombination and segregation or conversion events which result in the appearance of biril - homogenotes. Derepressed expression of the luc genes fused to bio occurs in these homogenotes which may then ferment lactose and form red colonies in the area of phage infection. The second test scores complementation of the galactose sensitive (Gal’) phenotype of a (gal-bio)ddl fusion strain bearing the birA301 mutation. Phages bearing biril mutations cannot complement for repression, while bir-l+

ROLE OF bir.4 IX bio REGULATION 473

phages frequently form (:alR lysogens, detected as red papillae. The third test indicates the presence of a birA allele able to complement the growth defect of BM4062. Introduction of bir.l+ or any of several birA mutant genes masks the severe growth defect of the birA8:i mutation in this strain in a pattern consistent with the previously reported characterization of altered minimum biotin growth requirement in birA mutant strains (Barker & Campbell. 1980). Together, these three tests provide a reliable indication of the sllelic form of OirA prrsent on putative mutation-bearing phages. Furthermore, the characterization of BHS enzyme produced upon infection with phages bearing birA mutant alleles, as described in the accompanying paper. provides evidence that the alleles detected on the phages are identical to those that occur in the chromosome of the host from which they are derived.

(c) Complementation analysis of birA mutations

Heterogenotes were constructed by lysogenizing the bir.4 mutant derivatives of BM2661. which carries the @(bioFC-lad?) fusion, with the A781birA - imm434 phages. Lysogens were isolated by streaking from areas of growth of the phage on lawns of the host and testing colonies for imm434 phage release. Integration of the phage may occur ljia the 9.2 kb of bact,erial homology or the approximately 34 kb of homology with the immh prophage present at the site of the bio-Zac fusion. Selection for Bir+ recombinants during this isolation was avoided by adding excess biotin to all media. Cultures of heterogenotes were grown as described in Results section (a) and assayed for fl-galactosidase after lysis with chloroform and SDS as described by Miller (1972). The frequency of cells in those cultures which had lost thr imm434 prophage was estimated by streaking for single colonies and testing 60 for imm433 phage release by picking into an immh lawn. The maximum frequency of segregants olbserved was about :i”/b. In instances where the contribution of segregants to the p- galactosidase activity observed was significant, corrections were applied based on the determined segregant frequency and the fl-galactosidase activity in a pure culture of cells of the segregant (i.e. host) type.

(d) PuriJicution of holoenzyme synthetase

The init,ial steps of purification of the holoenzyme synthetase preparations used in this work were as described in the preceding paper except that 9 I of cells infected with Abir=l cl857 Sam7 were extracted and a 3 cm x 90 cm column of Ultrogel AcA34 (LKB Instruments Inc.) was used for the gel filtration step. Peak fractions from the subsequent phosphocellulose column were pooled and loaded on a 2 cm x 18 cm column of hydroxylapatite (Bio-Rad, DXA grade) equilibrated with 10 mm-potassium phosphate buffer (pH 7.4), 10% (v/v) glycerol, @l mM-Na,EDTA, 10 mM-p-mercaptoethanol (PC buffer). After washing with several column volumes of buffer, a linear gradient of 001 M to I.0 M-potassium phosphate (pH 7.4) in PC buffer was applied. Assays of wash and gradient fractions with a crude extract source of apocarkwxylase revealed that all of the enzyme acbtivitv was bound to the column under the initial buffer conditions and was eluted with approximately @2 M-phosphate. Peak fractions were dialyzed and loaded on a 1 cm x 16 cm column of DEAE-cellulose (Cellex-D, Bio-Rad) equilibrated with PC buffer. Fractions collected during the loading and wash of the column contained most of the enzyme activity and were free of a potent exonuclease which copurified through the hydroxylapatite step. The enzyme was concentrated by adsorption to a 3 ml column of hydroxylapatite and elution with 0.4 M-pOtaSSiu!TI phosphate. The specific activity of the eluate (2 ml total from 2 Ibatches treated separately after the hydroxylapatite fractionation) was approximately SO& fold relative t.o the starting material and lO,OOO-fold that. of a crude ext,ract of uninfected hir.4 + <YGlS.

(e) Operator protection assay

Binding of biotin repressor to the biotin operator was monitored by the resulting protection of a Tag1 restriction site immediately adjacent to the presumed operator site (Otsuka & Abelson, 1978: Barker et al., unpublished results). The final 40 pl reaction

474 I). F. BARKER ANI) A. M. (‘AMPHELL

contained 6 mM-Tris (pH 7.4), 50 mivr-NaCl, 6 mM-Mg(:l,, 6 mM-j?-mercaptoethanol, 100 pg swine skin gelatin/ml, 2pg of plasmid pDB1 DNA, 3 to 6 ~1 of purified holoenzyme synthetase, corepressor (biotin, biotin plus ATP or biotinyl 5’.adenylate, see Results) and 8 units of TagI, prepared by the method of Greene rt al. (1978). The mixture was incubated for 30 min at 37°C prior to the addition of the Tag1 (2 ~1) and digestion was continued for 3 to 3.5 h. Following phenol and subsequent ether extractions, the DNA fragments were precipitated with ethanol, resuspended in water and separated on a polyacrylamide gel. Plasmid pDB1 is a derivative of pACYC184 (Chang & Cohen, 1978) and carries a 4.0 kb RamHI fragment of hbiol DNA which includes the biotin operator-promoter region. Plasmid pDB2, which was present as a control in some experiments, is identical to pDBl except that it carries the biopZ31 mutation, which is an insertion of IS1 in the hio control region (Barker rt nl., unpublished results). Biotinyl Gadenylate was synthesized and purified by DEAE chromatography ((‘hristner & Coon. 1970). The preparation used was about 802& pure.

(f) Recombinant DS.4 mdhds

Bacteriophage DNA for restriction analyses and cloning was extracted by the SDS lysis method of Cameron et al. (1977) from phage preparations that had been banded in a CsCl equilibrium gradient, then dialyzed against 10 mM-Tris (pH 7.3), 10 mM-MgSO,, @I mM- Na*EDTA and diluted to an d,,, value of 1 to j. Plasmid DNA was extracted (Gerry it al., 1973) from overnight cultures of .5 to 1000 ml to which 100 pg chloramphenicol/ml or 200 pg spectinomycinjml had been added to an ,-I,,, value of about 1. Buffer conditions for digestion with restriction enzymes were those recommended by the supplier or 100 miv-Tris (pH 7.5), 50 mM-N&l, 10 mM-Mel,, 6 mq%mercaptoethanol and 100 pg gelatin/ml for RamHI, BglI, Bg111, EeoRl. HirLdIII, SalI or XhoI. ,411 digestions were at 37°C except P&I and &2~3A (30°C) and Taql (65°C). Enzymes were obtained from New England Biolabs, BRL or Miles or prepared by published methods ((ireen? tt al., 1978). Ligations were in 25 to GOpI volumes containing 1 pg of each restricted DNA, and 20 mlcl-Tris (pH 7.6), 10 mM- MgC’l,, 10 mM-diothiothreitol, 0.5 m&f-4TP (neutralized with Tris base) and 1 unit of T4 DSA ligase (Biolabs). Plasmids pBR322 (Bolivar (4 al., 1977). pA(‘YC184 (Chang & Cohen. 1978). pHUB4 and pRK248 clts (Ulrich-Bernard rt nl., 1979) were obtained from R. (iunsalus and (:. Zurawski of this department. Cells were made competent for transformation by the method of Dagert & Ehrlich (1979).

Determination of DNA fragment sizes greater t,han I .O kb was by electrophorrsis in CbS?q, agarose (Bio-Rad), employing a digest of hclR57 8’am7 \tith RamHI plus EcoRI as size controls (Daniels d al., 1980). Lengths of shorter fragments were determined by 8% polyacrylamide gels with Tag1 and 8’auSA digests of pBR322 as sizca markers (Sutcliffe, 1978). For accurate measurement of fragments in the 0.5 to 2.0 kb range. 1.4% agarose gels with both sets of size standards were employed. Electrophorrsis buffer for agarose gels was described by Dugaiczyk d al. (I 975,) and for acrylamide gels by Peacock & Dingman (1968).

To prepare plasmid DNA mutagrnized with the kanamycin and nromypin resistance transposon Tn5 (see Results section (d)), strain BM6010 carrying pBA II was grown to 2 x lo* t,o 4 x IO8 cells/ml and infected with XNK467 by addition of phage to 1 ml of culture to give a multiplicity of 5 to IO. After 30 min adsorption at room t,emperature, 4 ml of fresh L-broth \vas added and the infected cells were incubated at 30°C’ for 2 h. A 1 -ml portion was diluted into L-broth containing 25 pg kanamycin/ml and inrubated overnight. The KmR cells were diluted 20.fold into L-broth with kanamycin. and plasmid DNA was amplified and prepared as usual. Plasmids carrying TnS comprised 0.01 to 0.1 “/b of the DN4 in these preparations. as judged by the relative yield of CmR and KmR transformants.

3. Results (a) Hiotitr operott expwssiotr itr hir?l hrtrrogrtlotrs

The ability of nine different bird alleles to complement in operon repression was examined by constructing heterogenotes in a strain that expresses the lac genes

ROLE OF birA IN bio liEGIJL.4’1’ION 475

under bio control and monitoring /?-galactosidase activity. Derivatives of strain BM2661 were constructed that bear 37 binary combinations of wild type and mutant bir=l alleles. In each heterogenote, one hirA allele is present at thr chromosomal location, and the second is on a XbirA prophage. Mutations were crossed onto the hbirrl phage and heterogenote strains constructed as described in Materials and Methods.

In choosing pairs of alleles for study. we specifically tested complementation between mutations with phenotypes indicative of a change in the b%o repressor function and those that we have shown to cause structural alterations in holoeneymr synthetase. We also included birA mutations with phenotypes rcprcasentative of the variety of defects observed with respect to altered biotin rcquirtment, uptake, repression and BHS. Combina,tions of alleles in which comF)lementabior1 for growth could be scored were included in order to show that a bir.4 gcbne on the prophage is expressed normally.

The properties of alleles used in this analysis are presented in Table 2. Four of thr mutat’ions. hir.-l206, hirrl303, bir.4301 and bir.4352, may be classified as affecting biotin repressor on the basis that, in vice, operon expression is not repressible by up to I m.tl-biotin, while the minimum biotin requirement of the mutant strains is not detectably altered. Two of these mutations, birA4206 and bir.4303, produce no BHS activity detectable in the irr Gtro assay. We assume that. in ?si?To, the BHS activity is near normal but that the same alteration that results in the repressor binding deficiency irr r+c*o increases the lability or changes some other property of the BHS enzyme which results in the failure to detect activity ijz vitro. bir.4301 and bir.-l,‘I,i% specify a RHS activity that is detectable i,n Gtro and only slightly less active than wild-t>rpr enzyme. The mutations in each of these pairs may be distinguished from

TABLE 2

Proprrties of birA mutatiorls

Allele

p-Galactosidaset Growth at 13 ‘(‘7 Biotin nt intiicitteti biotin at indicxtetl h&enzyme

conc~entration biotin concentration synthetwhe:

14 nM 11 nM 1m I*M 1 m!bl 1.1 nM -tl IIM 1.1 +M 30 (‘ 11 (‘

II7 W’l (Pi 0% + + + NDjj IN I”5 I42 + + + NI) 11’1) 137 133 + + + SI) SI) IO” 110 + + + SI) 1 (2 155 11!) + + + SI) X3 137 130 + + + 1M 21 18 10 + + + NI) I 23 1 I3 91 - + + NI) 2ti 44 05 SI) - + Nl) 111 3H 04 SI) - - Xl) Ii3 . 114 105 Sl) - -

1 CM) 100 loci :3x 91) (il

0 0 SI) 0 100 10 3 3 I 3X 8 0 0 0 0

NI)II SI)

t I)ata from Barker & (:ampbell (1980). $ Data from Barker R- Vampbell (1981). normalized to give wild-type values of 100 u/ rrcch, t~~~p~rulur~. $ XI). not determined. /I Infection of bird + cells with phage bearing this mutation results in no increase in ISHS activitv HS

mrasurrd at X’( vomparrd to a 20- to IO-fold increaxr in HHS activity with a 6ir.A + phage.

47ti I). F. BARKER ASI) A. M. ('AMPHELL

one another on the basis of measurements of biotin uptake presented in the accompanying paper. At 31 Y”, uptake rates of the &A206 and hir=1301 strains are consistently twofold less than for bir=l303 and hirA3:i2 strains, respectively. indicating that in each case the mutations are non-identical.

The lesions in biri1361, birA707 and hir=lgl cause structural alterations in biotin holoenzyme synthetase resulting in marked thermosensitirity or decreased specific activity of this enzyme in vitro. In GVO, hir.4361 and hir=l707 cause no marked increase in minimum biotin reyuirement, while hr.491 shows a significantly elevated requirement at 43°C’. Also, &A91 and birA,?GI are not repressible b) biotin concentrations up to 1 mM. while biril707 is partially repressed by 41 nM-

biotin but resistant to further repression by higher biotin concentrations. The last two mutations studied, bir.4215 and hir.-lX79, cause greater incrrases in the minimum biotin growth requirement at 43°C’ than bir.492: however, both allow complete repression at 30°C when sufficiently high biotin concentrations are present in the growth medium.

The results of assaying j3-galactosidase in cells bearing the 37 pairs of bir=l alleles tested from the 121 possible combinations with a given hr.4 allele on the prophage. or in the host grown with four different levels of biotin supplementation, are

presented in Table 3 with control assays of host cells without a h781hirA prophage. These represent 37 of the 55 combinations without regard to the location of the hirA alleles. Also shown in Table 3 are the results of testing the biotin growth requirement of certain heterogenotes. Where cells that have lost the hir3 prophage can contribute significantly to the observcad fl-galactosidase activity of heterogenote cultures. the assay values have been corrected by determining the frequency of such segregants in the culture and subtracting their contribution. We assume that in the vast majority of cases the cells that have lost, the h7Xlhir.4

prophage will retain the chromosomal hir=l allele. This has been verified 1,~ examining segregants from heterogenotes in which hir=121$ is the chromosomal allele (Table 3, lines 31 to 37). All of the segregants tested. at least two from each heterogenote. exhibited the characteristically elevated biotin requirement of the parental bir=l21-i strain. Although the genotype and /Sgalactosidase activity of the segregant cells are therefore predictable, the estimates made for the frequency of segregants, usually I to So/ of cult’ure. may be in error by 3%. For those heterogenotes where a pure culture of segregant cells makes 200 unit’s of fl- galactosidase, a possible error is six units. In heterogenote cultures where the host cell alone makes less than 200 units, for example bir.4707 and hirL-121:i (lines 22 and 31), the possible error is proportionately less.

Every mutant hirA allele tested is recessive to wild type. For those mutations that have been crossed onto h&A, this is shown by assays of partial diploids in which the chromosomal allele is wild type and the prophage allele mutant. Repression is normal in each case. (Compare Table 3 lines 2 to 7 with line 1.) WC know that hir=l is expressed when present on a prophage because the elevated growth requirement of hir.-1215 and hirA-lX79 strains is relieved by the presence of a prophage bearing other hir.4 alleles (lines 31 to 44). For the remaining hirA alleles. a bird + prophage establishes normal repression in every case (lines 23,28, 32,39). In the two cases where both possible configurations of bird ’ and a hir.4 - allele have

TABLE 3

Host allele

(‘omplementatiox for repression and growth’ hy selected pairs of birA mutations

fl-Galactosidaset Growth at 43 (‘: l’rophage allele + none 41 nix to5 PM 1 rnM 4.1 nM 41 IlM 4.1 pal

1x1 13 I 181 14X 88

152 18X

229 13X 221 206 240 19% 203

290 253 27X “51 230 208 “89

“38 216 154 342 176

112 176 268 194

109 164 I88 172 14X 175 173

243 127 161 194 154 174 146

1.3 0.5 2.0 I ,4 1.5 1.8 3.”

210 5.9

149 86

181 182 Ii”

“39 119

198 224 299 104 199

10 04 7 9

I 6

213 <l 24

164

83 04

155 7.3

30 35 47

211 <l 27 18

146 7%

139

1.1 03 1 .o 12 w9 1.3 1.6

240 5.0

IO9 93

133 229 13;

“30 1-X

177 215 214 213 192

7.5 0.5 (i 7 9

338 <l 14

145

9.5 0.6 3 .o I .ti 5.9 5.0 9.3

I22 <l

i.3 2.4

44 27 43

1 .o (F5 I 5 1.1 09 1 ?I 15

“27 5-i

112 106 162 197 136

213 6.7

179 214 “19 204 “21

7 0% (i

7

x

Zl)ti <l 10

146

0.8 0.4 0.7 04 07 1.0 09

0.7 Cb4 Wti

0.7 04 Wti

1 +i

SI)

SI)

Nl)

XI)

- - + + + + + + +

(+) + + + + + + + +

(+) + + - - - + + +

+ Cl) ‘+ + + + + + .+ +

(+) + + t (‘ukures were pregrown overnight in tryptone broth supplemented with the indicated amount ot

hiotin. t,hen diluted l/50 into fresh medium and assayed in log phase after 5 to 6 mass doubling. : Single colonies of heterogenote strains were streaked on glucose minimal plates supplemented with

(W”,, vasein amino acids and the amount of biotin indirated. $ ND. Sot detjermined. +. Indicates colonies appearing at the same time as for the bir;t ’ control strain : ( + ). indicate:,

volonies showing noticeable growth retardation: -, indicates no growth.

478 I). P. BARKER ANL) A. M. (‘AMPHEl,I,

been tested, hiril3,58 (lines 3 and 9) and hirA436/ (lines 7 and 16). greater variation in the repressed levels of fi-galactosidase is observed in t,he heterogenotes with bir.4 + on the prophage. This result is consistent, with our analysis of t,hr contribution of segregant cells t’o /3-galactosidase activity.

The /3-galactosidase assays from all of the mutant combinations tested reveal no instance of complementation to produce completely normal repression of operon expression. There are, however. several instances of partial complem~ntation. The pairs that show the st,rongest effects are bir=lYl with t~irrl.301 (line 29), Oir.421:i with bir.4352 and birA4301 (lines 33 and 34) and hr.4879 with hir.4352. hir.4301 and hir.4303 (lines 40. 11, 43). Smaller but probably significant effects are seen in the combinations of hir.4#79 with bir=l206 and bir.4361 (lines 12 and 44). (See Table 2 for fi-galactosidase assays of strains bearing mutations Dir=1301. hir.4206 and Oir=l303 alone.) The failure to observe any case of complete complemrntation is not due to the error associated with the occurrence of non-lysogenic segregants. If for each case of partial complementation observed, we assume that the maximum possible error has occurred and subtract’ it from the value reported in Table 3. a significant residue of derepressed expression always remains. -Also. for all allele pairs showing strong partial complementation, at least one other allele exists that fails to complement either one. The birA361 mutation. for examplr. fails to complement hirA9I. hir.4301, hir.4303, hirA421:i or hir.4352, although some combinations of these mutations show partial complementation. All of the cases of partial complementation may therefore hr classified as intracistronic and are not inconsistent with the conclusion that all of the hir.4 mutations affect the synthesis of a single po1ypeptide.

(‘omparison of the published restriction site map of Adrip (Horos & Sain. 1977 : Lindahl rt al., 1977) with our mapping of A781bir.4 Sam7 indicates that the 9.2 kh EcoRl fragment in the bacterial region of Adrifdl8 is the E’coRl fragment cloned in h781hirL4 (Fig. 1). The location of bir.4 in the hdrifd18 segment is consistent with genetic mapping data for ha4 (Pai & Yau, 1975) and t,he finding by Prakash & Eisenberg (197X) that hiotin repressor activity is amplified upon induction of a hdrifd18 lysogen. With appropriate single and double digests of h781hira4 Sam7

DNA, we have confirmed the absence of Hind111 sites from the bacterial segment and the locations of single BarnHI, Snll and Smnl sites as predicted from the published maps (data not shown). Under t,he conditions of electrophoresis used for the initial mapping. fragments of less than 0.5 kh were not detected. and the SMXI sites that are about 0.3 kl) apart (Lindahl rt N/.. 1977) appeared as a single sit,e. Electrophoresis of a SmaI digest of h78lbir.4 Sam7 in Sob polyacrylamide revealed a fragment of about 210 bp which must originate from the bacterial segment. as /\DNA contains no SnlnI fragment of this length (Daniels rt nl.. 1980). Digests with RglTI and appropriate double digests revealed a single site of cleavage for this enzyme at a point in the bacterial DSA 2.1 kl) from the RnmHJ site and 1.8 kh from the Sn~a1 site (Fig. 1).

Ligation of h781bir.4 Sam7 DNA cut with Sal1 and RgZII w&h pBR322 DNA cut

470

1 I t t t II SmoI

I t ECORI

X drifd18

L 1 pBA2

1 I pEAll,pBA22,pBA2

FI(:. I, Alignment of the genetic and physical map of the bacterial DXA region of hdriJ+‘lR with the bacterial segments of’h781birA and derivative plsxmids bearing the birA gene. Only bacterial DNA from hr/r</“ZX is shown. The locations of recognition sites for restriction enclonudeases and the positions of genes involved in transcription and translation are from published reports and shown approximately to wale (LindahI r/ rtl.. 1977 : Horos & Sain. 1977; Yamamoto & Nomura. 1979: Taylor & Burgess, 1979: Rossi & Idandy. 1979: I)anieln et al., 1980). The locations of restriction sites shown for X781birA hare either been cwdirmed to be in agreement with the published data for AdriJdZR or established by this work (/QUI site. see trst). The direction of transcription of’bir-4 is from right t,o left on the Figure as is true for all other genes shown. with the possible exception of the gene for protein r. The determination of the IcJc.ation and direction of transcription of birA and the construction of plasmids bearing this ,genc are clrswibd in the te.xt

\\ it’h Snll and HrtnlHl and screening of ApRTcS transformants of BM6038 resulted in detection of several Bir+ strains. Plasmid 11NAs prepa,red from six strains were digest,ed with RamHI. WI. Sol1 plus EcoRl and Sal1 plus Rglll. The simplest plasmid type. occurring twice, was 9.2 kb long with single Harr/HI. EcoRl, Snll and RgllI sites (Fig. 2). Each of the six plasmids included a 1.9 kb WI-RglIl fragment. The occurrence of any Rglll site in these plasmids is unexpected. since ligation of fragment ends left by HamHI and RgEll creates a hybrid site recognized by neither enzyme. Analysis of a Rglll digest of one 9.2 kb plasmid, designated pBA2. revealed a 60 bl) fragment. indicating that at least Tao small Rg!Tl fragments, presumably from the /\ I)XA of hhir.4. are incorporated into the plasmid in addition to the 4.9 kl) Srrll-Hglll fragment. The identity of the larger fragment hvith the ,%/I Hgll I bacterial segment of /\781 bir-4 was confirmed by showing that pBA2 has no S~nl site and digestion with HnmHl plus HglTl generates the expected 2.1 and 7.1 kh pieres (Fig. 2).

To locate the hr.4 gene more precisely, the 2.1 kb Rnn/H I-ngll I piece from pBA2 \vas cloned into the pA(“Y(‘l81 RamHI site as indicated in Figure 2. Several (‘mR. Tc’, Al? transformants of 9116038 proved to be Bir+, Digestion of plasmid 11X.4 from eight, of these with Hnn~Hl plus BcoRl revealed four with l.!) kb fragments identical to the shorter RnmHI-EcoR1 fragment of pA(‘Y(‘IX4. and a second fragment of 4.7 kb. Digestion with RglIl of one of the four. pBA11, revealed no site. confirming the map shown in Figure 2. Since the size of pA(‘Y(‘18-l cut with NamHI \vas estimated t,o be 4.3 kb in the same experiment. the calculated length of the

480 I). F. BARKER AND A. M. CAMPBELL

x bir A ~1857 Sam 7

EC0 RI

FIG. 2. Summary of construction of plasmids bearing t,he birA gene. For conditions of endonuclease digestions and ligation see Materials and Methods. Phenotypic properties of these plasmids are discussed in the text.

The bacterial DNA portion ofh78lbirA is drawn as a thick line. The restriction sites in this region and the Sal1 site in the phage J)NA relevant to discussion of the cloning procedure are shown. The distances shown between plasmid restriction sites are approximate and, where closely spaced restriction sites are shown. exaggerated for clarity. The I’&1 sites in the birA region have been demonstrated and mapped in pBAl1, pBA22 and pBA26 and are shown in their presumed location in pBA2 and AbirA The origin of the short RgZII fragment in pBA2 is described in the t,ext.

RamHI-RgZII insert is 2.3 kb, in fair agreement with mapping data from h781 birA ~\‘am7 and pBA2. Transformation with pBAl1 of BM4240. which carries the hjril8:i allele and recill, gave Bir+ transformants, indicating that the plasmid carries and expresses a functional birA gene.

The RamHI-RgZII fragment which includes bir.4 has also been cloned in both orientations into the RamHI site downstream from the hP, promoter on plasmid pHUB4 (Ulrich-Bernard et al., 1979). Plasmid DNAs were cut and ligated as indicated in Figure 2, then used to transform BM6038 bearing the plasmid pRK248clts, which provides the h repressor necessary to prevent lethal expression of XPL on pHUB4 and derivatives of it. Bir+ KmR transformants were selected on L-agar with kanamycin. About half of the Bir+ transformants formed normal-sized

HOLE OF birA IN bio REGI’LATION 481

colonies, while the other half grew more slowly. Plasmid DNA was prepared from four of each type of transformant and digested with RamHI plus SalI. Three of four large colony formers contained identical plasmids, one of which was designated $A22, and one of four small colony formers contained a plasmid, pBA26, which &owed the same 13 kb BarnHI-Sal1 fragment as pHUB4. The other RamHI-Sal1 fragment of pBA26 is 8.5 kb long while the RamHI-Sal1 fragments of pBA22 are 64 and 3.4 kb. Since the parental plasmid, pHUB4, is 7.7 kb, pBA22 and pBA26 apparently contain insertions of a 2.1 kb RamHI-RgZII fragment in opposite orientations. Subsequent mapping of the P&I sites in these two plasmids confirmed this difference (Fig. 2. data not shown). The remaining four plasmids from the experiment apparently contained additional insertions of the very short Hglll fragment or the larger RamHI-BgZII fragment of pBA2 as evidenced by absence of a HamHI site or conferral of ampicillin resistance, respectively, and were not st,udied further.

(c) Pharacterization of plasmid-mediated expression of birA

All of the plasmids that carry at least the 2.1 kb RamHI-Z3gZII segment h781bir.4 complement the growth defect of birA85. Measurements of BHS specific activity in strains bearing the different plasmids reveal a wide range of expression of bir=l. however (Table 4). The most significant difference is seen between the plasmids in which the bir.-l gene is cloned in opposite orientations downstream from hi’,. In pBA22. biril must be transcribed from Al’,, since inactivation of the thermolabile X

TABLE 4

Riot& holoenzyme synthetase activity in strains

bearing birA ’ plasmids

Strain Growth

condition? Spe<+ic

activity$ Relative act,ivity

BM6010 37°C BMtiOlO/pBA% 37°C BM6010/pBAll 37°C BM6010/pRK~48rlt.s~. pBAZ% 33°C BM6OlO/pRK~48rlts. pBAB;! Induced BM6OlO/pRK~48cZl.s. pBAS6 33°C BM6010/pRK248clt.s. pBAa6 Induced

0.043 (2) (1) 0.503 ( 1) 11.7 0.138 (1) 3.2 1.12 (1) “ti 6.4 (3) 149 04)56 ( 1) 1 .3 omi (2) 0.6

t Cultures were grown in L-broth at the indicated temperature. Induced cultures were gown at 33°C to an A,,, value of 1.2, then shifted to 43°C by addition of an equal volume of warm broth and incubated 15 min at 43°C. The cultures were then shaken at 37°C for 3 h. Extracts were prepared as described in the accompanying paper.

$ CTnits of activitv are pmol 1 ‘Y’]biotin incorporated into protein/mg protein per min. An exogenous source of unbiot,inylated prot,ein was added as described in the accompanying paper. Numbers in parentheses indicate the number of determinations averaged to arrive at the value reported.

ij A derivative of BM6010 bearing pRK248cIls, which produces a thermolabile h repressor. waci prepared by transformation and selection for Tcs. (In the absent! of h repressor, expression of hP, on pHUB4 and its derivatives is lethal.) pBAPd and pBAZ6 were subsequently introduced bj transformation and selection for KmR at 29°C.

48” 1). F. HAKKEK AXI) A. M. (‘AMPHELL

repressor results in a sixfold increase in BHS activity. Also. there is significant, expression of bird even at low temperature. This expression is eightfold more than from pBAl1, which includes the same bacterial fragment as pBA22. The simplest, explanation for this result is that t,he /\ repressor provided hy pRK238clts is insufficient to repress completely the AI-‘, promoters present on all copies of pHA22. In contrast to pBA22. the presence of pBA26 in a hir.4 ’ strain makes very little difference in the BHS activity (Table 4). The poor ability of this plasmid to complement bir=l85 indicates that the total expression of hit-.-l product from multiple copies of pBA26 is less than from a single chromosomal copy and that, each plasmid-borne gene is expressed at a sma.ll fraction of the rate of a chromosomal gene. This effect may be due to the convergence of t,ranscription from hl’, and the hir,-l promoter on this plasmid, in analogy with the negation of trp promoter expression by an opposing hl’, as reported by Ward & Murray (1979). The observation of apparent residual expression of hljL on pK.U2 under the repressing conditions employed here indicates that there is significant, transcription opposing the hir=l promoter on pBA26 even at low temperature. Alternatively. the natural biril promoter may have been weakened or separated from the hir.4 coding sequences by the cloning procedure. At, any rat’e. the inducibility of OirA expression in pBA22 strongly indicates that, the correct orientation of transcription of the gene for biotin holoenzyme sgnthetase is from the H~rrHl sit’e in MirA t,o\vard the RqII I as indicated in Figure I.

If the hir-4 product also functions as the hio repressor. t’hen the large excess of this protein that is present in cells bearing bird plasmids should affect expression of the bio operon. To test this prediction. BM2661 was t,ransformed with t,he hr.4 +

plasmids and appropriate vector plasmids as controls. The progeny strains \vercl grown in media with hiotin concentrations ranging from X.1 x ICY4 to 24 x 10P1 ptl

and assayed for p-galactosidase. The data plotted in Figure 3 sho\v t’hat the hio

operon is indeed superrepressed when a plasmid that, expresses hir=2 at a high rate, as indicated by the BHS assay, is present. At all of the biotin concentrations tested. the cells bearing plasmids pBA2. pRAl 1 or pBA22 show lower levels of fl- galactosidase than cells bearing the respective parental plasmids. For pBA26. no difference in repression is observed. as predicted from the very low levels of BHS synthesized from this plasmid. An estimate of the amount of repressor protein in cells containing the bir.4 ’ plasmids relative to cells w&h a single chromosomal copy of hir=l can be made by comparing the maximally repressed levels of operon expression. since the proportional decrease in bhis basal level is approximately the same as the increase in repressor concentration. The amount, of repressor in cells bearing pBA2. pBAI1. pRA22 and pBA26 relative t’o c~ells with the parental plasmids is then calculated to be 8, 12. 5 and 0~9. rrsprctivel~-. (‘omparison of thesv values with the relative BHS specific activities presentSed in Table 1 reveals qualitative agreement between repressor and BHS levels. Differences in culture conditions (see Table 3 and Fig. 3) and the source of /\ rtyressor (pRK218clts wrsus

hc+ prophage) in strains containing pBA22 and pB&426 in t,he two experiments could account for the quantitative variations otwerved. and we conclude that the hir.4 ’ plasmids express the hio repressor co-ordinately \vith hiotin holoenzymt synthrtasr.

(a) (b)

VI<:. 3. Superrepression of hio operon expression in cells bearing birA+ pla8mids. I)erivat.iws ot I3MZt$(il carrying the plasmids indicated were grown at 30°C in glucwse minimal medium supplemented with 04”,, casein amino acids anti various biotin ooncent,rations, Selwtion for antibiotic resistance was maintained by addition of ampicillin. chloramphenirol and kanamycin to cdt.ures carrying pRK322, pA(‘Y(‘184 anti pHVR4 anti their derivatives. respedively. After overnight growth to saturation. the cdturrs wrre diluted at least SO-fold and assayed for fl-galartosidase aft,er 5 to ti doublings. (a) (0) pHIt:W: (0) pKA”: (A) pA(‘TC184: (A) pRAl1. (b) (m) pHIW: (V) pHA4ti: (0) pHA22.

(d) llqfinition of thf birA gene by insertion rind

deletion nlv&stiorra

A more precise definition of the bir*+l genetic locus has been approached 1)~ analyzing the properties of derivatives of the &-A plasmids with well characterized deletions or Tn:i insertions. Restriction mapping by single and double digests of pBkJ2 and pBA11 revealed two PvuII and two PstI sites located in the bacterial segment of these plasmids as indicated in Figure 4. pBA22 contains no other f’~rll and only one other I’d1 site, permitting construction of derivative plasmids deleted for the small I’~11 and I’stI fragments by complete digestion with f’r~/~Il and partial digestion with f’stl, respectively. followed by ligation. The f+/jII-reduced plasmid. pBA10. shows an appropriate decrease in the size of the smaller RanjHl- 5’011 fragment relative to pBA22 and is cleaved once by f’vjrr II. indicating that it, is t,he product of a simple ligation. The small Ha,mH I-%x/l fragment of the PstI- reduced plasmid, pBA36, is also shorter. and the two large f’stl fragments are identical to those of pBA22, again indicating a simple ligat,ion and that no change in the relative orientation of the f’stl fragments has orcurred, since this would result in a gross change in the RrrmHI-Sal1 fragment pattern (Fig. 1).

484 Il. F. BARKER AND A. M. CAMPBELL

Born HI-Bg/II hybrid

I r

PstI

200 -

400 -

600 -

800 -

1000 -

1200-

1400 -

1600-

1800-

2000 -

-0 pBA 203

a pEIA242

PSf I

Pst I

-0 pt3A 237

Pvu II

HimIt

Bgl I

BornHI

pBA36

pBA40

BornHI-Bgll[ hybrid

PSf I Pa-f I

BomHI

PstI

PIG. 4. Restriction site map of pBA22 and the bacterial RamHI-BgZII fragment that includes birA. showing locations of deletions and Tn5 insertions in derivatives of pBA22 and pBAl1, Locations of restriction sites were determined by appropriate single and double digests of pBA22 and pBAl1, and comparison with digest patterns of pHUB4 and pACYC184, respectively. Fragment lengths were det,ermined as described in Materials and Methods. pBA36 and pBA40 are derivatives of pBA22, deleted for the material indicated by the cross-hatched regions. pBA203. pBA221, pBA242 and pBA237 are derivatives of pBAl1 which carry insertions of Tn5 at the indicated sites. Locations shown were determined by subtracting the known lengths from the ends of Tn5 to the first internal I’~11 or f’s/1 sites, 1365 bp and 660 bp, respectively (Rothst,ein et OZ., 1980). from the lengths of the junction fragments generated by these enzymes. The distance from the Tn5 end to the first internal PvuII site was also independently confirmed. Irnits shown for pBA22. upper part of Figure. are kb: and for BumHI-BgZII fragment, lower part of Figure, are bp.

Derivatives of pSAl1 carrying insertions of Tn5 were isolated from pBAl1 DNA prepared from cells that were infected with hNK467 as described in Materials and Methods. DNA from 12 independent preparations was used to transform BM6020 or BM4084 to KmR, and 37 plasmids that carried Tn5 were isolated, including 14 that were unable to complement the birA302 mutation in BM4084 for repression and 23 Bir+ plasmids that could complement. These were screened initially by digestion with RamHI plus Sal1 to determine whether insertions were located in the 2.4 kb fragment which includes the &A gene. All of the 14 Bir- plasmids and six of the Bir+ plasmids lacked this fragment. (Thirteen of the Bir+ plasmid

ROLE OF birA IN bio REGULATION 4%

preparations showed fragments whose sizes and total length indicated the presence of both pBA11 and a T&carrying derivative, possibly as a dimeric structure.) All of the six Bir’ plasmids were independent and non-identical, while the Bir- plasmids defined seven different insertion sites. Analyses of RamHI plus Xhol digests of these 13 unique plasmids permitted relat,ive ordering of the insertion sites. since Xh,oI makes symmetrical cuts in the inverted repeat, portion of Tn,i. eliminating any complication of interpretation due to possible differences in orientation of the inserts. The Bir+ and Bir- insertions were found to be clustered ill two different regions of the bacterial DNA segment, with the Bir- clustt~r proximal to the RwLHI site. The two insertions that define the end of the Bir- cluster as well as two Bir ’ insertions close to one end of that) cluster have been mapped precisely with respect to the PstI and P~u11 sites (Fig. 4).

The complementation properties of the deleted and TnG-carrying plasmids. shown in Table 5, clearly define a single region of DSA in which such lesions completely eliminate plasmid-mediated complementation of all classes of hir*-l

mutations. Both of the deleted plasmids, pBA36 and pBA40, have lost the abilit) to complement the whole spectrum of bird mutations for growth or repression. Measurements of BHS confirm that plasmid-mediated expression of this enzyme is eliminated by these deletions (Table 5). The Tn5 insertions define the limits of the critical region. Plasmids pBL4237 and pBA242 carry insert,ions that destroy all complementation activity, while pBA22 1 and pB.4203 show normal (~ornpl~~mentation. Therefore, either the 3’ terminus of the hiril gene lies &+,wren

TABLE 5

(‘wmplementation propertiex of birA pla,smids with

dqfined deletions and Tn5 ivwrtionst

l’lasmitl pBAdO3 pBAtiI pHAd42 pHA36 pHA40 p&A237 pHA22 I,esim 203 : :Tn5 “21: :Tn5 242: :Tn5 AI’Sll A I’w/ 1 I 237 : :Tn5 None Htspression

r~omplrmantation~ + + - - - - + (:rowth

~~omplernentation~ + + - - - - + Holoenzymc

synt~hrtasell ND7 ND ND 0-03H 0929 ND 6.4

t See Pig. 4 for location of the lesion in each plasmid. 1 Each plasmid was t,ransformed into recA - @(bioFC-lacZJ501 derivatives bearing birA353, birAJOI.

hirA;?OA. hir.430.1. birA215, birARii to test complementation or birA+ as control. Complementation for repression was svoretl by color of KmR transformants on MacConkey indicator plates containing 4-1 ~LW biotin. The pHUH4 plasmid vector was also included as plasmid control (not shown).

- . Indicates no instance of complementation observed : + indicat,es complement,ation of all birA rrrut,ants examined.

$ KmR transformants of the recA- @(bioFClacZ)501, strain with either birAR5 and bir.4215 mutations from the experiment described in j were streaked on MacConkey plates with kanamycin plus either X2 nM or 8.2 p.v-biotin added and the plates were incubated at 43°C”. Neither concentration supports’the growth of the bir-485 mutant. while birA215 grown onlv with the higher biotin supplement,at,ion. - Indicates failure to complement the growth defect ofl either mutation: + both mrrtatitms cotnplemmted,

11 Biotin holoenzyme synthetasr measurements of vultures induced. extracted and assayed as Iltvribed in the legend to Table 4.

’ NI). Sot det,ermined.

486 I). F. BARKER AND A. M. ('AMPBELL

the points of insertion in pBA242 and pBA221 or else that part of its protein product coded by DNA distal to the insertion site in pBA22I is not needed for either BHS or repressor function. The orientation of bir.4 transcription is leftward in Figure 4. The Tn:i in pBA237 must lie in t’he hir.4 gene or between the hir.4 genr and its promoter. If the natural hir.4 promoter is present on pBAl1 and expressed, then it must span or lie to the right of the pB.4237 insert.

(e) In vitro binding of puri;jkd hiotin holomzyyn,r synthetnw to bio operator

To confirm the genetically derived conclusion that the biril product is hifunctional, we have examined the ability of highly purified biotin holoenzyme synthetase to bind specifically to biotin operator DNA. -4 preparation of this enzyme, purified approximately 10,000-fold as described in Materials and Methods, was tested for its ability to protect a site in the DNA immediately adjacent to the bio operator from Tag1 cleavage. Figure 5 shows that specific protection of a Tnql site flanked by Tag1 fragments of length 980 and 240 bp is observed when purified BHS is present. One end of the 240 bp fragment is known to contain the bio operator sequence. DNAs hearing operator constitutive mutations that are single base changes in this sequence are not protected by BHS Tag1 cleavage (Barker rt al., unpublished results), indicating that specific binding of bio repressor to operator is responsible for the observed protection.

Repressor protein binding to operator DNA under these conditions is dependent, on added corepressor. In initial experiments. 50 PM-b&in was found to be weak]) effective, and concentrations of 2 pM and 0.1 pM are less effective and ineffective. respectively (Fig. 6). These concentrations are well above the 1 tlM presumed to be physiologically relevant, assuming a.11 intracellular space of lo- l2 cm3/celI and approximately 1000 biotin molecules per cell in the presence of excess extracellular biotin (Campbell rt al.. 1972). Addition of ATP to the reaction resulted in a dramatic reduction in the minimum biotin concentration required to provide significant protection. At’ a concentration of 2 mM-AT]‘. protection is observable when biotin is present at 10 nM (Fig. 6). The implication of this finding is that biotin and .4TI’ together or the BHS reaction intermediate biotinyl 5’-adenylate may act, as corepressor.

To test this, biotin,vl 5’.adenylate was synthesized and tested for corepressor activity in the protection assay. As Figure 6 shows, this compound alone functions as corepressor with a minimum effective concentration between 50 nM and 0.5 pM.

In control experiments where biotin plus AMP. these being the major contaminants in the biotinyl-AMP preparation, were tested for corepressor activity, no protection was observed (data not shown). These results are in agreement with the findings of Prakash 8r Eisenberg (1979) that highly purified biotin repressor. prepared by a different protocol than we have employed, synthesizes biotinyl 5’-adenylate, and that this compound, at 1 nM, provides effective corepressor function in a DSA filter binding assay of bio repressor. The comparably effective biotin concentration was 1 PM. The apparent discrepancy between the concentrations of biotin and biotinyl- AMP observed to be effective in these two experiments may be due to differences in

1374 1220

980

240

665

,475

-258

-207

PIG:. 5. Protwt.ion of the TnqI site adjacent to biotin operator by purified biotin holoenzymr synthetase. The trasic- rrwtion mixture was as described in Materials and Met,hods. Lltne (1). wmplete: (2), without BHS : (3). 7’nql digest of pDB1 under standard condit~ions: (1). ‘I’rcqI digest of pl)B:! whi1.h carries an inwrtion of ISI in the biotin operator-promoter region; (5) and (6). pBR321 cut with S’ov3.4 and TagI. respwtiwly. I’nder these electrophoresis conditions, 1?,, acrvlamide. the 980 bp 7’qI fragment ccmigr&trs with another fragment. Note the decrease in intensity of this band in lane (I) twsus lane (2).

sensitivity of the respective assays or the binding buffers employed: however. in both cases, it, is clear that’ hiotinyl-AM]’ is greater than lOO-fold more effective as corepressor than biotin.

The effect of added pyrophosphate on specific protection by the bio repressor provides a further indication that the active site which converts hi&in plus ATI’ t,o biotinyl-AMP and pyrophosphate is involved in the synthesis and action of corepressor. At a concentration of 0.1 mM, sodium pyrophosphate inhibits I)rotection when corepressor is supplied by 1 PM-biotin plus 04 mM-ATI’. This inhibition is reversed by raising the biotin and ATl’ concentrations to 2.50 ~CLM and 20 mM. respectively. or by adding pyrophosphatasc to the reaction. The same pyrophosphatr concentration inhibits protection lvhen 5 PM-biotin.yi-AM I’ is supplied as corepressor. and this inhibition is reversed by a Wfold increase in t)iotin>~l-AMP or addition of pyrophosphatase (data not shown). This apparent]?

488 D. F. BARKER AND A. M. CAMPBELL

240

FIG. 6. Corepressor dependence of bio repressor protection of ‘l’nq1 site adjacent t,o bio operator. Reaction conditions as described in Materials and Methods with the following additions to serve as corepressor. Lane (l), 50 PM-bio; (2), 2 @Wbia; (3), 0.1 PM-hio; (4). 2.0 mM-A’I’P: (5). 2.0 mwATP plus 0.1 pwbio: (6), 2.0 miwATP plus 10 nM-bio: (7), 2.0 mM-ATP plus 1 niwbio; (8), 2.0 mwATP plus @l nwbio; (9), 5piwbio-AMP; (IO), 0.5+-bio-.4MP; (11). 50 niwbio-AMP: (12). 5 nwbio-AMP.

specific and reversible effect of pyrophosphatc suggests that displacing the equilibrium of the biotin acylation reaction in favor of biotin plus ATP results in loss of corepressor function.

4. Discussion

We have presented several lines of evidence that a single gene resides at the hir.4 locus of Escherichia coli and that the product of birA functions as the bio repressor and also as the biotin holoenzyme synthetase. Our observations confirm the findings of Prakash & Eisenberg (1979) that biotinyl 5’-adenylate is an effective corepressor in vitro, and therefore possibly the physiologically important corepressor in, viva. Our results are consistent with their finding that biotin operator binding activity copurifies with the enzyme that synthesizes biotinyl 5’-adenylate, biotin holoenzyme synthetase. Below, we review the evidence which indicates that

ROLE OF birA IN bio REGULATION 489

the bir=l gene encodes a multifunctional protein with these activities and discuss the apparent advantages of this regulatory arrangement.

(‘omplementation analysis of nine biril mutations clearly shows that no pair of alleles fully complements to repress biotin operon expression, and indicates that the mutations lie in a single gene. We infer that the nine mutations are different missense changes that incompletely inactivate the bir.4 product rather than null mutations. since each mutant strain possesses unique properties with respect to residual operon repression. biotin growth response, BHS activity and biotin uptake. Although we have not shown that each of the mutations is a single point rhange. the frequency at which they occur (and revert in those cases where revertants may be scored by selecting growth on low biotin) is consistent with that conclusion (unpublished observations). Since the mutations are recessive to a wild-

type birA gene. the failure to observe complementation among them is strong evidence that a single gene is affected.

The apparent cases of intracistronic complementation may reflect the properties of enzyme molecules with combinations of different mutant bir.4 subunits. indicating that the bit-A product forms a multimer. The most striking instances of such complementation involve combinations of the mutations bir=L301 and hir.4352, which each specify BHS with nearly wild-type properties. and the mutations birA91, bir.4215 and bir-4879, which apparently have severe defects in it/ ni~o BHS activit.v as evidenced by their increased biotin growth requirement (Table 2). This may indicate that the birA product contains two domains in a manner analogous to that observed for the lac operon repressor (Ogata & Gilbert. 1979). If this is the case, one domain, affected by mutations biril91, birL421S and birAiR79. may be primarily involved in BHS activity and binding of the effectol molecule biotinyl 5’.adenylate, while the second domain, damaged by mutations bir.4301 and bir.4352, binds specifically to biotin operator. If this is true, then int,rractions between different bir.4 mutant enzyme forms could, in some instances. be mediated by diffusion of biotinyl5’-adenylate from one mutant form where it is synthesized to a second mutant protein that lacks synthetic capacity but ma\ serve effectively as repressor when provided with biotinyl 5’.adenylate.

The properties of strains bearing multicopy plasmids that include the biril gent’ lend further support to the conclusions derived from the complementation data. Three of the four plasmids characterized, pBA2, pBAl1 and pBA22. mediate a high Ie\.el of expression of biotin holoenzyme synthetase activity and also cause superrepression of bio operon expression in Go. indicating an increase in the concentration of bio repressor molecules in cells that contain these plasmids. The fourth plasmid, pBA26, carries the same bacterial genetic material as pBAl1 and pBA22. but the bird gene is apparently transcribed weakly. In cells containing this plasmid. very little BHS is produced from the plasmid-borne bir.4 genes and no superrepression is observed.

The locations of Tn5 insertions and deletions that inactivate all complrmentation activity of the bira4 ’ plasmids are consist,ent with the view that,

t,he co-ordinate expression of repressor and holoenzyme synthetase is not due to co- ordinate transcription of separate genes. The deletion and insertion mutations and tht, HamHI end of the cloned fragment define a 1.6 kb region that must include

490 1). F. BARKER AND A. M. CAMPBELL

coding sequences for both BHS and repressor, since all of the lesions that fall into a @9 kb segment of this region completely eliminate both of these activities while those outside of it do not affect either activity (Fig. I. Table 5). The simplest interpretation of these results that is consistent with the complementation data is that the BHS and repressor coding sequences are completely overlapping, although the possibility that the genes overlap extensively but not completely cannot be formally ruled out.

Our data also do not entirely eliminate the possibility that BHS and repressor are determined by non-overlapping genes of the same operon. We have shown that all of our hirrl mutations lie in a single complementation group, and that some of these mutations (bird361 and hirA92) increase the thermosensitivity of BHS. Amplification of the biril gene enhances production of BHS. Repressor activity copurifies with this BHS. Prakash &, Eisenberg (1979) showed that amplification of the hir=l segment in AdrifdIR ~I857 Sam7 enhances production of repressor, and that purified repressor can synthesize biotinyl 5’.adenylate. These results indicate that, if the repressor is coded by a separate gene, the repressor must be either very similar to BHS or closely associated with it. Because we have not demonstrated that any of our mutants forms an altered repressor protein. it is technically possible that all our b&-A mutations could be polar on a downstream gene that codes for repressor. However, such a gene would have to lie entirely between the sites of Tn.5 insertion in pBA242 and pBA221 (Fig. 4), a distanctl of about 260 nucleotides, because the insertion in pBA242 prevents BHS synthesis. whereas that in pBA221 allows formation of functional repressor.

Final verification of the nature of the bir.4 product must await more precise delineation of the boundaries of the gene, or demonstration of the homogeneity of a completely pure preparation. For the remainder of this discussion, we will ignore this uncertainty and assume that a single polypeptide has both BHS and repressor activity.

The dual function of the birA product as a regulatory and catalytic element apparently provides E. coli with an economical and sensitive mechanism for control of biotin biosynthesis. Accumulation of the corepressor molecule? biotinyl 5’. adenylate, can occur only when there are relatively few unbiotinylated acetyl-(‘oA carboxylase molecules in the cell, since, in the presence of this substrate, the biotin moiety will be transferred to it and AMP released. Therefore, repression of the biotin biosynthetic enzymes is directly sensitive to the cellular need for biotin as measured by the concentration of unbiotinylated acetyl-CoA carboxylase. (If synthesis of the carboxylase proteins is regulated by a system sensitive to the rate of cell growth, then biotin biosynthesis could be responsive to this rate without requiring any additional specific regulatory mechanism.) Of course. since biotin itself is required for biotinyl 5’-adenylate synthesis, the regulatory mechanism is also sensitive to the availabilit,v of free biotin.

If we assume that the binding of biotinyl-AMP to the catalytic site involved in biotin holoenzyme synthetase activity is necessary and sufficient to permit binding of the repressor to operator, then it is probable that the corepressor is an enzymr- bound intermediate. Prakash & Eisenberg (1979) have shown that biotinyl 5’- adenylate is very tightly. but not, covalently, bound to the site of its synthesis. The

ItOLE OF bir.4 IN bio REGULATION 49 I

evident advantage of this arrangement is that little or no ATI’ utilization is necessary to maintain an intracellular pool of biotinyl 5’-adenylate, since free diffussion is not required for corepressor to reach its site of action.

It is interesting to compare these features of biotin operon regulation with the mechanism of attenuation that has been shown to be important in the regulation of many amino acid biosynthetic operons (Lee et nl.. 1978: Zurawski et al.. 1978: ,Johnston it al.. 1980). One of the most striking features of this mechanism is that the polypeptide encoded in the leader region of a particular operon always contains a greater than statistically expected proportion of the amino acid or amino acids s>-nthtasized by the enzymes of that operon. The successful incorporation of the given amino acid into the hypothetical leader peptide is therefore a direct indication of the availability of the appropriately charged transfer RNA. In all cases, thtb postulated leader peptide translation negatively affects transcription of the operon. The analogy with biotin regulation lies in the fact that the critical taffector molecule is, in each case, an intermediate in the reaction which may be considered the final step of biosynthesis, incorporation of the biosynthetic product into it.s site of action within the cell. It is interesting to ask to what extent this apparent analogy reflects the competitive success in evolution of regulatory mechanisms that respond to such intermediates in contrast to the possibility that activating enzymes may have been the first biosynthetic enzymes elaborated in thtb (xourse of evolution and that early, possibly autogenous. regulatory systems involving these enzpmes and their products were later modified to accommodate thta regulation of newly acquired biosynthetic genes.

We thank M. (‘asadaban, K. Shaw, N. Kleckner, R. Cunsalus and (:. Zurawski for strains, phage and plasmids. One of the authors (D.B.) thanks R. Gunsalus for advice 011 protein purification and recombinant DNA technology and stimulating discussions. We are indebted to (‘. Yanofsky for advice and access to a wide variety of restriction nucleases and (icluipment. This research was supported by grant AI-08573 from the National Institute of Allergy and Infectious Diseases. One of us (D.B.) was a predoctoral fellow of the Xational Science Foundation.

REFERENCES

Barker. D. F. & Campbell, A. M. (1980). J. Buctwiol. 143, 789-800. Barker. D. F. 8r Campbell, A. M. (1981). J. Mol. Riol. 146, 451-467. Bolivar. F., Rodriguez, R. L., (Greene, P. J., Betlach. 11. C.. Heyneker, H. I,., Boyer, H. W.,

(‘rosa. J. H. & Falkow, S. (1977). Geftr 2, 95-l 13. Boros, I. &) Sain, B. (1977). Mol. Biol. hp. 3, 451-457. (‘ameron. J. R.. Philippsen, P. & Davis, R. W. (1977). Sucl. Acids Res. 4, 1429-1448. (‘ampbell, A., del Campillo-Campbell. A. & Chang. R. (1972). Proc. Sat. dead. Sci.. IT.A”..J

69, 676-680. (‘ampbell, A., Chang. R., Barker, D. & Ketner, (:. (1980). J. Bactrriol. 142, 102.51028. (‘hang, A. C”. Y’. & Cohen, S. N. (1978). J. Bacterial. 134, 1141-11~56. (‘hristner, J. E. & Coon, M. J. (1970). In M&wds in Entymology (McC’ormick, D. B. &

Wright, L. E., eds), vol. 18, pp. 38ct390, Academic Press, New York. Dagert, M. 8: Ehrlich. S. D. (1979). Gene 6, 23-28.

Daniels. D. L., deWet, J. R. & Blattner, F. R. (1980). J. lpirol. 33, 390-400. Dugairzyk, A.. Boyer. H. W. & Goodman, H. M. (1975). J. Mol. Rio/. 96. 171-184.

492 I). F. BARKER AND A. M. CAMPBELL

Greene, P. J., Heyneker, H. L., Bolivar, F., Rodriguez, R. L., Betlach, M. C., Covarrubias. A. A., Backman, K., Russell, D. J., Tait, R. & Boyer, H. W. (1978). Xucl. Acids Res. 5, 2373-2380.

Guerry, I’., Leblanc, D. J. & Falkow, S. (1973). J. Ructeriol. 116, 1064-1066. Johnston, H. M., Barnes, W. M., Chumley, F. (i., Bossi. L. & Roth, J. R. (1980). Proc. Sat.

Acad. Sci., U.S.A. 77, 508512. Ketner, G. & Campbell, A. (1975). J. Mol. Riol. 96, 13-27. Lee, F., Bertrand, K., Bennett, G. & Yanofsky. C. (1978). J. Mol. Riol. 121, 193-217. Lindahl, L., Yamamoto, M., Nomura. M., Kirschbaum, *J. B., .41let, B. & Rochaix. J.-D.

(1977). J. Mol. Biol. 109, 23-47. Miller, J. H. (1972). Experiments in Molecular Genetics, Cold Spring Harbor. New York. Ogata, R. T. & Gilbert. W. (1979). J. Mol. Biol. 132, 709-728. Otsuka, A. & Abelson, J. (1978). Suture (London), 276, 689-694. Pai, C. H. & Yau, H. (1. (1975). Cunad. J. Microbial. 21, 11161120. Peacock, A. C. & Dingman, C. W. (1968). Biochemistry, 7, 668-674. Prakash, 0. & Eisenberg, M. (1978). J. Bucteriol. 134, 1002~1012. Prakash, 0. & Eisenberg, M. A. (1979). Proc. Sat. -4cad. Sci.. r’.S.=l. 76, 5592-5595. Rossi, J. J. & Landy, A. (1979). CeZl, 16, 523-534. Rothstein. 6. J., Jorgensen, R. A., Pestle, K. & Reznikoff, W. S. (1980). Cell, 19, 79Ft805. Sutcliffe, J. G. (1978). N&. Acids Res. 5, 2721-2728. Taylor, W. E. & Burgess, R. E. (1979). Oenje, 6, 331-365. Ulrich-Bernard, H., Remaut, E., Hershfield, M. y’., Das, H. K.. Helinski, D. R., Yanofsky,

(‘. & Franklin. N. (1979). Gene, 5, 59-76. Ward, D. F. & Murray, N. E. (1979). J. Mol. Biol. 133. 249-266. Yamamoto, M. & Nomura, M. (1979). J. Bacterial. 137, 584-594. Zurawski, G., Brown, K., Killingly, D. & Yanofsky. C. (1978). Proc. Sat. ;Icad. Sci., fi.S.3.

75, 427 l-4275.