A Rapid, Efficient Method for Purifying DNA-binding Proteins

THEJOURNAL OF BIOLOGICAL CHEMISTRY (0

Vol. 266, No. 14, Issue of May 15, pp. 9153-9160, 1991 Printed in U.S.A.

1991 by The American Society for Biochemistry and Molecular Biology, Inc.

A Rapid, Efficient Method for Purifying DNA-binding Proteins DENATURATION-RENATURATIONCHROMATOGRAPHY EXTRACTS*

OF HUMANANDYEASTMITOCHONDRIAL

(Received for publication, October 11, 1990)

Robert P. Fisher$, Thomas Lisowskys, Gail A. M. Breenll, and David A. Clayton From the Departmentof Developmental Biology, Stanford University School of Medicine, Stanford, California 94305-5427

We describe a novel method for the purification of DNA-binding proteins. Isolated mitochondria were lysed in boiling sodium dodecyl sulfate-containing buffer, the extracts werechromatographedon hydroxylapatite in the presence of sodiumdodecyl sulfate, and DNA-binding activities wereidentified after adding a large excessof nonionic detergent (Triton X-100) and assaying fractions by a gel retardation procedure. Fractions containing DNA-binding activity were bulk renatured and chromatographed on phosphocellulose in the presence of Triton X-100. When applied to human mitochondria, the technique resulted in the purification to homogeneity of fully functional mitochondrial transcription factor 1 (mtTFl), the major activator of mammalian mitochondrial transcription. Moreover, the yield of mtTFl purified by this method was at least 25 times higher than that obtained by conventional nondenaturing chromatographies. When yeast mitochondria were subjected to the same protein isolation scheme, a 19-kilodalton putative yeast homologue of mtTFl was purified to homogeneity. These results suggest that the denaturation-renaturation approach may bea valuable general method for the identification and high yield purification of DNA-binding proteins.

An unusual featureof the factor is its affinity for DNA lacking specific binding signals; binding t o promoter DNA may be distinguished on the basis of stabilization of mtTFl-DNA complexes by contactswith specific basesinthemtTF1binding sites (2,3). Analysis of these contacts, in both human and murine mtTF1-promoter complexes, by methylation interference, revealed differences between the binding determinants recognized by the two mammalianfactorswithin their respective homologous DNA targets. However, heterologous bindingandtranscriptionalactivationexperiments demonstrated remarkable flexibility in sequence recognition. Methylation interference analysis also indicated that mtTFl contacts sites in both major and minor grooves in roughly equalnumbers,unlikemostDNA-bindingproteins, which interact predominantly or exclusively with major groove determinants, and also in contrast to the minority that exhibits a preference for the minor groove (4,5). T o determine whether such unorthodox behavior reflects variation on apreviouslydescribed mode of protein-DNA interaction, or indeed a fundamentally different mechanism, will require direct structural analyses both of mtTFl and of the nuclear gene that encodes this mitochondrial transcriptional regulator. A prerequisite for these investigations is the ability to obtain large amounts of pure mtTFl protein. Unfortunately, while our previously published method of isolating m t T F l from human tissue culture cells by conventional ion exchange chromatographies resulted electrophoretically in pure preparations(l),the yields were typically 50%) of m t T F l activityafter gel isolation (1). ance with 18 U.S.C. Section 1734 solely to indicate this fact. j: Supported by Medical ScientistTrainingProgramGrant While electrophoresis inpolyacrylamide gels is a precise way GM07365-14from the National Institute of General Medical Sci- to analyze proteinsdenatured insodium dodecyl sulfate ences. (SDS), it is not amenable tolarge scale preparative purifica5 Supported by a postdoctoral fellowship from thescience commit- tion. To achieve that objective, we have employed hydroxyltee of theNorthAtlanticTreatyOrganization via theDeutsche apatite chromatography in the presenceof SDS, a technique Akademische Austauschdienst (DAAD). ll Supported in partby Grant GM41738 from the National Institute that has been shown to separate proteins and polypeptide of General Medical Sciences. Permanent address: Molecular and Cell subunits not easily separated by other means (6). Coupled Biology Programs, The University of Texas a t Dallas, Richardson, with direct lysis of isolated mitochondria into boiling SDS, T X 75083-0688. hydroxylapatite chromatography effected the separation of

’ The abbreviations used are: mtTF1, mitochondrial transcription factor 1; SDS, sodium dodecyl sulfate; LSP, light-strand promoter; HSP, heavy-strand promoter; DTT, dithiothreitol; PMSF, phenylmethylsulfonyl fluoride.

R. P. Fisher, T . Lisowsky, G. A. M. Breen, and D.A. Clayton, unpublished results.

9153

9154

Purification of DNA-binding Proteins

Purification of Human mtTFl by Denaturation-Renaturation Chromatography SDS-Hydroxylapatite Chromatography-Mitochondria were isolated from human KB tissue culture cells bythe sucrose step-gradient procedure of Bogenhagen and Clayton (9). After recovery of purified mitochondria by centrifugation, the isolatedorganelles were washed once or twice with ice-cold 0.1 M NaP04, pH6.8,0.25 M sucrose, 15% glycerol, 1 mM dithiothreitol (DTT) and pelleted by centrifugation (15 min at15,000 rpm in a Beckman JA-20 rotora t 4 "C). Mitochondria were then resuspended in -50 ml of boiling 0.1 M NaP04, 2% SDS, 20 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride (PMSF). The volume of resuspension was determined by the amountof protein present in the mitochondrial fraction andwas calculated to result in an -10-fold mass excess of SDS to protein. The mitochondriawere gently mixed until a clear solution was obtained, boiled for 5 min, diluted 10-fold with 0.1 M NaP04, 1mM DTT, 0.5 mM PMSF (minus SDS), and loaded onto a hydroxylapatite (Bio-Rad) column equilibrated in 0.1 M NaP04, 0.1% SDS, 1 mM DTT. Best results were obtained when 100 mg or less of protein were applied to an -60-ml column. The column was washed extensivelywith 0.1 M NaP04, 0.1% SDS, 1mM DTT, and proteins were then eluted witha linear gradient of 0.1-0.5 M NaPO, in 0.1% SDS, 1mM DTT. Fractionswere assayed for conductivity and protein concentration (10). and mtTFl activity was identified by assaying fractions, diluted 10-fold with abuffer containing 2% Triton X-100,9 mM Tris-C1, p H 8.0, 45 mM NaC1,O.l mM EDTA, 45% glycerol, 1 mM DTT, for DNA binding using the gel retardation method. Renaturing Phosphocellulose Chromatography-Fractions containing mtTFl activity eluted from hydroxylapatite in the presence of SDS were pooled and diluted 3-fold with ice-cold buffer containing 4% Triton X-100,20 mM Tris-C1, p H 8.0,0.2 mM EDTA, 2 mM DTT, 0.25 mM PMSF, 15% glycerol, and loaded onto a 25-ml phosphocellulose (Whatman) column equilibrated in buffer A (10 mM Tris-C1, pH 8.0, 0.1 mM EDTA, 1 mM DTT, 7.5% glycerol, 0.1% Triton XMATERIALS ANDMETHODS 100) containing 0.1 M NaC1, a t a flow rate of 20 ml/h. The column was washed extensively with buffer A containing 0.1 M NaC1, and proteins were eluted with a linear 0.1-1.0 M gradient of NaCl in DNA Templates for in Vitro DNA-binding and TranscriptionAssays buffer A. Fractions were assayed for conductivity, protein concentraThe construction of recombinant templates containing the light tion, and forDNA binding by gel retardation. The mobility shift (L)-strand andheavy (H)-strand promoters (LSP and HSP) of human characteristic of DNA binding by mtTFl was produced by fractions mtDNA, and the preparation of runoff transcription templates and eluted with anionic strength corresponding to 0.64-0.70 M NaC1. For end-labeled fragments to assay DNA binding from these clones, have long term storage and analytical assays of mtTFl activity, the peak been previously described (1-3,7,8). Briefly, runoff transcription was fractions were pooled and dialyzed against buffer B (10 mM Tris-C1, programed either by plasmid pKB741SP, containing both the LSP p H 8.0, 50 mM NaCl, 0.1 mM EDTA, 1 mM DTT, 50% glycerol), and and HSP, linearized with EcoRI, or by plasmid H5'A-60, containing kept at -20 "C in the presenceof 5 mM DTT. SDS-Polyacrylamide Gel Purificationand Analysis of mtTFla n intact HSP only, also linearized with EcoRI. The LSP-containing fragment used in both gel retardation and DNaseI footprinting assays Recovery of mtTF1 from denaturing polyacrylamide gels and renawas the BamHI-Ball fragment of plasmid L5'A-56, labeled with 32P turation of activity following denaturation by 6 M guanidine hydrochloride were carried out as previously described (1). For some experat the BamHI site, -56 base pairs upstream of the transcriptional iments, an additional concentration of mtTFl protein purified from initiation site. polyacrylamide gels was achieved by centrifugation in a Centricon 10 A labeled DNA fragment containing the yeast mtDNA origin of (Amicon) ultrafiltration apparatus for -2 h at 6000 rpm ina Beckman replication,ori5, with its associated transcriptional promoter, was JA-20 rotor.Foranalysis of chromatographicfractions by SDSderivedfrom plasmidH3ori5(thekind gift of R. A. Butow) by polyacryamide gel electrophoresis, we used the method of Laemmli digestion with AuaII and HindIII, and used for gel retardation and (ll),followed by Coomassie Blue stainingof gels to visualize protein DNase I protection assays of DNA binding by yeast mitochondrial bands. proteins. Fractionation of Yeast Mitochondria by Denaturation-Renaturation Chromatography Assays of DNA-binding andPromoter-specific Transcription Mitochondria were isolated from 8 liters of exponentially growing DNA binding was detected either by the gel retardation assay or Saccharomyces cereuisiae of the triple protease-deficient strain, BJ926 by protection of end-labeled promoter-containing fragments from (correspondingto 12g of cells, 1 g of mitochondria, -50 mg of DNase I digestion, as previouslydescribed (1-3). Toquantitate mitochondrial protein), by previously described methods (12). The mtTF1 activity, we performed gel retardation experiments, including isolatedorganelleswere then lysed in boiling SDS solutionand autoradiography of dried gels to localize unbound free fragment and fractionated exactly asdescribed above for human tissue culturecell complexes of mtTFl and the LSP-containing fragment. Both bands mitochondria. The yeast protein pl9/HM elutes from hydroxylapatite were excised from the gel; radioactivity in each was determined by with a NaP04 concentration of -0.33 M and from phosphocellulose scintillation spectrometry in nonaqueous liquid, or quantitation was with a NaCl concentration of -0.58 M. by the use of a gel scanner. One unit of DNA-binding activity is defined as the amount of mtTFl required to shift 1 fmol of labeled RESULTS fragments at factor/DNA ratios that result in most of the fragments MitochondrialTranscriptionFactor 1 Is a Major DNAremaining unshiftedwhile a small fractionundergoes a singlebinding binding Protein of Human Mitochondria-Mitochondrial TF1 event. I n vitro transcription was carried out as previously described has been purified to homogeneity from human (1)or mouse (1-3).

m t T F l from the bulk of SDS-denatured mitochondrial proteins without themajor losses incurred during nondenaturing extractionandpurification procedures. Chromatographic fractionscontaining m t T F l activity were identifiedafter using the nonionic detergent Triton X-100 in large excess to remove SDS from its protein-bound state and sequester it in mixed detergent micelles. Most remarkably, following bulk renaturation of hydroxylapatitechromatographicfractions containing mtTF1 by Triton X-100 addition, conventional phosphocellulose chromatography could be used t o purify the factor essentially to homogeneity in a single step.When applied to SDS lysates of yeast mitochondria, the same sequence of chromatographies purifies to homogeneitya 19kDa protein with striking similarities to mtTF1. Human mtTFlpurified by this sequence of denaturing and renaturing chromatography steps retains all of the functions previously ascribed to the factorpurified under native conditions, includingsequence-specific DNAbindingandtranscriptional activation, aswell as DNA bending and wrapping activities recently associated with mtTFL3 Careful quantitation of the yield of 25-kDa mtTF1 protein reveals at least a 25-fold improvement over the previously published procedure. Moreover, it indicates that mtTFl isa moderately abundant protein of mammalian mitochondria thatmay play an important structuralrole in the physical organization and topology of mtDNA as well as a key regulatory role in the control of gene expression.

(3)4 mitochondrial extracts by a sequence of ion exchange

R. P. Fisher, T. Lisowsky, G. A. M. Breen, and D. A. Clayton, submitted for publication.

M. A. Parisi and D.A. Clayton, unpublished results.


9155

TABLEI Recovery of mtTFl by conventional ion-exchange chromatography step Purification Recoveryb Yield”

exposure to the ionic detergent SDS; the majority of mtTFl molecules could reform the active conformation(s)needed for promoter recognition, binding, and transcription activation following treatments known todisruptproteinsecondary units % structure and folding. Even more encouraging was the demCrude Triton X-100-KC1 lysate 4532 onstration that “renaturation”of m t T F l DNA-binding activ4727 104 S-130 502 11.1 ity after boiling in SDS, as measured by the highly sensitive DEAE-Sephacel pool 112 2.5 Phosphocellulose pool (but nonspecific) gel retardationassay,didnot requirea 10.3 0.2 MonoQ FPLC pool guanidine hydrochloride denaturation-renaturation step, but 1 unit = 1 fmol of DNA fragment bound in 1:l complex; starting could be achieved quantitatively by adding a large (5-fold) material equals 20 liters of human KB cells a t 6 X 105/ml = 1.2 X excess (over the concentrationof SDS) of the nonionic deter10’” cells. gent Triton X-100 to the SDS-denatured protein. I, Relative to starting material. We therefore reasoned that purification under denaturing conditions would have the advantage of minimizing lossesdue chromatographies (DEAE-Sephacel,followed by phosphocel- to protein-protein, protein-nucleic acid and, conceivably, prolulose, followed by MonoQ fast performance liquid chroma- tein-lipid interactions (Fig. 1). Moreover, any proteolysis or tography). A typical purification starting with50-70 liters of denaturation due to extensive manipulation in dilute solution tissue culture cells yields 1-5 pg of pure mtTFl protein. This could be reduced. Rapid lysis of isolated mitochondriain material was used to determine the amino-terminal sequence boiling 2% SDS, followed by chromatography on hydroxylin of human mtTFl(NHz-SSVLASXPKKPVSSYLRFSKE)5 apatite columns in the presence of 0.1% SDS afforded a rapid, hopes of designing a complementary oligonucleotide probe for efficient means to separate mtTFl from the ofbulk denatured screening cDNA libraries. However, the sequence obtained is mitochondrial proteins that moreover could be performed composed almost entirely of amino acid residues for which entirely a t room temperature. After elutionof proteins witha the DNA codons are extremelydegenerate. linear gradientof NaPO,, aliquots of each fractionwere mixed Alternative strategies to obtain design or a molecularprobe with Triton X-100 and assayed for DNA-binding activity in for the mtTFl cDNA gene or sequence were needed however, a gel retardation assay using as a probe a ”P-labeled DNA the quantitiesof m t T F l generated by the previously published LSP. A DNAfragment containing the human mitochondrial method were severely limiting from botha practical and cost binding activity that produces mobility shifts characteristic viewpoint. We attempted to quantitate mtTFl present at the different stages of purification, with the aim of optimizing of m t T F l reproducibly elutes with a N a P 0 4 concentration of extraction and purification efficiency. It was possibleto detect 0.30-0.33 M,whereas themajority of mitochondrial protein is column untiltheNaP04concentration m t T F l activity in unfractionatedhigh salt nonionic detergent retainedonthe M.’ Indeed, when careistakennotto reaches 0.35-0.40 lysates of human mitochondria using the gel retardation assay without including any nonspecific competitor DNA to inhibit overload thehydroxylapatite column (see“Materialsand Methods”), virtually all mitochondrial proteins are retained or bind nonspecific DNA-binding activities.’ This in turn at enabled us to determine the amount of mtTFl present at this 0.1 M NaP0, and completely removed by elution with 0.5 and subsequent stagesof purification (Table I). The yield of MITOCHONDRIA m t T F l at the phosphocellulose-chromatography stage of purification is roughly 1-10% of the starting activity. Further losses during MonoQ fast performance liquid chromatography are variable but may be as high as 50-90%. Theseresultshavetwo majorimplications. On the one SDS-lysate hand, our ability to detect DNA binding by mtTFl in total lysates of mammalianmitochondriawithoutinterference from nonspecific DNA-binding proteins (andindeed, without Hydroxylapatite as inclusion of competitor DNA, taking any precautions, such against such interference) together with the revised higher 0.1-0.5 M Napi estimates of theabundance of m t T F l derived fromthis analysis, indicate that mtTFl aismoderately copious protein inmammalianmitochondria,presentinmultiple copies/ mtTFl (-0.3 M Napi ) mtDNA genome (see below), and may in fact be the major DNA-binding proteinof the organelle. The second conclusion is that our previously published purification scheme (1)results Phosphocellulose in theloss of the bulk of mtTFl present at the outset. Purification of mtTFl underDenaturing Conditions-A 0.1 -1.O M NaCl strategy toimprove the yield was suggested by mtTFl’s resilience in denaturation-renaturation experiments.T o confirm that mtTFl activity resided in the major 25-kDa polypeptide rntTFI (-0.7 M NaCI) species, we have previously employed the method of Hager FIG. 1. Purification of human mtTFl by sequential denaand Burgess (13), involving denaturation with SDS, electro- turing and renaturing chromatographies. Mitochondria isolated phoresis in SDS-polyacrylamide gels and extraction from the from human KB tissue culture cells by t,he method of Bogenhagen and Clayton (9) were lysed in boiling SDS-buffer (see “Materials and gel, denaturation with guanidine hydrochloride, and subsequent renaturation by dilution. Recovery after this cycle of Methods”), and the lysatewas chromatographed on hydroxylapatite denaturation and renaturationwas >50%. Clearly the protein columns. The hulk of mtTFl activity, as measured by the gel retardation DNA-binding assay, is eluted with 0.30-0.33 when is quite resistant to physical perturbations and particularlyto these fractions arepooled, renatured by the additionMofNaP04; the nonionic

1 1 1 i 1

Parisi, M. A,, and Clayton, D. A. (1991) Science, in press.

detergent Triton X-100, and chromatographed on phosphocellulose, mtTFl elutes with 0.64-0.70 M NaCI.

Proteins DNA-bindingof Purification

9156

M NaP04, yet no other proteins capable of binding the LSPcontaining fragment with sufficient stability to cause a mobilityshiftaredetected.This could indicate a failure to renature other activities or their true absence from isolated mitochondria.Theseexperiments were carriedoutinthe presence of competitor DNA (poly dI-dC); when competitor was omitted, no additional discrete mobility shifts are seen, but on occasion we have observed the formation of very large complexes or aggregates containing thelabeled fragment that do not enter thegel. When nondenatured mtTF1-containing protein fractions are chromatographed on phosphocellulose, mtTFl elutes with a salt (either NaCl or KCl) concentration of -0.64-0.70 M (1, 8). When the hydroxylapatite fractions containing the bulk of mtTFl activity detected by gel retardation are pooled and renatured by the additionof ice-cold buffer containing Triton X-100 and chromatographed onphosphocellulose in the presence of the nonionic detergent, virtually identical chromatographic behavior is observed. Fig. 2 is a representative SDSpolyacrylamide gel from thetwo-column purification method. The initial crude SDS-lysate of total mitochondria (40 pg) was electrophoresed in lane I; the pooled SDS-hydroxylapatite fraction containing mtTFl (40 pg) was analyzed in lane 2 and compared with an equal amount of protein that flowed M

1

2

3

4

"25kDa

through thephosphocellulose column (lane 3 ) . Analysis of the fraction containing the peakof mtTFl after chromatography on phosphocellulose (20 pg) reveals a single major polypeptide species of -25 kDa, the apparent molecular mass of mtTFl (l), and thevariable appearance of some minor contaminant species of lower molecular mass (lane 4 ) . Some purifications have resulted in preparations that arefree of these contaminants and that appear to be electrophoretically pure even when gel lanes areoverloaded and/or stained withsilver (data not shown). T o confirm that the 25-kDaspeciespurified nearlyto homogeneity by sequential chromatography on hydroxylapatiteand phosphocellulose was authenticmtTFlandthus identical to the protein purifiedby the three-column nondenaturing procedure ( I ) , we subjected it to both structural and functional tests. The amino-terminal sequences of the 25-kDa proteins purifiedby the two protocols were identical. Gel isolation followed by guanidine hydrochloride denaturation and renaturationassigned to the25-kDa species DNAbinding and transcriptional function indistinguishable from that of the mtTFl purified by the earlier method (see below). The two methods differ in one important respect, however. An estimate of the yield of mtTFl protein purified by denaturation-renaturation chromatographywas derivedby electroeluting the 25-kDa band shown in Fig. 2, lane 4, and performing acid hydrolysis and amino acid analysis on the pure protein. From a gel lane containing a total of 20 pg of protein as measured by a dye-binding assay (lo), and representing approximately 4.2 x 10" human tissue culturecells, 2.4 pg could be recovered from this gel slice. Significant losses probably occur in theelectrophoresis, staining, andelectroelution steps; this representsa minimal estimateof yield and purity.Nevertheless, the denaturation-renaturation scheme has resulted routinely in an-25- to 50-fold improvement in theefficiency of m t T F l recovery over the nondenaturing method. Moreover, the abundanceof mtTFl in human KBcells can be estimated to be a t least 15 protein monomers/l6.5-kilobase mitochondrial genome, based on an mtDNAcopy number of -104/cell (9). Denaturation-Renaturation Chromatography as a General

Method for Purifying DNA-binding Proteins: Identification of a Putative Yeast Homologue of mtTF1-Although the denaturation-renaturation chromatographic procedure solved the problem of purifying mtTFl insufficient quantitiesfor internal peptide fragment sequencing, the general applicability of the method remained be totested. We have taken two separate approaches to thisquestion. First, we have applied the method to other proteins. Not surprisingly, mtTF1also canbe isolated FIG. 2. Denaturation-renaturation chromatography:analy- from mouse tissue culture cell mitochondria by the method sis by SDS-polyacrylamide gel electrophoresis. Fractions from described here: Moresignificantly, we have detectedthe different stages of purification were analyzed by electrophoresis in DNA-binding activity of a mitochondrial transcription ter12% polyacrylamide gels, with visualization by Coomassie Blue staining. 40 pg of the unfractionated mitochondrial lysate were electro- minationfactorinSDS-hydroxylapatitechromatographic fractions upon TritonX-100 renaturation! phoresed in lane 1; chromatography on hydroxylapatite in the presence of SDS produces a fraction enriched in lower molecular weight These investigations havefocused on previously characterpolypeptides (lane 2 ) , including mtTFl (arrow). When this fraction ized proteins; a more critical questionwas whether this stratis bulk renatured by the addition of Triton X-100 and loaded onto a egy could identify and purifypreviously unrecognized factors phosphocellulose column, virtually all the protein flows through, as or DNA-binding activities. Toward this end, isolated yeast indicated by comparison of lane 2, containing 40 pg of the pooled mitochondria were subjected to boiling SDS lysis and subsehydroxylapatite fractions that constitute the phosphocellulose load, quent denaturation-renaturation chromatography(Fig. 3), in and lane 3, containing40 pg of protein that flowed through the phosphocellulose resin a t 0.1 M NaCI. The major difference between the hope of identifying a functional homologue of mtTFl or of a band migratingwith an apparent other proteins capable the two fractions is the depletion of interacting with the yeast transcripM , of 25,000 upon adsorption to phosphocellulose. Mitochondrial tional promoter contained within the replication origin seTF1 elutes from phosphocellulose with 0.64-0.70 M NaCI; 20 pg of quence, ori5 (14). WhenSDS-hydroxylapatitechromatothe pooled peak fraction are electrophoresed in lane 4 . Molecular mass marker proteins,with the sizes in kDa indicatedby the numbers graphic fractions were assayed for binding to a labeled DNA closely resema t left, were electrophoresed in lane M . The 24-kDa marker protein, fragment containing this promoter, an activity PMSF-treated bovine trypsinogen,consistentlymigrateswith apparent M , of -28,000 in our gel system.

an

'J. F. Hess and D. A. Clayton, unpublished results.

Purification of DNA-binding Proteins I

2

9157

3

67.

29

20-19kOa .IC0

14-

4

""'imnn?

,

,

:

,

n , g >I,.. met, I, 511

,

1 *

:,

.,

:

:.

:

:.

.". .

,

1 r,!91HtA rnOD,'!" I h d i

FIG. 3. Denaturation-renaturation chromatography of yeast mitochondrial lysates identifies a puwaspurified by a modification of the tative mtTFl homologue. A, a yeastprotein,designatedpl9/HM, denaturation-renaturation chromatography procedure designed to purify mammalian mtTF1. Isolated mitochondria were lysed directly in boiling 2.0% SDS, 0.1 M NaP04, pH 6.8, 1 mM DTT. The lysate was chromatographed on hydroxylapatite in the presence of SDS, and fractions were assayed for DNA-binding activity, after addition of a large excess of Triton X-100, using a gel retardation assay. The fractions containing the bulk of DNA-binding activity were pooled, renatured by Triton X-100 addition,andchromatographedon phosphocellulose in the presence of 0.1% Triton X-100. Fractions containing DNA-binding activity were identified bygel retardation assays andpooled. Fractions from all stages of purification were subjected to 15% polyacrylamide gel electrophoresis in the presence of SDS; proteins were visualized by staining with Coomassie Blue. The unfractionated SDS lysate was electrophoresed in lane 1; the pooled fractions containing DNA-binding activity after chromatography on hydroxylapatite and phosphocellulose were analyzed in lanes 2 and 3,respectively. Molecular weight markers are indicated a t left. B, fractionation of yeast mtDNA-binding activities by denaturation-renaturation chromatography. Fractions eluted from hydroxylapatite were assayed for DNA-binding activity usinga labeled fragment containing the yeast mtDNA replication origin oris. The fractions pooled and electrophoresed in A, lane 2, correspond to fractions 39 and 40, eluted with 0.32-0.35 M NaPO,. An additional DNA-binding activity eluted a t higher ionic strength (0.36-0.43 M NaPO,) in fractions 42-46. C , fractions eluted fromphosphocellulose in thepresence of 0.1% Triton X-100 were assayed for DNA-binding activity as in panelR. Peak fraction 21, eluted with 0.58 M NaCI, was analyzed for protein content in lane3 of panel A.

bling mtTFl was identified (Fig. 3B, fractions 39-41). Sub- result is that mtRNA polymerase and another activity nearly sequent chromatography onphosphocellulose in the presence or exactly cochromatograph on hydroxylapatite, but are sepof Triton X-100 (Fig. 3C) results in the purification to near arated upon phosphocellulose chromatography, and thereby homogeneity of a 19-kDa protein (Fig. 3A, lane 3). Probing a lose their ability to bindDNA. nitrocellulose blot of the fraction electrophoresedin lane 3 of Functional Characterization of mtTF1 Purified by DenaturFig. 3A with a radioactive oris DNA fragment confirms the ation-Renaturation Chromatography-A second crucialtest of DNA-bindingfunction of the19-kDaprotein; however, the purification was the demonstration that mtTFlpurified DNase I protection assays failed to reveal specific binding to after SDS denaturation retainedall functions of the nondenoris.' Nonetheless, structural and functional homology with atured factor.A prerequisite for thepurificationstrategy mtTFl is suggested by the demonstration that the two pro- outlined above was the restoration of DNA-binding activity teins have virtually indistinguishable amino acidcomposiby Triton X-100 addition; we were able to show that this tions and are both capable of introducing negative supercoils restoration was quantitative by comparing the DNA-binding into relaxed closed circular DNA molecules. Moreover, we activity of an mtTFl preparation both before and after an haveidentified specific binding sites for the yeast protein SDS denaturation-TritonX-100 renaturation cycle.2 Furtherwithin both human and mouse mtDNA transcriptional con- more, the specific activities of mtTFl isolated byconventional trol regions.3 or denaturation chromatographies were similar(Table 11). In the chromatographic profile of DNA-binding proteins Fig. 4A shows titrations of mtTFl purified by conventional eluted from hydroxylapatite, an additional protein-DNA comchromatography(lanes 1-6), by denaturation-renaturation plex is produced by a fraction eluted at higher ionic strength chromatography (lanes 7-12), and by denaturation-renatur(Fig. 3B, lanes 42-46). DNase I protection analysis of this ation chromatographyfollowed by gel isolation and guanidine complex revealed specific binding to the oris promoter (data hydrochloride denaturation and renaturation (lanes13-18) in not shown), producing a footprint similar to that caused by a gel retardation DNA-binding assay with a radioactive LSPthe yeast mtRNA polymerase in combination with its acces- containing fragment. All three preparations produce qualitasory specificity factor (15). When this fractionwas renatured tively similar distributions of the fragment in an ascending with TritonX-100 and chromatographed on phosphocellulose, ladder-like pattern as the protein-DNA ratio increases; these the sole DNA-binding activity detectedby gel retardation was data were used to estimatevolumes containing equal amounts the 19-kDa protein, plS/HM, which by itself does not foot- of m t T F l for the experiments shown in Fig. 4 (R-D). Speciprint thepromoter.' The mostplausible interpretation of this ficity of DNA binding, as indicated by the ability to produce


400 410 1

52 4 3

6

7

8

3 1110

12

13

14

15 16

17

18

420-

430

440

1

2

3

4

5

6

i

8

9

1

0

--m=

(Dl +LSP

U "

1

2

3 4 5

6

7

8 91011121314

-I D.P I I H.P I

L 25kDa-l

a d d

FIG. 4. Functional comparisons of mtTFl purified by conventional and denaturation-renaturation chromatographies. A, DNA-binding. The gel retardation assay was used to quantitate mtTFl purified by the previously published method of sequential DEAE and phosphocellulose chromatographies under nondenaturing conditions ( D - P ,lanes 1-6), by denaturation-renaturation chromatography (H-P, lanes 7-12), or by denaturationrenaturation chromatography followed by gel purification, guanidinehydrochloride denaturation, and renaturation of the 25-kDa mtTFl polypeptide (25 kDa, lanes 13-18). All three fractions produce qualitatively similar patterns of shifted DNA fragments with increasing amounts of protein added; the distributions produced by the different fractions at different concentrationswere used to establish equivalent volumes of the three preparationsfor use in subsequent comparative experiments. H, specific binding to the upstream control region of the human LSP. The same three fractions were used to generate DNase I footprints on a labeled DNA fragment containing an intact human mtDNA LSP. Equal amounts of crude DNA-binding activity (as determined by the experiment shown in panel A ) ,purified by conventional chromatography(D-P,lanes 2-4), by denaturation-renaturation chromatography (H-P, lanes 6-8), or by gel isolation (25kDa, lunes 10-12), produce indistinguishablefootprints.The arrow indicates the start site and direction of transcription from the human LSP. C, activation of L-strand transcription. The ability of mtTF1 purified by conventional chromatography (D-P, lanes ]-.I), by denaturation-renaturation chromatography ( H - P , lanes 5-8), and by gel isolation (25 kDa, lanes 9 and IO) to stimulate transcription from the human mitochondrial LSP by mtRNA polymerase was determined using an in vitro runoff transcription assay. The full-length runoff transcript of -416 nucleotides is denoted by the arrou~at the right. Numerous smaller species represent processed or prematurely paused or terminated LSP-dependent transcripts. D, activation of Hstrand transcription. Same as C, except that the transcription template contains only the intact human HSP. Similar amounts (as determined by the DNA-binding titration experiment depicted in panel A ) of the three different preparations of mtTFl produce similar levels of HSP activation.

Purification of DNA-binding Proteins TABLEI1 Comparison of conventional and denaturation-renaturation chromatographies

9159

defect seen at the phosphocellulose stage. The defect does not appear to reflect the copurification or introduction of general transcriptional inhibitors, since H-strand transcription is unProcedure Specific activity” Yield affected. We have not ruled out the presence of LSP-specific units/pg protein unitslliter of KB celk inhibitors in the final fraction; however, extensive biochemical analyses of mitochondrial extracts have failed to identify 2.4 (1.76) 0.273 (0.205) DEAE-phosphocellulose 7.1 (4.6) 0.336 (0.217) such inhibitory factors. Perhaps a more likely explanation is SDS-hydroxylapatitephosphocellulose that the nonionic detergent effects only a partial restoration ’ 1 unit = 1 fmol of DNA fragment bound in 1:l complex. In each of native protein structure, while the complete disruption of case, the first number represents the estimate based on quantitating secondary structure induced by guanidine hydrochloride is radioactivity in bound complexes with a Molecular Dynamics Com- necessary to allow all domainsof the protein subsequently to puting Densitometer, multiplying byappropriate factors (1 X for complex 1, 2 X for complex 2, etc.) and then summing the “binding regain their active conformation. The notion that we may events” measured by this analysis. The second number (in parenthe- have crudely dissected out an LSP-specific transcriptional ses) is based on quantitating only complex 1, at low protein DNA activation domain of mtTFl deserves some consideration in ratios (using only two data points in each case). The SDS-hydroxyl- the light of our previous studies of the differential response apatite-phosphocellulose preparation used was chosen for highest of the two promoters to fluctuations in factor concentration purity, rather than yield, which has been routinely higher in subse- in uitro, as well as some recent discoveries about the effects quent purifications. These data indicate only that the two methods of mtTFl binding on DNA conformation. of purification yield mtTFl of comparable specific activity. The LSPof human mtDNA appears to be maximally activated by a single mtTFl-DNA binding event, whereas HSP discrete regions of protection from DNase I digestion, or footprints, on LSP-containing fragments, is not impaired by stimulation requires factor binding at much higher stoichithe denaturingsteps of the purification (Fig. 4B).Equal ometries (1).Moreover, mtTFl is capable of profound alteramounts of DNA binding activity afford similar degrees of ations in DNA conformation at both low and high factor DNA protection to the mtTF1-binding site of the LSP, regardless ratios, inducing both sequence-specific bends and apparently seems likely of the purification method (compare lanes 2-4 with lanes 6-8 nonsequence-specific negative ~upercoiling.~ It that such physical perturbations of DNA structure are inand lanes 10-12). In contrast, while mtTFl purified by SDShydroxylapatite and Triton X-100-phosphocellulose chroma- volved in transcriptional activation at both promoters. Howtography is able to stimulate transcriptionof the human LSP, ever, the data presented here may point to a fundamental it was clearly defective in this property when compared to mechanistic difference between the two major promoters of mtTFl purified by DEAE-Sephacel and phosphocellulose mammalian mtDNA. The two methods of purifying mtTFl chromatography under nondenaturing conditions (Fig. 4C, yield proteins that are indistinguishable in their ability to compare lanes 1-4 with lanes 5-8). In fact, over the range of bind, bend, and supercoil DNA, and in their stimulatoryeffect concentrations shown, the “denatured factor actually inhib- on HSP transcription, yet that differ markedly in their effects its LSP-specific transcription; at lower concentrations the on in vitro transcription initiating at the LSP. The identical stimulatory effect is more apparent, whereas equivalent footprinting behavior, moreover, appears to rule out an artiamounts,as defined by the gel retardation assay, of the factual difference in DNA sequence specificity. This leaves “native” mtTF1clearly stimulate transcription in an additive the intriguing possibility that mtTFl contains a domain (or fashion. Interestingly, when the fraction containing mtTFl domains) required for protein-protein interactions that selecpurified by denaturation-renaturation chromatography is sub- tively influence L-strand transcription. Transcription of the jected to more rigorous denaturation by guanidine hydrochlo- H-strand, on the other hand, may be predominantly deterride as part of the gel isolation procedure, the defect appears mined by the conformational effects of factor binding. Seemto be corrected, and full transcriptional activity is restored ingly consistent with this notion is the fact that mtTFlbinds (Fig. 4C, lanes 9 and 10). Stimulation of transcriptional ini- only weakly to the HSP(1-3), and theobservation that such tiation at the HSP of human mtDNA does not appear to be weak binding is sufficient to bend the HSP and other DNA as sensitive to differences among the various preparations of ~equences.~ The ability to obtain large amounts of pure mtTFl has mtTF1, as demonstrated by the results of the runoff transcription experiment shown in Fig. 4 0 . Both the native (lunes 2- resolved the impasse described earlier. Several tryptic peptide 6)and denatured (lanes 7-11) proteins are capable of stimu- fragments of mtTFl have now been isolated and sequenced, and oligonucleotides designed to be complementary to the lation and inhibition at roughly equivalent concentrations. DNA sequences potentially encoding these fragments have DISCUSSION been used to isolate a cDNA clone encoding human mtTF1.’ Chromatography on hydroxylapatite in the presence of Thusthe firststep in the detailed analysis of structuredenaturing concentrations of SDS has been previously em- function relationships for this key regulator of mitochondrial ployed in the compositional analysis of complex macromolec- gene expression, the determination of complete gene and ular structures, such as virus particles (6) and the eukaryotic derived amino acid sequence, is well within reach. An obvious nuclear matrix (16). We have adapted this procedure to facil- next step will be isolation and characterization of the murine itate therecovery of a transcriptionaleffector protein in pure, mtTFl gene; comparisons with its human homologue should biologically active form. No irreversible denaturation of func- help elucidate the evolution of sequence specificity of these tional protein domains of mtTFl occurs during the purifica- unusual molecules. Even in advance of these breakthroughs, the improved tion sequence; gel isolation of mtTFl purified by SDS-hydroxylapatite and Triton X-100-phosphocellulose chromatogra- purification scheme has afforded valuable new insights into phiesresults in a factor preparation that is functionally mtTFl structure and function. The denaturation-renaturaidentical to mtTFl that has never been denatured. However, tion chromatography procedure, when applied to yeast mitotemporary denaturation of a domain (or domains) required chondria, results in the purification of a 19-kDa protein with for optimal LSP activation, but apparently not for H-strand structural and functional similarities to mammalian mtTF1. transcription, may explain the LSP-specific transcriptional The two proteins have very similar amino acid compositions

9160

Purification of DNA-bindingProteins

and produce nearly identical mobility shifts ingel retardation DNA-binding assays. Moreover, both proteins are capable of introducing negative supercoils in relaxed closed circular DNA molecules,probably by wrapping the DNA double helix about a protein core in a structure analogous to the nucleosome, and both appear to be related to the histone-like proteins of b a ~ t e r i a . ~ Successful application of denaturation-renaturation chromatography to yeast mitochondrial extracts may indicate the general value of the technique; it could be particularly useful in the identification and purification of DNA-binding activities from organellar or membrane-bound sources, and may prove an attractive alternative to conventional chromatography for proteins that display a high affinity for nonspecific DNA sequences. These factors may not be amenable to purification by specific DNAsequence affinity and may also resist efficient extraction from tissue or cell culture sources rich in endogenous DNA. The data presented above suggest that relatively harsh conditions may be used to disrupt macromolecular interactions during purification, without jeopardizing the subsequent recovery of fully active DNA-binding proteins. Acknowledgments-We thank Dr. Paul A. Fisher of the Department of Pharmacologic Sciences, State University ofNew York, Stony Brook Health Sciences Center, for many helpful suggestions on the use of SDS-hydroxylapatite chromatography. We are also indebted to M. A. Parisi of this laboratory for assistance in the

preparation of this manuscript and for the communication of unpublished data. REFERENCES 1. Fisher, R. P., and Clayton, D. A. (1988) Mol. Cell. Biol. 8 , 34963509 2. Fisher, R. P., Topper, J. N., and Clayton, D. A. (1987) Cell SO, 247-258 3. Fisher, R. P., Parisi, M. A., and Clayton, D. A. (1989) Genes & Deu. 3,2202-2217 4. Pabo, C. O., and Sauer, R. T. (1984) Annu. Reu. Biochem. 5 3 , 293-321 5. Yang, C.-C., and Nash, H. A. (1989) Cell 57,869-880 6. Moss, B., and Rosenblum, E. N. (1972) J. Biol. Chem. 247,51945198 7. Chang, D. D.,and Clayton, D. A. (1984) Cell 36,635-643 8. Fisher, R. P., and Clayton, D. A. (1985) J. Biol. Chem. 2 6 0 , 11330-11338 9. Bogenhagen, D., and Clayton, D. A. (1974) J. Biol. Chem. 2 4 9 , 7991-7995 10. Schaffner, W., and Weissman, C. (1973) Anal. Biochem. 56,502514 11. Laemmli, U. K. (1970) Nature 227,680-685 12. Faye, G., Kujawa, C., and Fukuhara, H. (1974) J. Mol. Biol. 88, 185-203 13. Hager, D. A., and Burgess, R. R. (1980) Anal. Biochern. 109,7686 14. de Zamaroczy, M., Faugeron-Fonty, G., Baldacci, G., Goursot, R., and Bernardi, G. (1984) Gene (Arnst.) 32,439-457 15. Schinkel, A. H., Groot Koerkamp, M. J. A., and Tabak, H. F. (1988) EMBO J. 7, 3255-3262 16. Filson, A. J., Lewis, A., Blobel, G., and Fisher, P. A. (1985) J . Biol. Chem. 260,3164-3172