Proc. Natl. Acad. Sci. USA Vol. 91,
pp.
4514-4518, May 1994 BiochemistryThe Kruppel-associated box-A (KRAB-A) domain of zinc finger proteins mediates transcriptional repression
(transcriptionfactors/Kid-1/ZNF2/kidney)
RALPH
WITZGALL, EILEEN O'LEARY,
ALEXANDERLEAF, DILEK ONALDI,
ANDJOSEPH
V.BONVENTRE*
MedicalServices,Massachusetts GeneralHospital, 149 13thStreet,Charlestown,MA02129; and Department ofMedicine, Harvard Medical School, Boston, MA 02115
Contributed byAlexanderLeaf,February 3,1994
ABSTRACT We have previously reported the cloning, sequencing, and partial characterization of Kid-i, a zinc finger-encoding cDNA from theratkidney. The Kid-i protein andapproximately one-third of all other zinc
finger
proteins containahighly conservedregion
of -75 aminoacidsattheir NH2 terminus namedKrtippel-associated box(KRAB),which is subdivided into A andBdomains. Theevolutionary conser- vation, wide distribution, and genomic organization of the KRABdomainssuggestanimportantroleof thisregion
in the transcriptionalregulatoryfunction of zincfinger proteins.The functional significance of the KRAB domainwasevaluatedby studying transcriptional activities of yeast GAL4-rat Kid-i fusion proteins containing various regions of the non-zinc- finger domain of Kid-i. Transcriptional repressor activityof GAL4-Kid-1 fusion proteins mapsto the KRAB-A domain.The KRAB-Adomain ofanotherzincfinger protein, ZNF2, also hasrepressor
activity.
Sitedirected mutagenesisofcon- served amino acidsinthismotifresults indecreasedrepressor activity.Thus,wehaveestablishedafunctional signifncefor theKRAB-A domain, a consensussequence common in zinc finger proteins.Alimited numberofstructural motifs have been described for eukaryotictranscription factors.Acommonmotifis thezinc fingerstructure,which ishighlyconserved and found in many different species (1). Though the crystal structure ofaco- crystalbetween the three zincfingersof zif-268
(Egr-i)
and theirbindingsite has beenelucidated(2),little isknownabout theinteraction betweenzincfingers
and DNA andhowthese protein-DNA interactions areinvolved in the regulation of transcription.Approximately one-third ofall zincfingerproteinscontain anevolutionarily conservedregionof about 75 aminoacidsat theirNH2terminus-the
Krippel-associated
box(KRAB),a sequence motif of hitherto unknown function. Similar to regions found in a subset of homeodomain proteins (the pairedbox and POU domain), the KRAB domainisrich in charged amino acids(3). Because ofthe potential a-helical structureofKRABdomains, ithas beenproposedthatthis domainmediates protein-proteininteractions andfunctions as atranscriptional regulatory domain (3), but this hypothesis hasnotbeenpreviouslytested.Kid-i
isa ratzincfingergene that is expressedprimarily in the kidney (4).Kid-i
mRNA levels are regulated in renal ontogeny and during repair after ischemic and folic acid insults to thekidney(4).Kid-i
encodes a68-kDa protein with 13zincfingers.When COS orLLC-PK1cells are transfected with fusion constructs encoding the non-zinc-finger region of ratKid-i
and the yeast GAL4DNA-binding domain, chlor- amphenicol acetyltransferase (CAT) activity from cotrans- fectedreporter constructscontaining GAL4 binding sites andeither a minimal promoter or a simian virus 40 (SV40) enhanceris strongly repressed(4). Wedesigned experiments to test the hypothesis that the transcriptional repressor activityresided in the KRABdomainof Kid-i (Fig. 1). Our mutagenesis studies with the Kid-i protein define the KRAB-Adomainas astrongtranscriptionalrepressormotif.
METHODS
Expression Plasmids.Plasmidsencoding chimeric proteins of theDNA-binding domain (amino acids 1-147) of GAL4 and various parts of thenon-zinc-finger region ofKid-i (desig- nated Kid-iN) were constructed from
pBXGi/Kid-lN
in senseorientation(4)(designatedpBXGi/Kid-lN,s),encod- ingaGAL4-Kid-i fusion proteincontainingtheentirenon- zinc-finger region of Kid-i (Fig. 2). pBXGi/Kid-i,AB was madebyvector self-ligationaftercuttingpBXGi/Kid-lN,s withBamHI. To constructpBXGi/Kid-i,A(-)
lackingthe Kid-iKRAB-Adomain,pBXGi/Kid-lN,s
wasdigested with EcoRIandfilled inby usingtheKlenowfragment.
The insert wassubsequentlyremoved bydigestion withXba I, and the vectorfragment
wasisolated. Insert was prepared by digest- ingpBXG1/Kid-lN,swith NcoI,fillingin theoverhangwith theKlenowfragment,andreleasingtheinsert withXbaI. To constructpBXG1/Kid-1,AandpBXG1/Kid-1,B,fragments
containing the Kid-i KRAB-A- or B-encoding sequences (designatedKid-iA
andKid-iB)
were PCR-amplifiedfrom pBXGi/Kid-i,AB withprimers containingan EcoRI site or Xba I site. PCRproductsweredigestedwith EcoRI andXba I andcloned into the EcoRI/Xba I-cutpBXGi
vector. To constructpBXG1/ZNF2,A,
the exonencoding theKRAB-A domain ofZNF2 (designated ZNF2,A) was PCR-amplified from human genomic DNA with primers containing EcoRI andXba I sites andsubclonedintoEcoRI/XbaI-cutpBXGi.pCDM8/Kid-l,f',inwhich
Kid-l,t
signifies codingregion for the full-lengthKid-i
protein, was engineered by ligating BstXI-adapters to aKid-i
cDNAfragment extending from nucleotide312 to2123 [as numbered in the cDNA (4), where theputativestartcodon is at 312and the stop codon is at 2040]and ligating it into BstXI-cut pCDM8. PCR products were sequenced in theirentiretyby thechain-terminationmethod.
Cloningsitesofallconstructs weresequenced (5).
Site-Directed Mutagenesis. APCR-based strategy (6)was usedtoconstruct mutants withsingle amino acid changes in the KRAB-Adomain of
Kid-i
(Kid-1,A). Two PCR reactions were run withpBXG1/Kid-1,A as a template. In the first reaction,oneprimer, "F," complementary to the 3'-endof GAL4, was paired with a primer containing the desired Abbreviations: KRAB,Krfippel-associatedbox;SV40, simian virus 40; CAT, chloramphenicol acetyltransferase. PBS, phosphate- bufferedsaline; PCR,polymerase chain reaction.*To whomreprintrequests shouldbe addressed at: Massachusetts General Hospital East, Suite 4002, 149 13thStreet, Charlestown, MA02129.
4514 Thepublicationcostsof this article weredefrayed in part by page charge payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C.§1734 solelytoindicate this fact.
12 53 Kidl-Krab
Znf2-Krab Hpf4-Krab Htfl-Krab Kox8-Krab Znf43-Krab Htf9-Krab Htfl2-Krab HtflO-Krab Htf34-Krab Znfl4l-Krab Brcl744-Krab Ldrl52-Krab Koxl-Krab Hpfl-Krab Hzf7-Krab Znf74-Krab Znf8-Krab Znf41-Krab Hzf3-Krab P18-Krab Znf7-Krab Zfp36-Krab
VSVfL
ESVI GPLI
GPLE GPLI GLLI .IPQ ELLI EAVI EAVI TLVI ESFS ESFS ESVS ASVS ELVI ELV1E EWI DSVA
DXV DXV I I AIE
I V I
VvI
VKI
VD k S D
SV D
.D k
D Vy DV
DVrv
DV ..E VVSV
a V
flTRE*
TD SL SL SL CL SL SL SP SL SP SE TE TRE TQ TQ TQ TQE SK IR TS SRE
EW EW EW EW EW EW EW EW EW E EW EW EW EW EW EW EW EW EW TQELB F QC QC Qc Qc
HC QC QC HC QI ,GI Qi Qi c f G AC
QC
A
1J DLSF
VPI DTP DTI DTP DII DTP DTP DTP DTP DPE DLP DPP DTI DP' DP' DSE DPI DPI EPI GPI DPC GP'
PGN
I
iREf
LA LA LA LA LA LA LA LA LA LA LA LA LA LA LA
LI
KE
RN RN RN
RN
RD
RD
RNRN
RD RD
RDRN KDKD
RD
RDWD RDWD
RE RN
|VMLENSNLA
SM|VMLEkNSIV
SLVMLENRNLV FL
VMLENRNLV FL
VMLENRNLV FL
VMLENRNLV FL
VMLENRHLV FL
VMLENRNLV FL
VMLE RNLA FL
VMLE SNLV FL
VMLE RNLV SL
VMLE RNVV SV
VMLE RNLL SV
VMLE KNLV SL
VMLE SSLV SL
VMLE SSLV SL
VMLE QNLL AL IVMLEFGYLL SI VILE SHLL SV
VMLE GNVF SL VMLE GNVT SL VMLE SSVA GL IRNLD CI consensus esvtFeDVav eFsleEWqcL dpaQrnLYrd VMLENyrnlv sl
r i d t 1 t n s f
FIG. 1. Comparisons of KRAB-A regions of various zinc finger proteins. The second exon ofKid-1 encodes amino acids12through 53.A consensussequence ispresentedatthe bottomofthealignmentwithcapitallettersindicatingat least90%oidentityamong the various proteins andwith small lettersindicatingatleast25%identitybut less than90%o.When more than one aminoacid is common, the less frequent one is listed on the second line of the consensus sequence.
mutation. For the second reaction, theprimer"R,"comple- mentary to the 3' end ofKid-1,A waspairedwith aprimer overlappingthe region ofthe mutation. Theproducts from these two PCRreactions overlapped over a stretch of at least nine nucleotides and were used in athirdPCR reaction with primers "F" and"R," which in
50%o
of the casesyielded afull-length Kid-1,A fragment with the desired nucleotide change(s). PCR productsweredigestedwith EcoRI and Xba Iand cloned into theEcoRI/Xba I-cutpBXG1 vector. The absenceof additional mutations was verified by sequencing theentire PCRproduct. Mutations are named by listing the wild-type amino acid first, followed by the position inKid-i
(Fig. 1)and the amino acidtowhichit is changed.Transfection Protocols and CAT Assays. COS cells were plated 2 dayspriortotransfection at a densityof 2.5 x105cells per100-mm dish. Fortransfections,cells wereexposed to 20pg of total DNA in 5 ml of
DMEM/10i
NuSerum(Collaborative BiomedicalProducts,Bedford,MA)/400pg
ofDEAE-dextran perml/0.1 mM chloroquine. One microgram ofaluciferase- expressing plasmid, poLucSV/T1 (7), was included in all cotransfection experimentstonormalize transfection efficien- cies. Three to4hr after the addition of DNA, mediumwas removed,and cells were shocked for 2 min at room temperature with 10%o dimethyl sulfoxide in phosphate-buffered saline (PBS). After shock treatment, cells were washed once with PBS, and newmediumwasadded.Forty-eight hours after transfection, cells were washed twice with PBS, scraped with a rubber policeman into a microcentrifuge tube, and pelleted. The cell pellet was re- suspended in 200 ,u of 0.25 M Tris chloride (pH 7.8) and subsequently broken up by subjecting it to a freeze/thaw cyclethreetimes in adry
ice/ethanol
bath and 37°Cwater bath. The supernatantwas assayed for CATandluciferase activitiesaccordingtostandardprotocols(4). CATactivityis expressedastheratio of monoacetylated[14C]chlorampheni- col/(monoacetylated
plusnonacetylated)[14C]chlorampheni-
col andnormalizedtoluciferaseactivity.
Protein Gels and Western Blots. SDS/PAGE gels and Westernimmunoblotswereperformedaccording to standard protocols (8). An anti-GAL4 antibody(obtainedfrom S. A.
Johnston andK.Melcher,University ofTexasSouthwestern Medical Center) was used at a 1:1000 dilution. Immune complexes were detected with the Renaissance light-de- tection kit from DuPont.
RESULTS
To define the region in the Kid-i protein that is able to mediate its repressoreffect on transcription, we generated a set ofexpression plasmidsencoding chimeras ofthe DNA- binding region of GAL4 with various regions ofthe NH2 terminus ofKid-i (Fig. 2). A plasmid that encoded GAL4- (1-147) andonly the KRAB-A and -B domains(pBXG1/Kid- 1,AB) induced a very similar level oftranscriptionalrepres- sion of CAT activity of the cotransfected pGSSV-BCAT reporter plasmid as the sense construct encoding the full- length non-zinc-finger region of
Kid-i
(Kid-lN,s) (Fig. 3).The mean residual CAT activities in cells expressing Kid- 1N,s or Kid-1,AB, were 16.7 ± 1.9o and 6.6 ± 1.1%, respectively, when compared to cells transfected with an antisense construct ofKid-iN (pBXG1/Kid-1N,as), which servedas a negative control (Fig. 3). CATactivity in cells cotransfected with theantisenseexpression plasmidpBXG1/
Kid-lN,as was equal to the activity obtained from cells transfected with pBXG1, which encoded only the DNA binding region of GAL4 (data not shown). Transfection of pBXG1/Kid-1,Aresulted inrepressionoftranscriptionto15
+ 1.5% of antisense CAT activity, a degree ofrepression almost identical to that seen with the plasmid encoding Kid-lN,s. Conversely, when the KRAB-A region was de- letedfromtheNH2 terminus of
Kid-i
[pBXG1/Kid-1,A(-)]or when the construct encoded onlythe KRAB-B domain (pBXG1/Kid-1,B), transcriptional repression was almost completely absent. To demonstrate thatalack of repressor
4516 Biochemistry: Witzgalletal.
Reporter Plasmid pG5SV-BCAT
V V'l'S-T-'I ! A K7,ff.Thii 0l.l Iuh11Th
5GAL4 binding sites
SV40-Enh El b TATA p r o m o t e r
Expression Vectors pBXGliKid-lN constructs
SV40-ori GAL4 (1-147)
CAT
_._h.AIL~bjw _..
Kid-!N.s Kid-1iNas
sense. anti-sense.
tull length fuil length sense (1-195)
liaufi ..m
Cl- 1 AB Kid- .A.l- K1RAB-Ae sense without
K R4E-A
Kid-1 Nas
1l 95 A" \\\\\
antisense
1 7 3
AB
Kid-1,AB Kid-1,A(-)
1 5 54
A
55 73
B
Kid-1 ,A
U
0 c ao
to
Sn
5
50 <
25)q12 9
FLI
Kid-1,B
FIG. 2. Reporter andexpression plasmids.pG5SV-BCATcontains fiveGAL4 bindingsites upstream of theSV40enhancer and the TATA boxfrom the EIB promoter, which drive a CAT gene. In theexpression vectors, transcription is driven by the SV40 enhancer. The DNA bindingdomainof GAL4(aminoacids1-147)liesattheNH2terminus offusion proteins with various portions ofKid-i. "Sense (1-195)"
contains the first 195 amino acids ofKid-i (includingthefirstpairof cysteinesof the zincfinger domain).Intheantisense"as" construct, the orientation of theKid-ifragmentis reversed. In the "AB"chimera, theKRAB-Aand -B domains(aminoacids1-73)arefusedtoGAL4. In
"A(-),"amino acids53-195,lackingthe KRAB-Adomain,arefused to theGALA DNA-binding domain.Mutants "A" and "B"code for fusionproteinsof the GAL4DNA-binding domainandthe KRAB-A and -Bdomains ofKid-i, respectively.
activity of pBXG1/Kid-lN,asorpBXG1was notsimplydue tolack ofexpressionof theprotein,wetransfected COScells with 3pg ofeither
pBXGi/Kid-lN,s
orpBXG1/Kid-1,A
or 15,ug
ofpBXG1orpBXG1/Kid-lN,as. CAT activities were 21%(pBXG1/Kid-1N,s), 31%(pBXG1/Kid-1,A),
and108%(pBXG1) of the CAT activity in cells transfected with pBXG1/Kid-lN,as. Expression of proteinsencodedbythese plasmids was confirmed by Western blot analysis with an anti-GAL4 antibody (Fig. 4).
Similarto ourpreviousobservations with the
pBXG1/Kid-
1N,s plasmid (4), thepBXG1/Kid-1,A
construct repressed transcriptionin adose-dependent fashion(Fig. 5) with vir- tually identical efficiency as thefull-length non-zinc-finger regionconstruct(4). Theplasmid,pBXG1/ZNF2,A,
encod- ing a fusion protein of GAL4-(1-147) with the KRAB-A region of ZNF2,was aseffectivearepressoraspBXG1/Kid-
1N,s(Fig.6). This indicates that the repressoractivityof the KRAB-Adomain isnotspecific toKid-1.Toestablish thatbindingtotheGAL4siteonthereporter wasnecessary fortranscriptional inhibition, the cDNA en- codingthefull-length Kid-1 protein(i.e.,including all 13 zinc fingers)wassubcloned into the expressionvectorpCDM8, yieldingthe constructpCDM8/Kid-l,e. Thebinding site for the
Kid-i
protein is undefined, but likely distinct from the GAL4bindingsite. Cotransfection ofpCDM8/Kid-l,(
withpG5SV-BCAT
did not resultinareduction ofCAT activity(Fig. 7).
Point mutations in the KRAB-A region of
Kid-i
were gen- eratedby PCR to evaluate the importance of individual amino acids, whichwerehighlyconserved in all KRAB-A domains for which sequence information was available (Fig. 8). Each of the mutantsrepressedtranscription from the CAT gene to a lesser extentthan thewild-type KRAB-A domain. In the case of theKid-lA Kid-1 B
KRAB-A KRAB-E
sense AB A A B ExPression plasrnia
FIG. 3. Thin-layer chromatography assays of CAT activities in cells transfected with thepBXG1 expression constructsencoding fusionproteinsof theDNA-bindingregionof GAL4 with Kid-lN,s, Kid-1Nas, Kid-lAB,Kid-1,A(-), Kid-1,A,andKid-1,B. Expres- sionoffusion proteinscontaining Kid-lN,s, Kid-1,AB,orKid-1,A results in markedrepressionof CATactivity,whencomparedwith the CAT activity observed in cells transfected with the control antisense construct (pBXG1/Kid-1N,as). In contrast, when the KRAB-Adomain ismissingfrom theNH2 terminus ofKid-i [Kid- 1,A(-)orKid-1,B], CATactivityis similar to thatobserved inthe presence of the control antisense plasmid. (LowerRight) Graph providing quantitative CATactivity,with CATactivityobtained with theantisensepBXG1/Kid-lNbas plasmidtaken as100%.The num- bers above thestandarderrorbars indicate thenumber ofindepen- dentexperiments.
Phe-16, Val-19, and Leu-31mutations,nodifferencewasseen betweenaconservative
(Phe
-+Ala,Val Ala,orLeu -*Ala)
andanonconservative(Phe-+Asp,
ValGlu,
orLeutoGlu)
Supernatantsa---a*tso-- Pel let s --
ka0a tt m r- m F-~r-t--i|flF- --
43- _4-O
29o
I* ...io*
1 8- _-.
m iw
4w _.v:_._"pM ~ FIG. 4. Western blot with GAL4 antibodies. COS cells were transfected with 3
pg
ofeitherpBXG1/Kid-lN,sorpBXG1/Kid-1,A orwith 15pg ofpBXG1orpBXG1/Kid-1N,as.Pelletsobtained after subjectingthe transfected cells tofreeze/thawcycles were resus- pendedin 2001d
ofix PBS andsonicated;50jud
of 5x SDS samplebuffer containing 625 mM Tris HCl, 12.5% SDS, 50%1b glycerol, 12.5% 2-mercaptoethanol, and 0.05% bromophenol blue was added, and thesamples were boiled for 5 min. Aliquots of the supernatants andpellets corresponding to equal amounts of luciferase activity were runon a 13%gel and transferred to polyvinylidene difluoride membranes (Immobilon P, Millipore). GAL4 and GAL4 fusion proteins weredetected with an anti-GAL4antibody. Some of the smallerproteinsleaked out into thesupernatantduringthefreeze/
thawprocess. Arrowspoint to the fusion proteins. TheGAL4/Kid- 1N,sband is above an unrelated band similarin size.
.S-.:
__II1.|..l:-.:.:.
,-ssirsssosowsoiI;
Proc. Natl. Acad. Sci. USA 91
(1994)
stop Kid-1 Ns
F. -, -.1,
.. .. .. .. ..V///Ii.,
-.,.? ..IIXA
Y:o
To
I
Kc
c 75'
00
I0- -50
<25S
0.0 2.5 5.0 7.5 10.0
pBXG1/Kid-1,A, tLg
100% 8.4% 251%
I~~~~~~~~~..ii Li. L.11,.,,..~ I
pBXG1/Kld-1N,as pBXG1/Kid-1,A pCDM8/Kid-1,l antil-sense KRAB-A
full length FIG. 5. Quantitation of CAT activitiesfromthin-layer chroma-
tographyassaysof cells transfectedwith variousamountsofpBXGl/
Kid-1,A and 3pgof thepG5SV-BCATreporterplasmid.There isa
dose-dependent inhibition of CAT activitywithincreasingamounts ofpBXG1/Kid-l,Atransfected.CATactivitieswereunchangedwith equivalentamountsof transfectedplasmidencodingGAL4/Kid-lN antisense (pBXG1/Kid-lN,as;datanotshown).
mutation. For the Tyr-39 mutations, a difference was seen
between an alanine-substitutedmutant andamutantwith an
acidic residue. Theloss ofrepressor activity was most pro-
nouncedforTyr-39-+ Asp. Tyr-39-* Ala and Glu45-* Ala
showed the strongest repressor effects among all mutants examined. While therewassomevariation inexpression,each of themutantswasexpressedatalevelsimilartothatofthe wild-type KRAB-A domain as determined by Western blots withanti-GAL4 antibodies(Fig. 8).
DISCUSSION
Our dataprovidefunctional evidence for theimportance of the KRAB-A domain, a highly conserved motif found in
manyzincfinger proteins. Ithas been estimated thatapprox-
imatelyone-third of the hundreds of zincfingergenesinthe mammalian genome are members of the KRAB family (3).
With thiswidespreadconservation of the KRABdomain, it ispossiblethattranscriptionalrepressoractivityisassociated withmanymembers ofthe zincfinger family.
100% 14.8% 8.8%
Kid-1Nlas Kid-i N,s ZNF2,A
anti-sense sense KRAB-A full length full length
FIG.6. Thin-layerchromatography assaysofCAT activities in cellstransfected with eitherpBXG1/Kid-lN,sorpBXG1/Kid-1N,as
or anexpressionconstructencoding the GAL4 DNA-binding domain and theKRAB-Adomain of ZNF2. The KRAB-Adomain from the human zincfinger proteinZNF2repressestranscriptiontoanextent similartothatseenwithKid-lN,s.
u
C) .+ a)
:= .
C) =
0_1) 2 5-
° n
FIG.7. Thin-layer chromatographyassaysof CAT activitiesin cells transfected with an expression construct encoding the full- length Kid-i proteinin pCDM8(pCDM8/Kid-1,e)orwithKid-1N,as
or Kid-1,A in pBXG1. pCDM8/Kid-l,t, which encodes Kid-1,e withoutaGAL4 DNA-bindingdomain,hasnoeffectonCATactivity whencotransfected withpG5SV-BCAT. The pCDM8vectorisvery
similar in design to the pBXG1 vector in that it can replicate extrachromosomallytoveryhighcopynumbers in COS cellsbecause of itsSV40originofreplication.
Atpresentwedonotunderstandhow the KRAB-Adomain
repressestranscription. Our experiments withpCDM8/Kid- 1,( suggest that the KRAB-A domain has to be bound (indirectly through aDNA-binding domain) to DNA tobe abletorepress transcription. It doesnot sufficetooverex- press a KRAB-A-containing protein like Kid-i to repress
transcription, arguing against a "squelching effect" by KRAB proteins. This lack of "squelching effect" infers specificity, since therearepotentially large numbers of zinc finger proteins of the KRAB family expressedinagiven cell with the specificity conferred by the DNA-binding charac- teristics ofthe specific zinc finger protein. Inaddition, the
repressoreffect of theGAL4/Kid-1,A construct cannotbe attributedto"competition" with other factors for bindingat the GAL4 binding sites because thereare no knownmam-
maliantranscription factorsthatcanbind toGAL4binding sites, and a GAL4/Kid-1,A(-) mutant had only a minor
repressoreffect. We very likely can also rule out apurely
"stericmodel" ofrepressioninwhich the KRAB-Adomain
I I -v
ant wt E45AL31AL31EV19AV19EY39AY39DF16AF16D
Fax:-r s i1-
FIG. 8. CATactivities and Western blotofGAL4-(1-147)-Kid- 1,A fusion proteins with point mutations in the KRAB-A domain.
CATactivitiesareshownasthepercentageof theCAT activity from cells transfected withpBXGl/Kid-lN,as("anti");n=5-10for each construct.Western blotswereperformedasdescribed in Fig.4.Only thepellet fractionsareshown. Themutantswereall expressed, with mostexpressedatlevelsgreaterthanorequaltolevelsof wild-type Kid-lN,s (wt).
N..~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
I,- cbv. cb,0
I
kV# Iraqi
T T
I .,V/
'r
m
4518 Biochemistry: Witzgall etal.
prevents the interaction of a trans-acting domain with the basal transcription factors because the GALA binding sites in our reporterconstructlieupstreamofthe SV40 enhancer and not between the SV40 enhancer and the TATA box and because largerproteins suchas
GAL4/Kid-1,A(-)
areonly weak repressors. The finding that GAL4/Kid-lN,s is also able to represstranscription fromabasalpromoter(4)argues against the hypothesis that the KRAB-A domain "masks"theactivator region of a positively acting transcription factor.
At present, we suggestthe followingtwomodels forrepres- sion by the KRAB-A domain. Binding of a KRAB-A- containingprotein maylead to a change in local "chromatin structure" and thus impair binding of other transcription factors to their binding sites. It is also possible that the KRAB-A domain, either alone or together with another interacting protein, prevents "assembly" of the basal tran- scription factors or "locks" them in an inactive state and therefore inhibits transcription.
Each of the point mutationswegenerated showedalossof repressoractivity although to various extents, emphasizing the importanceof the absolutelyconserved residues in the KRAB-A domain.Glu-45 -* Ala andTyr-39-- Alawerethe mutants with the highest remaining repressor activity. In threeoffourresidues examined(Phe-16,Val-19, and Leu-31), nodifference was detected between a conservative and a nonconservative mutation, whereas inthe- other case (Tyr- 39), aconservativemutation affected the repressoractivityto alesserextentthananacidic residue substitution. Amongall the mutants, the Tyr-39 -* Asp mutant
repressed
CAT activity least. The introduction ofanacidic amino acid for Tyr-39 may result in the loss of interaction with the natural partnerofthe KRAB-Adomain.The KRAB-Adomainisencodedbyasingleexonin the rat Kid-] gene (9) and the humanZNF2 (10) and ZNF4S (11) genes. Acorrespondingarrangementcanbe assumed for the human ZNF43gene, whereonealternatively splicedmRNA species has been identified that lacks the KRAB-A region (12). Hence,by alternative splicing,cells mayproducezinc finger proteins withorwithout the KRAB-Atranscriptional repressordomain.
While there are at least four well-defined types of tran- scriptional activationdomains, serine/threonine-rich, acidic, proline-rich, andglutamine-rich (13), to ourknowledge the only other well-defined motif that has been postulated to mediate transcriptional repressoractivity is analanine-rich domain found in four transcriptional repressors from Dro- sophila:
Krippel
(14),engrailed(15),even-skipped(16), and AEF-1 (14). In other cases where arepressor domain hasbeendelineated, such as Egr-1 (17), SRF (18), andE4BP4 (19),noobviousconsensus sequencemotifs have been iden- tified. Our data indicate that the KRAB-A domainrepresents awidely distributed transcriptionalrepressormotif.
Wethank Karen Dellovofor artwork and Xia-Mei Liu for technical assistance. We thank Dr. Ptashnefor thepBXG1andpG5SV-BCAT vectors.This work was supported by National Institutes of Health Grants DK 39773 and DK 38452. Partsof these data were presented inthe 26th Annual Meetingof the American Society of Nephrology and have been published (20). Protein sequence alignments were constructed with the PILEUP program ofthe Genetics Computer GroupoftheUniversity of Wisconsin.
1. Klug, A. & Rhodes, D. (1987) Trends Biochem. Sci. 12, 464 469.
2. Pavletich,N.P.&Pabo,C.0. (1991) Science 252, 809-817.
3. Bellefroid,E.J.,Poncelet, D. A., Lecocq, P. J.,Revelant,0.
&Martial,J. A.(1991)Proc.Natl. Acad. Sci. USA 88, 3608- 3612.
4. Witzgall, R., O'Leary, E., Gessner, R., Ouellette, A.J. &
Bonventre, J. V.(1993)Mol. Cell.Biol. 13, 1933-1942.
5. Sanger, F.,Nicklen, S. & Coulson, A. R. (1977) Proc. Natl.
Acad. Sci. USA 74, 5463-5467.
6. Aiyar, A. &Leis,J.(1993)BioTechniques11,366-368.
7. Shelley,C. S. & Arnaout, M. A. (1991) Proc. Natl. Acad. Sci.
USA88, 10525-10529.
8. Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D. D., Seidman, J. G., Smith, J. A. & Struhl, K. (1987) Current ProtocolsinMolecular Biology (Wiley,NewYork).
9. Witzgall, R., Volk, R., Yeung, R. & Bonventre,J. V. (1994) Genomics, in press.
10. Rosati,M.,Marino, M., Franze, A., Tramontano, A. & Grim- aldi,G. (1991) Nucleic Acids Res. 19, 5661-5667.
11. Constantinou-Deltas, C. D.,Gilbert,J., Bartlett, R. J.,Herb- streith, M., Roses, A. D. & Lee, J. E. (1992) Genomics 12, 581-589.
12. Lovering, R. & Trowsdale, J. (19910 Nucleic Acids Res. 19, 2921-2928.
13. Mitchell,P.J.&Tijian,R.(1989) Science245,371-378.
14. Licht,J. D.,Grossel,M.J.,Figge,J.& Hansen, U. M. (1990) Nature(London) 346, 76-79.
15. Jaynes, J. B.& O'Farrell, P. H. (1991) EMBO J. 10, 1427-1433.
16. Han, K.& Manley,J. L.(1993) Genes Dev. 7, 491-503.
17. Gashler, A. L., Swaminathan, S. & Sukhatme, V. P. (1993) Mol. Cell. Biol. 13, 4556-4571.
18. Johansen, F.-E. & Prywes, R. (1993) Mol. Cell. Biol. 13, 4640-4647.
19. Cowell, I. G., Skinner,A. & Hurst, H. C. (1992) Mol. Cell.
Biol. 12, 3070-3077.
20. Witzgall,R., O'Leary, E.,Volk, R., Yeung, R. & Bonventre, J.V.(1993) J. Am. Soc. Nephrol. 4, 908.
Proc. Natl. Acad. Sci. USA 91