Proc. Nati. Acad. Sci. USAVol. 90, pp. 1107-1111, February 1993Biochemistry

Multiple mechanisms are implicated in the regulation of NF-KBactivity during human cytomegalovirus infectionTIMOTHY F. KOWALIK*, BRET WING*t, J. STEPHEN HASKILL*t*, JANE C. AZIZKHAN*§¶1IALBERT S. BALDWIN, JR.*,§,**, AND ENG-SHANG HUANG*,t,§,tt*Lineberger Comprehensive Cancer Center, §Curriculum in Genetics, and Departments of **Biology, *Obstetrics and Gynecology, tMicrobiology andImmunology, VPediatrics, IIPharmacology, and ttMedicine, School of Medicine, University of North Carolina Chapel Hill, NC 27599-7295

Communicated by Mary Ellen Jones, September 28, 1992

ABSTRACT Infection-induced activation of the humancytomegalovirus major immediate early enhancer/promoterhas been shown to be regulated primarily by transcriptionfactor NF-KB cis elements. However, the mechanism(s) bywhich human cytomegalovirus induces NF-mcB activity is un-known. A study was therefore undertaken to determine howthis virus would affect normal NF-#cB regulation. Viral infec-tion of fibroblasts resulted in the specific stimulation of pro-moters containing major histocompatibility complex NF-#cB ciselements fused upstream of the chloramphenicol acetyltrans-ferase reporter gene. Electrophoretic mobility shift assays ofnuclear extracts derived from mock- and virus-infected cellsshowed dramatic and sustained increases in DNA-bindingproteins specific for these NF-KB sequences. Experiments usingMAD-3 lKB, a specific inhibitor of NF-ucB, and antibodiesdirected against rel family members demonstrated that theinduced binding activities contained p5O and p65 proteins butnot c-rel. Northern analysis indicated maximal levels of p5OmRNA by 4 h postinfection, whereas p65 and MAD-3 IicBmRNA accumulation peaked at 48-72 h postinfection, suggest-ing different regulatory mechanisms for p5O and p65/IicBgenes. Electrophoretic mobility shift assays with deoxycholate-treated cytoplasmic extracts demonstrated a 3- to 4-fold de-crease in the cytosolic stores of NF-#cB binding activity by 4 hpostinfection. Western blots probed with antibodies directedagainst MAD-3 IicB or pp4O (a protein isolated from chickenwith sequence and biochemical properties similar to those ofMAD-3 IncB) indicated that a cross-reactive peptide of 39 kDawas no longer detectable after 24 h postinfection. These resultsdemonstrate that the activation and maintenance of nuclearNF-icB DNA binding and enhancer activities upon humancytomegalovirus infection occurs by multiple mechanisms.

Human cytomegalovirus (HCMV) is a ubiquitous herpesvi-rus associated with developmental abnormalities and is oftena pathogen in immunocompromised hosts (for review, seeref. 1). HCMV gene expression is coordinated into threetiers: immediate early (IE), early, and late. The major IEproduct, IE1, is under the control of a powerful enhancer/promoter, the major IE (MIE) promoter. Regulation of thispromoter is complex, entailing positive, negative, and celltype-specific enhancer elements (2-5). Of these, the 18-bprepeat element has been shown to be central to the activationof the MIE promoter in productive infections of humanfibroblasts (3, 6, 7). Furthermore, the IE1 gene productpositively autoregulates the MIE promoter via the 18-bprepeats (6). Each of the four 18-bp repeats contains a con-sensus sequence for transcription factor NF-KB (GG-GACTTTCC or GGGATTTCC), and previous work hasdemonstrated an HCMV infection-induced DNA-binding ac-tivity specific for the 18-bp repeat that is similar to the NF-KB

activity observed in phorbol 12-myristate 13-acetate- andphytohemagglutinin-stimulated Jurkat cells (7). How HCMVinfection alters the normal cellular regulation of this apparentNF-KB activity is not known.NF-KB is a pleiotropic DNA-binding activity involved in

inducible gene expression. It is one of several related DNAbinding activities, which are composed of proteins in the relfamily (for review, see ref. 8). NF-KB was originally de-scribed as an enhancer activity specifically associated withactivated expression of the K light chain in differentiated Bcells (9). NF-KB DNA binding activity is composed of twoproteins, p65 and p50, whereas another activity, KBF-1, iscomposed of only p50 homodimers (10). Other rel familyproteins include c-rel and p49 (pSOB), both of which interactwith NF-KB enhancer sequences (8).

In most cell types, including fibroblasts, NF-KB is in aninactive state associated with IKB (11, 12). Originally, twoforms of IKB were identified, IKBa and -,8 (13), each capableof inhibiting NF-KB DNA-binding activity. Recently, acDNA (MAD-3) isolated from human monocytes whose geneproduct has properties of IKB has been identified. At thistime, it is unclear whether MAD-3 is the a, the f3, or anadditional form of IKB, but partial peptide sequence datafrom IKBa suggests that the MAD-3 gene product is IKBa(14-19). In the current model, IKB sequesters NF-KB in thecytosol in an inactive complex, which can respond to appro-priate stimuli such as tumor necrosis factor or cytokines (20).NF-KB activation occurs by an ill-defined mechanism thatreleases NF-KB from IKB, permitting the translocation ofNF-KB into the nucleus where it can stimulate the transcrip-tion of promoters containing NF-KB cis elements. The utilityof this system is in its ability to rapidly and transientlyrespond to signals without de novo protein synthesis.We report here that HCMV infection results in sustained

nuclear NF-KB activity and transactivation of promoterscontaining NF-KB enhancer sequences. This activation isinitiated by the release of cytosolic stores of NF-KB. Theincreased nuclear NF-KB activity is apparently maintainedover the course of virus infection by increasing the mRNAlevels for p50 and p65. In addition, steady-state levels ofMAD-3 IKB protein are diminished to undetectable levels atlater times of infection even though MAD-3 mRNA is in-duced. Together, these results suggest that HCMV utilizesmultiple mechanisms to activate and sustain nuclear NF-KBactivity necessary for a productive viral infection.

MATERIALS AND METHODSVirus and Cells. HCMV Towne strain was propagated and

titrated as described (21). Normal human foreskin fibroblasts

Abbreviations: HCMV, human cytomegalovirus; MHC, major his-tocompatibility complex; CAT, chloramphenicol acetyltransferase;IE, immediate early; EMSA, electrophoretic mobility shift assay;hpi, hour(s) postinfection; NHF, normal human foreskin fibroblasts;MIE, major IE; GST, glutathione S-transferase.

1107

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

1108 Biochemistry: Kowalik et al.

(NHF) were kindly provided by W. Kaufman (University ofNorth Carolina at Chapel Hill) and maintained as described(22). Cells were infected with HCMV at a multiplicity ofinfection of 5 and absorbed for 2 h at 370C.

Transfection and Chloramphenicol Acetyltransferase (CAT)Assays. DNA was transfected into cells (5 jig) by the Lipo-fectin method (BRL) as described by the manufacturer. Aftertransfection, duplicate plates were infected with HCMV; thecells were incubated for 48 h and assayed for CAT activity asdescribed (22, 23). The c-fos minimal promoter-CAT con-structs (A56) containing three copies of wild-type (TGGG-GATTCCCCA) or mutant (TGCGGCTTCCCQA) NF-KBsequences from the major histocompatibility complex (MHC)class I enhancer have been described (24).

Nuclear and Cytosolic Extract Preparation. Nuclear andcytosolic extracts of mock- and HCMV-infected NHF cellswere prepared as described (22, 25).

Electrophoretic Mobility Shift Assays (EMSAs). EMSAs ofnuclear extracts and radiolabeled DNA probes were per-formed as described elsewhere (24, 26). The NF-KB wild-typeand mutant probe sequences were those used in the promoter-CAT constructs. In the indicated experiments, extracts werepreincubated for 10 min at room temperature with eitherglutathione S-transferase (GST)-MAD-3 IKB or specific anti-bodies prior to the addition ofprobe and subsequent EMSAs.

Bacterial Expression ofMAD-3 IKB. A cDNA representingMAD-3 IKB was digested with EcoRI, ligated in-frame intothe EcoRI site ofpGEX-3X (Pharmacia), and transfected intoEscherichia coli. Protein expression and affinity purificationwere as described (27). Functional activity of purified GST-MAD-3 IKB was confirmed by specifically inhibiting theDNA binding ofheteromeric complexes containing either p65or c-rel (ref. 16; T.F.K., unpublished observations).Northern Blot Analysis. Total RNA from mock- and

HCMV-infected NHF cells was resolved by formaldehyde/agarose electrophoresis and blotted to nitrocellulose mem-branes. Blots were hybridized with cDNAs representing p65(ref. 28; a gift from C. Rosen, Roche Institute, Nutley, NJ),plO5 (pSO) (ref. 29; kindly provided by A. Israel, InstitutPasteur, Paris), and MAD-3 IKB (14).Western Blot Analysis. Western blots were made and

processed as described (15).

RESULTSHCMV Infection Enhances Promoter Transactivation via

NF-KB Elements. To demonstrate that HCMV infection re-sults in the specific transactivation of promoters containingcharacterized NF-KB cis elements, NHF cells were trans-fected with the appropriate promoter-CAT constructs (24).As shown in Fig. 1, a promoter containing essentially only aTATA sequence (A56) was sufficient for transactivation byHCMV infection. The ability to activate promoters contain-ing only a TATA box upon HCMV infection has beenobserved previously (30-32). The combination of wild-typeMHC NF-KB-promoter-CAT construct (MHC) and virusinfection resulted in a 12-fold stimulation of CAT activityover that observed with A56 or a plasmid bearing mutantNF-KB-CAT (MHC mut) sequences. Specific viral transac-tivation through the 18-bp repeat of the HCMV major TEenhancer/promoter region (which contains an NF-KB bind-ing sequence) has also been observed (6, 7).HCMV Infection Induces Nuclear DNA-Binding Proteins

Specific for NF-cB Sequences. To determine if the observedenhanced promoter transactivation was associated with in-creased nuclear DNA-binding proteins, nuclear extracts frommock- and HCMV-infected fibroblasts were analyzed fortheir ability to bind to NF-KB sequences. EMSAs of mock-and virus-infected nuclear extracts using MHC NF-KB probesequences are shown in Fig. 2. The mock-infected extracts

S

Reporter plasmid

HCMV infection

°o Acetylation

56

Io-ln+D.40.1 0 5.4

IF*o I

MHC mu, MrHc(

0.153 57 0.22 if 4

FIG. 1. HCMV infection results in enhanced promoter transac-tivation via NF-KB cis elements. NHF cells were transfected induplicate with the appropriate CAT construct. Halfofthe plates weresubsequently infected with virus (+), and all of the plates wereincubated for an additional 48 h. Harvested cells were processed andassayed (50 ,ug) for CAT activity. Acetylated forms of chloramphen-icol were resolved by thin-layer chromatography, identified byautoradiography, and quantitated by liquid scintillation spectrome-try.

contained low levels of proteins capable of binding NF-KBsequences by this assay; two complexes (II and III) wereapparent. However, large increases in retarded DNA wereobserved by 12 h post-infection (hpi); at least four majorcomplexes of different mobilities were obtained (complexesI-IV). These complexes are better differentiated in theshorter exposures shown in Fig. 3. All four complexesremained over the course of infection, although by 65 hpi thelevels of DNA binding had diminished somewhat (Fig. 2).

r Extract: Nuclear NHIF----+ HCMV

- Probe MHC NF- B

Free probe* jJ

FIG. 2. EMSA of nuclear extracts (7 M~g) from mock-infected orinfected NHF cells using double-stranded MHC NE-KB DNA as theprobe. Nuclear extracts were prepared from uninfected (mock) orinfected NHF cells at the times indicated. The four major complexesobserved in this and other EMSAs (Fig. 3) are numbered (I-IV), andfree probe is indicated. Lanes 2-4 have been overexposed to showthe presence of complexes II and III in lane 1.

Proc. Natl. Acad Sci. USA 90 (1993)

Proc. Natl. Acad. Sci. USA 90 (1993) 1109

A-Extract: 12 hpl

Probe K-- -- MHC NF-KB -----

Competitor H-MHC NF-KBF-F- MHC rnut

x x- x o 0

LO I- LO

X x- X Q 02

LO10

Sm111IV -* Is a am _ i a

1 2 3 4 5 6 7 8

Br-Extract: 12 hpi-----F-Probe: MHC NF-KB---i

MAD-3 IK:B - + +

KBF-1 - - + - +(t. rel - _ +

IVl. " |||1+ihifl;

Ills

1 2 3 4 5

FIG. 3. Characterization of the major complexes observed inEMSAs of HCMV-infected nuclear extracts. Radiolabeled MHCNF-KB probe was used in each assay. (A) Competition assays usingwild-type and mutant competitor DNAs. Either a 0- (-), 5- (5x), 10-(10x), or 50- (50x) fold excess of unlabeled competitor was incu-bated with nuclear extract prior to the addition of radiolabeled probeto the binding reaction. (B) Identification of proteins in retardedcomplexes. Binding reactions were preincubated with nuclear ex-tract alone (lane 1) or in the presence of GST-MAD-3 IKB (200 ng;lane 2), anti-KBF-1 antiserum (antibody 3; kindly provided by A.Israel, Institut Pasteur, Paris; lane 3), or anti-rel antiserum (gift fromN. Rice, Frederick Cancer Center, Frederick, MD; lane 4). In lane5, a combination of IKB and anti-KBF-1 antiserum was used toenhance the KBF-1 supershift (Ills). After the appropriate preincu-bations in A and B, MHC NF-KB probe was added and assayed as inFig. 2.

This is in contrast to the more transient nuclear NF-KBbinding activity observed in serum-stimulated mouse 3T3fibroblasts (24, 33).

Competition experiments with excess wild-type or mutantNF-KB sequences in EMSAs are shown in Fig. 3A. Theaddition of a 50-fold excess of unlabeled MHC NF-KB DNAcompletely blocked by competition all but one (complex IV)ofthe shifted complexes (Fig. 3A, lane 4). Complex I was alsopoorly blocked by wild-type NF-KB sequence; it requiredmore than a 10-fold excess of unlabeled competitor. How-ever, the formation of the other two complexes (II and III)was readily blocked by a 10-fold excess of unlabeled DNA(Fig. 3A, lane 3). Competition with excess mutant NF-KBsequences resulted in minimal decreases in complexes I, II,and IV and increased binding of complex III (Fig. 3A, lanes6-8). The several minor complexes migrating with mobilitiesslower than complex I were blocked by competition withboth wild-type and mutant NF-KB sequences and were notfurther characterized.

IKB and antibodies were added to DNA-binding reactionsto determine which proteins were involved in specific bindingto the wild-type NF-KB sequence and presumably associated

with enhanced CAT activity afterHCMV infection. Inclusionof bacterially expressed GST-MAD-3 IKB in the DNA-binding reaction resulted in the disappearance of complex IIfrom nuclear extracts of virus-infected cells (Fig. 3B, lane 2).This suggested that complex II contained either p65 or c-relbecause the DNA binding of these proteins has been shownto be inhibited by the MAD-3 protein (14-16). Increasedlevels of complex III were also observed in this experiment.Similar results have been described elsewhere (15, 16). Theaddition of antibodies that specifically supershift complexescontaining c-rel had no effect on any of the complexesobserved by EMSA (Fig. 3B, lane 4). Inclusion ofan antibodythat specifically supershifts p50/pSO DNA complexes(KBF-1) resulted in the disappearance of complex III (Fig.3B, lanes 3 and 5). As there was little complex III in theinfected nuclear extracts, the supershifted material was moreapparent after prolonged exposure of the gel to x-ray film.That complex II was NF-KB (p50/p65), or a very similarcomplex, was inferred by its dissolution by MAD-3 1KB, thelack of interaction with the rel antibody, and comigrationwith the NF-KB complex observed in phorbol ester-stimulated HeLa and Jurkat cell nuclear extracts (T.F.K.,unpublished observations).The composition of complexes I and IV is not clear. The

competition experiments suggested that they were less spe-cific than complex II and III. In fact, they comigrate withnonspecific DNA-binding activities observed in phorbol es-ter-stimulated Jurkat cells. However, their exact identity hasyet to be determined.HCMV Infection Increases the Levels of mRNAs Encoding

p50, p65, and MAD-3 bIcB. Because the major HCMV infec-tion-induced nuclear DNA-protein complex contained p50and p65, Northern blots of total RNA from mock- andHCMV-infected cells were probed with radiolabeled cDNAsrepresenting these two genes (Fig. 4). Maximal levels of p50RNA were observed at 4 hpi and remained elevated over thecourse of virus infection. This contrasts with the patternobserved with p65 gene expression, where RNA levels grad-ually increased over time and were maximal at 48-72 hpi,suggesting that these two genes are differently regulated uponHCMV infection. For comparison, blots were also probed forexpression of MAD-3 IKB mRNA. Accumulation of thistranscript displayed kinetics similar to p65 except that in-creasing amounts of RNA were still being observed at latetimes of infection (72 hpi). The increased levels of RNAencoding p50 and p65 at these later times were likely the

p50U

p65

FIG. 4. HCMV infection increases steady-state levels of RNAsrepresenting p5O, p65, and 1KB. Mock- and HCMV-infected NHFcells were harvested, and total RNA was prepared. The 0 hpi timepoint was harvested immediately after virus absorption. A Northernblot ofRNA from 5 x 105 cells per time point was sequentially probedwith nick-translated cDNAs encoding p5O (p105), p65, and MAD-31KB. RNA integrity was monitored by ethidium bromide staining andUV visualization of duplicate lanes in the same agarose gel (notshown).

Biochemistry: Kowalik et al.

1110 Biochemistry: Kowalik et al.

source of proteins for the sustained nuclear NF-KB bindingactivities. This is true provided that the IKB transcriptspresent at these time points were not translated or that theresultant IKB protein was rapidly inactivated. These resultsdiffer from previous studies in other systems where themRNA accumulation pattern for the p50 or IKB gene wastransient (14, 34-36).

Virus Infection Results in Decreased Cytosolic Stores ofp50/p65. Because the delayed expression ofp65 RNA did notcompletely correlate with the early nuclear p5O/p65 activityand the concomitant activation of MAD-3 IKB expressionwould, in principle, sequester this DNA-binding activity,cytosolic extracts of mock- and HCMV-infected cells wereassessed for the presence of p50/p65 stores. Upon treatmentof these extracts with deoxycholate (11), two complexes (Icand I1c) were observed by EMSA in all of the times pointstested (Fig. 5, lanes 1-5). Virtually no complexes wereobserved without deoxycholate treatment (Fig. 5, lane 7).Both complex Ic and complex 1Ic were reduced 3.5-fold in the4 hpi cytosolic extract and remained suppressed relative tomock-infected extracts in the later time points. A 3- to 4-foldreduction between the mock and 4 hpi time points was highlyreproducible among assays and individual sets of cytosolicpreparations (T.F.K., unpublished observations). The mo-bility of complex HIc was similar to the complex II observedin infected nuclear extracts. Mobilities similar to complex Icwere not observed in any of the nuclear extracts. Theinclusion ofGST-MAD-3 1KB in the binding reaction resultedin disruption of both cytosolic complexes (Fig. 5, lane 6).Together, these results suggest that the initial activation ofnuclear NF-KB binding activity is at least partly due to therelease of cytosolic stores ofNF-KB and the apparent nucleartranslocation of these components upon HCMV infection.

Virus Infection Alters the Levels of a 39-kDa Protein ThatCross-Reacts with Anti-MAD-3 IcB and Anti-pp4O Antisera. AWestern blot of mock- and HCMV-infected cell lysates wasprobed with a polyclonal antiserum generated against anN-terminal peptide of MAD-3 1KB (Fig. 6A). On the basis ofcompetition experiments with excess peptide, one protein of39 kDa reacted specifically with the antibody (Fig. 6A). Adramatic decrease of the steady-state levels of this protein

r-- Extract: Cytoplasmic+ - HCMV

- Probe: MHC NF-KB----m

-Q Z- - ao - Tv N 0 < 00C

,*- C14 R r-_C

Icy d-

1 2 3 4 5 6 7

FIG. 5. Characterization ofNF-KB binding activity in cytosols ofmock- and HCMV-infected NHF cells. Cytosols were prepared asS100 extracts after cell lysis by Dounce homogenization, and theextracts (3 gg) were analyzed by EMSA with the MHC NF-KBsequence as the DNA probe. In lanes 1-5, cytosolic extracts werepretreated with deoxycholate and Nonidet P-40 (11) prior to theiraddition to the binding reactions. The EMSA shown in lane 6 wasprocessed in a manner similar to lane 1 except that GST-MAD-3 IKB(200 ng) was included in the binding reaction. The EMSA ofmock-infected cytosolic extract in lane 7 was not pretreated withdetergents. Results similar to that observed in lane 7 were alsoobtained with infected cytosolic extracts without detergent pretreat-ment (not shown).

A

B

- _1 U

FIG. 6. Western blot assays for IKB in mock- and HCMV-infected NHF cells. (A) Western blot probed with a polyclonalantiserum developed against a peptide representing the N terminusof MAD-3 IKB (antibody 9). Total proteins (20 lug) from mock- andvirus-infected NHF cells were resolved by SDS/10%7 PAGE, elec-troblotted onto a nitrocellulose membrane, and probed with antibody9. (B) Western blot using an anti-pp4O polyclonal antiserum (a giftfrom H. Bose, University of Texas, Austin). Mock and 48 hpisamples were the same as in A. Lane 3 contains purified MAD-3 1KB(500 ng) expressed in bacteria (ref. 15; a gift from C. Rosen, RocheInstitute, Nutley, NJ).

occurred subsequent to 24 hpi. By 48-72 hpi, this proteinspecies could no longer be detected in this assay. In addition,antiserum generated against pp40 (17), a protein isolated fromchicken and having sequence and biochemical propertiessimilar to MAD-3 IKB (17-19), was found to cross-react withMAD-3 IKB (Fig. 6B). This antibody also recognized asimilarly sized protein in uninfected cell lysates, but little ornone was visible in lysates derived from cells 48 h afterHCMV infection (Fig. 6B). The disappearance of IKB pro-tein(s) from infected cells, even though its mRNA levels haveincreased, could explain the reduced cytosolic stores ofNF-KB (Fig. 4) and the presence of sustained levels ofnuclearNF-KB at later times of infection (Fig. 2). However, IKBwould need to be inactivated through an additional mecha-nism(s) at the earlier time points (compare Fig. 5, lane 2 withFig. 6A, lane 3). It seems plausible that the turnover of IKBat later times of infection could be signaled by this initialinactivation. Finally, at time points where little or no IKBprotein could be detected by the MAD-3 IKB and pp4Oantibodies, there were still significant cytosolic stores ofNF-KB, suggesting that there could be undetectable levels ofMAD-3 IKB remaining, or possibly the presence of immuno-logically distinct IKB species that were resilient to inactiva-tion upon HCMV infection.

DISCUSSIONIn this study, we have shown that NF-KB is highly inducedin the nuclei of HCMV-infected fibroblasts. This activationcorrelates with the specific transactivation of a promotercontaining NF-KB sequences. Induction of nuclear NF-KB byvirus infection is apparently the result of several factors: (i)releasing the majority of the cytosolic stores of NF-KB, (ii)increasing the levels of mRNAs encoding p50 and p65, and(iii) removing IKB activity from the cell. The first twomechanisms have been hypothesized as key control points inthe regulation of NF-KB. Temporally, cytosolic NF-KB ac-tivity is released very early in the infection cycle. Aftertranslocating to the nucleus, it activates expression of theHCMV MIE genes and probably the gene encoding p50(p105), each of which contains NF-KB binding elements in its

Proc. NatL Acad Sci. USA 90 (1993)

Proc. NatL. Acad. Sci. USA 90 (1993) 1111

promoter (7, 37). We and others have observed nuclearNF-KB activity by 3-4-hpi (ref. 7; T.F.K., unpublishedobservation) and, as herpesvirus IE genes are defined asbeing transcriptionally active in the absence of protein syn-thesis, the contribution of NF-KB to the activation of theHCMV MIE promoter is further substantiated. That theHCMV MIE promoter is dependent, at least initially, uponNF-KB in a productive infection has been demonstrated (7).This activation ofthe HCMV MIE promoter by NF-KB wouldsubsequently develop into a positive feedback loop in whichone of the gene products of the MIE region, IE1, stimulatesgreater MIE promoter activity through the multiple 18-bprepeats contained within this promoter. The activation ofp65and MAD-3 IKB genes at later times suggests that theirpromoters are regulated by different transcription factors.Cloning of these promoters is needed to facilitate studies onthe separate and coordinate regulation of the p5O (p105), p65,and MAD-3 IKB genes.

Recently, additional components involved in the NF-KBsystem have been cloned. The mRNA of one factor, p49(p100) or pSOB (38, 39), is not induced by HCMV infection(T.F.K., unpublished observation). Inoue et al. (40) identi-fied in lymphoid cell lines a spliced version of p105 mRNAresulting in a protein with IKB activity. Although the 2.6-kbmRNA for this protein was not observed in either mock- orHCMV-infected fibroblasts (Fig. 4), a role for this IKB cannotbe ruled out at this point. Rivibre et aL (41) have demon-strated that the human immunodeficiency virus protease canprocess the inactive p105 precursor to an active 50-kDa form.Activation of this NF-KB component by such a mechanismneeds to be addressed in the HCMV system.

Viral infection results in the disappearance of at least oneform ofIKB. The mechanism for this turnover is not apparent,but it could possibly occur by signaling degradation throughthe "PEST" sequence in the C-terminal half of MAD-3 1KB(14). Regardless of the mechanism, there is a discrepancybetween the late turnover of MAD-3 IKB and the earlyactivation of cytosolic NF-KB. This suggests that an addi-tional mechanism(s) to inactivate IKB may be involved.Phosphorylation is a means by which IKB can be inactivatedin vitro (42) and, as HCMV has been shown to activatesignaling pathways associated with cytosolic kinases (43),inactivation of IKB by phosphorylation in infected cellsremains a possibility.

We would like to acknowledge Shu-Mei Huong for technicalassistance and Robert Scheinman and Michael Wade for helpfuldiscussions. This work was supported by Public Health ServiceGrants CA52515 (A.S.B), CA21773 (E.-S.H), and A112717 (E.-S.H.)from the National Institutes of Health and by the March of Dimes(1-FY91-0286) (J.C.A.). A.S.B. was also supported by an AmericanCancer Society Junior Faculty Award (JFRA-309) and a March ofDimes Basil O'Conner Award. T.F.K. was supported by a DamonRunyon-Walter Winchell Cancer Research Fund Fellowship (DRG-1097).

1. Huang, E.-S. & Kowalik, T. F. (1992) in Molecular Aspects ofHuman Cytomegalovirus Disease, eds. Becker, Y. & Darai, G.(Springer, Heidelberg), in press.

2. Boshart, M., Weber, F., Jahn, G., Dorsch-Hasler, K., Fleck-enstein, B. & Schaffner, W. (1985) Cell 41, 521-530.

3. Stinski, M. & Roehr, T. J. (1985) J. Virol. 55, 431-441.4. Nelson, J. A. & Groudine, M. (1986) Mol. Cell. Biol. 6,

452-461.5. Ghazal, P., Lubon, H. & Hennighausen, L. (1988) Mol. Cell.

Biol. 8, 1809-1811.6. Cherrington, J. J. & Mocarski, E. S. (1989) J. Virol. 63, 1435-

1440.7. Sambucetti, L. C., Cherrington, J. M., Wilkinson, G. W. G. &

Mocarski, E. S. (1989) EMBO J. 8, 4251-4258.

8. Gilmore, T. D. (1990) Cell 62, 841-843.9. Sen, R. & Baltimore, D. (1986) Cell 46, 705-716.

10. Fan, C.-M. & Maniatis, T. (1991) Nature (London) 354, 395-398.

11. Baeuerle, P. A. & Baltimore, D. (1988) Cell 53, 211-217.12. Baeuerle, P. A. & Baltimore, D. (1988) Science 242, 540-546.13. Zabel, U. & Baeuerle, P. A. (1990) Cell 61, 255-265.14. Haskill, S., Beg, A. A., Tompkins, S. M., Morris, J. S., Yu-

rochko, A. D., Sampson-Johannes, A., Mondal, K., Ralph, P.& Baldwin, A. S. (1991) Cell 65, 1281-1289.

15. Beg, A. A., Ruben, S. M., Scheinman, R. I., Haskill, S.,Rosen, C. A. & Baldwin, A. S. (1992) Genes Dev. 6, 1899-1913.

16. Duckett, C. S., Perkins, N. D., Kowalik, T. F., Schmid,R. M., Baldwin, A. S., Huang, E.-S. & Nabel, G. J. (1992)Mol. Cell. Biol., in press.

17. Davis, J. N., Bargmann, W. & Bose, H. R. (1990) J. Virol. 64,584-591.

18. Davis, N., Ghosh, S., Simmons, D. L., Tempst, P., Liou,H.-C., Baltimore, D. & Bose, H. R. (1991) Science 253, 1268-1271.

19. Kerr, L. D., Inoue, J.-I., Davis, N., Link, E., Baeuerle, P. A.,Bose, H. J. & Verma, I. M. (1991) Genes Dev. 5, 1464-1476.

20. Bauerle, P. A. & Baltimore, D. (1991) in Hormonal ControlRegulation of Gene Expression, eds. Cohen, P. & Foulkes,J. G. (Elsevier Biomedical Press, Amsterdam), 409-432.

21. Huang, E.-S. (1975) J. Virol. 16, 298-310.22. Wade, M., Kowalik, T. F., Mudryj, M., Huang, E.-S. &

Azizkhan, J. C. (1992) Mol. Cell. Biol. 12, 4364-4374.23. Gorman, C., Moffat, L. F. & Howard, B. (1982) Mol. Cell.

Biol. 2, 1044-1051.24. Baldwin, A. S., Azizkhan, J. C., Jensen, D. E., Beg, A. A. &

Coodly, L. R. (1991) Mol. Cell. Biol. 11, 4943-4951.25. Dignam, J. D., Lebovitz, R. M. & Roeder, R. G. (1983) Nu-

cleic Acids Res. 11, 475-489.26. Baldwin, A. S. (1990) DNA Protein Eng. Tech. 2, 73-76.27. Smith, D. B. & Johnson, K. S. (1988) Gene 67, 31-40.28. Ruben, S. M., Dillon, P. J., Schrech, R., Henkel, T., Chen,

C. H., Maher, M., Baeuerle, P. A. & Rosen, C. A. (1991)Science 251, 1490-1493.

29. Kieran, M., Blank, V., Legeat, F., Vandkerckhove, J., Lott-seich, F., Le Bail, O., Urban, M. B., Kourilsky, P., Baeuerle,P. A. & Israel, A. (1990) Cell 62, 1007-1018.

30. Biegalke, B. J. & Geballe, A. P. (1991) Virology 183, 381-385.31. Hagemeier, C., Walker, S., Caswell, R., Kouzarides, T. &

Sinclair, J. (1992) J. Virol. 66, 4452-4456.32. Walker, S. M., Hagemeier, C., Sissons, J. G. P. & Sinclair,

J. H. (1992) J. Virol. 66, 1543-1550.33. Olashaw, N. E., Kowalik, T. F., Huang, E.-S. & Pledger,

W. J. (1992) Mol. Biol. Cell 3, 1131-1139.34. Bours, V., Villalobos, J., Burd, P. R., Kelly, K. & Sienbenlist,

U. (1990) Nature (London) 348, 76-80.35. Sporn, S. A., Eirman, D. F., Johnson, C. E., Morris, J., Mar-

tin, G., Ladner, M. & Haskill, S. (1990) J. Immunol. 144,4434- 441.

36. Tewari, M., Dobrzanski, P., Mohn, K. L., Cressman, D. E.,Hsu, J.-C., Bravo, R. & Taub, R. (1992) Mol. Cell. Biol. 12,2898-2908.

37. Ten, R. M., Paya, C. V., Israel, N., Le Bail, O., Mattei, M.-G.,Virelizier, J.-L., Kourilsky, P. & Israel, A. (1992) EMBO J. 11,195-203.

38. Schmid, R. M., Perkins, N. D., Duckett, C. S., Andrews,P. C. & Nabel, G. J. (1991) Nature (London) 352, 733-736.

39. Bours, V., Burd, P. R., Brown, K., Villalobos, J., Park, S.,Ryseck, R.-P., Bravo, R., Kelly, K. & Siebenlist, U. (1992)Mol. Cell. Biol. 12, 685-695.

40. Inoue, J.-I., Kerr, L. D., Kakizuka, A. & Verma, I. M. (1992)Cell 68, 1109-1120.

41. Riviere, Y., Blank, V., Kourilsky, P. & Israel, A. (1991) Nature(London) 350, 625-626.

42. Ghosh, S. & Baltimore, D. (1990) Nature (London) 344, 578-582.

43. Albrecht, T., Boldogh, I., Fons, M., Lee, C. H., AbuBakar, S.,Russell, J. M. & Au, W. W. (1989) Subcell. Biochem. 15,157-202.

Biochemistry: Kowalik et al.