Proc. Nati. Acad. Sci. USAVol. 91, pp. 11884-11888, December 1994Biochemistry

Inducible phosphorylation of IKBa is not sufficient for itsdissociation from NF-KB and is inhibited by protease inhibitorsTIMOTHY S. FINCO*t, AMER A. BEG*t§, AND ALBERT S. BALDWIN, JR.*t$¶*Lineberger Comprehensive Cancer Center, tCurriculum in Genetics and Molecular Biology, and tDepartment of Biology, University of North Carolina,Chapel Hill, NC 27599

Communicated by Thomas P. Maniatis, July 27, 1994

ABSTRACT The ubiquitous transcription factor NF-KB isregulated by its cytoplasmic inhibitor IKB. A variety of cellularstimuli cause the dissociation of NF-#cB from IKB, allowingNF-KB to translocate to the nucleus and regulate gene expres-sion. Although the activation ofNF-KB in vivo is associated withthe phosphorylation and degradation of IKBa, it has remainedunclear how each of these events contributes to this process.Recently, studies utilizing protease inhibitors have suggestedthat the proteolysis of IKBa is a necessary event in theactivation of NF-ucB. We demonstrate in this study that theseand an additional protease inhibitor also completely repressinducible phosphorylation of IKBa. This surprising resultsuggests a more complex role of proteases in NF-KB activation.In addition, data presented here indicate that many of theseinhibitors also directly modify NF-KB and inhibit its DNAbinding activity. Due to the pleiotropic effects of these proteaseinhibitors, it is difficult to conclude from their use how IKBaphosphorylation and degradation contribute to NF-KB activa-tion. In the present study, a more direct approach demon-strates that phosphorylation of IKBa alone is not sufficient forNF-KB activation.

The NF-KB/Rel transcription factor family regulates a vari-ety of genes whose products are involved in diverse biolog-ical processes including cell growth, inflammation, and im-mune responses (1, 2). Members of this family, which in-cludes the proteins NFKB1 (p105/p50), NFKB2 (p100/p52),RelA (p65), c-Rel, and RelB, are present within cells asfunctionally-distinct homo- and heterodimers (1, 2). Induciblecomplexes of this family are collectively referred to asNF-KB. Typically, inactive NF-KB is found in the cytoplasmcomplexed to its inhibitor, IKB (3, 4). Treatment of cells witha large array of inducers, including tumor necrosis factor a(TNFa), interleukin 1,3, and lipopolysaccharide (LPS), re-sults in the dissociation of NF-KB from IKB. This allowsliberated NF-KB to translocate to the nucleus and modulatetarget gene expression (1-4). Initial in vitro studies on themechanism of NF-KB activation have suggested that phos-phorylation of IKBa may be sufficient for its release fromNF-KB (5, 6). Consistent with this model, a phosphorylatedform of IKBa can be detected after treatment of cells withvarious inducers of NF-KB (7, 8). However, the role of IKBaphosphorylation in vivo has yet to be determined. It was alsoobserved in vivo that IKBa was rapidly degraded after itsphosphorylation (7-10). The ability of activated NF-KB todirectly stimulate IKBa gene expression (11-13) results in thereaccumulation of inactive NF-KB-IKBa complexes in thecytoplasm.More recent studies have addressed whether events in

addition to phosphorylation may be required for NF-KBactivation. In particular, the role of IKB proteolysis has beeninvestigated (10, 11, 14-18). It has been demonstrated that

IKBa is extremely unstable unless complexed with NF-KBfamily members (10, 15). Thus, it was proposed that thedegradation of IKBa observed in vivo was a consequence ofits dissociation from NF-KB and not a prerequisite for acti-vation. However, other work has suggested that degradationof IKBa was necessary for NF-KB activation. It was foundthat pretreatment of cells with a variety of serine proteaseinhibitors prevented not only the degradation of IKBa butalso the activation of NF-KB (11, 16-18). Here we demon-strate that many of the protease inhibitors utilized in theseaforementioned studies have additional effects, including anability to directly modify NF-KB and block its interactionwith DNA. Significantly, all of the protease inhibitors effec-tive in the repression of IKBa degradation also prevent itsinducible phosphorylation. This dramatic result suggests thatproteases may regulate multiple events necessary for NF-KBactivation. Since these protease inhibitors interfere with boththe phosphorylation and degradation of IKBa, it is difficult toascertain from their use how each of these events contributesto NF-KB activation. As reported here, the finding thatinduced phosphorylation of IKBa does not cause its rapiddissociation from NF-KB is an important step in elucidatingthe mechanism(s) of NF-KB activation.

MATERIALS AND METHODSCell Culture and Reagents. HeLa S3 and Jurkat T cells were

grown as described (8). THP-1 cells were grown in RPMI1640 medium containing 10%o (vol/vol) fetal calf serum,penicillin (100 units/ml), and streptomycin (100 pg/ml). LPS,3,4-dichloroisocoumarin (DCIC), 7-amino-1-chloro-3-tosylamido-2-heptanone ("N1'-p-tosyl-L-lysine chloromethylketone," Sigma; TLCK), L-1-tosylamido-2-phenylethyl chlo-romethyl ketone ("N-tosyl phenylalanine chloromethyl ke-tone," Sigma; TPCK), and N-benzoyl-L-tyrosine ethyl ester(BTEE) were obtained from Sigma; 4-(2-aminoethyl)benze-nesulfonyl fluoride (AEBSF) was from Boehringer Mann-heim; TNFa was from Promega; and okadaic acid was fromLC Laboratories (Woburn, MA). TPCK was dissolved inethanol; TLCK was dissolved in 20 mM sodium phosphate;DCIC, BTEE, and okadaic acid were dissolved in dimethylsulfoxide; AEBSF was dissolved in H20; and LPS wasdissolved in phosphate-buffered saline (PBS). These solventshad no detrimental effect in vivo or in vitro on NF-#cBactivation or its ability to bind DNA. Concentrations ofprotease inhibitors used and time ofincubation were not toxic

Abbreviations: LPS, lipopolysaccharide; TNFa, tumor necrosisfactor a; EMSA, electrophoretic mobility shift assay; TPCK, L-1-tosylamido-2-phenylethyl chloromethyl ketone; TLCK, 7-amino-1-chloro-3-tosylamido-2-heptanone; DCIC, 3,4-dichloroisocoumarin;BTEE, N-benzoyl-L-tyrosine ethyl ester; AEBSF, 4-(2-aminoethyl)-benzenesulfonyl fluoride; DOC, deoxycholate; DTT, dithiothreitol;NP-40, Nonidet P-40.§Present address: Department of Biology, Massachusetts Institute ofTechnology, Cambridge, MA 02139.$To whom reprint requests should be addressed.

11884

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.

Dow

nloa

ded

by g

uest

on

Janu

ary

21, 2

021

Proc. Nadl. Acad. Sci. USA 91 (1994) 11885

to cells and are detailed in Fig. 1 unless otherwise indicated.Okadaic acid was used at 1 pM, LPS was at 10 ,g/ml, andTNFa was at 10 ng/ml. Bacterially expressed ReIA was a giftfrom C. Rosen (Human Genome Sciences, Rockville, MD).

Preparation of Extracts and Electrophoretic Mobility ShiftAssays (EMSAs). Nuclear and cytoplasmic extracts wereprepared as described (8). Samples standardized by proteinconcentration (19) were analyzed by EMSA as detailed (20).Probe sequences were as follows: NF-KB, 5'-CAGGGC-TGGGGATTCCCCATCTCCACAGTTTCACTTC-3';AP-1, 5'-TTCCGGCTGACTCATCAAGCG-3; OCT-1, 5'-TTCACGCGGTAATGAGATGGGTT-3'; SRE, 5'-CCTT-TACAACAGGATGTCCATATTAGGACATCTGCGT-CAGCAG-3'. Deoxycholate (DOC) activation of NF-KB invitro was performed by treating cytoplasmic extracts with0.8% DOC for 5 min followed by the addition ofNonidet P-40(NP-40) to 1.6% (4). For in vitro experiments, nuclear andcytoplasmic extract or 1 A4 ofbacterially expressed RelA wasincubated for 1 h at 40C in binding buffer (20) lackingdithiothreitol (DTT) but containing the indicated concentra-tions of protease inhibitors. For some experiments DTT wasalso added. Cytoplasmic extracts were then incubated withDOC/NP-40. Subsequently, DTT, dI/dC, and labeled probewere added and samples were analyzed by EMSA.

Immunoprecipitatlon and Western Blot Analysis. Polyclo-nal RelA antibody (Rockland, Boyertown, PA; 2 jul of serum)was added to 400 /lg of cytoplasmic extract and the volumewas adjusted to 200 pI with lysis buffer (10 mM Hepes, pH7.6/60 mM KCl/1 mM EDTA/0.3% NP-40/1 mM DTT/1mM phenylmethylsulfonyl fluoride). The samples were incu-bated for 2 h at 40C with mixing. The extract was then addedto 20 pd of protein A-Sepharose previously blocked with 1%bovine serum albumin and incubated for another 2 h at 40Cwith mixing. Pelleted protein A-Sepharose was washed twicewith 500 A4 of lysis buffer and once with lysis buffer lackingNP-40 and then resuspended in 20 p4 of buffer (withoutNP-40). DOC was then added to 0.8% to release IKBa. UsingDOC instead of boiling to dissociate IKBa from NF-KBsignificantly reduced the background on Western blots. Thesupernatant, containing IKBa previously bound to RelA, wasthen analyzed on Western blots. For Western blot analysis,samples were separated by SDS/PAGE, transferred to ni-trocellulose, and probed with the IKBa antibody (Rockland).

RESULTSProtease Inhibitors Prevent NF-KB Activation as Assayed by

EMSAs. Consistent with previous results (16), pretreatmentof HeLa or Jurkat T cells with the serine protease inhibitorTPCK, TLCK, BTEE, orDCIC prevented subsequentTNFainduction of NF-KB as quantitated by EMSAs of nuclearextracts (Fig. 1A). We demonstrate here that another serineprotease inhibitor, AEBSF, had a similar effect. Other pro-tease inhibitors, including chymostatin, leupeptin, and Na-benzoyl-L-arginine ethyl ester, did not prevent NF-KB acti-vation when tested at various concentrations (data notshown). To demonstrate that NF-KB from cells treated withprotease inhibitors retains the capacity to bind DNA, EMSAsof corresponding cytoplasmic extracts were performed aftertreating extracts with the disrupting agent DOC. DOC acti-vates NF-KB in vitro by causing its dissociation from IKB (4).Surprisingly, many ofthe protease inhibitors (TPCK, TLCK,and DCIC) suppressed the ability ofDOC-activated NF-KB tobind DNA (Fig. 1B). As a control, it was confirmed that DOCcan dissociate IKB from NF-KB in extracts from cells treatedwith any of the protease inhibitors (data not shown). BTEEand AEBSF did not negatively affect NF-KB binding (Fig.1B). Pretreatment of cells with lower concentrations ofprotease inhibitors for shorter periods yielded similar results.However, more consistent results could be obtained at the

AExtract:TNFu v:

HeLa NE

Protease Inh: - - L Ad to

NF-KB + U

ris --

BExtract:DOC:TNF:RC

Protease Inh: -

NF-KB -*O -SUE

cExtract:TNFu.:Protease Inh

lurkalt NE

.. '': .i-,

-. <

* - m

HeLa GE JurkatCJE

a\ :' LL- tii) {..

-. iL, ci

.f 5an

HeLa NE+ - 4- 4

J frj L Cl

--C

AP-i "ON I

OCT-1 are"oo

SRE W0.

FIG. 1. In vivo effects of protease inhibitors on NF-KB activationand DNA binding. HeLa and Jurkat T cells were pretreated withprotease inhibitors and then stimulated with TNFa for 15 min. Finalconcentration of inhibitors: TPCK, 50 paM; TLCK, 300 tM; BTEE,2 mM (HeLa) or 1 mM (Jurkat); AEBSF, 1 mM (HeLa) or 2 mM(Jurkat); DCIC, 45 ,uM. The preincubation time prior to TNFaaddition was 30 min for TPCK, TLCK, BTEE, and DCIC and was180 (HeLa) or 90 (Jurkat) min for AEBSF. (A) Nuclear extracts(NEs), isolated from cells treated as indicated above each lane, wereassayed by EMSA. Two major NF-KB complexes are induced byTNFa in HeLa cells but only one was induced in Jurkat cells. Inh.,inhibitor. +, Added; -, not added. (B) Cytoplasmic extracts (CEs)corresponding to the NEs used in A were either directly assayed orwere first treated with 0.8% DOC to dissociate IKB from NF-KB andthen assayed. Two NF-KB complexes are induced by DOC. InA andB, NF-KB is indicated by a large arrow and the nonspecific band (ns)is indicated by a small arrow. Free probe is not shown. (C) HeLaNEs used in A were examined by EMSA for binding of othertranscription factors. The complexes specific for each probe werecompeted with a 100-fold excess identical unlabeled probe but notwith a 100-fold excess KB probe (data not shown). Only specificcomplexes are shown. The identity of each probe is indicated to theleft.

higher concentrations. EMSAs of nuclear extracts from cellstreated with only TPCK, TLCK, or DCIC indicated thatthese agents also suppressed basal NF-KcB DNA bindingactivity, including that of NFKB1 homodimers (data notshown).We also examined the effect ofthese protease inhibitors on

the binding of other transcription factors. AP-1 DNA bindingactivity in nuclear extracts from cells treated with TNFa andTPCK, TLCK, or DCIC (but not BTEE and AEBSF) wasalso inhibited (Fig. 1C). However, not all proteins wereaffected similarly. The binding of Oct-i and a protein com-plex that specifically binds the serum response element werenot significantly altered by the various protease inhibitors.Therefore, the inhibitory activities of TPCK, TLCK, andDCIC on NF-KB and AP-1 DNA binding potential appearspecific. These results imply that TPCK, TLCK, and DCIC,in addition to their antagonism of protease activity, may

Biochemistry: Finco et al.

." iu J.'Il ..L T.

- '-.

Dow

nloa

ded

by g

uest

on

Janu

ary

21, 2

021

Proc. Natl. Acad. Sci. USA 91 (1994)

directly modify NF-KB and alter its ability to interact withDNA. However, it is also conceivable that these agentsindirectly affect NF-KB and its DNA binding potential. Toexplore these possibilities, we tested the effects of theseinhibitors on NF-KB binding in vitro.

Modification of NF-cB by Certain Protease Inhibitors Re-presses its Ability to Interact with DNA. If the proteaseinhibitors directly modify NF-KB in vivo, a similar effect maybe reproduced in vitro. Alternatively, if the protease inhibi-tors initiate a cellular response that indirectly suppressesNF-KB DNA binding, an effect in vitro would not be expectedsince most cellular processes are inactive under these assayconditions. Nuclear extracts from TNFa-stimulated cells,containing activated NF-KB free of IKB, were investigatedfirst. In vitro treatment of these extracts with increasingconcentrations ofTPCK, TLCK, or DCIC inhibited the DNAbinding of activated NF-KB (Fig. 2A). Untreated cytoplasmicextracts containing inactive NF-KB complexed to IKB werethen analyzed. These extracts were treated with DOC afterincubation with protease inhibitors to allow analysis ofDNAbinding by NF-KB. As shown in Fig. 2B, TPCK, TLCK, andDCIC also inhibited the DNA binding potential of cytoplas-mic NF-KB. BTEE and AEBSF, protease inhibitors that didnot alter the binding of NF-KB when added in vivo (Fig. 1B),also had no effect in vitro (Fig. 2A andB and data not shown).Finally, the effects of these protease inhibitors on the bac-terially expressed RelA subunit ofNF-KB were examined. Inthese experiments, the possible contribution of other cellularproteins in mediating the effects of TPCK, TLCK, or DCICon NF-KB DNA binding was eliminated since only ReLA andexogenously added bovine serum albumin were present.Consistent with the above results, TPCK, TLCK, and DCICinhibited the ability of RelA to bind DNA (Fig. 2C). Theprotease inhibitors do not directly modify the KB DNA probeused in the EMSA since prolonged incubation of the probewith concentrations of TPCK, TLCK, or DCIC that com-pletely suppressed NF-KB binding in vivo and in vitro had noinhibitory effect when the probe was subsequently diluted (tosubstantially lower the concentration of protease inhibitor)and used in an EMSA (data not shown). Furthermore, asdetermined by Western blot analysis, NF-KB is not degradedin vivo or in vitro in the presence of TPCK, TLCK, or DCIC(data not shown). Thus, these results strongly suggest thatmany of the protease inhibitors previously utilized in studiesof NF-KB can directly interfere with the ability of thistranscription factor to bind DNA (also see Discussion).

In contrast to the results presented here, other investiga-tors (17, 21) have reported that TPCK does not inhibit theDNA binding activity of NF-KB in vitro. This discrepancymay be a result of the assay conditions in which the exper-iments were performed. We have found that DTT suppressesthe in vitro inhibitory activity of TPCK toward NF-KB DNAbinding (Fig. 2D). It is possible that other investigators didnot detect the inhibitory effects of TPCK in vitro becauseDTT was present. As our in vitro results are identical to thoseobserved in vivo, we believe that the in vitro conditions usedin this study more closely reflect protease inhibitor activitywithin cells.

Protease Inhibitors Block Both Phosphorylation and Degra-dation of IucBa in Vivo. Pretreatment of cells with TPCKsuppresses degradation of IKBa by inducers of NF-KB in-cluding phorbol 12-myristate 13-acetate, interleukin 1p3, andLPS (11, 15-17). We demonstrate here that the other testedserine protease inhibitors, in addition to TPCK, also re-pressed TNFa-induced degradation of IKBa in HeLa cells(Fig. 3A). It is well established that TNFa-induced degrada-tion of IKBa is preceded by its phosphorylation (8, 22).Therefore, in the absence of degradation, phosphorylatedIKBa would be expected to accumulate in cells. However,newly phosphorylated IKBa, as detected on a Western blot

N F--B _ AO

SO4------- - h-

B i.}e. );By Lf i' _

''

z fi ->e >R.; s C'_ A\! ;: It

j_ -

l<._. . .... .. _ .,._ ^ ._.=. ^ - .__.

,., 2

:., ._ _

: s 8 o

NF-[NB -O- a

C

i*IIA - 4

D50 ttM TPCKmM DTT

NF-,. B _-*25

msaa

FIG. 2. In vitro effects of protease inhibitors on NF-KB DNAbinding. (A) Protease inhibitors prevent DNA binding of activatednuclear NF-KB in vitro. Nuclear extracts from Jurkat cells stimulatedwith TNFa for 15 min were incubated at 40C for 1 h alone or withTPCK, TLCK, BTEE, or DCIC in the absence of DTT. DTT wasthen added and samples were analyzed by EMSA. The final con-centration of protease inhibitors is indicated. ns, Nonspecific band.(B) Protease inhibitors prevent DOC-induced binding of cytoplasmicNF-KB in vitro. Experiments using untreated Jurkat cytoplasmicextracts were performed as described in A except for the in vitroactivation of NF-KB by DOC after the 1-h incubation at 40C withprotease inhibitors. The final concentration of protease inhibitor foreach sample is shown. (C) Protease inhibitors prevent DNA bindingof the bacterially expressed RelA subunit of NF-KB. Experimentswere performed as in A by using 1 A4 of RelA and the indicatedconcentrations of protease inhibitors. (D) DTT suppresses the in-hibitory effect of TPCK on NF-KB binding. As indicated, HeLanuclear extracts from cells treated with TNFa for 15 min wereincubated for 1 h at 40C alone, with 50 /AM TPCK, or with 50 pMTPCK and the concentrations of DTT indicated. Afterwards, DTTwas added to 5 mM in all samples, which were then analyzed byEMSA. NF-KB and the nonspecific (ns) band are labeled to the left;free probe is not shown.

and based on its reduced mobility during SDS/PAGE (7, 8,22), was not observed in samples stimulated with TNFa for15 min in the presence of protease inhibitors (Fig. 3A). Thisresult indicates that protease inhibitors may prevent thephosphorylation of IKBa in addition to its degradation. An-other explanation is that IKBa is rapidly dephosphorylated bya cytoplasmic phosphatase in the absence of degradation. Todistinguish between these possibilities, cytoplasmic extractsfrom HeLa cells treated with TNFa for shorter periods oftime were examined. Stable phosphorylated IKBa can bedetected by Western blot analysis at these earlier time points

11886 Biochemistry: Finco et al.

if

a, C)

1.

M -7.1 ...

.1, .11 -:;,

Lb ..;f14 ,,

Dow

nloa

ded

by g

uest

on

Janu

ary

21, 2

021

Proc. Natl. Acad. Sci. USA 91 (1994) 11887

ATNFa& (min): - 15 15 15 15 15 15

Protease Inh.: - - TPCK TLCK BTEE AEBSF DCIC

Ik&L 0-No_ . i. _ ___

BTNFa (min):

Protease Inh:

5 5 5 5 5

TPCK TLCK BTEE AEBSF

ASource CF CF: VkTNFu(Rel A peptideIKBu peptide

IKB(it- I

lkB( /_ ...... .

IKBet ../

BSource:Inducer

5

DCIC

i' p

-~~~~~~4_. ..4.+ i

.I

Ci:L .t I

IKB-CJ)) - LPSIKB( ~/

wOkaclaic acid

IkBal)

FIG. 3. Protease inhibitors prevent TNFa-induced phosphoryla-tion and degradation of IKBa. (A) Protease inhibitors block IKBadegradation. HeLa cells, untreated or preincubated with the indi-cated protease inhibitors, were induced with TNFa for 15 min.Cytoplasmic extracts were then isolated and analyzed on a Westernblot by using a polyclonal IKBa antibody. Protease inhibitor con-centrations and time of preincubation are as in Fig. 1. IKBa isindicated at the left and the addition of TNFa (for 15 min) andprotease inhibitors is also indicated. No other bands were specificallyrecognized by the antibody. (B) Protease inhibitors block IKBaphosphorylation. Experiments were performed as inA except TNFainduction was for 5 min. The phosphorylated form of IKBa inducedby TNFa, which has a reduced mobility on SDS/PAGE, is indicatedto the left. Samples were treated as indicated above the blot.

(8). Treatment of cells with TNFa alone for 5 min resulted inthe appearance ofphosphorylated IKBa. However, this formof IKBa was not present when cells were first preincubatedwith any of the protease inhibitors and then stimulated withTNFa (Fig. 3B). Essentially identical results were obtained inJurkat cells (data not shown). Thus, these serine proteaseinhibitors repress both the phosphorylation and degradationof IcBa. Because the inhibitors block both phosphorylationand degradation, it is difficult to assess from their use howeach event contributes to the activation of NF-KB.

Phosphorylated IncBaRens Associated with NF-#cB. Tobetter understand the mechanism(s) responsible for NF-KBactivation, we determined in vivo whether induced phosphor-ylation of IKBa was sufficient for its dissociation fromNF-KB. Specifically, we ascertained whether the phosphor-ylated form of IKBa observed after stimulation ofHeLa cellswith TNFa for 5 min could be coimmunoprecipitated with anantibody to the RelA subunit of NF-cB. The results, shownin Fig. 4A, indicate that phosphorylated IKBa remains asso-ciated with NF-KB. Neither unphosphorylated nor phosphor-ylated IicBa was detected when the RelA antibody used in theimmunoprecipitation or the IKBa antibody used in the West-em blot was preincubated with its corresponding peptide,demonstrating the specificity of the assay.We also investigated whether similar results could be

observed for other inducers ofNF-KB. Studies have observedthat LPS activation of NF-KB in the monocytic cell lineTHP-1 involves the phosphorylation and degradation ofIKBa(ref. 7 and Fig. 4B). As shown in Fig. 4B, the phosphorylatedform of IKBa induced by LPS also retains association withNF-KB. Okadaic acid, a specific inhibitor of the serine/threonine phosphatases 1 and 2A, induces NF-KB DNAbinding activity in Jurkat T cells (23). This activation ismediated primarily through the phosphorylation and degra-dation of IKBa (T.S.F. and A.S.B., unpublished data). Con-sistent with results obtained using TNFa and LPS, thephosphorylated form of IxBa induced by okadaic acid alsoremains complexed to NF-KB (Fig. 4B). It is evident fromthese results that under a variety of conditions the induciblephosphorylation ofIKBa does not cause its dissociation from

IKsB xf\._ -,

IKB~t I/

FIG. 4. Phosphorylation of IKBa is not sufficient for its disso-ciation from NF-KB. (A) TNFa-induced phosphorylation of IKBadoes not cause its dissociation from NF-KB. NF-KB complexes fromcytoplasmic extracts of either untreated cells or cells treated withTNFa for 5 min were immunoprecipitated with a ReIA polyclonalantibody. Washed complexes, containing NF-KB and associatedproteins, were then treated with 0.8% DOC to release IKBa bound toNF-KB. Liberated IKBa was subsequently analyzed on a Westernblot with an IKBa polyclonal antibody. Also shown is 50 ug ofcytoplasmic extract from untreated cells or cells stimulated withTNFa for 5 min, neither of which underwent immunoprecipitation.Sample treatment, including controls for specificity, was as indicatedabove the blot. (B) Phosphorylated IKBa, induced by LPS orokadaicacid, remains associated with NF-KB. Experiments were performedas in A except cytoplasmic extracts from THP-1 cells treated 1 hrwith LPS or Jurkat T cells treated 1 h with okadaic acid were used.As indicated, both cytoplasmic extracts and immunoprecipitationsusing the RelA antibody were analyzed on a Western blot with theIKBa antibody. The various forms of IKBa are indicated to the leftand the inducer is indicated to the right. CE, cytoplasmic extract; IP,immunoprecipitation.

NF-KB. These data strongly suggest that other events signif-icantly contribute to the activation of NF-KB.

DISCUSSIONIn this report we demonstrate that protease inhibitors usedthus far in studies of NF-KB (11, 16-18) have previouslyuncharacterized properties that may complicate conclusionsbased on their use. For example, TPCK, TLCK, and DCICinterfere with NF-KB DNA binding potential. It is known thatthese inhibitors block protease activity by direct alkylation oracylation of the protease active site (24-26). Moreover,TPCK, TLCK, and DCIC alter the activity of proteins inaddition to proteases such as glycogen phosphorylase b,TFIIIC, and Ets-1 through direct modifications (27-29).Furthermore, NF-KB DNA binding activity is redox-sensitive and can be inhibited by alkylation within its DNAbinding domain (30, 31). Based on these observations and thedata in this report, it is highly likely that the effects elicitedin vivo and in vitro by these agents on NF-KB DNA bindingoccur through a direct modification of NF-KB and not by anindirect effect. It is also apparent that NF-KB can be modifiedwhen free or complexed to IKBa. An awareness of theseeffects on NF-KB DNA binding will be beneficial duringrelevant experiments and in the interpretation of results.BTEE and AEBSF, protease inhibitors that do not appear tointerfere with NF-KB DNA binding activity, may be bettersuited for studies of NF-KB.We also show that all protease inhibitors that repress IKcBa

degradation also completely prevent its phosphorylation.Therefore, it is difficult to conclude whether the suppressiveeffects of these inhibitors on NF-KB activation are due to

Biochemistry: Finco et A

Dow

nloa

ded

by g

uest

on

Janu

ary

21, 2

021

Proc. Natl. Acad. Sci. USA 91 (1994)

their inhibition of IKBa phosphorylation or instead its deg-radation. Several models exist to explain how these proteaseinhibitors prevent IKBa phosphorylation. (i) The inhibitorsmay target a protease that regulates a kinase responsible forIKBa phosphorylation; inhibition of this putative upstreamprotease would repress kinase activation and, therefore,prevent IKBa phosphorylation. Thus, proteases potentiallyexist at two steps in the NF-KB signal transduction pathway,one upstream ofthe kinase and one that acts directly on IKBa.(ii) It is conceivable that the protease inhibitors directlyinterfere with a signal transduction component that is not aprotease but is essential for NF-KB activation. As describedin this report and others (27-29), TPCK, TLCK, and DCICcan directly modify and inhibit the activity of nonproteolyticproteins. Although BTEE and AEBSF do not effect NF-icBDNA binding, they may alter the activity of proteins withinthe cell besides proteases. This second model includes thepossibility that direct modification ofNF-KB and/or IKBa bycertain protease inhibitors may prevent their phosphoryla-tion and/or activation. (iii) Phosphorylation and proteolysisare coupled events. This model would be relevant in situa-tions where the kinase and protease directly interact, forexample, as part ofa multiprotein complex. Interaction oftheprotease inhibitors with the protease could cause a confor-mational change in the complex that then alters kinaseactivity. (iv) A combination of the above models may alsoexplain how all of these mechanistically distinct proteaseinhibitors prevent IKBa phosphorylation; although the inhib-itors are targeting different proteins in the signal transductionpathway, the outcome is identical.The identification ofinhibitors that block IKBa degradation

but allow phosphorylation may clarify the role of proteolysisin NF-KB activation. Recently, an inhibitor of NF-KB acti-vation that targets the proteasome has been shown to preventdegradation ofIKBa without altering its phosphorylation (32).Intriguingly, two ofthe protease inhibitors used in the presentstudy, DCIC and AEBSF, repress proteolytic activitieswithin the proteasome (33, 34). However, they must bealtering additional cellular processes due to their ability toinhibit IKBa phosphorylation. DCIC and AEBSF may inter-act with the proteasome in a manner that also inhibits IKBaphosphorylation or an additional target ofDCIC and AEBSFwithin the cell may be the proposed upstream protease.

Previous studies have suggested that the phosphorylationof IKBa is responsible for its dissociation from NF-KB, afterwhich free IKBa is rapidly degraded (5, 6, 10, 15). However,we show in this report that in vivo phosphorylation of IKBaalone is not sufficient for its dissociation from NF-KB. Thisimportant result indicates that additional cellular events arerequired for NF-KB activation. Perhaps other phosphoryla-tion events, either on IKBa or NF-KB, are necessary. Con-sistent with this idea, we have observed that the ReLA subunitofNF-KB is also phosphorylated after treatment of cells withinducers of NF-KB (unpublished observation). Thus, withinan NF-KB-IKBa complex, phosphorylation of NF-KB orIKBa alone may have no consequence, but both phosphor-ylation events together would cause the dissociation ofIKBa.Another possible role for inducible phosphorylation of IKBais that it serves as a signal for a subsequent essential event.For example, phosphorylated IKBa may be a better substratefor a cytoplasmic protease. The proteolysis of phosphory-lated IKBa complexed to NF-KB would then result in theactivation of NF-KB.

Elucidation of the signals required for NF-KB activation isimportant in deciphering how this family of transcriptionfactors is regulated. An understanding of this complex pro-cess may lead to the discovery of pharmacologic modulatorsof NF-KB activity that are beneficial in the treatment ofimmune and inflammation disorders, in the regulation oftumorigenesis, and in the treatment of AIDS.

We thank T. Maniatis for sharing data prior to publication and J.Cheshire for critical reading of the manuscript. This research wassupported by grants from the National Institutes of Health to A.S.B.(CA 52515 and AI35098) and by a grant from the Arthritis Founda-tion.

1. Grilhi, M., Chiu, J. J.-S. & Lenardo, M. J. (1993) Int. Rev.Cytol. 143, 1-62.

2. Baldwin, A. S. (1994) in Transcription: Mechanisms and Reg-ulation, eds. Conaway, R. C. & Conaway, J. W. (Raven, NewYork), pp. 443-457.

3. Baeuerle, P. A. & Baltimore, D. (1988) Science 242, 540-546.4. Baeuerle, P. A. & Baltimore, D. (1988) Cell 53, 211-217.5. Ghosh, S. & Baltimore, D. (1990) Nature (London) 344, 678-

682.6. Shirakawa, F. & Mizel, S. B. (1989) Mol. Cell. Biol. 9, 2424-

2430.7. Cordle, S. R., Donald, R., Read, M. A. & Hawiger, J. (1993) J.

Biol. Chem. 268, 11803-11810.8. Beg, A. A., Finco, T. S., Nantermet, P. V. & Baldwin, A. S.

(1993) Mol. Cell. Biol. 13, 3301-3310.9. Brown, K., Park, S., Kanno, T., Franzoso, G. & Siebenlist, U.

(1993) Proc. Natl. Acad. Sci. USA 90, 2532-2536.10. Sun, S.-C., Ganchi, P. A., Ballard, D. W. & Greene, W. C.

(1993) Science 259, 1912-1915.11. Chiao, P. J., Miyamoto, S. & Verma, I. M. (1994) Proc. Natl.

Acad. Sci. USA 91, 28-32.12. de Martin, R., Vanhove, B., Cheng, Q., Hofer, E., Csizmadia,

V., Winkler, H. & Bach, F. H. (1993) EMBO J. 12, 2773-2779.13. Bail, 0. L., Schmidt-Ullrich, R. & Israel, A. (1993) EMBO J.

12, 5043-5049.14. Fan, C. M. & Maniatis, T. (1991) Nature (London) 354, 395-

398.15. Scott, M. L., Fujita, T., Liou, H.-C., Nolan, G. P. & Balti-

more, D. (1993) Genes Dev. 7, 1266-1276.16. Henkel, T., Machleidt, T., Alkalay, I., Kronke, M., Ben-

Neriah, Y. & Baeuerle, P. A. (1993) Nature (London) 365,182-185.

17. Mellits, K. H., Hay, R. T. & Goodbourn, S. (1993) NucleicAcids Res. 21, 5059-5066.

18. Miyamoto, S., Chiao, P. J. & Verma, I. M. (1994) Mol. Cell.Biol. 14, 3276-3282.

19. Bradford, M. M. (1976) Anal. Biochem. 72, 248-254.20. 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., Jr. (1991) Cell 65, 1281-1289.

21. Machleidt, T., Wiegmann, K., Henkel, T., Schutze, S., Bae-uerle, P. & Kronke, M. (1994) J. Biol. Chem. 269, 13760-13765.

22. Sun, S.-C., Ganchi, P. A., Beraud, C., Ballard, D. W. &Greene, W. C. (1994) Proc. Natl. Acad. Sci. USA 91, 1346-1350.

23. Thevenin, C. V., Kim, S.-J., Rieckmann, P., Fujiki, H., Nor-cross, M. A., Sporn, M. B., Fauci, A. S. & Kehrl, J. H. (1990)New Biol. 2, 793-800.

24. Schoellmann, G. & Shaw, E. (1%3) Biochemistry 2, 252-255.25. Harper, J. W., Hemmi, K. & Powers, J. C. (1985)Biochemistry

24, 1831-1841.26. Shaw, E., Mares-Guia, M. & Cohen, W. (1%5) Biochemistry 4,

2219-2224.27. Rusbridge, N. M. & Beynon, R. J. (1990) FEBS J. 268, 133-

136.28. Cromlish, J. A. & Roeder, R. G. (1989) J. Biol. Chem. 264,

18100-18109.29. Fisher, R. J., Koizumi, S., Kondoh, A., Mariano, J. M.,

Mavrothalassitis, G., Bhat, N. K. & Papas, T. S. (1992)J. Biol.Chem. 267, 17957-17%5.

30. Toledano, M. B. & Leonard, W. J. (1991) Proc. Natl. Acad.Sci. USA 88, 4328-4332.

31. Matthews, J. R., Wakasugi, N., Virelizier, J.-L., Yodoi, J. &Hay, R. T. (1992) Nucleic Acids Res. 20, 3821-3830.

32. Palombella, V. J., Rando, 0. J., Goldberg, A. L. & Maniatis,T. (1994) Cell 78, 773-785.

33. Djaballah, H., Harness, J. A., Savory, P. J. & Rivett, A. J.(1992) Eur. J. Biochem. 209, 629-634.

34. Orlowski, M. & Michaud, C. (1989) Biochemistry 28, 9270-9278.

11888 Biochemistry: Finco et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

21, 2

021