Immunity, Vol. 13, 677–689, November, 2000, Copyright 2000 by Cell Press

NF-kB Activation by the Pre-T Cell ReceptorServes as a Selective Survival Signalin T Lymphocyte Development

stable antigen (HSA or CD24) can be separated throughtheir expression of CD44 and the IL-2 receptor a chain(CD25) into four consecutive developmental stages: DNCD441CD252 (stage I), DN CD441CD251 (stage II), DNCD442CD251 (stage III), and DN CD442CD252 (stage

Reinhard E. Voll,1,2,6,9 Eijiro Jimi,1,2,4,6

Roderick J. Phillips,1,2,4 Domingo F. Barber,2,10

Mercedes Rincon,1,7 Adrian C. Hayday,2,8

Richard A. Flavell,1,3,4 and Sankar Ghosh1,2,3,4,5

1 Section of ImmunobiologyIV) (Figure 1A) (Godfrey et al., 1993; Hoffman et al., 1996).2 Department of Molecular BiophysicsTCR b chain rearrangement takes place in the stage IIIand Biochemistrythymocytes, and, in cells that express a functional TCR3 Department of Molecular, Cellular,b chain, the b chain associates with the pre-Ta chainand Developmental Biologyto assemble a pre-TCR complex (Figure 1A).4 Howard Hughes Medical Institute

Pre-TCR expression occurs during the transition ofYale University School of Medicinestage III to stage IV cells (Hoffman et al., 1996). TheNew Haven, Connecticut 06520majority of stage III cells are small in size and eitherhave not completed b chain rearrangement or have out-of-frame rearrangements (“E” cells). Approximately 15%Summaryof stage III cells, however, are large in size, are often inS phase, and show in-frame b chain rearrangementsActivation of the transcription factor NF-kB and pre-T(“L” cells) (Figure 1A). Therefore, the L fraction repre-cell receptor (pre-TCR) expression is tightly correlatedsents cells that have productively rearranged their bduring thymocyte development. Inhibition of NF-kB inchain and express the pre-TCR (Hoffman et al., 1996).isolated thymocytes in vitro results in spontaneousThe assembly of the pre-TCR serves as the first check-apoptosis of cells expressing the pre-TCR, whereaspoint in T lymphocyte development and is critical forinhibition of NF-kB in transgenic mice through expres-further proliferation and differentiation (von Boehmersion of a mutated, superrepressor form of IkBa leadsand Fehling, 1997). In mutant mice lacking the recombi-to a loss of b-selected thymocytes. In contrast, thenase proteins RAG1 or RAG2 or in TCR b- or pre-Ta-forced activation of NF-kB through expression of adeficient mice, none of which can form a pre-TCR, a/bdominant-active IkB kinase allows differentiation toT cell development is significantly arrested at the stageproceed to the CD41CD81 stage in a Rag12/2 mouseof stage III thymocytes (Mombaerts et al., 1992a, 1992b;that cannot assemble the pre-TCR. Therefore, signalsShinkai et al., 1992; Fehling et al., 1995). Therefore, suc-emanating from the pre-TCR are mediated at leastcessful TCR b chain rearrangement is essential for thy-

in part by NF-kB, which provides a selective survivalmocyte precursors to proceed to the next develop-

signal for developing thymocytes with productive b mental stage.chain rearrangements. The signals originating from the pre-TCR that influ-

ence the differentiation of developing T lymphocytesIntroduction remain to be fully characterized. Recent studies have

shown that the pre-TCR exits the endoplasmic reticulumStages of thymocyte differentiation have been charac- and is most likely expressed on the cell surface (O’Sheaterized through the expression of specific cell surface et al., 1997). Nevertheless, the functioning of the recep-markers. The most immature thymocytes, which lack ex- tor appears to be independent of a specific ligand, sincepression of both coreceptors CD4 and CD8 (CD42CD82 mutant forms of pre-Ta lacking the extracellular domaindouble-negative [DN] cells), develop into an interme- are also functional in transgenic mice (Irving et al., 1998).diate CD41CD81 double-positive (DP) stage before The requirement of CD3 components in the expressionmaturing into CD41 or CD81 single-positive (SP) cells and functioning of the pre-TCR suggests that signalingand exiting the thymus. It is now clear that DN thymo- from it might resemble that from the TCR (von Boehmer

et al., 1998). In support of such a hypothesis, mutantcytes represent a heterogeneous population that canmice lacking signaling molecules involved in TCR signal-be further fractionated into distinct developmentaling, such as p56lck, SLP-76, and LAT, all exhibit a blockstages by using the differential expression of additionalin T cell differentiation at the stage of b chain selectioncell surface markers. DN cells that express the heat-(Anderson et al., 1994; Pivniouk et al., 1998; Zhang etal., 1999).5 To whom correspondence should be addressed (e-mail: sankar.

Our current study was initiated by attempts to explainghosh@yale.edu).the constitutive activation of the transcription factor NF-6 These authors contributed equally to this work.kB in thymocytes (Ivanov and Ceredig, 1992; Zuniga-7 Present address: Department of Rheumatology, University of Ver-

mont, Vermont. Pflucker et al., 1993; Weih et al., 1994; Sen et al., 1995).8 Present address: Department of Immunobiology, Guy’s Hospital, Transcription factors of the NF-kB/Rel family are evolu-London, SE1 9RT, United Kingdom. tionarily conserved proteins that play important roles in9 Present address: Medizinische Klinik III, Immunologie Labor,

development and host defense (reviewed in Ghosh etGluckstrabe 41, 91054, Erlangen, Germany.al., 1998). Although constitutively active NF-kB can be10 Present address: Laboratory of Immunogenetics, National Institutedetected in developing T and B lymphocytes, it is un-of Allergy and Infectious Diseases, National Institutes of Health,

Rockville, Maryland 20852. clear whether NF-kB plays an important role in lympho-

Immunity678

NF-kB in Thymocyte Survival679

cyte development. Studies of mice lacking various NF- ResultskB/Rel family members, including p50, p52, c-Rel, andRelB, failed to reveal marked impairment of lymphocyte NF-kB Activation in DN Thymocytes Coincides

with Expression of the Pre-TCR Complexdevelopment (Ghosh et al., 1998). Even adoptively trans-ferred hematopoietic precursors from a p50/p65 double- Activation of NF-kB in thymocytes has been described

earlier using electromobility shift assays (Ivanov andknockout mouse could develop into mature B and Tcells if wild-type stromal cells were simultaneously Ceredig, 1992; Zuniga-Pflucker et al., 1993; Weih et al.,

1994; Sen et al., 1995). In agreement with these findings,transferred, suggesting an essential role of NF-kB instroma rather than in developing lymphocytes them- we could also detect constitutive NF-kB DNA binding

activity in extracts prepared from spleen, bone marrow,selves (Horwitz et al., 1997). Only the p50/p52 double-knockout mouse shows severe impairment of T and B and thymus (Figure 1B). To determine whether the ob-

served DNA binding activity correlated with NF-kB tran-lymphocyte development; however, adoptively trans-ferred bone marrow cells lacking p50 and p52 develop scriptional activity, we created transgenic mice bearing

an NF-kB-dependent luciferase reporter gene. Two kBnormally into T cells. Although it is unlikely that thesecells are intrinsically deficient in development (Franzoso sites derived from the k light chain intronic enhancer

were placed upstream of a minimal fos promoter toet al., 1997), the lack of an obvious phenotype may bedue to complementation between different Rel family control the expression of a luciferase reporter gene.

Several founders were tested for tissue-specific lucifer-proteins.In the present study, we have examined the regulation ase expression and showed a very similar pattern of

expression (data not shown). In most tissues, the pres-of NF-kB in developing thymocytes. We have found thatthe highest levels of constitutive NF-kB activity are ob- ence of nuclear DNA binding activity detected by EMSA

correlated with luciferase reporter activity, thus indicat-served in stage III L and stage IV cells, which correspondto the populations that express the pre-TCR (Hoffman ing that the constitutive NF-kB in thymus is transcrip-

tionally active (Figure 1C). To further explore the regula-et al., 1996). We have also found that stable expressionof components of the pre-TCR in a T cell line that ex- tion of NF-kB in the thymus, we fractionated developing

thymocytes into DN, DP, and CD4 and CD8 SP cells.presses CD3 but not the TCR results in constitutiveactivation of NF-kB, suggesting that the pre-TCR most Analysis of the DNA binding activity from nuclear ex-

tracts revealed that the majority of the NF-kB activitylikely directly activates NF-kB. To explore the possiblebiological role of NF-kB in developing thymocytes, we was in the DN population (data not shown). To further

characterize the DN subpopulation containing activatedhave generated transgenic mice expressing a dominant-active form of IkBa, a superinhibitor of NF-kB. Analysis NF-kB, we fractionated DN thymocytes based on the

expression of CD44 and CD25 surface markers (Figureof T cell populations from such mice reveals a selectivereduction in the number of stage IV cells. Vice versa, 1A). Analysis of the different subpopulations showed

that the majority of the NF-kB DNA binding activity wasforced activation of NF-kB in transgenic mice express-ing a constitutively active form of IkB kinase b (IKKb) in the stage III and IV cells (Figure 1D). As a control, we

tested the binding of the constitutive Oct-1 transcriptionleads to a significant increase in the number of stageIV cells. More importantly, expression of active IKKb in factor, which did not exhibit the same difference among

the three fractions. We could not isolate sufficient num-transgenic mice bypassed the block in differentiation inRag12/2 thymocytes, allowing a significant fraction of bers of stage I or stage II cells to carry out comparable

EMSA analysis, but the extract from CD441 cells (con-cells to differentiate into DP cells. Furthermore, isolatedstage III L and stage IV cells were found to be hypersen- taining both stage I and II cells) was completely devoid

of any NF-kB DNA binding activity (data not shown).sitive to apoptosis upon inhibition of NF-kB in vitro.Therefore, taken together, our results suggest that pre- Further separation of stage III cells into small prese-

lection E cells and large postselection L cells showedTCR signaling leads to the activation of NF-kB and activeNF-kB provides a selective survival signal for thymo- that most of the NF-kB DNA binding activity was in

the L cell fraction (Figure 1E). Once again, Oct-1 DNAcytes that have undergone productive “b selection.”

Figure 1. NF-kB Activity Peaks in CD442CD251 L Cells and CD81 Thymocytes

(A) Schematic depiction of thymocyte development based on expression of different cell surface markers (Hoffman et al., 1996).(B) Nuclear NF-kB DNA binding activity in tissues. Nuclear extracts from spleen, bone marrow, brain, liver, kidney, and thymus were analyzedfor NF-kB DNA binding activity by electromobility shift assay (EMSA).(C) Luciferase assay of tissue extracts from NF-kB luciferase reporter mice. Spleen, bone marrow, brain, liver, kidney, and thymus extractsof NF-kB luciferase transgenic mice were analyzed for luciferase activity. Light units per 100 mg of tissue extract are shown.(D) NF-kB nuclear DNA binding activity in thymocyte subsets. Nuclear extracts were prepared from sorted thymocyte subsets of wild-typemice. NF-kB DNA binding activity of 1.5 mg nuclear extract was analyzed by electromobility shift assay.(E) Peak of NF-kB DNA binding activity in stage III L cells, which express the pre-T cell receptor. Nuclear extracts of E, L, and CD41/CD81

thymocytes from wild-type mice were analyzed for NF-kB DNA binding capacity by electromobility shift assay.(F) Supershift analysis of nuclear extracts from L cells reveals predominantly p50-p65 heterodimers. Nuclear extracts from L cells wereincubated with antibodies against p50, p52, p65, RelB, c-Rel, and c-myc before electromobility shift assay.(G) Peak of luciferase activity in stage III and stage IV cells, coinciding with expression of the pre-T cell receptor (middle panel). After depletionof CD4- and CD8-positive cells, stage III thymocytes of NF-kB luciferase reporter mice were further separated, according to size, into smallE cells and large “b-selected” L cells (right panel) (see flow chart on the left-hand side of the figure). The TCRab1, CD4, or CD8 SP cells weresorted as described in Experimental Procedures. Equal numbers of sorted cells were used to determine the luciferase activity. Error barsrepresent the standard deviations.

Immunity680

binding was used as an internal control. The DNA bind- RAG2, TCR b, and pre-Ta-deficient mice and allows thedifferentiation of DN stage III cells into DP thymocytesing complexes consisted predominantly of p50/p65 het-

erodimers, as revealed by supershift analysis (Figure (Levelt et al., 1993; Jacobs et al., 1994; Shinkai andAlt, 1994; Fehling et al., 1995). Most likely, anti-CD3e1F). To determine if the NF-kB DNA binding activity

in the DN thymocytes correlated with transcriptionally treatment provides signals that normally originate fromthe pre-TCR complex. Therefore, we investigated whetheractive NF-kB, we sorted thymocytes of NF-kB luciferase

transgenic mice into the different developmental stages anti-CD3e antibodies can induce NF-kB activity inRAG1-deficient DN stage III thymocytes. RAG1-defi-(Figures 1A and 1G). Initial analysis revealed that the

majority of the luciferase activity was detected in the cient mice were intraperitoneally injected with eitheranti-CD3e antibody 145-2C11 in PBS or PBS alone.DN and the CD8 SP thymocytes (Figure 1G). The DN

thymocytes were further fractionated into stage I, stage Stage III thymocytes were sorted 12 hr after injection,and nuclear extracts were analyzed by electromobilityII, stage III, and stage IV cells. The majority of luciferase

activity in DN thymocytes was detected in the stage III shift assay. As shown in Figure 2C, there was a stronginduction of NF-kB in stage III thymocytes from anti-and stage IV cell populations (Figure 1G, middle panel),

similar to the pattern observed in the DNA binding analy- CD3e-treated mice. This result demonstrates that anti-CD3e treatment, which overcomes the early develop-sis (Figures 1D and 1E). We then resorted thymocytes

from luciferase-transgenic mice to specifically obtain mental block in RAG1-deficient thymocytes, is alsoassociated with strong activation of NF-kB in stage IIIstage III E and L cells and found significantly greater

luciferase activity in stage III L cells (Figure 1G, right thymocytes.To directly determine whether signaling from the pre-panel). These results therefore show that NF-kB is acti-

vated in developing thymocytes that are known to express TCR can lead to NF-kB activation, we stably transfecteda T cell line, 4G4, which expresses CD3 and lck butthe pre-TCR and suggest that the constitutive NF-kB ac-

tivity observed in thymocytes might be due to signaling lacks expression of the TCR chains, with TCR b, eitheralone or with one of the pre-Ta isoforms, pre-Taa orfrom the pre-TCR.pre-Tab (Barber et al., 1998). As shown in Figure 2D,representative stable clones expressing TCR b togetherNF-kB Activation in Stage III Thymocytes Iswith any of the pre-Ta splice variants, pre-Taa (clone 1)Dependent on Expression of the Pre-TCRor pre-Tab (clone 2), displayed strongly increased consti-Mice deficient for the TCR b chain or recombinationtutive NF-kB DNA binding activity compared to the pa-activating gene RAG1 cannot form the pre-TCR com-rental cell line (compare Figure 2D, lanes 2 and 3, toplex, and T lymphocyte development is partially or com-lane 1). Stimulation with anti-CD3 did not significantlypletely blocked at the DN stage (Mombaerts et al.,increase NF-kB DNA binding activity in these cells (data1992a, 1992b). In RAG1-deficient mice, T cell develop-not shown). Transfection of the b chain alone did notment is completely arrested at stage III, whereas, ininduce constitutive activity (Figure 2D, lane 4). As a con-TCR b–deficient mice, low numbers of DP cells can betrol, the level of the constitutive transcription factordetected and the numbers of TCR gd–positive thymo-Oct-1 was unaltered in the different clones (Figure 2D).cytes are, in fact, increased. To further investigate theThe specificity of the DNA–protein complexes for kBpotential role of pre-TCR signaling in activation of NF-sequences was demonstrated by competition with wild-kB in stage III thymocytes, we analyzed the NF-kB DNAtype and mutant oligonucleotide probes, while super-binding activity in nuclear extracts of stage III thymo-shift analysis indicated that the DNA binding complexescytes from TCR b–deficient and RAG1-deficient mice.primarily consisted of NF-kB (data not shown). TheseAs shown in Figure 2A, the NF-kB DNA binding activitydata demonstrate that expression of the pre-TCR com-was dramatically decreased in nuclear extracts of stageplex is capable of inducing constitutive NF-kB nuclearIII cells from both mutant mice compared to wild-typetranslocation in T cell lines and that the activation ofmice. As a control, we tested the DNA binding activityNF-kB appears to occur in the absence of stimulationof Oct-1, which was not similarly affected (Figure 2A).by an exogenous ligand.To analyze the NF-kB transcriptional activity in DN thy-

mocytes from TCR b–deficient mice, we crossed thesemice with NF-kB-luciferase reporter gene transgenic Inhibition of NF-kB Activation in Transgenic Mice

Expressing a Superinhibitor of NF-kBmice. As shown in Figure 2B, stage III thymocytes fromTCR b–deficient mice display significantly less NF-kB- To determine whether the activation of NF-kB in DN

thymocytes had a functional role, we generated micedependent luciferase activity than those from TCR b(wt/wt) littermates. The loss of NF-kB activity is even that expressed a mutated IkBa in thymocytes under

control of the proximal lck promoter to inhibit NF-kBmore pronounced in the few stage IV thymocytes, whichare still detectable in TCR b–deficient mice and probably activation (Figure 3A). To prevent signal-induced degra-

dation, serines 32 and 36 of IkBa, which are phosphory-express TCR gd. Therefore, these results suggest thatexpression of the TCR b chain, and probably the assem- lated by the IkB kinases IKKa and IKKb, were mutated

into alanines. In addition, the sites of ubiquitination, ly-bly of the pre-TCR, is essential for efficient activationof NF-kB in the stage III and stage IV subsets. sines 21 and 22, were changed to arginines (Figure 3A).

Since TCR signaling (and most likely pre-TCR signaling)Pre-TCR signaling appears to be critically dependenton the CD3 complex, in particular, CD3g and CD3e (De- involves the activation of tyrosine kinases, we also mu-

tated tyrosine 42 of IkBa into phenylalanine. TyrosineJarnette et al., 1998; Haks et al., 1998). It has beendemonstrated that anti-CD3e treatment can partly over- phosphorylation at position 42 has been reported to

cause dissociation of NF-kB from IkBa, thereby leadingcome the block in early T cell development in RAG1,

NF-kB in Thymocyte Survival681

Figure 2. Absence of NF-kB Activation inStage III Thymocytes of TCR b–Deficient andRAG1-Deficient Mice

(A) Decreased DNA binding activity in nuclearextracts from TCR b–deficient and RAG1-deficient mice. Stage III thymocytes from TCRb–deficient, RAG1-deficient, and wild-typemice were sorted. In addition, stage III cellswere obtained from wild-type mice. Nuclearextracts were prepared and analyzed by elec-tromobility shift assay for NF-kB DNA bindingactivity.(B) Decreased NF-kB transcriptional activityin stage III and stage IV thymocytes from TCRb–deficient mice. TCR b–deficient mice werecrossed with NF-kB luciferase reporter mice.Thymocytes of NF-kB-luc1/TCR bo/o and NF-kB-luc1/TCR bwt/wt littermates were sortedinto stage III and stage IV subsets, and equalnumbers of sorted cells were used to deter-mine luciferase activity. Results shown aremeans of three independent experiments,normalized to cell number. Error bars repre-sent standard deviations.(C) Activation of NF-kB in stage III cells ofRAG1-deficient mice by anti-CD3 treatment.RAG1-deficient mice were intraperitoneallyinjected with 100 mg of the anti-CD3e anti-body 145-2C11. After 12 hr, thymocytes wereprepared, and stage III cells were sorted.Equal amounts of nuclear extracts from theuntreated and treated mice were analyzed byelectromobility shift assay for NF-kB DNAbinding activity.(D) The T cell line 4G4, which lacks either aTCR or pre-TCR, was stably transfected withTCR b, either alone or in combination withone of the pre-TCR isoforms pre-Taa or pre-Tab. The stable cell lines have been previouslydescribed (Barber et al., 1998). Nuclear ex-tracts (5 mg) were prepared from unstimu-lated cells and were analyzed by electromo-bility shift assay using labeled kB and Oct-1probe. Representative results from at leastthree independent experiments are shown.

to NF-kB activation without IkB degradation (Imbert et and #28 (data not shown). The constitutive NF-kB nu-clear DNA binding activity was markedly decreased butal., 1996). Furthermore, serines and threonines within

the C-terminal PEST-region of IkBa were mutated into not abolished in thymocytes of these double-transgenicmice (Figure 3C). We then analyzed the influence of thealanines to lower the rate of basal protein turnover (Fig-

ure 3A) (Schwarz et al., 1996). Transient transfection expression of the superinhibitor IkBa on T cell develop-ment. The thymus size was not significantly affected,studies in Jurkat T lymphocytes activated with PMA plus

PHA showed that this superinhibitor of NF-kB was an and the number of total thymocytes was not markedlydecreased in 3- to 5-week-old mice. As reported pre-efficient repressor of NF-kB activity (data not shown).

This superinhibitor of NF-kB was cloned downstream viously by other groups, cytofluorometric analysis re-vealed a significant decrease of CD8 SP cells in all trans-of the proximal lck promoter to achieve high levels of

thymocyte-specific expression. The expression con- genic lines (Figure 3D) (Boothby et al., 1997; Esslingeret al., 1997; Hettmann et al., 1999). This decrease instruct contained at its 39 end human growth hormone

genomic sequences including introns to promote effi- CD8 SP cells was even more dramatic in peripheral Tlymphocytes (data not shown). Interestingly, our earliercient expression (Figure 3A). Seven mouse lines trans-

genic for the superinhibitor IkBa were obtained. Four analysis had shown that, among SP cells, the CD8 SPcells had greater transcriptionally active NF-kB (Figureof these mouse lines showed relatively high levels of

expression of the transgene, as revealed by immunoblot 1G). The reduction of the numbers of CD8 SP thymo-cytes clearly correlated with the amount of transgeneanalysis of thymocytes (Figure 3B). The amount of the

transgenic superinhibitor IkBa was highest in mouse expression. Further studies are underway to explore themolecular basis for the reduction in the number of CD8lines #7, #12, and #28. Since the superinhibitor has to

compete with endogenous wild-type IkBs, we further SP thymocytes (E. J., R. E. V., and S. G., unpublisheddata).increased the amount of the superinhibitor by crossing

the mouse lines #12 and #28 (Figure 3B, right) and #7 Besides the effect on SP cells, the relative and total

Immunity682

Figure 3. Generation of Transgenic Mice Expressing a Constitutively Inhibiting IkBa in Thymocytes

(A) The IkBa superinhibitor transgene. IkBa was mutated at the positions indicated, and the coding sequence of the hemagglutinin tag was attachedat the 59 end. This superinhibitor of NF-kB was cloned behind the proximal lck promoter, which is active in thymocytes. For increased expression,a part of the human growth hormone sequence containing introns and a polyadenylation signal was cloned 39 of the mutated IkBa.(B) Expression levels of the mutated IkBa in transgenic mouse lines. Thymocytes of transgenic mouse lines were tested for expression of thetransgenic IkBa by immunoblot. Staining with anti-IkBa visualizes both the slower migrating transgenic and the endogenous IkBa.(C) Reduced NF-kB nuclear DNA binding activity in thymocytes of transgenic mice expressing a mutant IkBa. The transgenic lines #12 and#28 were crossed, and nuclear extracts from thymocytes of 4-week-old male littermates, which were either nontransgenic, single transgenic(#28), or double transgenic (#12 3 #28), were analyzed by electromobility shift analysis.(D) Decrease of CD81 thymocytes in transgenic mice expressing a mutated IkBa. CD4 and CD8 expression on thymocytes of nontransgeniclittermates, single-transgenic mice and double-transgenic mice of the lines (#12 3 #28) were analyzed by flow cytometry.(E) Decreased numbers of stage III and stage IV subsets in mice expresssing a mutated IkBa in thymocytes. Thymocytes were stained withanti-CD4-Quantumred, anti-CD8-Quantumred, anti-HSA-PE, anti-CD44-biotin/Avidin-Texas red, and anti-CD25-FITC and analyzed by flowcytometry. In the upper panel, a gate was set on HSA1 cells negative for CD4 and CD8 (DN thymocytes). These DN cells were further analyzedaccording to their expression of CD44 and CD25, as shown in the lower panel.(F) The average number of cells in the indicated fractions from the DN population from transgenic mice (hatched bars) expressing the IkBa

superinhibitor compared to nontransgenic controls (solid bars) is shown. The results from the transgenic mice are the average from 14 mice(seven were #12 3 #28 crosses, while seven were from #7 3 #28 crosses).(G) Wild-type or mutated IkBa transgenic mice received 4 i.p. injections of 1 mg BrdU. Thymocytes were stained with anti-CD4-APC, anti-CD8-Quantumred, anti-CD25-PE, and anti-BrdU. The percentage of BrdU-positive cells was measured in CD251 or CD252 cells by flowcytometry, after electronically gating on CD42CD82 cells. The results are from four independent experiments.

NF-kB in Thymocyte Survival683

Figure 4. Expression of Constitutively Active IkB Kinase Leads to Increase in Numbers of Stage IV Cells

(A) The constitutively active IKKb transgene. FLAG-tagged IKKb was mutated at the positions indicated and cloned behind the thymocyte-specific proximal lck promoter.(B) IkB kinase activity (left panel) and NF-kB DNA binding activity (right panel) in thymocytes from transgenic mice expressing the dominant-active form of IKKb.(C) Increased numbers of DN stage IV subset in mice expressing the constitutively active form of IKKb in thymocytes. Thymocytes werestained with anti-CD4-Quantumred, anti-CD8-Quantumred, anti-HSA-PE, anti-CD44-biotin/Avidin-APC, and anti-CD25-FITC and analyzed byflow cytometry, as described in Figure 3. The cell numbers of stage III E, L, and stage IV cells from 16 different transgenic mice from twodifferent founders were tabulated, and the results are shown graphically. Solid bars are nontransgenic mice, while hatched bars representIKKb transgenic mice.

numbers of DN, HSA-positive thymocytes were de- DN cell numbers from 14 mice that were double trans-genic for the superinhibitor IkBa (seven #12 3 #28creased on average z50%, depending on the amount

of transgene expressed (Figure 3E, upper panel). More crosses and seven #7 3 #28 crosses) is representedgraphically in Figure 3F. Remarkably, the reduction ofdetailed analysis showed a marked reduction of the

stage IV cells (Figure 3E, lower panel). The reduction in cell numbers in mice expressing the superinhibitor of

Immunity684

Figure 5. Expression of Constitutively ActiveIkB Kinase in RAG12/2 Mice Leads to Rescueof a Fraction of the Cells to the Double-Posi-tive Stage

(A) Immunoblot analysis of thymocytes fromwild-type, RAG12/2, and RAG12/2 crossed todominant-active IKKb trangenic mice. Totalextracts (30 mg) were immunoblotted with an-tibodies to RAG1 and IKKb.(B) Rescue of a fraction of DP cells in RAG12/2

mice expresssing a dominant-active IKKb

transgene in thymocytes. Thymocytes werestained with anti-CD4-FITC and anti-CD8-PEand analyzed by flow cytometry.(C) The average cell numbers of total and DPthymocytes from 24 different transgenic mice(12 RAG12/2, 12 RAG12/2 3 IKKb) were tabu-lated, and the results are shown graphically.

NF-kB is observed only in populations with elevated were established, expressing this constitutively activeIKKb in thymocytes under the control of the proximallevels of constitutive NF-kB transcriptional activity (in-

cluding CD8 SP cells, Figure 3D). The loss in stage IV lck promoter. Analysis of such transgenic mice revealedthat the transgenically expressed IKKb was active andcells in mice transgenic for the superinhibitor IkBa is

most likely due to increased apoptosis rather than de- NF-kB DNA binding activity in thymocytes was markedlyincreased (Figure 4B). Fractionation of the thymocytescreased proliferation, since in vivo BrdU-labeling experi-

ments did not reveal a significant influence of the IkB indicated a significant increase (to z2-fold) in the cellnumbers of the stage IV population (i.e., the same popu-transgene on proliferation (Figure 3G).lation that is most decreased when NF-kB is inhibited byexpression of the mutant IkBa) (Figure 4C). Interestingly,

Activation of NF-kB by Expression of Constitutively these transgenic mice also exhibit a significant decreaseActive IkB Kinase b Leads to Increased Numbers in the numbers of CD4 SP cells relative to CD8 SP cells,of Stage IV Thymocytes which is exactly the opposite to the phenotype of theIf activation of NF-kB by the pre-TCR helps in survival mutant IkBa transgenic mice (data not shown).of b-selected thymocytes, then induced activation ofNF-kB in cells that fail to assemble the pre-TCR shouldlead to their survival, at least transiently. To test this Expression of Active IkB Kinase b Prevents Death

of Pro-T Cells that Fail to Express TCR Chainshypothesis, we generated a constitutively active mutantof IKKb, in which serines 177 and 181 in the activation The failure to rearrange the TCR b chain and, conse-

quently, the failure to assemble the pre-TCR in miceloop were replaced by glutamic acid residues (Mercurioet al., 1997) (Figure 4A). Several transgenic mouse lines lacking the RAG1 gene leads to a complete block at

NF-kB in Thymocyte Survival685

stage III of thymocyte differentiation. As discussedabove, it has been observed that injection of RAG12/2

mice with a-CD3e forces the progression of thymocytesto the DP stage, most likely by mimicking the signalingfrom the pre-TCR. Since we have found that treatmentwith a-CD3e also leads to the activation of NF-kB in DNcells (Figure 2C), we wanted to see if expression of thedominant-active IKKb transgene in RAG12/2 mice wouldlead to the rescue of at least a fraction of the developingthymocytes to the DP stage. We therefore crossedRAG12/2 mice with transgenic mice expressing the ac-tive IKKb (Figure 5A). Analysis of progeny with the appro-priate genetic background (i.e., RAG12/2 3 IKKb [active])revealed a significant number of cells were now in theCD41CD81 DP stage (Figure 5B). The absolute numberof DP cells varied among individual RAG12/2 3 IKKbmice, ranging from 5 3 105 to 9 3 105 (Figure 5C). Thisresult therefore suggests that forced activation of NF-kBin thymocytes that lack the pre-TCR provides a survivalsignal that allows a portion of the cells to mature intothe DP stage. Our earlier results had indicated that NF-kB does not significantly affect the proliferative capacityof developing thymocytes (Figure 3G). Therefore, thecells that survive and develop into DP cells in theRAG12/2 3 IKKb (active) mice probably do not receivethe growth stimulatory signals that likely emanate fromthe pre-TCR and thus do not undergo the tremendousproliferative burst that is seen during normal devel-opment.

Loss of Bcl-2 Expression in Stage III Land Stage IV CellsIf NF-kB was functioning to prevent apoptotic cell death

Figure 6. Antiapoptotic Function of Bcl-2 Is Selectively Replacedin DN stage IV thymocytes, then its inhibition might ren-by Activation of NF-kB during b Selection

der NF-kB-dependent cells sensitive to apoptosis. We(A) Selective hypersensitivity to pharmacological inhibition of NF-therefore incubated DN CD441, stage III E and L, stagekB in L cells compared to E cells. Stage III E and L cells, stage IV,

IV, DP, and CD4 SP thymocytes with pyrrolidone dithio- DP, and CD4 SP cells were sorted and cultured in the presence ofcarbamate (PDTC), a highly effective pharmacological the NF-kB-inhibitor PDTC (pyrrolidine dithiocarbamate) (100 mM) for

12 hr.inhibitor of NF-kB. As shown in Figure 6A, stage III L(B) Loss of Bcl-2 expression in stage III L cells and in the stage IVand stage IV thymocytes were markedly more sensitivesubset. Stage III E and L cells, stage IV, and DP thymocytes wereto apoptosis induced by PDTC, as compared to CD441,sorted. Total cell extracts (25 mg/lane) were analyzed by immunoblotE, DP, and CD4 SP cells, which display relatively lowfor expression of Bcl-2. Representative results of one of three inde-

levels of endogenous NF-kB activity. pendent experiments are shown.In early pro-T cells, survival appears to be crucially

dependent on Bcl-2 expression driven by signaling fromthe IL-7 receptor and c-kit (Akashi et al., 1998). However, expressing cells. Interestingly, the level of Bcl-2 is also

significantly less in CD8 SP cells versus CD4 SP cellsvery little Bcl-2 is detectable in stage IV and DP thymo-cytes (Sentman et al., 1991; Strasser et al., 1991). In (Figure 6B), which is consistent with the finding that

inhibition of NF-kB by expressing the superrepressorDP thymocytes, antiapoptotic function appears to bemostly dependent on Bcl-xL expression, which is absent IkBa leads to a preferential loss of CD8 SP cells. There-

fore, based on these results, we propose that the pre-in DN cells (Hettmann et al., 1999) (data not shown).Therefore, it is unclear which factors protect stage III L TCR activates NF-kB, whose antiapoptotic function may

substitute for Bcl-2 and selectively rescue cells express-and stage IV cells from apoptotic cell death. To investi-gate the involvement of antiapoptotic factors in greater ing the pre-TCR from apoptotic cell death.detail, we examined the level of Bcl-2 protein in devel-oping thymocytes. Immunoblot analysis showed mark- Discussionedly decreased Bcl-2 protein levels in stage III L cellscompared to E cells, and a further decrease was found In this study, we have identified a previously unappreci-

ated role for NF-kB in T lymphocyte development. Ourin stage IV thymocytes (Figure 6B). Therefore, loss ofBcl-2 protein coincides with NF-kB activation in stage studies have revealed that NF-kB is most active in stage

III L and stage IV thymocytes. These stages representIII L cells. These findings raise the possibility that NF-kB-induced antiapoptotic factors substitute for Bcl-2 populations of developing T lymphocytes that express

the pre-TCR and have undergone selection for produc-and provide a selective survival advantage to pre-TCR-

Immunity686

Figure 7. Model of the Role of NF-kB duringb Selection: Selective Rescue from ApoptoticCell Death

(A) Schematic depiction of the expression ofBcl-2 (immunoblot) and NF-kB (luciferase andEMSA) during T lymphocyte development.The expression of Bcl-2 is constitutive in earlyDN cells, whereas activation of NF-kB occursonly in the subset of cells expressing thepre-TCR.(B) A model showing the proposed role of pre-TCR-induced NF-kB as a selective survivalsignal in the development of a/b lineage Tlymphocytes.

tive b chain rearrangements. Data presented here pro- of the pre-TCR is soon followed by a dramatic decreasein levels of Bcl-2 and an increase in NF-kB activity ac-vide strong evidence that the activation of NF-kB in

these thymocyte subsets is most likely due to signaling commodates both of these outcomes (Figure 7A). Bcl-2,which plays an important antiapoptotic role in earlierfrom the pre-TCR. Activation of NF-kB in the b-selected

cells appears to provide an antiapoptotic survival signal, develomental stages, is also antiproliferative, as sug-gested by studies on the stage IV subset of thymocyteswhich may be particularly important, since these cells

soon dramatically downregulate the expression of Bcl-2, of Bcl-2 transgenic mice (O’Reilly et al., 1996, 1997).Therefore, the pre-TCR-induced substitution of the anti-the major antiapoptotic factor in the preceding develop-

mental stages (Figure 7A). Pre-TCR-activated NF-kB, apoptotic function of Bcl-2 by the antiapoptotic activityof NF-kB, relieves the block in proliferation and allowstherefore, functions as an inducible antiapoptotic signal

and represents a novel utilization of an inducible tran- cells with productive TCR b chain rearrangements tomultiply rapidly.scription factor to allow selective survival of a fraction

of cells in a developmental pathway. Therefore, these Signaling from the pre-TCR is required for progressionto stage IV cells and subsequent maturation to DP thy-findings establish a novel conceptual framework within

which to consider not only lymphocyte development mocytes of the a/b T cell lineage. In contrast, the pre-TCR is not required and may even inhibit the develop-but also selection events in developmental biology in

general. ment of g/d T cells and thereby appears to be involvedin cell fate decisions (Figure 7B) (Aifantis et al., 1998).The pre-TCR plays a crucial role in T cell development

by allowing maturing thymocytes to “sense” the genera- In addition, the pre-TCR has been implicated in allelicexclusion at the TCR b locus (Aifantis et al., 1997) and intion of a functional b chain. It has been assumed that

signaling from the pre-TCR would have two major con- the release of a proliferative block in stage III thymocytes(Hoffman et al., 1996). However, besides stimulating pro-sequences: first, it would provide some kind of survival

advantage, and, second, it would give a growth or prolif- liferation, another important role for the pre-TCR mightbe to selectively rescue pre-T cells with a functionaleration advantage (von Boehmer and Fehling, 1997; von

Boehmer et al., 1998). Our finding that the expression TCR b chain rearrangement from apoptotic cell death

NF-kB in Thymocyte Survival687

(von Boehmer and Fehling, 1997). Here, we show that Golgi (O’Shea et al., 1997). In addition, pre-TCR signalingthe pre-TCR most likely activates the transcription factor requires the CD3 complex, which further supports theNF-kB, which has been shown to act as a potent anti- notion that pre-TCR functions at the cell membrane (Ber-apototic factor in various situations (reviewed in Barkett ger et al., 1997; Cheng et al., 1997). Proximal signalingand Gilmore, 1999). It has been suggested that NF-kB events depend on the presence of either lck or fyn, whichprotects cells from death by inducing the expression can partially replace each other in genetically modifiedof genes such as the caspase-8 inhibitors c-IAP1 and mice (Groves et al., 1996). In RAG1-deficient mice, thec-IAP2 as well as other antiapoptotic factors, including pre-TCR signal can be replaced by anti-CD3 treatment,Bfl/A1 (Barkett and Gilmore, 1999). NF-kB activation has transgenic expression of activated lck, or g-irradiationbeen shown to counteract apoptosis induced by TNF, (reviewed in von Boehmer et al., 1998). Noticeably, allFas ligation, cytotoxic drugs, and other stimuli (reviewed these stimuli are also known to activate NF-kB. More-in Barkett and Gilmore, 1999). In fact, our data show over, we have observed that anti-CD3 treatment stronglythat L cells rapidly undergo apoptotic cell death when induces NF-kB activity in RAG1-deficient stage III thy-treated with pharmacological inhibitors of NF-kB (Figure mocytes, preceding their progression to stage IV (Figure6A). Therefore, the reduction of cell numbers in the stage 2C). Therefore, various treatments that induce NF-kBIV fraction in mice expressing a mutant IkBa in thymo- activation can release the developmental block at stagecytes is consistent with an antiapoptotic role for NF-kB. III cells in genetically modified mice, which cannot formOur data from Ika transgenic mice are also consistent a functional pre-TCR.with a recent study of inhibition of NF-kB in fetal thymus In summary, our results suggest that pre-TCR signal-organ culture by infection with a recombinant adenovi- ing causes activation of the transcription factor NF-kB,rus expressing Ika (Bakker et al., 1999). In that study, which selectively rescues pre-TCR-positive cells fromdevelopment until the stage III stage was normal, but apoptotic cell death by replacing the antiapoptotic func-cell numbers of consecutive developmental stages were tion of Bcl-2, the dominant antiapoptotic mechanismstrongly decreased due to increased apoptosis. Also, during earlier stages of T cell development. Therefore,in that in vitro system, proliferation was not markedly along with induction of proliferative pathways, pre-TCRaffected, similar to the results of our in vivo BrdU-label- signaling also makes b-selected cells more resistant toing experiments in mutant IkBa transgenic mice (Figure cell death by activating NF-kB. These findings shed new3G). Therefore, inhibition of NF-kB predominantly af- light on an important developmental checkpoint in lym-fects the survival of b-selected thymocytes rather than phocyte development, and it is likely that NF-kB servestheir proliferation. a similar role in other developmental systems.

The less than complete reduction in cell numbers ofstage IV cells is most likely due to incomplete inhibition Experimental Proceduresof NF-kB by the transgenic IkB. In addition, other signal-

Miceing pathways emanating from the pre-TCR may still beB10.A mice were obtained from NIH. TCR bo/o (Mombaerts et al.,sufficient to confer some survival advantage to the pre-1992a) and RAG12/2 (Spanopoulou et al., 1994) were originally ob-TCR-expressing thymocytes. In addition, the prolifera-tained from Jackson Laboratories (Bar Harbor, ME). All mice weretive capacity of DN thymocytes in IkBa transgenic mice maintained under specific pathogen free conditions in a climate-

is mostly unaffected (Figure 3G), thus further explaining controlled environment with 12 hr light/dark cycles.why the marked alteration of stage IV cell numbers inmutant IkBa transgenic mice does not lead to significant Generation of NF-kB-Luciferase Transgenic Micechanges in the overall number of T cells. Hence, the A 2.0 kb HindIII–XhoI fragment from the NF-kB reporter gene pBIIx-

luc was microinjected into fertilized (C57/BL6xSJL)F2 eggs, as pre-stage IV thymocytes from the mutant IkBa transgenicviously described. The NF-kB-luc transgene contained the fireflymice that survive may continue to proliferate rapidly andluciferase gene, driven by two NF-kB sites from the k light chainultimately fill up the DP cell compartment. Similarly, theenhancer in front of a minimal fos promoter. Two expression-positiveactivation of NF-kB in the IKKb transgenic mice en-founder lines were identified by slot blot hybridization (R. J. P. et al.,

hances the survival capacity of cells in the stage IV unpublished data). Several founder lines showed the same tissue-compartment. However, these cells lack the proliferative specific expression pattern, which was correlated with NF-kB DNAsignals that would come from the pre-TCR, and hence binding activity, as detected by EMSA. Routine testing was per-

formed by PCR from tail DNA, using the primers 5LUC (CGCGGAAmost likely do not expand and are lost from the DPTACTTCGAAATGTC) and 3LUC (CTTAGGTAACCCAGTAGATCC),population. An additional explanation for the mainte-resulting in a 500 bp fragment. The lines were backcrossed severalnance of DP cell numbers in the mutant IkBa transgenictimes onto B10.A background (NIH). NF-kB-luc transgenic micemice may result from the inhibition of NF-kB-dependentwere crossed and backcrossed with TCR b–deficient mice.

apoptosis that is observed in the DP thymocytes, atleast after anti-CD3 treatment (Hettmann et al., 1999). Generation of Superinhibitor IkBa Transgenic MiceThe mechanism underlying this observation has not Several point mutations were introduced into the human IkBa cDNAbeen clarified but may be due to NF-kB-dependent clone MAD3 by PCR-aided mutagenesis. This NF-kB superinhibitorupregulation of Fas ligand expression (Kasibhatla et al., was then cloned into the plasmid p1017 (gift from R. Perlmutter)

behind the proximal lck promoter, which confers thymocyte-specific1999). Therefore, the reduction of stage IV cells may beexpression. A 4.8 kb NotI fragment containing the proximal lck pro-compensated for by decreased apoptosis in DP thymo-moter, the superinhibitor, and a part of the human growth hormonecytes. Further work will be necessary to clarify thesegenomic sequence including the poly A tail was microinjected into

issues. fertilized (C57/BL6xSJL)F2 eggs, as previously described. FoundersThus far, little is known about signaling events from were screened by slot blot analysis of tail DNA using a 32P-labeled

the pre-TCR complex. The pre-TCR signal appears to BamH1–XhoI fragment of the transgene. For routine testing, tail DNAwas analyzed by PCR, using a forward primer (GAAGGAGACTGTGdepend on the release of the pre-TCR from the ER/cis-

Immunity688

GTTGAGTGG) binding within the proximal lck promoter and a re- ml of DNase I for 15 min at 378C. Fixed cells were washed andstained with anti-BrdU-FITC (Becton Dickinson) and analyzed byverse primer binding to the HA sequence of the transgene (TAGTCG

GGGACGTCGTAGG), thereby resulting in a 500 bp PCR fragment. flow cytometry.

In Vitro IkBa Kinase AssayGeneration of Mice Transgenic for a Constitutive Active IKKbTotal thymocyte extracts were prepared by TNT buffer containingThe IKKb cDNA was kindly provided by Dr. M. Karin. Using PCR-protease inhibitors and incubated with anti-FLAG M2 beads at 48Caided mutagenesis, point mutations were introduced into the IKKb

overnight. Immunoprecipitates were washed, and their kinase activ-gene, which resulted in a substitution of serines 177 and 181 byity was measured by using 2 mg of GST-IkBa as substrate in theglutamic acid. This constitutively active form of IKKb containing apresence of ATP.sequence coding for a FLAG tag was cloned into the plasmid p1017,

and a NotI fragment was microinjected into fertilized (C57/BL6xSJL)F2 eggs, as described for the mutant IkBa. In Vitro Apoptosis Assay

Thymocyte subsets were sorted as described. Cells were centri-fuged and resuspended at a concentration of 0.5 3 106 /ml in Bruff’sFluorescence-Activated Cell Sorting/Analysismedium supplemented with L-glutamine and 10% fetal calf serum.Thymus lobes were harvested at z8 a.m. and ground into single-Cells were cultured in the presence of dexamethasone (10 mg/ml),cell suspensions between two pieces of nylon mesh. Thymocytespyrrolidine dithiocarbamate (100 mM), or vehicle only. After 12 hr,were centrifuged and resuspended in phosphate-buffered salinecells were stained with Annexin V-FITC and propidium iodide (R&D(PBS) containing 2% fetal calf serum. Cells were kept on ice through-Systems) and analyzed by flow cytometry.out staining and sorting. For sorting of CD41/81, CD41, and CD81

thymocytes, cells were preincubated with Fc-Block (PharMingen)for 10 min and subsequently stained with anti-CD4-Quantumred Acknowledgments(Sigma) and anti-CD8a-PE (PharMingen). For sorting of double-neg-ative cells, the CD41/81 double-positive cells were depleted using We would like to thank Tom Taylor for expert cell sorting, Debbiegoat anti-rat–coated magnetic beads (PerSeptive Biosystems), Buskus for expert technical assistance during generation of trans-which had been preincubated with anti-CD4 (GK1.5) and anti-CD8 genic mice, and Crystal Bussey and Iris Douglas for their invaluable(TIB 210) monoclonal antibodies. After three rounds of depletion technical assistance. We also thank Drs. Michael May, David Schatz,using a magnet (PerSeptive Biosystems), cells were preincubated Mark Shlomchik, and Eric Pamer for critical reading of the manu-with Fc-Block (PharMingen) for 10 min and subsequently stained script. This work was supported by the Howard Hughes Medicalwith anti-CD4-Quantumred (Sigma), anti-CD8-Quantumred (Sigma), Institute and a grant to S. G. from the National Institutes of Healthanti-HSA-PE (PharMingen), anti-CD25-FITC (PharMingen), and anti- (AI 33443). R. E. V. was supported by a fellowship from the DeutscheCD44-biotin (PharMingen) for 30 min. Cells were washed and incu- Forschungsgemeinschaft. D. F. B. was supported by a fellowshipbated with Avidin-Texas red for 20 min. After a final washing step, from Ministerio de Educacion y Ciencia of Spain.thymocytes were resuspended in PBS/2% fetal calf serum andsorted on a FACStar Plus flow cytometer equipped with a dual laser Received June 29, 2000; revised October 16, 2000.system (Becton Dickinson).

For cytoflourometric analysis, thymocytes were resuspended in ReferencesPBS/2% fetal calf serum containing 0.05% sodium azide. Stainingwas performed as described above; however, Avidin-APC (Phar- Aifantis, I., Buer, J., von Boehmer, H., and Azogui, O. (1997). Essen-Mingen) was used, and cells were analyzed on a FACSCalibur flow tial role of the pre-T cell receptor in allelic exclusion of the T cellcytometer (Becton Dickinson). receptor beta locus. Immunity 7, 601–607.

Aifantis, I., Azogui, O., Feinberg, J., Saint-Ruf, C., Buer, J., and vonIsolation of TCRab1, CD4, or CD8 SP Cells Boehmer, H. (1998). On the role of the pre-T cell receptor in abSingle-cell suspension of total thymocytes was divided in two differ- versus gd T lineage commitment. Immunity 9, 649–655.ent tubes. Cells in one tube were incubated with anti-CD4 antibody

Akashi, K., Kondo, M., and Weissman, I.L. (1998). Role of interleu-conjugated with magnetic beads to deplete CD4-positive cells (con-kin-7 in T-cell development from hematopoietic stem cells. Immunol.taining CD4SP and DP cells, remaining cells were CD8SP and DNRev. 165, 13–28.cells). Another was incubated with anti-CD8 antibody conjugatedAnderson, S.J., Levin, S.D., and Perlmutter, R.M. (1994). Involvementwith magnetic beads to deplete CD8-positive cells (containingof the protein tyrosine kinase p56lck in T cell signaling and thymo-CD8SP and DP cells, remaining cells were CD4SP and DN cells).cyte development. Adv. Immunol. 56, 151–178.The remaining cells in the two tubes were combined and stained

with anti-TCR b–FITC, anti-CD8-PE, and anti-CD4-Quanturmred. Bakker, T.R., Reed, D., Renno, T., and Jongeneel, C.V. (1999). Effi-CD4SP or CD8SP cells were further separated according to the cient adenoviral transfer of NF-kB inhibitor sensitizes melanoma toexpression of TCR b. tumor necrosis factor-mediated apoptosis. Int. J. Cancer 80,

320–323.Analysis of Luciferase Activity Barber, D.F., Passoni, L., Wen, L., Geng, L., and Hayday, A.C. (1998).Approximately 0.5 3 106 sorted thymocytes were centrifuged and The expression in vivo of a second isoform of pTa: implications forlysed in 25 ml of passive lysis buffer (Promega) for 20 min. After the mechanism of pTa action. J. Immunol. 161, 11–16.centrifugation at 16,000 3 g, the supernatant was added to 100

Barkett, M., and Gilmore, T.D. (1999). Control of apoptosis by Rel/ml luciferase substrate (Promega) and analyzed in a luminometer

NF-kappaB transcription factors. Oncogene 18, 6910–6924.(Berthold).

Berger, M.A., Dave, V., Rhodes, M.R., Bosma, G.C., Bosma, M.J.,Kappes, D.J., and Wiest, D.L. (1997). Subunit composition of pre-TAnalysis of Nuclear NF-kB DNA Binding Activitycell receptor complexes expressed by primary thymocytes: CD3dPreparation of cell extracts and carrying out of EMSA analysis wasis physically associated but not functionally required. J. Exp. Med.as described previously (Ghosh and Baltimore, 1990)186, 1461–1467.

Boothby, M.R., Mora, A.L., Scherer, D.C., Brockman, J.A., and Bal-Bromodeoxyuridine Assaylard, D.W. (1997). Perturbation of the T lymphocyte lineage in trans-To determine cell proliferation, wild-type or IkBa-mutated trans-genic mice expressing a constitutive repressor of nuclear factorgenic mice received four i.p. injections of 1 mg BrdU in PBS. Thymo-(NF)-kB. J. Exp. Med. 185, 1897–1907.cytes were stained with anti-CD4-APC, anti CD8-Quantumered, and

anti CD25-PE. Cells were then resuspended in 360 ml of ice-cold Cheng, A.M., Negishi, I., Anderson, S.J., Chan, A.C., Bolen, J., Loh,D.Y., and Pawson, T. (1997). The Syk and ZAP-70 SH2-containing70% ethanol. After 30 min incubation on ice, cells were washed and

incubated in 500 ml of PBS containing 1% formaldehyde and 0.01% tyrosine kinases are implicated in pre-T cell receptor signaling. Proc.Natl. Acad. Sci. USA 94, 9797–9801.Tween-20 at 48C overnight. Cells were then treated with 50 units/

NF-kB in Thymocyte Survival689

DeJarnette, J.B., Sommers, C.L., Huang, K., Woodside, K.J., Em- et al. (1992a). Mutations in T-cell antigen receptor genes alpha andbeta block thymocyte development at different stages. Nature 360,mons, R., Katz, K., Shores, E.W., and Love, P.E. (1998). Specific

requirement for CD3e in T cell development. Proc. Natl. Acad. Sci. 225–231.USA 95, 14909–14914. Mombaerts, P., Iacomini, J., Johnson, R.S., Herrup, K., Tonegawa,

S., and Papaioannou, V.E. (1992b). RAG-1-deficient mice have noEsslinger, C.W., Wilson, A., Sordat, B., Beermann, F., and Jongeneel,mature B and T lymphocytes. Cell 68, 869–877.C.V. (1997). Abnormal T lymphocyte development induced by tar-

geted overexpression of IkBa. J. Immunol. 158, 5075–5078. O’Reilly, L.A., Huang, D.C., and Strasser, A. (1996). The cell deathinhibitor Bcl-2 and its homologues influence control of cell cycleFehling, H.J., Krotkova, A., Saint-Ruf, C., and von Boehmer, H.entry. EMBO J. 15, 6979–6990.(1995). Crucial role of the pre-T-cell receptor a gene in development

of ab but not gd T cells. Nature 375, 795–798. O’Reilly, L.A., Harris, A.W., and Strasser, A. (1997). bcl-2 transgeneexpression promotes survival and reduces proliferation of CD3-CD4-Franzoso, G., Carlson, L., Xing, L., Poljak, L., Shores, E.W., Brown,CD8- T cell progenitors. Int. Immunol. 9, 1291–1301.K.D., Leonardi, A., Tran, T., Boyce, B.F., and Siebenlist, U. (1997).

Requirement for NF-kB in osteoclast and B-cell development. Genes O’Shea, C.C., Thornell, A.P., Rosewell, I.R., Hayes, B., and Owen,Dev. 11, 3482–3496. M.J. (1997). Exit of the pre-TCR from the ER/cis-Golgi is necessary

for signaling differentiation, proliferation, and allelic exclusion inGhosh, S., and Baltimore, D. (1990). Activation in vitro of NF-kB byimmature thymocytes. Immunity 7, 591–599.phosphorylation of its inhibitor IkB. Nature 344, 678–682.Pivniouk, V., Tsitsikov, E., Swinton, P., Rathbun, G., Alt, F.W., andGhosh, S., May, M.J., and Kopp, E.B. (1998). NF-kB and Rel proteins:Geha, R.S. (1998). Impaired viability and profound block in thymo-evolutionarily conserved mediators of immune responses. Annu.cyte development in mice lacking the adaptor protein SLP-76. CellRev. Immunol. 16, 225–260.94, 229–238.Godfrey, D.I., Kennedy, J., Suda, T., and Zlotnik, A. (1993). A devel-Schwarz, E.M., Van Antwerp, D., and Verma, I.M. (1996). Constitutiveopmental pathway involving four phenotypically and functionallyphosphorylation of IkBa by casein kinase II occurs preferentially atdistinct subsets of CD3-CD4-CD8- triple-negative adult mouse thy-serine 293: requirement for degradation of free IkBa. Mol. Cell. Biol.mocytes defined by CD44 and CD25 expression. J. Immunol. 150,16, 3554–3559.4244–4252.Sen, J., Venkataraman, L., Shinkai, Y., Pierce, J.W., Alt, F.W., Bura-Groves, T., Smiley, P., Cooke, M.P., Forbush, K., Perlmutter, R.M.,koff, S.J., and Sen, R. (1995). Expression and induction of nuclearand Guidos, C.J. (1996). Fyn can partially substitute for Lck in Tfactor-kB-related proteins in thymocytes. J. Immunol. 154, 3213–lymphocyte development. Immunity 5, 417–428.3221.Haks, M.C., Krimpenfort, P., Borst, J., and Kruisbeek, A.M. (1998).Sentman, C.L., Shutter, J.R., Hockenbery, D., Kanagawa, O., andThe CD3gamma chain is essential for development of both theKorsmeyer, S.J. (1991). bcl-2 inhibits multiple forms of apoptosisTCRab and TCRgd lineages. EMBO J. 17, 1871–1882.but not negative selection in thymocytes. Cell 67, 879–888.Hettmann, T., DiDonato, J., Karin, M., and Leiden, J.M. (1999). AnShinkai, Y., and Alt, F.W. (1994). CD3 epsilon-mediated signals res-essential role for nuclear factor kB in promoting double positivecue the development of CD41CD81 thymocytes in RAG-2-/- micethymocyte apoptosis. J. Exp. Med. 189, 145–158.in the absence of TCR b chain expression. Int. Immunol. 6, 995–1001.Hoffman, E.S., Passoni, L., Crompton, T., Leu, T.M., Schatz, D.G.,Shinkai, Y., Rathbun, G., Lam, K.P., Oltz, E.M., Stewart, V., Mendel-Koff, A., Owen, M.J., and Hayday, A.C. (1996). Productive T-cellsohn, M., Charron, J., Datta, M., Young, F., Stall, A.M., et al. (1992).receptor b-chain gene rearrangement: coincident regulation of cellRAG-2-deficient mice lack mature lymphocytes owing to inability tocycle and clonality during development in vivo. Genes Dev. 10,initiate V(D)J rearrangement. Cell 68, 855–867.948–962.Spanopoulou, E., Roman, C.A., Corcoran, L.M., Schlissel, M.S., Sil-Horwitz, B.H., Scott, M.L., Cherry, S.R., Bronson, R.T., and Balti-ver, D.P., Nemazee, D., Nussenzweig, M.C., Shinton, S.A., Hardy,more, D. (1997). Failure of lymphopoiesis after adoptive transfer ofR.R., and Baltimore, D. (1994). Functional immunoglobulin trans-NF-kB-deficient fetal liver cells. Immunity 6, 765–772.genes guide ordered B-cell differentiation in Rag-1-deficient mice.

Imbert, V., Rupec, R.A., Livolsi, A., Pahl, H.L., Traenckner, E.B.,Genes Dev. 8, 1030–1042.

Mueller-Dieckmann, C., Farahifar, D., Rossi, B., Auberger, P.,Strasser, A., Harris, A.W., and Cory, S. (1991). bcl-2 transgene inhib-Baeuerle, P.A., and Peyron, J.F. (1996). Tyrosine phosphorylationits T cell death and perturbs thymic self-censorship. Cell 67,of IkB-a activates NF-kB without proteolytic degradation of IkB-a.889–899.Cell 86, 787–798.von Boehmer, H., and Fehling, H.J. (1997). Structure and functionIrving, B.A., Alt, F.W., and Killeen, N. (1998). Thymocyte developmentof the pre-T cell receptor. Annu. Rev. Immunol. 15, 433–452.in the absence of pre-T cell receptor extracellular immunoglobulinvon Boehmer, H., Aifantis, I., Azogui, O., Feinberg, J., Saint-Ruf, C.,domains. Science 280, 905–908.Zober, C., Garcia, C., and Buer, J. (1998). Crucial function of theIvanov, V., and Ceredig, R. (1992). Transcription factors in mousepre-T-cell receptor (TCR) in TCR b selection, TCR b allelic exclusionfetal thymus development. Int. Immunol. 4, 729–737.and ab versus gd lineage commitment. Immunol. Rev. 165, 111–119.

Jacobs, H., Vandeputte, D., Tolkamp, L., de Vries, E., Borst, J., andWeih, F., Carrasco, D., and Bravo, R. (1994). Constitutive and induc-Berns, A. (1994). CD3 components at the surface of pro-T cellsible Rel/NF-kB activities in mouse thymus and spleen. Oncogenecan mediate pre-T cell development in vivo. Eur. J. Immunol. 24,9, 3289–3297.934–939.Zhang, W., Sommers, C.L., Burshtyn, D.N., Stebbins, C.C., DeJar-Kasibhatla, S., Genestier, L., and Green, D.R. (1999). Regulationnette, J.B., Trible, R.P., Grinberg, A., Tsay, H.C., Jacobs, H.M., Kes-of fas-ligand expression during activation-induced cell death in Tsler, C.M., et al. (1999). Essential role of LAT in T cell development.lymphocytes via nuclear factor kappaB. J. Biol. Chem. 274, 987–992.Immunity 10, 323–332.

Levelt, C.N., Mombaerts, P., Iglesias, A., Tonegawa, S., and Eich-Zuniga-Pflucker, J.C., Schwartz, H.L., and Lenardo, M.J. (1993).mann, K. (1993). Restoration of early thymocyte differentiation inGene transcription in differentiating immature T cell receptor(neg)T-cell receptor b-chain-deficient mutant mice by transmembranethymocytes resembles antigen-activated mature T cells. J. Exp.signaling through CD3 epsilon. Proc. Natl. Acad. Sci. USA 90, 11401–Med. 178, 1139–1149.11405.

Mercurio, F., Zhu, H., Murray, B.W., Shevchenko, A., Bennett, B.L.,Li, J., Young, D.B., Barbosa, M., Mann, M., Manning, A., and Rao,A. (1997). IKK-1 and IKK-2: cytokine-activated IkB kinases essentialfor NF-kappaB activation. Science 278, 860–866.

Mombaerts, P., Clarke, A.R., Rudnicki, M.A., Iacomini, J., Itohara,S., Lafaille, J.J., Wang, L., Ichikawa, Y., Jaenisch, R., Hooper, M.L.,