7/30/2019 Lympho Toxin Beta a Novel Member

1/10

Cell, Vol. 72, 847-856, March 26, 1993, Copyright 0 1993 by Cell Press

Lymphotoxin p, a Novel Member of the TNF FamilyThat Forms a Heteromeric Complex with Lymphotoxinon the Cell SutfaceJeffrey L. Browning,* Apinya Ngam-ek,Pornsri Lawton, l Janice DeMarinis, lRichard Tirard, E. Pingchang Chow,Catherine Hession, l Betsy DBrine-Greco, lSusan F. Foley, and Carl F. WaretBiogen Incorporated14 Cambridge CenterCambridge, Massachusetts 02142tDivision of Biomedical SciencesUniversity of California at RiversideRiverside, California 92521

Summary

The lymphokine tumor necrosi s f actor (TNF) has awell-defined role as an inducer of infl ammatory re-sponses; however, the function of the structurally re-lated molecule lymphotoxin (LTa) is unknown. LTa ispresent on the surface of activated T, B, and LAK cellsas a complex with a 33 kd glycoprotein, and cloningof the cDNA encoding the associated protein, calledlymphotoxin p (LTP), revealed it to be a type II mem-brane protein with significant homology to TNF, LTa,and the ligand for the CD40 receptor. The gene forLTP was found next to the TNF-LT locus in the majorhistocompatibili tycomplex (MHC), a region of the MHCwith possible linkage to autoimmune disease. Theseobservations raise the possibility that a surface LTa-LTP complex may have a specific role in immune regu-lation distinct from the functions ascribed to TNF.

IntroductionThe initiation of the immune response invol ves a complexarray of intercellular signals, usual ly soluble cytokinescoupled with a number of cell-cell contact-dependent sig-nals. The contact-dependent events, most notably activa-tion of the T cell receptor, lend specificity to the response,whereas the soluble mediators are generally responsiblefor maintenance of cell differentiation and proliferation.Tumor necrosis f actor (TNF) and lymphotoxin (LT) (alsocalled TNF-8) are related cytokines involved in many regu-latory activities (Fiers, 1991; Beutler, 1990; Paul and Rud-dle, 1988). Their roles in the immune system are somewhatof an enigma since in vivo experiments suggest very criti-cal functions (Jacob and McDevitt, 1989; Ruddle et al.,1990; Kossodo et al., 1992) yet the corresponding in vitrowork has not led to a very clear picture of their place in Tand B cell regulation (Tartaglia et al., 1991).

TNF is synthesized in response to various insults bya variety of cell types, including both hematopoieti c andnonhematopoieti c cells (Beutler, 1990; Spriggs et al.,1988; Jevnikar et al., 1991) and is generally regarded asone of the primary i nitiating events in the infl ammatorycascade. LT, in contrast, is made specifically by lympho-

cytes (Paul and Ruddle, 1988), and its biological role is notunderstood. Both genes lie closely spaced within the classIll region of the major histocompatibilit y complex (MHC)(Spieset al., 1986; Nedospasovet al., 1986; Mueller et al.,1987; Gardner et al., 1987) yet they are clearly i ndepen-dently regulated (Sung et al., 1988). In general, LT andTNF display simil ar spectra of activities in in vitro systems,although LT is often less potent (Browning and Ribolini,1989) or displays apparent partial agonist activity (An-drews et al., 1990). Moreover, the two known TNF recep-tors do not appear to discriminate between t he two mole-cules (Schall et al., 1990; Smith et al., 1990). Theseobservati ons suggested that LT was either a poorly redun-dant cytokine or that there were further facets, as yet un-known, t o this cytokine. Within this context, it was of inter-est that LT is found on the surface of activatedlymphocytes (Browning et al., 1991; Androlewicz et al.,1992; Ware et al., 1992; Abe et al., 1991, 1992; Miyake etal., 1992).A number of what appeared originally to be soluble cy-tokines or growth factors have now been shown to exist inmembrane-bound forms, e.g., transforming growth factora, TNF, and the kit ligand (Massague, 1990; Flanagan andLeder, 1990) and it is likely that the switching betweensoluble and membrane forms is an important regulatoryevent. In all of these cases, retention of a transmembraneregion underlies the membrane association. TNF is a typeII membrane protein similar to LT and the ligand for theCD40 receptor (Farrah and Smith, 1992; Hollenbaugh etal., 1992), and it is retained on the cell surface in bothmacrophages and T cells (Kreigler et al., 1988; Perez etal., 1990; Kinkhabwala et al., 1990; Ware et al., 1992).Surface LT does not result from the presence of the trans-membrane region, but rather was found associated with a33 kd integral membrane glycoprotein (Browning et al.,1991; Androlewicz et al., 1992). We hypothesized that thisunique complex represented a more relevant form of LTand imparts specificity relative to TNF.

In this paper we describe the cloning of the gene encod-ing this second protein in the surface LT complex, note itsresemblance to other members of the TNF-LT family, anddelineate its genomic location next to the TNF-LT locusin the MHC. Since this protein, ~33, forms a complex withLT, is structurally related to LT, and lies next to the TNF-LT locus in the genome, we have given the names LTaand LT6 to the original LT and the novel gene, respectively,as is typical of subunits of a single structure.

ResultsIsolation of the LTf3 cDNAThe previously defined p33 protein was purified by affinitychromatography, and both N-terminal and internal aminoacid sequences were obtained. A degenerate oligonucleo-tide based on the sequence EEEPET was used to screena cDNA li brary from phorbol myristate acetate (PMA)-

7/30/2019 Lympho Toxin Beta a Novel Member

2/10

Cell848

1

61

121

181

241

301

361

421

481

541

601

661

721

781841

CAGTCTCMTGGGGGCACTGGGGCTG~GGGCAGGGGT~GAGGCTCCAGGGGAGGGG~MO ALGL B G R G ER LO OR G S l8

CCCTCCTGCTAGCTGTGGCAGGAGCCACTTCTCTGGTGACC~GTTGCTGGCGGTGCCTALLLAVAGATS L V T L LL AV P 138

TCACTGTCCTGGCTGTGCTGGCCTTAGTGCCCCAGGATCAGGGAGGACTGGTMCGGAGATVLAVLAL V P Q D QG GL VT ET 58

CGGCCGACCCCGGGGCACAGGCCCAGCMGGACTGGGG~CAGMGCTGCCAGAGGAGGAD P G AQAQ Q G L G P Q XL P E E E 70

AGCCAGAAACAGATCTCAGCCCCGGGCTCCCAGCTGCCCACCTCATAGGCGCTCCGCTGAPET D L S P G L P AA H L I G A P L K 98

AGGGGCAGGGGCTAGGCTGGGAGACGACGMGGMCAGGCGTTTCTGACGAGCGGGACGCG Q G L G W E T T K E 0 A P L T S G TQ 118

AGTTCTCGGACGCCGAGGGGCTGGCGCTCCCGCAGGACGGCCTCTATTACCTCTACTGTCBSDAEGLALPQDGLYYLYCL 138

TCGTCGGCTACCGGGGCCGGGCGCCCCCTGGCGGCGGGGACCCCCAGGGCCGCTCGGTCAVGYRGRAPP GGGDPQGRS V T 158

CGCTGCGCAGCTCTCTGTACCGGGCGGGGGGCGCCTACGGGCCGGGCACTCCCGAGCTGCL R S S LYRA G G A Y G P GT P B L L 178

TGCTCGAGGGCGCCGAGACGGTGACTCCAGTGCTGGACCCGGCCAGGAGACMGGGTACGLEGAETVTPVLDPARRQGYG 198

GGCCTCTCTGGTACACGAGCGTGGGGTTCGGCGGCCTGGTGCAGCTCCGGAGGGGCGAGAP LWY T SV G P G G L VQ LR R G E R 218

GGGTGTACGTCMCATCAGTCACCCCGATATGGTGGACTTCGCGAGAGGGMGACCTTCTv Y v N I S HPDMVDFARGKT P P 238

TTGGGGCCGTGATGGTGGGGTGAGGGMTATGAGTGCGTGGTGCGAGTGCGTGMTATTGGAVMVG' 244

GGGGCCCGGACGCCCAGGACCCCATGGCAGTGGG~TGTAGGAGACTGTTTGG~TTGATTTTGAACCTGATG-TAAAGAATGGAAAGCTTCAGTGCTGCCGATAAAMAAttt***

Figure 1. Nucleotide and Predicted AminoAcid Sequence of the cDNA Encoding LTbThe putative membrane-spanning region isfound between amino acids 19 and 48. Theunderlined sequences were previously deter-mined by N-terminal amino acid sequencing ofthe affinity-purified protein and its tryptic frag-ments with spaces within one peptide, indicat-ing a tentative assignment.

activated II-23 cells, a human T cell hybridoma that dis-plays large amounts of surface LT upon phorbol ester acti-vation (Browning et al., 1991). The cDNA encodes for a240-244 amino acid sequence (molecular mass of 25-26kd) typical of a type II membrane protein (Figure l), andno identical sequences were found within the EMBL orGenBank data bases. This protein has been named LT8(Figure 1). Following a short 15-19 amino acid N-terminalcytoplasmic domain, there is an extensive stretch of 30hydrophobic amino acids that presumably acts as a mem-brane-anchori ng domain. Biochemical analyses were con-sistent with the presence of one or more methioni ne resi-dues within lo-20 amino acids from the N-or C-terminusand one or more cysteine residues (Browning et al., 1991).The cloned cDNA reveal ed the existence of one cysteineresidue in the extracellular domain and two methionineswithin t he last C-terminal 17 amino acids, in agreementwith the prior characterization. The protein possesses anN-linkedglycosylati on consensussequence, as previouslyexpected on the basis of carbohydrate analysis (Browninget al., 1991). The difference between the 33 kd size of thepreviously analyzed protein and the encoded 25 kd resultsat least partially from N-linked glycosylation and possiblysmall inaccuracies in sizing on a SDS-polyacrylamide gel.The 5end of the cDNA was difficult to determine despiteanalysis of many independent clones. Based on N-ter-minal amino acid analysis, the start site was believed toprecede the GLEG sequence. The cDNAs uniformly

lacked an ATG in the preceding codon yet did possessCTG in this position. It was postulated that a CTG transla-tional start was utilized by this gene, and, as shown below,a cDNA clone starti ng at the first CTG was expressed ina transient transfecti on experiment in a functional form.To define further the 5cDNA sequence, primer extensionanalysis was undertaken, and 125-l 28 bp extension prod-ucts were obtained (Figure 2) and sequenced. The com-plete cDNA sequence based on Maxam-Gilbert sequenc-ing of the primer extensi on product reveals a 5-8 bp 5untranslated stretch and an in-frame methionine yieldinga potential N-terminal amino acid sequence of MGALGLE.Overall, the difficulty in obtaining a full-length cDNA wasdue either to the very short nature of the 5 untranslatedend or the very GC-rich content of the Send. Both leucinesare encoded by CTG codons, and the proximity of the ATGto the 5end of the messenger RNA(mRNA) suggests thatit may not be functional and that most translation initiatesat one or both of the two CTG sit es. No evidence for minorGALGLE or LGLE N-terminal amino acid sequences wasfound, which is consistent with aCTG start site. Moreover,the ATG codon lacks the typical consensus elements sur-rounding normal initiating codons, further implicating aCTG initi ation site (Kozak, 1986). If the CTG is the relevantinitiating codon, the apparent lack of an N-terminal leucinesuggests processing of whatever amino acid is transferredto this CTG. The exact nature of the protein product arisingfrom a CTG start codon in eukaryotes is unclear.

7/30/2019 Lympho Toxin Beta a Novel Member

3/10

Lymphotoxin 9 Is a Component Of Cell Surface Lymphotoxin849

bp344220154

75

Figure 2. Primer Extension Analysis of LT9 RNAAutoradiograph of a denaturing polyacrylamide gel of the extensionproducts. Lane 1, labeled primer alone; lane 2, extension product usinguninduced II-23 RNA; lane 3, extension product usi ng PMA-i nducedII-23 RNA. The position of Hinfl pBR322 fragments is indicated.

Comparison with the TNF FamilyComparison of the LTf3 sequence with other proteinsknown to bind to members of the TNF receptor family re-veals considerable structural similarity (Figure 3). All threeknown ligands to members of theTNF-nerve growth factor(NGF) receptor family (TNF, LTa, and the CD40 ligand)are type II membrane proteins and share at least five largeregions of sequence conservation in the extracellular do-main, as indicated in Figure 3. Using the alignment in thisfigure, LT8 is 21%, 240/o, and 19% identical with TNF,LTa, and CD40 l igand, respectively, in the region definedby the fourth exon. This level of identity is basically similarto that found bet ween TNF and LTa. The positi on of theglycosylation site in LTf3 and CD40 ligand is identical butdiffers from the site in LTa. Both TNF and LTa are homotri-mers, whereas the quaternary structure of the CD40 ligandis unknown. The regions of homology between membersof the TNF family when in-laid into the crystal structure ofLTa (Eck et al., 1992) are found primarily on the internal8 strands A, H, C, and F, although the external-facing f3strands A and G show significant conservation (Figures4A and 48). A view of the base:of the LTa trimer reveals theconservation in A, F, C, and H strands where the interfacesform that stabilize the oligomeric structure (Figure 4C).The homology regions contain many of the contact resi-dues involved in forming t he trimer, except for three resi-dues (51,53, and 55) in the Astrand that have conserva-tive substitutions between LTa, LT8, and TNF, but notCD40 ligand. Additional conserved residues (L102, W104,L113, L125, and L127) are located on outer 8 strands butpoint inward, contributing to the interactions between theinternal and external sheets. Analysis of a space-fillingmodel of LTa (Figures 4D and 4E) reveal that most ofthe conserved regions create a large contiguous swath ofresidues on the internal sur face of the trimer. Only a small

portion of the residues in these homology regions arefound at the solvent-accessible surface near the interac-tion crevice of the two subunits (except near the base)where the receptors are thought to bind, and thus the non-conserved residues may impart receptor specificity toeach ligand.Characterizati on of the LTP GeneIn light of the tandem arrangement of the TNF and LTagenes, a cosmid clone, 031 A, containing the human TNFand LTa locus was examined and found to contain the LT6gene. A 6 kb EcoRl fragment was sequenced and agreedwith the primer extension sequence confirming the lack ofan intron in the 5untranslated region (Figure 5). The exactlocation of the genomic EcoRl fragment was establishedby sequencing a Xhol fragment that linked the end of theEcoRl LT8 fragment and the 3end of TNF (Figure 6). Thelocalization of the LTf3 gene to within 2 kb of the TNF genewas not surprising in view of the proximity of the TNF andLTa genes (Nedwin et al., 1985). Thus, the LT8 gene issandwiched between the TNF and 8744 genes in the classIll region of the MHC. The 8744 gene is expressed in Bcells and in macrophages (Tsuge et al., 1987) and hadbeen found to be closely l inked to the TNF-LT locus (Spieset al., 1989). The LTf3 gene is contained within f our exonsand spans 2 kb in an arrangement very similar to that of

hTNFhTNPhLT-lhLT-phCDIOL

hTWhLT-ahLT-phCD4OL

hTNFhLT-aUT-phCDIOL

hTWhLT-ahm-phCDlOL

TNFhLT-ahLT-phcD4OL

hTNPhLT-nhLT-phCD4OL

.

Figure 3. An Amino Acid Sequence Comparison of Four Members ofthe Family of Ligands Binding to Members of the TNF-NGF ReceptorFamilyHomology regions are shown boxed with sequence identity indicatedby a dot and conserved sequences by a plus sign. Putative N-linkedglycosylation sites are underlined. The sequence for human CD40ligand was taken from Hollenbaugh et al. (1992).

7/30/2019 Lympho Toxin Beta a Novel Member

4/10

Figure 4. Homology Regions of the TNF-LT Cytokine Family Defined in the Crystal Structure of LTaIndividual 8 strands are labeled according to the convention of Sprang (Eck et al., 1992) where the A strand emerges from the amino-terminal end(residue 26) and the carboxyl terminus (residue 117) is at the end of the H strand. The conserved homology regions (HR) are defined, with theparentheses indicating the LTa sequence numbering and 8 strand assignment, as HR-1 (29-34; A), HR-2 (43-45; A), HR-3 (71-81; C), HR-4 (136-147; F), and HR-5 (165-171; H) and correspond to the boxed residues in Figure 3. The homology regions are colored gold in all panels.(A and B) A single subunit of LTu is shown (in schematic form using ri bbons) in the orientation the monomer would have in the trimer with the aminoterminus pointing downward (A) and the monomer rotated around the vertical axis, revealing the interior side of the subunit (B).(C) A bottom vi ew of the LTa trimer rotated 90 on the horizontal axis.(D) A space-filli ng model of the trimer, in which the individual subunits are colored blue, green, or gray. The amino-terminal residues protrude fromthe base as shown above.(E) The green-colored subunit has been removed, leaving a dimer that has been rotated around the vertical axis, exposing the interior.

TNF and LTa except for being oriented in the oppositedirection. Only one copy of the gene is present in the hu-man genome, as defined by stringent Southern analysis.The position of the intron-exon junction linking the lastlarge exon that encodes essentially all of the extracellulardomain and most likely the receptor-bind ing region is com-pletely conserved in all three genes, underscoring the im-portance of the fourth exon. The promoter region of LTf3contains putative TATA and CAAT elements at positions -26 and -180. The AU-rich motif f ound in the Buntranslatedregion of LTa and TNF that is involved in determiningmRNA stability is lacking in LT6 (Shaw and Kamen, 1986).

Expression of LTPNorthern analysis of II-23 cells showed hybridization ofthe LT6 cDNA to a 0.9-l .O kb mRNA, indicating that thecloned cDNA represents essentially all of the transcribedgene. The LT(3 gene was expressed at low levels in un-treated I I-23 hybridoma cells; however, upon cell activa-tion with phorbol ester, mRNA levels increased dramati-cally (Figure 7A). Thus, it is clear that expression of bothsubunits of the LTsurface complex a re induced upon acti-vation of this T cell hybridoma. The cell line Hut-78, whichconstitutively displays surface LT (Ware et al., 1992) ex-pressed low levels of LT6 mRNA in the absence of phorbol

7/30/2019 Lympho Toxin Beta a Novel Member

5/10

Lymphotoxin 6 Is a Component of Cell Surface Lymphotoxin651

~CtggCttCCtgCaagCtaCcttccctccctgggcctcagtttctctctctgctg~gcc~gaag~tgtcta~~gacccctttggttccaccctgagagcctgtCtCCC~CtCaaCttcttccccagttcag~gaacccaggcatccagctgccccaccccagctctgggt~~acaggaagctgggtgagggg~gcaggggtgtgcggaaagtcccagcc~ggtgtgcaggtctac&gggagggggtgggcccgtccctgaggtataa~agccccctgctctggctctgg~AG~TCAATGGGGGCACTGGGGC TOOAGOCCACCDOTOCOACOCTCCAOOOCA000(3TTCCCYGALGLEG RGGRLQ G R G S LLLAVAGAT SLVTLLOCTOGCOCMCCTATCACTCTC-~~-C~AG~CCCA~~A~~C~tgagtggctgcaacaggccctggtggagagttgta

LAVPITVLAVLALVPQDQGGLtcttgcggatgcttggctccctctggttgtgcctgtggtcttttgccccctctggctc~gctggctcggctgtccctggtgggg~tgtcttgtctctttgCtgactCtCtttCCatgttcctgtgatgttgtgcttgtgtcccgacata~gccccttgtgtctcctctcctcttcccg~ggt~catctgtttctccgccca~gt~cct~tgccttgcttgttctcccttct~aggtpcctgcagcttgggtcc~ctttc~gaggggtaggggtgaCatgagctgaatctgaactctg~~CtCtgggCaCtgtgaCCCCaCccaaCCagGTMC~~C~CGACCCC~ACA~CCA~MG

VTETADPGAQAQQG

100200300400500600700so0900

OlrCTOOgtaagagcagactgtctctccttccccgcttcagaccctcaggggctcccagctccctgctgcgtcccc~gatacctcttcctctagga~tcc~ 1000L GggCtCCCCltCCCtgCgCCCtgttCtCtC~~gggt~gCCtgCatgggtggCtgCCCtgCCCCC~atCgtgg~CtCtttgCCccttcCagffiTT~A~G 1100

P Q KCTOCCAOACGAOOACCCA~CA~~~A~CCC-~CCA~~CCACC~ATA~trraggacctccaagacctgaataagagtgtaaataatc 1200L PEE E PETDLSPGLPAAHLIGCgaaggttCCagttCtgCtCgccc~gagtccttcggctccatgattcc~gtgctcggtttcccacccgcttc~cgaccttttgtcgctcgtgccc~ctct 1300tacgctcgtccccgcagtgtagtttcttcttccctccggtgcaagc~aa~gccggcctggaggtcccc~ctacagcgttctgc~ccccacatccgtgttc 1400cctcggcccccaactcgcactoatcccc~g~~acagcacc~tccctcctcccccggcccggctcggctcccgcaggggctaaaagccgccacttcccc~ga 1500agtCCC~~gCCttt~gg~tCgC~ttCCCaag~gCgCgtCggCCCgtgtCtCCgCag~~~C~~M~A~TA~~~C~C~G 1600

APLKGQGL GWETTACMCACDCGTTTCTDAC~~~C~AG~~~C~C~~-~~CC~A~C~C~TATTACC~TAC~~~G~~TACC 1700EQAPLTSGTQP SDAEGLALPQDGLYYLYCLVGYRGGGGCCOCCCGCCCCCTGGCOCCOOOCACCCCCCAGGGCCGCTCGGTCACGCTGCGCAGCTCTCTGTACCGGGC GGGGGGCGCCTACGGGCCGGGCACTCC 1800G R A P PGGGDPQGRSVTLRSSLYRAGGAYGPGTPCoAcCTocTccTccAoooCOC~~C~~C~CAG~~CCC~CA~~CM~TAC~C~~~TA~C~~G~~~ 1900

ELLLEGAETVTPVLDPARRQGYGPLWYTSVGFGGOCCTOOTGCAOCTCCOOACCCOCCAGACOOTCTACOTCM -CGTGA 2000G LV Q L RRGERVYVNISHPDYVDPARGKTFFGAVMTGGTGGGGTGAGGGAATATGAGTGCGTGGTGCGAGTGCGTfSAATATTGG%GCCCGGACGCCCAGGACCCCATGGCAG~ TCTAGGAGACTGT 1100V G lTTGGAAATTGATTTTGAACCTGATGAAAA T~~TOCAMDCTTCAG~~C~T~g~tgctgagttgcgacacacgtcttaattcagggtggg 2200Figure 5. Genomic Sequence of Human LT6lntrons are indicated by lowercase type. Putative TATA and CAAT el ements in the promoter region are underlined.

stimulation. Molt-4 and THP-1 cells that do not displaysurface LT also do not express LTa or LTf3 mRNA. Thetime course of phorbol ester induction of the LT subunitsin II-23 cells was sli ghtly diff erent for LTa and LTf3 (Figure76) suggesting nonidentical regulation, although whetherthese diff erences stem from differences i n the promotorregions or in mRNA stability was not investigated.

Human peripheral blood lymphocytes (PBLs) culturedwith anti-CD3 or interleukin-2 (IL-2) expressed both LTmRNAs (Figure 9) confirming earlier observati ons (Abe et

al., 1992; Ware et al., 1992) that surface LT was found onactivated l ymphocytes. While there was variable loadingof RNA in this analysis, comparison of RNA from freshPBLs and from IL-2-activated PBLs clearly i ndicates induc-tion of mRNA by IL-2 treatment. This observation corre-lates well with the abundant expression of surface LT onlymphokine-acti vated killer cells (Abe et al., 1992; Ware etal., 1992). Interestingly, there was some expression of LTbmRNA in both freshly isolated PBLs and in resting cellsafter 24 hr in culture. Earli er fluorescence-acti vated cell

- centromere Teleomere -) 2 kb8144 LT.!?, -) + TNF + LT-a

Figure 6. Schematic Diagram of the Region ofChromosome 6 Encompassing the TNF-LTLocusThe restriction map shows sites for EcoRl (E),Xhol (X), and Hindlll (t-f) as determined by Ned-ospasovet al. (1966)andSpiesetal. (1969)andconfirmed in this work. The expanded region ofLT5 shows restriction sites for Bglll (B), Kpnl(K), Pstl (P), and Ncol (N).

a P e x 400bpNH HKI I I I II II

7/30/2019 Lympho Toxin Beta a Novel Member

6/10

Cell852

LT-a

LT-b

Actin

m . *. i

Figure 7. Northern Analysis of LTa and LT8B. Hours post PMA Induction Expression in Several Cell Lines(A) Specific expression of both LTgenes in Hut-0 0.5 1 2 4 6 10 20 78 and II-23 cells.kb

-2 4(B)Timecourseof PMAinductionof LTmANAsin II-23 cells.

sorting (FACS) experiments also indicated low levels ofsurface LT in freshly i solated unstimulated cells that variedfrom donor to donor. LTb mRNA was observed in thespleen and thymus, but not in lung, fetal or adult brain,heart, muscle, liver, kidney, or placenta, consistent withlocalization to lymphocytes (data not shown). Thus, theseexperiments suggest that LTa and LTb expression mayparallel each other, in agreement with the observation thatall cell types known to produce LTa and LTf3 also displaysurface LT (Ware et al., 1992). The B lymphoblastoid lineRPM1 1788 does not display appreciabl e surface LT(Wareet al., 1992) yet it secretes LTa well, raising the specula-tion that there may be poor expression of LTf3 and hencelittle diversion of LTa molecules to the cell surface.To test the hypothesis that LTf.3expression targets nor-mally secreted LTa to the cell surface, CHO cells constitu-tively secreting LTa were transiently transfected withclone 12 LTf3 cDNA. Surface LTa expression was assayedby staining with monoclonal anti-LTa, foll owed by FACSanalysis. Transfection with LTf3 ed to surface LTa staining(Figure 9A), whereas a LTf3 cDNA that contained a splicingerror (clone 4) result ing in a nonfunctional frameshift, wasunable to target LTa to the surface. Transfection of CHOcells not expressing LTa did not result in surface LTa ex-pression, confirming that this monoclonal antibody cannotrecognize the related LTf3 protein (Androlewicz et al.,1992). An experiment with COS cells transfected with LTf3alone or cotransfected with LTa and LTf3 cDNAs confirmedthat surface LT expression requires both genes (Figure9B). The cDNA cl one used in these transfecti on experi-ments contained two base pairs between the cloning l inkerand the first CTG prior to the postulated GLEG matureN-terminal sequence. This DNA can encode for proteinexpression, and, given the lack of an upstream methioninein this clone, at least one of the putative initiating CTGscan be fully functional, as has been demonstrated for asmall number of mostly regulatory proteins (Kozak, 1991).Subsequent experiments wherein the CTG was substi-

tuted with an ATG did not improve expression in atransienttransfection (C. H., unpubli shed data).DiscussionThe cloning of LTb and its ability to target LTa to the cellsurface confirms our previous suppositi on that surface LTwas a result of a heteromeric complex formed betweenthe normally secreted LTa and the earlier defined p33

-+ ++ CulturedCells

kb2.4

LT-a 1.35

0.24

2.4

35T+0.24

Actin 2.41.35

Figure 8. Northern AnalysisNorthern analysis of fresh human PBLs and PBLs cultured for 1 daywith media alone, anti-CDS, or IL-2.

7/30/2019 Lympho Toxin Beta a Novel Member

7/10

Lymphotoxin 9 Is a Component of Cell Surface Lymphotoxin853

A100

r---i

CHO dhfr-Control DNA Control DNA

oj ,,,, ,-; ,,,,,,, oj ;.&,.,..10 100 1000 4000 10 100 1000 4000

B Control DNA Control DNA50\ 5)--j + T-aDNAj ,-L!i- ,.I:!,_ 10 100 1000 4000 10 100 1000 4000

z 5ojj\ LT.p DNA 5oT-

1000 4000 10 100Relative Fluorescence

Figure 9. Expression of Surface LTu CHO and COS Cells in TransientTransfection Experiments(A) CHO cells either drhydrofolate reductase minus (dhfr) or stablytransfected with the LTa gene were transfected with either a controlcDNA or the LTS in pCDMt3 After 2.5 days, cells were analyzed byFACS for surface LTa. Broken lines indicate control immunoglobulinG, staining, with solid lines showing anti-LTa staining.(B) COS cells were transfected with control cDNA or LTS with or withoutLTa cDNA In pCDMf3 and analyzed as above.

molecule. The comparison of all three genes (TNF, LTa,and LT8) led to the definition of homolog y regions, as pre-viously described for the related CD40 ligand (Farrah andSmith, 1992; Hollenbaugh et al., 1992). TNF and LTa areknown to be homotrimers (Wingfield et al., 1987; Smithand Baglioni, 1987; Browning and Ribolini, 1989), and thehomology regions lie on the internal surfaces involved intrimer formation in theTNF and LTa crystal structures (Ecket al., 1992; Ec k and Sprang, 1989; Jones et al., 1989;Tavernier et al., 1989). This observation provides a struc-tural basis for the association between LTa and LT8 inthat this heteromeric complex most likely retains a trimericstructure similar to TNF and LTa, with the homology re-gions interacting in a heterotypic fashion. From chemical

cross-linking experiments with the heteromeric surfaceLTa-LT8 complex, LTf3 was believed to exist primarily as adimer, and hence the stoichiometry of the overall complexwas believed to be al:bZ (Androlewicz et al., 1992). Prelimi-nary data suggest that a small portion of the complex mayalso exist in an a*$, ratio (C. F. W., unpublished data).lmmunoprecipitation analysis indicated that LTa-LT8complexes were not secreted (Browning et al., 1991); how-ever, whether LTf3 homooligomers are formed and se-creted will be addressed with anti-LT8 monoclonal anti-bodies.

The concept of cell surface ligands as cytokines orgrowth factors has gained recognition primarily throughthe characterization of transforming growth factor a, TNF,and the kit ligand (Massagu e, 1990; Flanagan and Leder,1990; Wong et al., 1989) although membrane forms of anumberof factors have been described (Massague, 1990).The membrane form of LTa we have characterized differsfrom the above systems in that it does not retain its trans-membrane domain, but rather is anchored via LTD. Theexistence of a heteromeric complex of lymphokines is alsoreminiscent of signaling molecules in other areas, e.g.,cytotoxic lymphocyte maturation factor (Gubler et al.,1991), platelet -derived growth factor (Raines et al., 1990)and heteromeric inhibin-activin complexes (Vale et al.,1990). The existence of similar heteromeric configurationsof signaling molecules should be considered for othermembers of the TNF family. The elucidation of hetero-merit receptor forms may be especially important for het-eromeric ligands, as is the case for another member of theTNF receptor family, the low affinity NGF receptor thatinteracts with the trk proto-oncogene (Hempstead et al.,1991).

The restricted expression of LTa relative to TNF hastantalized workers in the field with the idea that LT hasspecific and important immunoregulatory functions (Pauland Ruddle, 1988; Ruddle a nd Homer, 1988). Delineationof the LTa-LT8 complex poses the possibility of immuno-regulatory activities unique to the complex that cannot bemimicked by the LTa homotrimer. We are hypothesizingthat the surface LTa-LTf3 complex binds to a unique recep-tor or combinations of receptors, leading to a high affinityinteraction and biologicall y relevant signaling. In support,preliminary data indicate that the major LT surface com-plex cannot bind to the two known TNF receptors (C. F. W.,unpublished data). I n this model, it is possible that therelatively poor activity of the LTa homotrimer relative toTNF in many systems indicates that the secreted LT phe-nomenon is only peripherally related to the true functionof LT. The tethering of soluble LTa to the cell surface viacomplexation with LT8 raises thespeculation that cell-cellcontact-specific signaling through LTa-LT[I is an im-portant aspect of immune regulation. The CO40 receptor-ligand pair represents a signaling mechanism whereby theT cell provides help to the B cell via a cell-cell contact(Armitage et al., 1992; Hollenbaugh et al., 1992). By draw-ing parallels with the CD40 system, one could postulatethat surface LTa-LT8 may be a component of the regula-tion of T cells or other immune cells such as LAK or NKcells and B cells. Alternatively, in keeping with the known

7/30/2019 Lympho Toxin Beta a Novel Member

8/10

Cell654

cytotoxic activities of TNF and LTa, either LTf3 or the LTa-LT8 complex may be involved in inducing programmed celldeath through a cell-cell contact-dependent mechanism.The programmed cell death observed in conjunction withhuman immunodeficiency virus (HIV) infection may in-volve aspects of the LT system (Ameisen, 1992).

The TNF receptor family has grown to a substantial size,inviting the speculation that several TNF-like ligands mayexist. The addition of the CD40 ligand, and now LT8, tothe family of TNF-related ligands reinforces this premise.The localization of the LTa gene to the TNF-LT locuswould suggest more extensive duplication of a primordialLT gene than was previously realized, and several TNF-related ligands could beclustered within the class Ill regionof the MHC. This region of the MH C may be associatedwith some autoimmune conditions such as insulin-depen-dent diabetes mellitus (Badenhoop et al., 1990; Porciot etal., 1991), and the potential disease linkage has inspiredseveral analyses of gene structure in the region. Aberrantregulation of TNF has been proposed to be involved withthe phenotype of the autoimmune NZB mouse (Jacob andMcDevitt, 1989; Jongeneel et al., 1990) and chronic TNFadministration can rescue the nonobese diabetic mouse(Jacob et al., 1990). In humans, TNF restriction-lengthfragment polymorphisms have been linked to various pop-ulations (Messer et al., 1991; Fugger et al., 1989; Dawkinset al., 1989; Webb a nd Chaplin, 1990) and these ana lysescan now be extended to this novel membe r of the region.Microsatellite DNAs have been described on both flanksof the TNF-LT locus; however, neither of the reportedregions was located within the LT8 gene (Nedospasov etal., 1991). Given t he localizati on of LT8 to this region, it ispossible that this gene or its receptor is dysfunctional incertain autoimmune diseases. Interestingly, the Fas re-ceptor is a member of the TNF-NGF family of receptors,and antibody binding to this protein can induce apoptosis(Itoh et al., 1991). A defective Fas receptor molecule inmice harboring the Ipr allele results in a lymphoprolifera-tive lupus-like disorder (Wantanabe-Fukunaga et al.,1992). Whether LTf3 or the LTa-LTf3 complex interactswith Fas or other orphan receptors in this family (Mallettand Barclay, 1991) can be readily addressed. These ob-servations point to a fundamental role for this family ofreceptors and ligands in immune regulation. Now with de-lineation of LT8 and the CD40 ligand, it is clear that a famil yof TNF-related ligands is emerging to complement the al-ready extensive family of TNF-NGF-type receptors. Thesereceptor-li gand interacti ons point toward an additionalarray of important regulatory elements within the immunesystem, overlaying the known regulatory cytokine systems.Experimental ProceduresAmino Acid Sequencing of LTPII-23.D7 cells (5 x IOO) were stimulated for 6 hr with 50 ngiml PMAand lysed by mtrogen cavitatron under conditions described pre-viously, and LT6 was purified by affinity chromatography essentially asdescribed previously (Browning et al., 1991). From the column eluate,roughly 2 pg was electrophoresed on a SDS-polyacrylamide gel andblotted onto ProBlott (Appli ed Biosystems), and the remaining 4-5 pgof material was similarly resolved and blotted onto nitrocellulose. TheProBlott was subjected to N-terminal amino acrd sequencing (Matsu-

daira, 1967) using gas phase techniques in an automatic sequencer(Applied Biosystems). The nitrocellulose blot slice was digested withtrypsin in situ as described (Aebersold et al., 1987) and tryptic frag-ments were resolved by narrow bore reverse-phase high pressureliquid chromatography and sequenced using liquid pulse sequencing.Cloning of the LTP cDNA and Genomic FragmentA 32-fold degenerate oligonucleotide, GTYTCNGGCTCYTCYTC, wasdesigned on the basis of the EEEPET sequence and used to probe acDNA library prepared in pCDM6 with poly(A) RNA from II-23.D7 cellsstimulated with PMA for 4 hr as described (Aruffo and Seed, 1967).Filters were washed with 3 M tetramethylammonium chloride at 5OC.Multiple clones were i solated and sequenced using dideoxynucleotidemethodology. To isolate the IT6 gene, a BstEll-Xmnl fragment of thecDNA was used to probe a Southern blot of the cosmid 031A providedby T. Spies (Spies et al., 1969). A 6 kb EcoRl fragment that cross-hybridized tothecDNAprobe wassubcloned into pNNlOS(aderivativeof pUC carrying a kanamycin-resistance gene) and sequenced. Sepa-rate Xhol fragments of this region of the cosmid were also subclonedinto pBluescript II and sequenced.Northern AnalysesTotal RNA was isolated by the guanidine-SDS-cesium pellet method,or poly(A) RNA was prepared with a Micro Fast Track kit (Invitrogen).Several human poly(A) RNA samples were purchased from Clontech.Blots were prepared from formaldehyde-agarose gels and probed witha 0.6 kb BstEll-Xmnl LT6 fragment, a 0.6 kb BarnHI-Seal fragment ofhuman LTa, or a fragment of 6.actin. Blots were hybridized at 65%in a modified Denhardts solution and washed wit h 0.5x SSC. 1%SDS at 65C. Human PBLs were isolated using Ficoll and cultured inRPM1 1640 with 10% fetal bovine serum, glutamine, and antibioticswith or without 100 nglml IL-2 or 10 nglml OKT3.Transient Expression of LTPCHO cells either dihydrofolate reductase mi nus or stably transfected withthe LTu gene as described (Browning and Ribolini, 1969; the cell linewas gift from Dr. W. Fiers) were transfected by electroporation, re-moved with Ca/Mg-free Hankssolution with 5 mM EDTA, and analyzedby FACS. Cell s were stained with 10 uglml of a monoclonal humanLTa antibody (Boehringer Mannheim) followed by an affinity-purifiedFabm goat anti-mouse fluorescein isothiocyanate-labeled antibody(Cappel). Cells were also stained with propidium iodine to allow exclu-sion of any dead cells. Cell s were analyzed after 2.5 days with FACStarPlus (Becton-Dickinson), and because of the low expression levels inCHO cells, only the relatively bright fluorescein isothiocyanate cellswere live gated. The data presented show only propidium-negativecells.

COS cells were transfected similarly either with 20 frg of clone 12LT6 cDNA, a complete clone starting at the first CTG, or with controlDNA that was clone 4, a LTP cDNA with an internal frameshift due toa missplicing event. Both DNAs were transfected either alone or inconjunction with 20 ng of a complete LTa cDNA clone i n pCDM6,previously isolated from the II-23 cDNA library (M. Ward and J. L.B., unpublished data). The COS cell data were obtained from 30,000events where only dead cells were excluded by a live gate.Primer Extension AnalysesAn oligonucleotide primer, GACAGTGATAGGCACCGCCAGCAACAA,was annealed to 20 pgof poly(A) RNAfrom PMA-stimulated II-23cellsfor 6 hr at 65C, precipitated, and extended at 37C as described(Wallner et al., 1986). The extension products were resolved on a 6%polyacrylamide denaturing gel, and the extension product was excisedand subjected to Maxam-Gilbert sequencing.AcknowledgmentsWewish to thank Jackie Ashookfor help with the FACSanalyses; MaryWard, Gary Jaworski, Lars Tragethon, and Christopher Tonkin forassistance during vari ous phases of this project; and the scientists atBiogen for molecular biology advice. The suggestion by Steve Fawelland Richard Cate that a CTG may be the initiating codon was mostwelcome. The gift of the TNF-LT cosmid by Thomas Spies is greatlyappreciated, as is the use of the CHO-LT line provided by Walter

7/30/2019 Lympho Toxin Beta a Novel Member

9/10

Lymphotoxin e Is a Component of Cell Surface Lymphotoxin655

Fiers. We are indebted to Stephen Sprang for providing the structurecoordinates of LTa and to Marie Green and Marilyn Yoder for ass,+tance in molecular modeling and computer graphics.Received December 7, 1992; revised January 15, 1993.ReferencesAbe, Y.. Miyake, M., Horiuchi, A., Kimura, S., and Hitsumoto, Y. (1991).Expression of membrane-associated lymphotoxinltumor necrosis fac-tor-6 on human lymphokine-activated killer cells. Jpn. J. Cancer Res.82, 23-26.Abe, Y., Horiuchi, A.. Osuka, Y., Kimura, S., Granger, G. A., andGatanaga, T. (1992). Studies of membrane associated and soluble(secreted) lymphotoxin in human lymphokine activated T-killer cells invitro. Lymphokine Cytokine Res. 17, 115-121.Aebersold, R. H., Leavitt, J., Saavedra, R. A., Hood, L. E., and Kent,S. B. H. (1987). Internal amino acid sequence analysis of proteinsseparated by one- or two-dimensional gel electrophoresis after in situprotease di gestion on nitrocellulose. Proc. Natl. Acad. Sci. USA. 84,6970-6974.Ameisen, J. C. (1992). Programmed cell death and AIDS: from hypoth-esis to experiment. Immunol. Today 13, 268-391.Andrew% J. S., Berger, A. E., and Ware, C. F. (1990). Characterizationof the receptor for tumor necrosis factor (TNF) and lymphotoxin (LT)on human T lymphocytes: TNF and LT differ in their receptor bindingproperties and the induction of MHC cl ass I proteins on a human CD4+T cell hybridoma. J. Immunol. 744, 2582-2591.Androlewicz, M. J., Browning, J. L., and Ware, C. F. (1992). Lympho-toxin i s expressed as a heteromeric complex with a distinct 33 kllaglycoprotein on the surface of an activated human T-cell hybridoma.J. Biol. Chem. 267, 2542-2547.Armitage, R. J., Fanslow, W. C., Strockbine, L., Sato, T. A., Clifford,K. N., Macduff, 8. M., Anderson, D. M., Gimpel, S. D., Davis-Smith,T., Maliszewski, C. R., Clark, E. A., Smith, C. A., Grabstein, K. H.,Cosman, D., and Spriggs, M. K. (1992). Molecular and biological char-acterization of a murine ligand for CD40. Nature 357, 60-82.Aruffo, A., and Seed, B. (1987). Molecular cloning of a CD28 cDNA bya high efficiency COS cell expression system. Proc. Natl. Acad. Sci.USA 84, 8573-0577.Badenhoop, K., Schwarz, G., Bingley, P., Trowsdale, J., Usadel, K. H.,Gale, E. A. M., and Bottazzo, G. F. (1990). TNF-a gene polymorphisms:association with type I (insulin dependent) diabetes mellitus. J. Immu-nogenetics 76, 455-460.Beutler, B. (1990). CachectinAumor necrosis factor and lymphotoxin.In Peptide Growth Factors II, M. B. Sporn and A. B. Roberts, eds.(Berlin: Springer-Verlag), pp. 39-70.Browning, J. L., and Ribolini, A. (1989). Studies on the differing effectsof tumor necrosis factor and lymphotoxin on the growth of severalhuman tumor lines. J. Immunol. 743, 1859-1867.Browning, J. L., Androlewicz, M. J., and Ware, C. F. (1991). Lympho-toxin and an associated 33 kDa glycoprotein are expressed on thesurface of an activated human T cell hybridoma. J. Immunol. 747,1230-l 237.Dawkins, R. L., Leaver, A., Cameron, P. U., Martin, E., Kay, P. H.,and Christiansen, F. T. (1989). Some disease associated ancestralhaplotypes carry a polymorphism of TNF. Hum. Immunol. 26, 91-97.Eck, M. J., aad Sprang, S. R. (1989). The structure of tumor necrosisfactor at 2.6 A resolution. J. Biol. Chem. 264, 17595-17605.Eck, M. J., Ultsch, M., Rinderknecht, E., de Vos, A. M., and Sprang,S. R. (1992). The structure of human lymphotoxin (TNF-!3) at 1.9 Aresolution. J. Biol. Chem. 267, 21 19-2122.Farrah, T., and Smith, C. A. (1992). Emerging cytokine family, Nature358, 26.Fiers, W. (1991). Tumor necrosis factor: characterization at the molec-ular, cellular and in vivo level. FEBS Lett. 285, 199-212.Flanagan, J. G., and Leder, P. (1990). The kit ligand: a cell surfacemolecule altered in steel mutant fibroblasts. Cell 63, 185-194.Fugger, L., Morling, N., Ryder, L. P., Platz, P., Georgsen, J.. Jakobsen,

B. K., Svejgaard, A., Dalhoff, K., and Ranek, L. (1989). Ncol restrictionpolymorphism fragment length polymorphism (RFLP) of the tumor ne-crosis factor (TNF-a) region in primary biliary cirrhosis and in healthyDanes. Stand. J. Immunol. 30, 30185-30189.Gardner, S. M., Mock, B. A., Hilgers, J., Huppi, K. E., and Roeder,W. D. (1987). Mouse lymphotoxin and tumor necrosis factor: structuralanalysis of the cloned genes, physical linkage and chromosomal posi-tion. J. Immunol. 139, 476-483.Gubler, U., Chua, A. O., Schoenhaut, D. S., Dwyer, C. M., McComas,W., Motyka, R., Nabavi, N., Wolitzky, A. G., Quinn, P. M., Familletti,P. C., and Gately, M. K. (1991). Coexpression of two distinct genes isrequired to generate secreted bioactive cytotoxic lymphocyte matura-tion factor. Proc. Natl. Acad. Sci. USA 88, 4143-4147.Hempstead, B. L., Martin-Zanca, D., Kaplan, D., Parada, L. F., andChao, M. V. (1991). High affinity NGF bi nding requires co-expressionof the trk proto-oncogene and the low affinity NGF receptor. Nature350, 676-663.Hollenbaugh, D., Grosmaire, L. S., Kullas, C. D., Chalupny, N. J.,Braesch-Anderson, S., Noelle, R. J., Stamenkovic, I., Ledbetter, J. A.,and Aruffo, A. (1992). The human T cell antigen gp39, a member ofthe TNF gene family, IS a ligand for the CD40 receptor: expression ofa soluble form of gp39 with B cell co-stimulatory activity. Eur. J. Immu-nol. 71, 4313-4321.Itoh, N., Yonehara, S., Ishii, A., Yonehara, M., Mizushima, S.-l.,Sameshima, M., Hase, A., Seto, Y., and Nagata, S. (1991). The poly.peptide encoded by the cDNA for human cell surface antigen Fas canmediate apoptosis. Cell 66, 233-243.Jacob, C. O., and McDevitt, H. 0. (1989). Tumor necrosis factor-a inmurine autoimmune lupus nephritis. Nature 337, 356-358.Jacob, C. O., Aiso, S., Michie, S. A., McDevitt, H. O., and Acha-Orbea,H. (1990). Prevention of diabetes in nonobese diabetic mice by tumornecrosis factor (TNF): similarities between TNF and interleukin-1.Proc. Natl. Acad. Sci. USA 87, 968-972.Jevnikar, A. M., Brennan, D. C., Singer, G. G., Heng, J. E., Maslinski,W.. Wuthrich, R. P., Glimcher, L. H., and Kelley, V. E. R. (1991).Stimulated kidney tubular epithelial cells express membrane associ-ated and secreted TNF-a. Kidney Int. 40, 203-211,Jones, E. Y., Stuart, D. I., and Walker, N. P. C. (1989). Structure oftumor necrosis factor. Nature 338, 225-228.Jongeneel, C. V., Acha-Orbea, H., and Blankenstein, T. (1990). A poly-morphic microsatellite in the tumor necrosis factor-a promoter identi-fies an allele uni que to the NZW mouse strain. J. Exp. Med. 171,2141-2146.Kinkhabwala, M., Sehajpal, P., Skolnik, E., Smith, D., Sharma, V. K.,Vlassara, H., Cerami, A., and Suthanthiran, M. (1990). A novel additionto the T cell repertory: cell surface expression of tumor necrosis factor/cachectin by activated normal human T cells. J. Exp. Med. 177, 941-946.Kossodo, S., Grau, G. E., Daneva, T., Pointaire, P., Fossati, L., Ody,C., Zapf, J., Piguet, P.-F., Gauillard, R. C., and Vasseli, P. (1992).Tumor necrosis factor-a is involved in mouse growth and lymphoiddevelopment. J. Exp. Med. 176, 1259-1264.Kozak. M. (1986). Point mutations define a sequence flanking t he AUGinitiator codon that modulates translation by eukaryotic ribosomes.Cell 44, 283-292.Kozak, M. (1991). An analysis of vertebrate mRNA sequences: intima-tions of translational control. J. Cell Biol. 115, 687-903.Kriegler, M., Perez, C., DeFay, K., Albert, I., and Lu, S. D. (1986). Anovel form of TNFlcachectin is a cell surface cytotoxic transmembraneprotein: ramifications for the complex physiology of TNF. Cell 53, 45-53.Mallett, S., and Barclay, A. N. (1991). A new superfamily of cell surfaceproteins related to the nerve growth factor receptor. Immunol. Today12, 220-223.Massague, J. (1990). Transforming growth factor-a, a model for mem-brane-anchored growth factors. J. Biol. Chem. 265, 21393-21396.Matsudaira, P. (1987). Sequence from picomole quantities of proteinselectroblotted onto polyvinylidene difluoride membranes. J. Biol.Chem. 262, 10035-10038.

7/30/2019 Lympho Toxin Beta a Novel Member

10/10

Cl?ll856

Messer, G., Spengler, U., Jung, M. C., Honold, G., Bloemer, K., Pape,G. R., Teirhmueller, G., and Weiss, E. H. (1991). Pol ymorphicstructureof the tumor necrosis factor (TNF) locus: an Ncol polymorphism in thefirst intron of the human TNF-8 gene correlates with a variant aminoacid in position 26 and a reduced level of TNF-8 production. J. Exp.Med. 773, 209-219.Miyake, M., Horiuchi, A., Kimure, K., Abe, Y., Kimura, S., and Hrtsu-moto, Y. (1992). Correlation between killing activity towards the murineL929 cell line and expression of membrane associated lymphotoxin-related molecule of human lymphokine-activated killer cells. Eur. J.Immunol. 22, 2147-2152.Mueller, U., Jongeneel, V., Nedospasov, S. A., Lindahl, K. F., andSteinmetz, M. (1987). Tumor necrosrs factor and lymphotoxin genesmap close to H-2D in the mouse major histocompatibility complex.Nature 325, 265-267.Nedospasov, S. A., Shakhov, A. N., Turetskaya, R. L., Mett, V. A.,Azizov, M. M., Georgiev, G. P., Korobko, V. G., Dobrynin, V. N., Filip-pov, S. A., Bystrov, N. S., Boldyreva, E. F., Chuvpilo, S. A., Chumakov,A. M., Shingarova, L. N., and Ovchinnikov, Y. A. (1986). Tandem ar-rangement of genes encoding for tumor necrosis factor (TNF-a) andlymphotoxin (TNF-8) in the human genome. Cold Spring Harbor Symp.Quant. Biol. 51, 611-624.Nedospasov, S. A., Udalova, I. A., Kuprash, D. V., and Turetskaya,R. L. (1991). DNA sequence polymorphism at the human tumor necro-sis factor (TNF) locus. J. Immunol. 747, 1053-1059.Nedwin, G. E., Naylor, S. L., Sakaguchi, A. Y., Smith, D., Jarrett-Nedwin, J., Pennica, D., Goeddel, D. V., and Gray, P. W. (1985). Hu-man lymphotoxin and tumor necrosis factor genes: structure, homol-ogy and chromosomal localization. Nucl. Acids Res. 73, 6361-6373.Paul, N. L., and Ruddle, N. H. (1988). Lymphotoxin. Annu. Rev. Immu-nol. 6, 407.Perez, C., Albert, I., DeFay, K., Zachariades, N., Gooding, L., andKreigler, M. (1990). A nonsecretable cell surface mutant of tumor ne-crosis factor (TNF) kills by cell-to-cell contact. Cell 63, 251-258.Porciot, F., Molvig, J., Wogensen, L., Worsaae, H., Dalboge, H., Baek,L., and Nerup, J. (1991). A tumor necrosis factor beta gene polymor-phism in relation to monokine secretion and insulin-dependent diabe-tes mellitus. Stand. J. Immunol. 33, 37-49.Raines, E. W., Bowen-Pope, D. F., and Ross, R. (1990). Platelet de-rived growth factor. In Peptide Growth Factors II* M. B. Sporn andA. B. Roberts, eds. (Berlin: Springer-Verlag), pp. 173-262.Ruddle, N. H., and Homer, R. (1988). The role of lymphotoxin in theinflammatory response. Prog. Allergy 40, 162.Ruddle, N. H.. Bergman, C. M., McGrath, K. H., Lingenheld, E. G.,Grunnet, M. L., Padula, S. J., and Clark, R. 8. (1990). An antibody tolymphotoxin and tumor necrosis factor prevents transfer of experimen-tal allergic encephalomyeliti s. J. Exp. Med. 772, 1193-1200.Schall, T. J., Lewis, M., Keller, K. J., Lee, A., Rice, G. C., Wong,G. H. W., Gatanaga, T., Granger, G. A., Lentz, R., Raab, H., Kohr,W. J., and Goeddel. D. V. (1990). Molecular cloning and expression ofa receptor for human tumor necrosis factor. Cell 67, 361-370.Shaw, G., and Kamen, R. (1986). A conserved AU sequence from the3 untranslated region of GM-CSF mRNA mediates selective mRNAdegradation. Cell 46, 659-667.Smith, C. A., Davis, T., Anderson, D., Solam, L., Beckmann, M. P.,Jerzy, R.. Dower, S. K., Cosman, D., and Goodwin, R. G. (1990). Areceptor for tumor necrosis factor defines an unusual family of cellularand viral proteins. Science 248, 1019-1023.Smith, R. A., and Baglioni, C. (1987). The active form of tumor necrosisfactor is a trimer. J. Biol. Chem. 262, 6951-6954.Spies, T., Morton, C. C., Nedospasov, S. A., Fiers, W., Pi ous, D., andStrominger, J. L. (1986). Genes for the tumor necrosis factors a and8 are linked to the human major histocompatibility complex. Proc. Natl.Acad. Sci. USA 83, 8699-8702.Spies, T., Blanck, G., Bresnahan, M., Sands, J., and Strominger, J. L.(1989). A new cluster of genes within the human major hi stocompatabil-ihy complex. Science 243, 214-217.Spriggs, D. R., Imamura. K.. Rodriguez, C., Sariban, E., and Kufe,D. W. (1988). Tumor necrosis factor expression in human epithelial

tumor cell Lines. J. Clin. Invest. 87, 455.Sung, S-S. J., Bjorndahl, J. M., Wang, C. Y., Kao, H. T., and Fu,S. M. (1988). Production of tumor necrosis factor/cachectin by humanT cell lines and peripheral blood T lymphocytes stimulated by phorbolmyristate acetate and anti-CD3 antibody. J. Exp. Med. 767, 937.Tartaglia, L. A., Weber, R. F., Figari, I. S., Reynolds, C., Palladino,M. A., and Goedell, D. V. (1991). The two different receptors for tumornecrosis factor mediate distinct cellular responses. Proc. Natl. Acad.Sci. USA 88, 9292-9296.Tavernier, J., van Ostade, X., Hauquier, G., Prange, T.. Lasters, I.,de Maeyer, M., Lewit-Bentley, A., and Fourme, R. (1989). Conservedresidues of tumor necrosis factor and lymphotoxin constitute theframework of the trimeric structure. FEBS Lett. 257, 315-318.Tsuge, I., Shen, F.-W., Steinmetz, M., and Boyse, E. A. (1987). Agene in the H-2S:H-2D interval of the major histocompatabili tycomplexwhich is transcribed in B cells and macrophages. Immunogenetics 26,378-380.Vale, W., Hsueh, A., Rivier, C., and Yu. J. (1990). The inhibinlactivinfamily of hormones and growth factors. In Peptide Growth Factors II,M. B. Sporn and A. B. Roberts, eds. (Berlin: Springer-Verlag), pp. 21 l-248.Wallner, B. P., Mattaliano. R. J., Hession, C., Cate, R. L., Tizard, R.,Sinclair, L., Foeller, C., Chow, E. P., Browning, J. L., Ramachandran,K. L., and Pepinsky, R. B. (1986). Cloning and expression of lipocortin,a phospholipase A2 inhibitor with potential anti-inflammatory activity.Nature 320, 77-81.Wantanabe-Fukunaga, R., Brannan, C. I., Copeland, N. G., Jenkins,N. A., and Nagata, S. (1992). Lymphoproliferation disorder in miceexplained by defects in Fas antigen that mediates apoptosis. Nature356, 314-317.Ware, C. F., Crowe. P. D.. Grayson, M. H., Androlewicz, M. J., andBrowning, J. L. (1992). Expression of surface lymphotoxin and TNF onactivated T, B and NK cells. J. Immunol. 749, 3881-3888.Webb, G. C., and Chaplin, D. (1990). Genetic variability at the humantumor necrosis factor loci. J. Immunol. 745, 1278-1285.Wingfield. P., Pain, R. H., and Craig, S. (1987). Tumor necrosis factoris a compact trimer. FEBS Lett. 27 7, 179.Wong, S. T., Winchell, L. F., McCune, B. K., Earp, H. S., Teixidd. J.,Massague, J., Herman, B., and Lee, D. C. (1989). The TGF-a precursorexpressed on the cell surface binds to the EGF receptor on adjacentcells, leading to signal transduction. Cell 56, 495-506.GenBank Accession NumbersThe accession numbers for the sequences reported in this paperare L11015 (human LT8 cDNA) and Lll 016 (human genomic EcoRlfragment).