The −308 tumor necrosis factor-α promoter polymorphism effects transcription

?Volcc~lila~ Imnlunolo,y~, Vol. 34. No. 5. pp. 391 39’). 1997 c 1997 Elsevier Science Ltd. All rights reserved Printed in Great Britain

Pergamon

nl6l-sx90 97 517.00+0.00

PII: SO161-5890(97)00052-7

THE

-308 TUMOR POLYMORPHISM KAREN

Department

NECROSIS FACTOR-a PROMOTER EFFECTS TRANSCRIPTION

M. KROEGER, LAWRENCE

of Biochemistry. The University

KYLIE S. CARVILLE J. ABRAHAM* of Western Australia

Australia.

and

Perth, Nedlands.

WA 6907.

AbstractpSince the tumor necrosis factor alpha (TNF-x) gene was found to be located in the central major histocompatibility complex (MHC) there has been much speculation concerning a genetic association between particular TNF alleles and disease susceptibility. A relationship between the MHC haplotype Al, B8, DR3. TNF-r expression levels and susceptibility to autoimmune disease has been suggested by several groups. The identification of the - 308 polymorphism and its association with the HLA Al, BS, DR3 haplotype have led to speculation that the polymorphism may play u role in the altered expression of TNF-c(. We have demonstrated that the region ( ~ 323 to -2X5) encompassing -308 in the TNF2 allele binds nuclear factors differently to the same region in the promoter of the more common TNFI allele. The G/A - 308 polymorphism affected the afinity of factor binding and resulted in a factor binding to TNFZ but not TNFI. The observed differential binding was shown to be functional, with the 38 bp region from TNF2 causing a two-fold greater activity of a heterologous promoter over that due to the same region in TNFI. To further substantiate the functional consequences of the TNF-a -308 polymorphism, we analysed both allelic forms of the TNF-cc promoter region (-993 to + 110) in a transient transfection assay, using luciferase as a reporter gene. The results showed that when present with the 3’UTR the -308A allelic form gave a two-fold greater level of transcription than the -308G form in PMA-stimulated Jurkat and U937 cells. This suggests that the -308 G/A polymorphism may play a role in the altered TNF-r. gene expression observed in individuals with the HLA Al, B8. DR3 haplotype. (’ 1997 Elsevier Science LSd.

K~J,I,~t,o~t/.s:TNF. polymorphism,

transcriptional

INTRODUCTION The location of the tumor necrosis factor alpha (TNF-x) gene in the major histocompatibility complex (MHC) has prompted speculation about the role of TNF in the etiology of MHC-linked diseases, particularly autoimmune diseases. Several groups have demonstrated a correlation between HLA genotype and TNF-r expression levels (Jacob et al., 1990; Abraham et al.. 1993~; Pociot et al.. 1993; Candore et al., 1994) suggesting that regulatory polymorphisms in or near the TNF locus may be responsible for the variations seen. For instance. work we have carried out establishes that B lymphoblastoid cell lines, which are homozygous for the A 1, B8. DR3 haplotype, are able to produce high levels of TNF-c( whereas other MHC haplotypes (e.g. A3, B7. DR15. DQ6) produce lower levels (Abraham et al.. 1993~). Similarly. Pociot et al. (1993) found in a survey of European MHC haplotypes that those carrying DR3 and TNF microsatellite allele a2 (an Al, B8, DR3-specific

*Author to whom correspondence labraham!n

cyllene.uwa.edu.au.

should be addressed.

E-mail:

regulation.

EMSA

marker, see Abraham et al.. 1993h) had the highest induced TNF-x levels in peripheral blood mononuclear cells. The recent identification of a number of biallelic polymorphisms in the promoter of the human TNF-rr gene (D’Alfonso and Richiardi, 1994; Wilson et ul., 1993; Hamann 6’1al., 1995) led to several studies to determine whether an association with various diseases can be found that is independent of other linked HLA markers. Evidence for a potential role for the polymorphism found at nucleotide - 308 (with respect to the transcriptional start site) came from a large controlled study in Gambia where the relative risk of cerebral Malaria is significantly elevated when the less common (TNF2) allele is present in homozygous form (McGuire et ul.. 1994). The TNF2 allele was originally described as an adenine nucleotide A at - 308 (Wilson rt al., 1992). The - 308 polymorphism consists of a guanine (G) residue (TNFI) in most MHC haplotypes and the TNF2 allele was associated with HLA B8 and HLA DR3 and the Al, B8, DR3 haplotype (Wilson et al., 1993). Considering the association of the A 1, B8. DR3 haplotype with a range of regulatory abnormalities, including responses to mitogen and various anti391

392

K. M. KROEGER

gens and also a wide range of diseases including many autoimmune diseases (McCombs and Michalski, 1982; Amer rt ul.,1986; Jacob et al.,1990; Pociot et al., 1993; Wilson et al., 1994) it is possible that the TNF2 polymorphism at - 308 may lead to altered TNF-a expression levels and so contribute to aberrant or inappropriate immune function. Indeed, a recent study has shown that individuals homozygous for the TNF2 allele have higher TNF-r levels than TNFI homozygotes (Bouma ct al., 1996). Therefore, the purpose of this study was to investigate whether the -308 polymorphism was able to effect expression of the TNF-c( gene. We have investigated whether the -308 polymorphism can effect: (1) transcription factor binding, (2) activity as an enhancer in a heterologous system and (3) transcription from the TNFx promoter.

MATERIALS

AND METHODS

Cell lines mu’ culture conditions Human T cell derived Jurkat E6-1 (TIB-152) B lymphoblastoid Raji (CCL 86) hepatoma-derived HepG2 (HB 806.5) epithelial-like HeLa (CCL2) and promonocytic U937 (CRL 1593) cell lines were obtained from the American Type Culture Collection and maintained in culture medium consisting of RPMI- 1640 supplemented with 2mM L-glutamine (Jurkat, Raji, U937) or Eagles MEM with non-essential amino acids, sodium pyruvate and Earles BSS (HepG2 and HeLa) supplemented with 100 keg/ml each of streptomycin and penicillin, and 10% fetal bovine serum. at 37 C with 5% CO?. Oligonucltw ticks

Sequences representing allelic forms dation assays

of the double-stranded oligonucleotides, -323 to -285 of the TNFl and TNF2 of the TNF-a promoter, used in gel retarwere as follows:

TNF I, (SGTTTTCAGGGGCATGGGGACGGGGTTCAGCCTCCAGGGT3’); TNF2, (SGTTTTCAGGGGCATGAGGACGGGGTTCAGCCTCCAGGGT3’); TNNF7.1, (SGCCCAGAAGACCCCCCTCGGAATCGGAGCAGGGAGGATGG3’). The TNFI and TNF2 oligonucleotides had a single 5’ G overhang to facilitate end-labelling with “P[dCTP].

Ekctrophoretic

niohilit~~ .sl?$ assq* (EMSA)

Nuclear extracts were prepared according to the method of Li et al. (1991). Approximately 8 x IO’ cells were incubated with or without PMA (20 ng/ml) for 2 hr in medium before extract preparation. Extracts were frozen in liquid N, and stored at -80 C. The protein concentrations of extract preparations were determined using the Bio-Rad protein assay kit. For EMSA. nuclear proteins (l2-2Opg) were preincubated for 10 min on ice

c’t N/

with Ipg of poly(dldC) (Pharmacia, Australia) in a binding buffer (4% Ficoll, 20 mM HEPES [pH 7.91, I mM EDTA. 1 mM DTT, 50 mM KCl) to give a final reaction volume of 20 htl. For competition assays unlabelled competitor oligonucleotide was incubated with nuclear extract for IOmin on ice prior to the addition of “Plabelled oligonucleotide. Nuclear proteins were then incubated with “P-labelled oligonucleotide (80 fmol) for 30min on ice then loaded onto a 6% polyacrylamide gel containing either 0.25 x Tris-borate,iEDTA (TBE) OI 0.25 x Tristaurine/EDTA, and electrophoresed at 15OV. Gels were dried, and exposed to X-ray tilm a~ - 80 C using intensifying screens.

Oligonucleotides TNFI and TNF2 were cloned into the luciferase reporter gene construct pGL3-Promoter (Promega Inc., Australia). The techniques for plasmid construction were performed as described (Sambrook et (I/., 1989). TNFI and TNF2 were end-filled and bluntend ligated into the Snlal site of pGL33Promoter, which contains the SV40 promoter upstream of the luciferase coding region and the SV40 3’ untranslated region. To determine both the number and orientation of inserts. plasmids were sequenced by the dideoxy chain terminator method. and analysed on an ABI 373 DNA Sequencer. Both allelic forms of the TNF-r promoter (-993 to + 110) were cloned into the luciferase reporter gene construct pGL2 Basic (Promega Inc.). The TNF-z promoter fragment (-993 to + 110) with a G at -308 was amplified from genomic DNA isolated from the B lymphoblastoid cell line (10th International Histocompatibility Workshop #908X, Abraham c/ 1993~) using crl., the PCR primers (forward, S’GTCAGGGAGCTCCTGGGAGATATG3’; reverse, S’GCCTGGAAGCTTGTCAGGGGATGTG3’). The resulting SucI Hi&II fragment wascloned into the Suc,I and HirrrlIlI restriction sites located upstream of the luciferase gene in pGL22Control to create the TNFY- 7”x”,‘Luc reporter gene construct. The TNF-2 ~? Lucl’3’UTR plasmid was constructed by replacing the SV40 3’ untranslated region (UTR) with the 3’UTR (+ 1957 to +2792) of TNF-x, following digestion at the fflM1 and BcrlrrHI restriction sites located in TNF,~ ~31)X(i,LUC, The

TNF_1

j”X’j/Luc

and TNF-s(- ““‘\,‘Luc/3’UTR plasmids were constructed by site-directed mutagenesis of the TNF-a-“‘K”/Luc and TNF-x “‘““:‘Luc:3’UTR constructs, respectively. Site-directed mutagenesis was performed to change the G at position -308 to an A. using the Transformer Site-directed Mutagenesis Kit (Clontech. U.S.A.). Two independently derived mutant constructs were isolated for each construct. To confirm the identify of the nucleotide at the -308 position. constructs were sequenced. Trmsfi~ctiom

cd

luc~fC~rrr.vc us.w~~.s

Jurkat and U937 cells were transfected phase of their growth by electroporation

during the log with plasmid

Effects of the - 308 TNF-x polymorphism DNA prepared and purified using the QIAGEN Maxiprep-500 Kit (QIAGEN, Australia). Following incubation with 15 pg of construct DNA plus 1Opg pCAT_ Control vector DNA (Promega Inc.) (to control for transfection efficiency), cells were electroporated using a BioRad gene pulser (370 V and 960 /lF for Jurkat or 240 V and 960 /IF for U937) and then distributed into two wells each containing 5ml of medium and incubated either unstimulated for 24 hr or induced for 24 hr with phorbol 12-myristate 13-acetate (PMA; 20 ng,‘ml). Cell lysates of transfected cells were then prepared and assayed using kits for luciferase and chloramphenicol acetyltransferase (CAT) as described by the manufacturer (Promega Inc.). All transfections were performed at least in triplicate and at least two independent preparations of plasmid DNA were used. and where indicated two individually isolated clones were used. Luciferase activity was normalised against CAT activity. Data were subjected to analysis of variance (ANOVA) with a correction for multiple comparisons. with 11 < 0.05 considered significant.

RESULTS

To confirm that the region surrounding -308 of the TNF-x promoter was able to bind transcription factors (Kroeger and Abraham. 1996) and to determine whether the G:A polymorphism effected binding, EMSAs were performed using double-stranded 38 bp oligonucleotides. TNFI and TNF2, that represented -323 to -285 of the G and A allelic forms, respectively. The EMSA results using Jurkat nuclear extracts indicated that both TNFl and TNF2 bound multiple proteins and shared three complexes (B. C and D: Fig. 1A, lanes 3 and 4). Complexes B, C’. D and an additional complex E were found to be specilic as a 500-fold molar excess of a nonspecific oligonucleotide (TNN F7.1). encompassing - 262 to -223 of the TNF-r promoter, failed to compete fol binding (data not shown). With respect to differential binding due to the -308 polymorphism, comparison of the EMSA complexes formed using oligonucleotides TNFl and TNF2 indicated that complex (E) was interacting uniquely with the TNF2 sequence (Fig. IA. lane 3). In addition, the proteins associated with the shared complexes (B, C and D) displayed different binding activities to TNFI compared with TNF? (Fig. I, C and D). Competition assays were performed using unlabelled TNFI and TNF2 oligonucleotides as the competitors. Unlabelled TNFl completely competed away proteins binding to labelled TNFl in complexes B and C at IOO-fold molar excess, and complex D at SOO-fold excess. and allowed binding of complex E (Fig. IA. lane 5). This suggests that complex E can bind TNFI. to a smaller extent and with low affinity. and increases when complexes B, C and D are removed (Fig. IC). Unlabelled TNF2 completely competed for protein complexes B. C and D binding to TN F I. when IOO-. 250- and 7SO-fold excess was used (Fig. IC).

on transcription

393

A complementary series of competitions using labelled TNF2 and unlabelled TNFl and TNF2 confirmed the magnitude of the binding differentials (Fig. ID). The competition assays indicated that proteins involved in the formation of protein complexes B. C and D were capable of binding to both TNFl and TNF2. and proteins in complex E were observed preferentially binding to TN F2. However. proteins in complexes C and D showed a greater ability to bind to TNFl than to TNF2. This may be due to competition of complexes C and D with complex E proteins for binding to overlapping binding sites on TNF2. The binding of complex E to TNF2. in cwnbination with the differential binding acti\,ities ofproteins in complexes C and D. may have an efl‘ect on the transcriptional regulation of TNF-a.

As TN F-x is also expressed in many different ccl1 types including B cells. hepatocytes and macrophages, we determined whether the same Jurkat-derived activities binding to TNFl and TNF2 were prcscnt in nuclear extracts from Raji, HepG2. HeLa and U937 cell lines. Comparison of EMSA complexes indicated that similar complexes were able to form for each cell type (Fig. 2). However. compared to Jurkat extracts (lanes 3 and 4). quantitative differences were seen. Complex Dl appears to be particularly prominent in the case of HepG2 and U937 extracts (Fig. 2. lanes 7. X, I I and 12). With respect to complex E. all cell types appeared to possess the binding activity and the results confirmed that rhc TNF? sequence had a greater attinity for complex E than the TNFI allelic sequence. The results suggested that all of the factors binding to the -30X region arc ubiquitous.

To substantiate the functional consequences of the polymorphism at position -30X of the TN F-x promoter. the TNFI and TNF2 oligonucleotidc sequences were cloned upstream of the SV40 promoter in the luciferase reporter vector pGL3-Promoter as either ;I single copy (pTNFI.1 and pTNF2.1) or a double copy (pTNFl.2 and pTNF2.2. see Fig. 3A). The resulting constructs were transfected into Jurkat and U937 cells and luciferase expression monitored in the presence or absence of PMA. The TNFI sequence, when present in single copy (in pTNFl. 1). caused a three-fold repression of the basal SV40 promoter activity in Jurkat cells (Fig. 3B). However, when two copies of the element Mere present in tandem (pTNFI .2). the level of repression decreased (to 1.4-fold) when compared to the parent plasmid. By contrast. in U937 cells a single copy of TNFI caused no significant change in SV40 promoter activity but an enhancement of transcription ( I .75-fold) was observed when two copies of the TNF I element ivere present (Fig. 3C). On the other hand, in Jurkat cells, the TNF? element when present in single copy (pTNF7. I ) caused no change in the activity due to the SV40 sequences alone, but

K. M. KROEGER

394

A


B

Oligo

TNF2 -

TNFl --

Competitor

TNF2

TNFl TNF2

-- TNFl

12345678

1

-I-,***

1)

TNF2

23 4 I*#rm@IL,

Fig. 1. EMSA of Jurkat nuclear extracts using TNFI and TNF2 ohgonucleotides representing the -323 to -285 region of TNF-c(. (A) Labelled double-stranded oligonucleotides TNF2 (lane I) and TNFI (lane 2) were incubated with 12 pg Jurkat nuclear extracts (TNFZ, lane 3; TNFI, lane 4) and analysed on a 6% polyacrylamide/0.25 x TBE gel. Major complexes are indicated with arrows (B, C, D and E). Competitions against labelled TNFI carried out using IOO- and 250-fold molar excess of unlabelled TNFl (lanes 5 and 6) and TNF2 (lanes 7 and 8) are shown. (B) Competitions against labelled TNF2 carried out using IOO- and 250-fold molar excess of unlabelled TNFI (lanes 1 and 2) and TNF2 (lanes 3 and 4). (C) The apparent binding affinities of complexes B, C. D and E for TNFI were determined, using IOO-, 250-, 500-, 750- and IOOO-fold excess competitor. from the results shown in panel A (500- to 1OOO-foldexcess not shown) and are expressed in terms of the amount of unlabelled competitor (TNFI or TNF2) required to completely compete out binding of complexes to labelled TNFI. (D) The apparent binding affinities of complexes B, C, D and E for TNF2 were determined. using IOO-, 250-. 500-, 750- and IOOO-fold excess competitor, from the results shown in panel B (and data not shown) and are expressed in terms of the amount of unlabelled competitor (TNFI or TNF2) required to completely compete out binding of complexes to labelled TNF?.

when present as two tandem copies (pTNF2.2) a 1.5-fold greater level of expression was seen when compared to the SV40 promoter alone (Fig. 3B). Similar results were seen in U937 cells, except a single copy of TNF2 served to increase I .4-fold the activity of the heterologous promoter (Fig. 3C). Following PMA stimulation for 24 hr, the levels of transcription achieved by all constructs were increased in both Jurkat (1.25-1.5 fold, Fig. 3B) and U937 (224 fold, Fig. 3C). Overall, although the parental SV40 reporter vector was PMA responsive (1.4-fold for Jurkat and 1.3-fold for U937), all four TNF constructs showed an additional responsiveness following stimulation. Although there is no significant difference between -fold induction of the equivalent TNFI and TNF2 constructs, for each construct the difference between uninduced and induced activity is significant @ < 0.01). The data indicate that the region surrounding -308 in TNFI and TNF2 confers a significant,

additional PMA-responsiveness to the heterologous promoter in both Jurkat and U937 cells. Significantly, a differential effect of the -308 polymorphism was seen when comparing the two -308 alleles. Comparison of the TNFl constructs with equivalent TNF2 constructs indicated that the presence of an adenine nucleotide at - 308 generally increased transcription at least two-fold above the levels seen when a G was present at this position. For instance, in unstimulated Jurkat cells, when a single copy of TNF2 was present (in pTNF2.1) the level of transcription from the SV40 promoter was 2.7-fold @ < 0.001) higher than for the TNFl sequence (in pTNF1. I, Fig. 3B). Following PMA stimulation the difference was increased to 3. I-fold (p < 0.001). Similar results were obtained in U937 cells but the differences were smaller (I ,2552.0-fold, p < 0.05) as shown (Fig. 3C). These results demonstrate that the - 308G/A polymorphism is sufficient to differentially

Effects of the -308 Cell type Oligonucleotide TNFl TNF2 Complex

TNF-z

polymorphism

Jurkat -----+-+-+-+-+-+ + -+ 1 234

Raji

+ 5

395

on transcription

HepG2

HeLa

u937

9

11

-+-+-+67

8

10

12

B, C, D, Dl,

Fig. 2. Similar binding activities are found in a variety of cell types. Labelled oligonucleotide probes TNF2 (lane I, unbound) and TNFI (lane 2, unbound) were incubated with nuclear proteins (20 l(g) prepared from Jurkat (lanes 3 and 4) Raji (lanes 5 and 6). HepG2 (lanes 7 and 8) HeLa (lanes 9 and 0.25 x Tris-taurine:‘EDTA gel. 10) or U937 cells (lanes II and 12) and run on a 6% polyacrylamide,

alter the transcriptional activity of the -308 element least in the context of a heterologous promoter. The - 308 pol~morphim TNF-x promoter

influence.s transcription

at

qf’ tkr

To assess whether the - 308 region functioned as a cisacting element in the context of the TNF-c( promoter and also to determine whether the - 308 polymorphism effected transcription, we constructed two plasmids containing both - 308 polymorphic forms of the TNF-a promoter region (-993 to + 110) fused to the luciferase reporter gene (see Fig. 4A). The constructs contained either a G (TNF-c~“~“/Luc) or an A (TNF-cc~3”8”/L~~) at position -308. Upon transfection into either Jurkat or U937 cells, no significant difference in activity was observed between the two allelic -308 constructs (Fig. 4A). Likewise, no differences were observed following PMA treatment although approximately two-fold (Jurkat) or 40-fold (U937) induction was seen; both constructs gave a similar level of expression in both cell types (Fig. 4A). indicating that the G/A polymorphism was not effecting transcription in these constructs. In view of the functional importance of the 3’UTR of TNF-(x, two further constructs were made by replacing the SV40 3’UTR contained in the pGL2-Basic luciferase vector with the TNF-x 3’UTR (+ 1957 to +2792) to generate constructs TNF-c”“~“/Luc/~‘UTR and TNFCY-“‘~*,ILUC~~‘UTR (see Fig. 4B). Jurkat or U937 cells were transiently transfected with both allelic forms of the TNF-x promoter/luciferase constructs and stimulated with PMA (20 ngjml) for 24 hr or left unstimulated. When either Jurkat or U937 cells were transfected with the TNF-c “‘X”/Luc/3’UTR construct the relative transcriptional activity was not significantly different from

ceils transfected with TNF-cc 3’1”A,‘Luc,‘3’UTR (Fig. 4B). However. significant differences were observed following 6, 12, 18 or 24 hr of PMA treatment. with both induction levels and differences being maximal at 24 hr (data not shown). Following 24 hr of PMA treatment, the TNFx ~‘“X”:‘Luc~3’UTR construct gave a I.7-fold (I, < 0.05) or a 2.1-fold (p < 0.01) greater level of expression than the TNF-Y~“~“,!Luc/~‘UTR construct (Fig. 4B) in Jurkat and U937 cells, respectively. The response of the TNF promoter constructs to lipopolysaccharide (LPS) was also tested in Jurkat and U937 cells (data not shown) and similar results were obtained to those generated using PMA stimulation, with the TNF2 construct showing a two-fold higher activity than the TNFl construct in the presence of the 3’UTR. Therefore. the polymorphism at -308 has a signiticant influence on transcription from the TNF-cc promoter but only when the TNF-a 3’UTR is present. The EMSA results suggest that a TNF2-specific nuclear factor (complex E) may be responsible for the increased activity observed when an A is located at - 308 in the TNF-x promoter.

DISCUSSION In this study we have shown that the polymorphism at -308 is able to differentially effect transcription of the TNF-r gene. The A nucleotide-containing allele (TNF2) may be responsible for a general increase in transcriptional activity compared to the -308G allele (TNFl). The synthesis of TNF-c( is known to be regulated, in part, at the transcriptional level (Beutler, 1992). The TNF-a promotor has been well characterised and contains a number of important regulatory elements that effect TNF-a transcription in response to various stimuli

396

K. M. KROEGER

et ul.

A pTNF1 .l TNFl b

TNFI -+

pTNF1.2

~wl&&ase~

pTNF2.1

TNFZ

TNFZ -4

pTNF2.2

*

C

B Jurkat

u937

d

2.5

s 0

17

2 8 ‘B Y z 6 SJ ‘m 3 a p!

b=

?? PMA24h

:

PMA24h

c

b T

1.5

ll3.d -I a

1

0.5

Uninduced

Uninduced

C

a

d T

b

0

N.

E

5

Fig. 3. Effect of the region containing the -308 polymorphism on a heterologous promoter. (A) Schematic diagram of luciferase reporter constructs used in transfections. The filled box represents the SV40 promoter, the grey box the luciferase gene and the boxes with the arrows represent the elements TNFl or TNF2 (- 323 to -285), with the direction of the arrows showing the orientation of the elements. (B) Relative expression of luciferase reporter constructs pTNFl_ I. pTNFl.2, pTNF2.1. pTNF2.2 following transfection into Jurkat cells, Following incubation in the presence or absence of PMA for 24 hr, transfected cells were assayed for luciferase activity, the results normalised with respect to transfection efficiency. The bars represent the mean (k SEM) of three independent experiments performed in duplicate, expressed relative to expression due to the SV40 promoter vector alone. Comparisons between equivalent TNFl and TNF2 values were made (indicated by the lower case letters above the bars) and the significance of the differences between TNFl and TNF2 driven expression are: a, p < 0.001 (ANOVA); b, p < 0.001; c, p < 0.001; d, p < 0.001. (C) Relative expression of luciferase reporter constructs pTNF I. I, pTNFl.2, pTNF2. I. pTNF2.2 following transfection into U937 cells. Data were treated as for (B) and represent the mean (+ SEM) of three independent experiments performed in duplicate. The levels of significance of the difference between TNFl and TNF2 driven expression are: a, p < 0.05; b, p < 0.05: c. /7 < 0.05; d, p < 0.001.

et al., 1989; Hensel rt al., 1989; Goldfield rt et al., 1992; Kramer rt al., 1994; Jongeneel. 1995). Some evidence for the role of the region encompassing -308 in the transcription of TNF-a has been provided by several researchers. Previously, Economou et al. (1989) have established by deletion analysis that the TNF-c( promoter region between -479 and -295 contains sequences that modulate transcription of a TNF/luciferase reporter gene. In addition, Fong et al. (1994) demonstrated that the region between - 351 and -280 makes a modest contribution to PMA-induced transcription of the TNF-a gene and we have shown previously that the - 323 to -285 region acts as an enhancer (Kroeger and Abraham. 1996). (Economou

al., 1991, 1993; Rhoades

In the current study, we have used three approaches to show that a single nucleotide difference at -308 of the TNF-x promoter was capable of effecting expression of the gene. Transcription factor binding studies identified multiple protein complexes binding to both the common TNFl allele and the less common TNF2 allele in the region surrounding - 308. However, the TNF l/TN F2 polymorphism appeared to affect the ability of nuclear factors to bind to the - 308 region. Similar protein complexes were observed binding to the -308 region when nuclear extracts from various cell types were compared. This suggests that the region surrounding the - 308 polymorphism plays a role in the regulation of TNF-cx in many different cell types. with similar trans-activating

Effects of the - 308 TNF-r

polymorphism

+llO

-993

TNF-cx -~~=‘A/Luc -308GIA

397

on transcription

TNF-(r

TNF-o: -308G’A/Luc/3’UTF

Luciferase

-308 G/A

ug37

u937

Jurkat 7

-308 G

a

-308 A

1

-308 A

T I

0

-308 G

? ?-308 ??-308

A

I ; I

A*

Fig. 4. The - 308 polymorphism influences transcription of the TNF-x promoter in Jurkat and U937 cells. (A) Schematic representation of the TNF-rm”‘XG/Luc and TNF-amm”‘XA:‘Lucconstructs and relative activities following transfection into Jurkat and U937 cells. Cells were either untreated or PMA treated for 24 hr as indicated and luciferase activity was measured. normalised for transfection efliciency, and expressed relative to the highest value for that cell type in panel B. Shown for both Jurkat and U937 transfections are the means ( f SEM) of four separate transfections, each performed in duplicate, using three different batches of DNA. The -308A” values represent the means obtained from one transfection experiment performed in duplicate of a second independently generated clone from the mutagenesis of the TNF-xmioX”,I Luc construct. There is no significant difference between the transcriptional activity of the -308G and ~ 308A constructs. (B) Schematic representation of the 30X4 TNF-g- “?:Luc:.?‘UTR and TNF-r /Luq3’UTR constructs and relative activities following transfection into Jurkat and U937 cells. Transcriptional activity is expressed relative to the highest each value for that cell type. The bars represent means ( f SEM) of four separate transfections. performed in duplicate, using three different batches of DNA. The -3OXA* values represent the tneans obtained from one transfection experiment performed in duplicate of a second independently generated clone from the mutagenesis of the TNF-rm’“XG/ Luc;.?‘UTR construct. There is a signiticant ‘“““/Luc/_?‘UTR and TNF-x “lh’, Luc, difference between the transcriptional activity of the TNF-r ?‘IJTR constructs following 24 hr PMA stimulation in both Jurkat (ANOVA. 11< 0.05) and 1J937 cells (/I < 0.01).

factors being involved. However, the transfection data indicate that additional factors, involved in the PMA response. are also required to produce the differential effect seen when comparing the two allelic forms of the -308 polymorphism in the context of the TNF-c( promoter and 3’UTR. Taken together, the results may indicate that a differential effect will be seen only in some cell types, including macrophages and T cells. The presence of a factor (complex E) binding strongly to the promoter region of the TNF2 allele, but weakly to the TNFl allele, may have an important differential effect on the transcriptional regulation of TNF-cr and may contribute to higher levels of TNF-a expression in individuals carrying this allele as part of the HLA Al, B8, DR3 haplotype (Abraham et al., 19930; Pociot et ul., 1993). Current

studies are aimed at the purification and identification of this protein/s. Two different approaches using reporter gene assays were also used to determine the significance of the - 308 polymorphism. We have shown that the region surrounding - 308 is capable of modulating expression of both the TNF-c( promoter and the heterologous SV40 promoter, confirming that this region may play a significant role in regulation. In the case of the SV40 promoter. the - 308 region in the TNF2 allele enhanced SV40 promoter activity two-fold above the equivalent region from the TNFl allele. This may be due to the presence of the factor(s) binding uniquely to TNF2 as shown in the EMSA experiments. The use of allelic TNFSI promoter reporter gene constructs provided further

398

K. M. KROEGER

evidence for the role of the - 308 polymorphism in TNFx gene transcription. The TNF2 allele when present in the context of the TNF-LY promoter and the TNF-x 3’UTR was shown to be responsible for higher inducible levels of transcription than the TNFI allele. This strongly suggests that the -308 polymorphism in the context of the TNF-c( gene can effect transcription, and may be responsible for elevated TNF-r expression levels observed in individuals carrying the TNF2 allele (Abraham et al., 1993a; Pociot et al., 1993; Candore et al., 1994). Interestingly, the differential effect was seen only in the presence of the TNF-(x 3’UTR. It is possible that the previously described interaction between the promoter and the 3’UTR of the TNF-z gene (Han et ul., 1991) may result from transcriptional regulatory elements being present in the 3’UTR. An intriguing question concerns the effect of the - 308 region when linked to the SV40 minimal promoter compared to that seen when the effects of the -308 region are assessed in the context of the TNF-a promoter itself. When present upstream of the SV40 promoter the two alleles of the -308 region act to differentially modulate transcription but the polymorphism does not diff‘erentially effect PMA responsiveness. In contrast, the two allelic forms differentially effect transcription from the TNF promoter only following PMA stimulation and only in the presence of the TNF 3’UTR. This paradox probably reflects the complex nature of the interaction between the different elements of the transcriptional machinery. However, the results do indicate that the - 308GiA polymorphism has the ability to differentially influence transcriptional activation in a number of different transcriptional contexts such as may occur in different tissue types. Similar reporter gene studies using -308 allelic TNF-x promoter/CAT constructs have recently been performed by two other groups. Stuber e’t al. (1996) found no significant difference in the transcriptional activity of the TNFl and TNF2 allele promoters in PMA-stimulated Jurkat T cells and an LPS-induced macrophage line. Our results also show no significant difference in reporter gene expression when either the TNFl or TNF2 TNF-a promoter alone is present. However, in the study by Stuber et a/. (1996) the activity of the two allelic forms of the TNF-s( promoter was not studied in the context of the 3’UTR. In contrast to our results, Brinkman et a/. (1996) found no significant difference in the transcriptional activity of the TNFl and TNF2 promoter region in the context of the TNF-x 3’UTR in Jurkat cells. although the TNF2 construct (TNFp”‘X”3’/CAT) was invariably higher, although not significantly so. We do not know why our results differ, but possible explanations are that our Jurkat cell line E6-1 may behave differently and the differences seen, when comparing the two alleles in our system, reach significance. However, it should be noted that the constructs differ; our constructs include an additional 374 bp of upstream promoter sequence but lack the recently described 3’ distal NF,B site (Kuprash et cd., 1995) included in the constructs of Brinkman et al. (I 996). The presence of this strong 3’ regulatory site may

et rrl

mask the allelic differences we have observed. Alternatively, additional regulatory elements located upstream of -619 (the extent of the constructs of Brinkman et al. (1996)) may interact, via nuclear factors, with the allelic forms of the -308 element and differentially activate transcription. In summary, we have provided evidence that the - 308 polymorphism does effect transcription of the TNF-r gene in vitro. The association between the -308G,1A TNF-‘2 promoter polymorphism and susceptibility to various diseases (Abraham et al., 1993a; McGuire et al., 1994; Bouma et a/.. 1996) is suggestive for the polymorphism being functionally relevant in vice. There is evidence suggesting that TNF-cr levels may play an important role in determining the susceptibility of individuals to autoimmune diseases such as insulin-dependent diabetes mellitus (IDDM) and systemic lupus erythematosus (SLE; Pociot et al., 1993; Jacob et al.. 1990; Wilson et al., 1994). The frequency of the TNF2 allele was shown to be higher in both IDDM and SLE patients than controls, with IDDM patients having a higher TNF-z secretory capacity compared with controls (Pociot rt a/., 1993; Wilson et al., 1994). It is possible that the increased production of TNF-x could contribute to the increased incidence of autoimmune diseases observed in individuals with an HLA Al, B8, DR3 haplotype. Studies in mouse support this contention. Recent evidence indicates that TNF-a expression levels can change the course of an immune response (Muller et ul.. 1994) which in turn may be attributable to polymorphisms in the mouse TNF-a gene. Work by McDevitt’s group using nonobese diabetic (NOD) mice indicates that administration of TNF-rr to newborn animals results in earlier onset and dramatically increased incidence of diabetes (Yang et ~1.. 1994). Presumably, humans carrying the TNF2 allele and producing higher basal and induced levels of TNF-x may be subject to a similar increased susceptibility to IDDM. Our results concerning the functional relevance of the -308 TNF-c( promoter polymorphism provide a possible explanation for the generalised immune dysfunction ascribed to the A I. BX. DR3 haplotype. Ac,lirzollk~~y!yrmerlt-This work was supported Health and Medical

Research

Council,

by the National Australia.

REFERENCES Abraham L. J.. French M. A. H. and Dawkins R. L. (1993~) Polymorphic MHC ancestral haplotypes affect the activity of tumor necrosis factor alpha. Clin. e-up. Immun. 92, 14~18. Abraham L. J.. Marley J. V., Nedospasov S. A., CambonThomsen A.. Crouau-Roy B., Dawkins R. L. and Giphart M. (19936) Microsatellite, RFLP and SSO typing of the TNF region. Hun~cm Inmun. 38, 17-23. Amer A.. Singh G., Darke C. and Dolby A. E. (1986) Impaired lymphocyte responsiveness to phytohaemagglutinin associated with the possession of HLA-B8/DR3. T~SSUCAntigrru 28, 193-198. Beutler B. A. (I 992) Application of transcriptional and posttranscriptional reporter constructs to the analysis of tumor

Effects of the - 308 TNF-x

polymorphism

necrosis factor gene regulation. il1?1. .I. Mrd. Sci. 303, l29133. Bouma G.. C’rusius J. B. A.. Oudkerk Pool M.. Kolkman J. J.. Von Blomberg B. M. E.. Kostense P. J., Giphart M. J., Schreuder G. M.. Meuwissen S. G. M. and Pena A. S. (1996) Secretion of tumour necrosis factor alpha and lymphotoxin alpha in relation to polymorphisms in the TNF genes and HLA-DR nlleles~-relevance for inflammatory bowel disease. Surtd. .I. Imtttutt. 43, 45&463. Brinkman B. M. N.. Zuijdgeest D.. Kaijzel E. L., Breedveld F. C. and Verwcij C. L. ( 1996) Relevance of the tumor necrosis factor alpha (TNF- ~0 -30X promoter polymorphism in TNF-r. gene regulation. J. It~futtttnutiott 46, 3241. Candore C.. Cigna D.. Gervasi F.. Colucci A. T.. Modica M. A. and Caruso C. (1994) frl c.ilro cytokine production by HLA-BX. DR3 positive subjects. Au/obmzutzity 18, 121-l 32. D’Alfonao S. D. and Richiardi P. M. (1994) A polymorphic variation in a putative regulation box of the TNFA promoter. /tlJttllttl(~,~f,tt~,/i~.S 39, 150 154. Economou J. S., Rhoades K., Essner R., McBride W. H., Gasson J. C. and Morton D. L. (1989) Genetic analysis of the human tumor necrosis factor x;cachectin promoter region in a macrophagr cell line. ./. (‘.q,. Mctl. 170, 321-326. Fong C. W.. Siddiqui A. H. and Mark D. F. (1994) Identification and characterisation of a novel repressor site in the human tumor necrosis factor x gene. ,Y&. Acid Res. 22, I108~-1111. Goldfield A. E.. McCaffrey P. G.. Strominger J. L. and Rao A. ( 1993) Identification of a novel cyclosporin-sensitive element in the human tumor necrosis factor x gene promoter. J. c’_vp. :l/lrt/. 178, 1365-1379. Goldfield A. E.. Strominger J. L. and Doyle C. (1991) Human tumor necrosis factor r gene regulation in phorbol ester stimulated T and B cell lines. J. ‘.x-p. Mctl. 174, 73-81. Hamann A.. Mantzoros C.. Vidal-Puig A. and Flier J. S. (1995) Genetic v,lri;tbility in the TNF-alpha promoter is not associated \\ith type II diabetes mellitus (NIDDM). Bioclwttt. RiophLY.Rcs. C’otwttw?.211, X33-839. Han J.. Hue/ G. and Beutler B. (1991) Interactive effects of the tumor- necrosis factor promoter and 3’-untranslated regions. .1. Ittrttturt. 146, 1843~~1848. Hensel G.. Meichle A.. Pfizenmaier K. and Kronke M. (1989) PMA-responsive 5’ flanking sequences of the human TNF gene. I,l,ttt/,/toXittc, Rcs. 8, 347 35 I Jacob C. 0.. Fronek Z., Lewis G. D.. Koo M.. Hansen J. A. and McDevitt H. 0. (1990) Heritable major histocompatibilit) complex class II-associated differences in production of tumor necrosis factor 7: relevance to genetic predisposition to systemic lupus crythematosus. Proc,. twtt7. Ad. SC;. L’..s.il. 87, 1233_~1237. Jongeneel C‘. V. ( 1995) Transcriptional regulation of the turn01 necrosis t actor alpha gene. Ittttttuttohioloy?, 193, 2 I O-2 16. Kramer B.. Meichle A.. Hensel G.. Charnay P. and Kronke M. (1994) C‘haractcrisation of a Krox-24!Egr-l-responsive element 111 the human tumor necrosis factor promoter. Bioc,hi/>~. Bio/+rs. .lcfl/ 1219, 4 I3 42 I Kroegcr K. M. and Abraham L. J. (1996) Identification of an AP-2 clement in the x 313 to x 285 region of the TNF- 2 gene. Hioc /~(,tii. titok. Bid. Ittt. 40, 43-5 I

on transcription

399

Kuprash D. V., Udalova I. A., Turetskaya R. L., Rice N. R. and Nedospasov S. A. (1995) Conserved kappa-B element located downstream of the tumor necrosis factor alpha gene--distinct NF,>-B binding pattern and enhancer activity in LPS activated murine macrophages. Otwt,qcw 11, 97- 106. Li Y.. Ross J.. Scheppler J. A. and Franza B. R. (1991) An itt rittw transcriptional analysis of early responses of the human immunodeficiency virus type I long terminal repeat to different transcriptional activators. Mol. C‘c,/l.Bio/. II, I XX3 -1893. McCombs C. C. and Michalski J. P. (1982) Lymphocyte abnormality associated with HLA-BX in healthy young adults. ./. c’s/). Mrd. 156, 936 94 1. McGuire W.. Hill A. V. S., Allsopp C. E. M.. Greenwood B. M. and Kwjatkouski D. (1994) Variation in the TNF- r promoter region association with susceptibility to cerebral malaria. ,Yaflirr 371, 508 5 I I. Muller K. M.. Lisby S.. Arrighi J. F.. Grau G. E.. Saurut J. H. and Hauser C. (I 994) H-1D haplotype-linked expression and involvement of TNF- I in Th:! cell-mediated tissue inflammation. ./. Itwmttt. 153, 3 16-314. Pociot F.. Briant L., Jongeneel c‘. V., Molwg J.. Worsaae H.. Abbal M.. Thomsen M.. Neurp J. and Cambon-Thomsen A. (1993) Association of tumor necrosis factor (TNF) and class 11 major histocompatibility complex alleles with the secretion of TNF-x and TNF-/I by human mononuclear cells: a possible link to insulin-dependent diabetes mellitus. I&r-. ./. htinrtt. 23, 224 23 I. Rhoades K. L.. Golub S. H. and Economou J. S. (IY92) The regulation of the human tumor necrosis factor CI.promoter region in macrophage. T cell. and B cell line\. ./. hiol. C‘ht. 267, 22 101-22 IO7. Sambrook J.. Fritsch E. F. and Maniait\ T. (IYXY) ~~IoIccu/~tt c’/otzity: .4 Laho~rrtoq~ Mmud. 2nd Edn. Cold Spring Harbor Laboratory Press. U.S.A. Stuber F.. Udalova I. A.. Book M.. Drutskaqa L. N.. Kuprash D. V.. Turetskaya R. L.. Schade F. U. and Nedospasov S. A. ( 1996) - 308 Tumor necrosis factor (TNF) polymorphism is not associated with survival in severe sepsis and is unrelated to lipopolysaccharide inducibility of the human TNF promoter. J. fttflrtttrtturfitrtt 46, 42 SO. Wilson A. G.. di Giovine F. S.. Blakemore ‘4. I. F. and Duf G. W. (1992) Single base change in the human tumor necrosis factor alpha (TNF- %) gene detectable by Ncol restriction of PCR product. Htrttt. ttrol. G‘cwt. I, 353. Wilson A. G., Gordon C.. di Giovme F. S.. de Vries N.. van de Putte L. B. 4.. Emery P. and Dufl’ G. W. ( 1994) A genetic association between systemic Iu~LI:,cl-bthematosus and tumor necrosis factor alpha. El,,,. ./. /tmtwt. 24, I9 I I95. Wilson A. G.. de Vries N., Pociot F.. di Giovinc F. S., van dcr Putte L. B. A. and Duff G. W. (1993) An allelic polymorphism \+ithin the human tumor necrosis factor 7 promoter region is strongly associated with HLA Al. BX. and DR3 alleles. J. (‘.x-,/7. Mctl. 177, 557 -560 Yang X. D.. Tisch R.. Singer S. M.. C‘ao Z. A.. Liblau R. S.. Schreiber R. D. and McDecitt H. 0. (1994) Effects of tumor necrosis factor alpha on insulin-dependent diabetes mellitus in NOD mice. I. The early development of autoimmunity and the diabetogcnic process. ./. ~‘.\-,~.121c,t/.1x0. 995 1004.