Molecular Brain Research, 11 (1991) 255-264 (~ Flsevier Science Publishers B.V. All rights reserved. 0169-328X/92/$03.50 ADONIS 0169328X9270336X

255

BRESM 70336

Structural and functio al identification of regulatory regions and cis elements surrounding the nerve growth factor gene promoter Santosh R. D'Mello and Gerhard Heinrich Section of Biomolecular Medicine, Evans Department of Clinical Research and Department of .Medicine, University Hospital, Boston University Medical Center, Boston, MA 02118 (U.S.A.)

(Accepted 21 May 1991) Key words: Nerve growth factor; Transcription regulation; Cis regulatory element; AP-1 element; Transcription factor; Deletion analysis; DNAse-I footprinting; Gene expression

The transcriptional mechanisms which contribute to the regulation of nerve growtit factor (NGF) production are still largely unknown. We previously expressed the NGF promoter region in transgenic mice to localize cis regulatory elements to within 5 kb of the promoter. To further map these elements, and to begin to study the corresponding transacting factors, we here assayed the effects of 5" deletions and point mutations and examined the binding of nuclear factors to the NGF promoter region using 1,929 cell fibroblasts. Sequential deletions delineated regions upstream from the promoter which stimulated and inhibited transcription. DNAse-1 footprintit'tg experiments identified four upstream segments, designated F2, F4, F6 and ~ , whieh bound L929 cell nuclear proteins. F2 and F4 mapped to stimulatory and F6 and F8 to inhibitory regions. Competition experiments using a heptanucleotide present in both F2 and F4 segments suiggested that they may be bound by related factors. Gel shift assays showed that the F8 binding proteins are less abundant in L929 cells than in NIH 3T3 fibroblasts and BI6 melanoma cells. In addition to the upstream segments, a downstream AP-I consensus sequence bound L929 nuclear proteins. Mutation of the AP-1 consensus sequence eliminated binding of nuclear proteins and reduced transcriptional activity. Our results indicate that transcriptional activator as well as suppressor regions surround the NGF gene promoter. The regulation of NGF production is likely to involve ,:is elements within these regions and transacting factors that bind to them.

INTRODUCTION Nerve growth factor (NGF) is a member of a structurally and functionally related family of polypeptides each of which is trophic for discrete sets of neurons ~4'!s" 17. NGF is secreted constitutively in minute amounts from targets of NGF-responsive neurons by a variety of cells including fibroblasts, smooth m u ~ ! e cells, Schwann cells, gila and neurons 4"!2"22. N G F ~:.roduction in neuronal targets is regulated developmentally and tissue-specificallys'29. The relative abundance of N G F m R N A in neuronal targets of adult rodents correlates with the relative intensity of innervation by l~]GF-responsive neurons 26, suggesting a critical ro~e of N G F in the regulation of neuro~target interactions. Compared with neuronal targets, N G F is secreted in amounts several orders of m a ~ i t u d e higher by guinea pig prostate ~°, bull seminal v~:sicle ~s, and mouse submandibular gland (SMG) 19. The specific cells which secrete N G F from the mouse SMG, the granulated convoluted tubular (GCT) ceils, aad the regulation of N G F gene expressioa during sexual maturation within these cells have been exten-

sively im, estigated 2a9`2s. In SMG but not ~,n neuronal targets there is an androgen-dependent sex difference in N G F gene expression 2'19"2s. As an important step toward elucidating the transcriptional mechanisms that generate this complex and contrasting pattern of basal N G F production, we previously cloned and structurally characterized the mouse and rat N G F gene promotor regions 31. We recently employed the cloned mouse N G F gene promotor region for expression studies in transgenic mice ~. For these experiments, the promotor including 5 kb of the 5" flanking region were fused to the human growth hormone (hGH) structural gene and allowed to stably integrate into the chromosal D N A of mice1. Analyses of expression of the transgene in the salivary gland revealed that it was expressed at high levels in G C T cells and regulated in parallel with the endogenous N G F gene during development and sexual maturation. The transgene m R N A also reflected the tissue distribution of N G F m R N A including the iarge difference in levels between salivary glands and neuronal targets. These results suggested that cis regulatory elements which mediate N G F gene expression

Correspondence: G. Heinrich, Biomolecu!ar Medicine, Evans 603, University Hospital, 88 East Newton S,.reet; Boston, MA 02118, U.S.A. Fax: (1) (617)638-6009.

256 are present in the transgene and are located within 5 kb of the promoter. To map and functionally characterize these cis elements and to begin to identify the transacting factors that act on them, we here report experiments in which we transiently expressed N G F / h G H fusion genes in L929 fibroblasts and examined the binding of nuclear proteins extracted from the. e cells to the N G F gene p r o m o t e r region. We used L929 cells for these experiments because they are derivatives of fibroblasts and as such represent a principal cell type which expresses the N G F gene in neuronal targets in vivo 4"12. In addition, L929 cells have b e e n extensively utilized to study N G F gene expression 6'27'3°. Finally, they are m o r e easily grown and transfected in tissue culture than G C T cetis and neurons. The results of our experiments indicate that distinct regions upstream and downstream from the p r o m o t e r have stimulatory and inhibitory effects on basal N G F gene transcription. Both the upstream stimulatory and inhibitory regions contain binding sites for nuclear factors. These binding sites may represent cis regulatory elements which play a role in the regulation of N G F gene transcription together with the transacting factors that interact with them. MATERIALS AND METHODS

Cell culture Cells were grown in Dulbecco's modified Eagle's medium (DMEM) containing either 10% horse serum (L929 cells) or 10% fetal calf serum (B16 melanoma and NIH 3T3 cells) and penicillin (100 units/ml), streptomycin (100 pg/ml), and glutamine (1 mM). Cells were fed every 3--4 days and passaged every 7-10 days. All cells were maintained in a humidified atmosphere of 93% 02/7% CO 2 at 37 °C.

Nuclear Froteins Conflueut cells were extracted using the procedure of Dignam et al. 7, and extracts stored at -70 °C. Protein content was estimated according to Bradford 5.

DNAse-I Jbotprinting DNAse-1 footprintin~ analyses we~'e performed as described by Jones et a1.13. EcoRI-BamHl (-250 to +125) and EcoRV-EcoRI (-500 to -250) restriction fragments were labeled each at a single end with 321,. Aliquots of 5 ng of each fragment were incubated separately with 20/~g of L929 nuclear extract on ice for 90 min and allowed to warm to room temperature over a 2 min period. The mixture was then digested with 15 ng of DNAse-1 (Sigma) for 1 min. The digests were fractionated on 6% denaturing polyacrylamide gels alongside G+A and C tracks of chemically cleaved frag~:~,'Is as markers.

ATT TVI'F and 5"-TCC AAA ACC CCT GGT TrA GAT GAA AAT GCA TGG. The probes representing the 'wild-type' or mutant AP-1 elements were similarly prepared from larger templates and a smaller primer complementary to the 3' end of the template. The sequence of the template for the 'wild-type' probe was AGC GCA TCG GTG AGT CAG GCT TCT CTG AGC CGA. The template for the mutant probe was AGC GCA TCG GTG AGe tgG GCT TCT C'rG AGC CGA. The common primer was TCG GCT CAG AGA AGC.

Cons~_ruction of plosmids The set of 5" deletion mutants lacking the APol element was derived from the construct previously described and expressed in L929 cells31 using restriction enzyme digestions and religations. The -39 deletion mutant was constructed using synthetic oligonucleotides and the sequence confirmed by DNA sequence analysis. The plasmids containing either 'wild-type' or mutant AP-I elements (p-750.AP-1 and p-750.MUT) were constructed by using the cloned mouse NGF promoter region as a template in polymerase chain reactions (PCR). The primers for the PCR were synthetic oligonucleotides. The downstream primers were for p-750.AP-l: 5"-GGATCC TATCG GGGCT CGGCT CAGAG AAAGCC TGAGTCA CCGATGC GC-3" and for p-750.MUT: 5"-GGATCC TATCG GGGCT CGGCT CAGAG AAAGCC TGAGctg CCGATGC GC-3". A common upstream primer was: 5"-TGCTA AGCTr CACGC AGGGA GCACAT-3". The amplified DNA was digested with HindHI, and ligated into a modified version of the promoterless hGH expression vector pDGH 24. This version was prepared by inserting two synthetic complementary oligonueleotides into the Aatll site in exon 2 of the hGH gene to reconstitute the N-terminus of hGH. The oligonucleotides have the sequences 5" GCTGCAG ATG GCG ACG GGAA GCA GGA CGT-3" and 5"CCT GCT TCC CGT CGC CAT CTG CAG CACGT-Y. Once the oligonucleotides were inserted, the plasmid was digested with Ball to cut in intron 1 of the hGH gene and with Hindlll to cut in a polylinker upstream of the hGH gene sequences. The intervening fragment was removed and replaced with the amplified DNA prepared am1 digested as described above. The pla~,mids containing the mouse metallothionein promoter (pYGH5) and, in addition, the NGF flanking regions (pYGH5sup), were derived from the plasmid pXGH5, originally constructed and described by Selden et al. 24. In pXGH5 the hGH structural gene is linked to the mouse metallothionein promoter and 1.5 kb of its 5" flank. The plasmid pYGH5 was generated from pXGH5 by digestion with BamHl and Kpnl, blunt-ending using the Klenow fragment of E. coil DNA polymerase I, and religation under conditions which favored intramolecular ligation. This maneuver removed an upstream 1 kb fragment thereby shortening the 5" flank to approximately 500 bps. To construct pYGH5-sup, pYGH5 was digested with Xbal and blunt-ended using the Klenow fragment of E. coil DNA polymera.,;e 1. An EcoRl/EcoRl fragment from the NGF gene 5" flank (-250 to -750), similarly blunt-ended, was ligated into the blund-ended Xbal site, and the orientation established using restriction enzyme digestions.

Transections of cells and assays of transcription activity L929 cells were transfected with 10/~g of DNA using the DEAEdextran methods as previously described 31. Medium was assayed for hGH using a two-site radioimmunoassay kit from Nichols Institute (San Juan Capistrano, CA) as previously described 31.

Gel ~hift assays Gel shift assays were performed using 32p-labeled double-stranded oligonucleotide probes as described by ,Sen and Baltimore 25. The probes representing the F8 segment in the suppressor region were prepared by annealing two oligonucleotides complementary at their 3" ends, and extending them in the presence of dATP, dTrP, dGTP, and either dCTP or [32p]dCTP using the Klenow fragment of E. coil DNA polymerase !, The sequences of the oligonucleotides were: 5"-CTC TGC ATC TGT GAC CTC CCC CCA CCA TGC

RESULTS

Deletion analysis identifies regions within the NGF 5" flanking region that affect N G F promoter activity The N G F / h G H fusion gene we used for the construe-

257

I +8

-5kb

F~

TT'~AA

F8 -:1011 - 2 l l O

F6 -184

1:4

-1ll3

-1111

.106

-|4

-39

I

-500 -5kb

.;¢ .',._-

r m

I

I I

--' 50

-7oo

Exon I

-4~l

-9O 20

I

hGH In m e d i u m ( n g l m l ) ~. ~. ~. ~.

3.30 26.2 35.6 22.4

+1.0.3 +1.0.05 +1.3.4 +/-2.4

_"-

3.65 5.37 3.01 1.56 =.150

+/-0.~8 +1.0.09 +!.0.17 +1-0.08

m m

promo~erleos hGH gone metollothioreln gone promoter

Fig. 1. 5" deletion analysis of the NGF promoter region using transient expression in L929 cells. The top line diagrammatically outlines the NGF/hGH fusion gone. The open boxes represent hGH gone exons 1-5, and the black box the first 8 nucleotides of oxen 1 of the mouse NGF gone. The TTAAA sequence of the promoter is represented as a cross-hatched box. Thin lines are 5" flank, introns, or 3" flank. Below the NGF/hGH fusion gone diagram and on the left is shown the set of 5" deletion mutants. The dark black lines indicate the relative extents of the 5" flank of each deletion mutant. The most upstream nucleotide position is indicated by a number at the end of each line. All deletion mutants terminated at position +8. The sites where protein binding was detected using DNAse-1 footprinting are indicated by ovals in the NGF/hGH fusion gone diagram. The binding sites are projected onto the 5" flanks in the set of deletion mutants. The number in each projection line indicates the nucleotide positions of the 5" and 3" borders of each footprint. Next to each deletion mutant is shown the concentration of hGH in the medium of L929 cells transiently transfected with the corresponding constructs. The procedure and other details are provided in the methods section. Each experiment was rel~ated several times using triplicate samples. The plasmids p~)GH and pXGH5, described by Selden et al.24 were used for the experiment labeled 'promoterless hGH gone' and 'metallothionein gone promoter', respectively. tion of transgenic mice extended from - 5 kb to + 8 of the mouse N G F gone 31. To map cis regulatory elements within this region that affect basal N G F gene transcription, we subjected the 5" flank to stepwise 5" deletion analysis beginning with the -5 kb construct expressed in the transgenic mice. To detect the effects of the deletions on basal transcription, the 5" deletion constructs were transiently expressed in L929 fibroblasts. The amount of h G H secreted into the culture medium was measured by radio immunoassay and used as an index of transcriptional activity. The constract containing 5 kb of 5" flanking region was expressed at relatively low levels. This low level of expression did not change significantly when constructs containing 750 and 500 bp of 5" flank were used, indicating that basal transcription is not affected by far upstream sequences. Transcription increased 8-fold when the 5" flank was deleted to -250 ~nd increa~=d further when deleted to -120. A small decrease of transcriptional activity was observed when the -120 to -90 region was deleted and expression dropped to initial low leve!s on deletion to -39 (Fig. 1). These analyses identified two regions that have significant effects on promoter activity, a suppressor region between -250 and -500, and an activato~- region between -39 a n d - 9 0 . Two additional regions that have relatively small inhibitory and stimuiatory effects lie in the -250 to -120, and the -120 to -90 regions, respectively.

L929 nuclear proteins bind segments in the functionally defined regions The transient expression assays identified several regions whose deletion affects basal transcription, suggesting that they contain cis regulatory elements which may bind transacting factors and mediate the observed effects. To identify segments bound by proteins, nuclear extracts from L929 cells were assayed for binding using DNAse-1 footprint analysis. As shown in Fig. 2 (lane ' + ' ) , three regions in the + 8 to -250 bp region were protected from DNAse-1 digestion indicating binding of nuclear proteins in the footprinted segments. The footprinted segments were designated from 3" to 5" as F2, F4 and F6. F2 extends from the --43 to -64, F4 from -105 to -119, and F6 from -163 to -189. The F2 and F4 segments lie in regions associated with transcriptional activation. Interestingly, both regions contain identical ' G G A G G G G ' heptanucleotides suggesting that the F2 and F4 regions may bind similar factors. To test this possibility, a 200-fold excess of a synthetic oligonucleotide representing the F2 footprint and containing the common heptanucleotide was used as a competitor in the binding reaction. As shown in Fig. 2A, lane 'C', the F2 oligonucleotide competed out protein binding to both the F2 and F4 regions. In contrast, binding to the F6 region was not affected. This result indicates that F2 and F4 are bound by factors which recognize the common heptanucleotide and therefore

258

-C

-189

---

F6

T C C C C A G G A C C C T T A C G G A C C C

-119 ..,J

F4

C C A G A G A G G G A

G T . . . . . . , ~ ,,-:~",~i~ 1 ~ , ~ o v ~ , ; ~

t~ 4" O)

O

w

-I-

-" - 2 9 6

,(T :G :T :G

!A

'C ~ -291

¢ ~ -285

:C :C :A I ~ ' ~

-64

G C C T G G A

G G

A G G G G C 'r A A A T A C

............

emmm

m

....

!~!i

A

,C :L,C ...-280 -277

C A T T T T C A T - - ~

B

-269

259

2

Fig. 3. Gel retardation assay using probes representing the -250 ~o -309 segment in the suppressor region of the NGF gene. A double-stranded oligonucleotide probe spanning nucleotides -250 to -309 of the NGF gene 5" flank and labeled with 32p was prepared as described in Materials and Pvlethods. The radiolabeled probe was incubated with nuclear extiacts, the mixtures subjected to electrophoresis on a 5% polya~ylamide gel and DNA/protein complexes visualized by autoradiography. A: assays using extracts from L929 cells. Lane l: extract plus labeled probe. Lane 2: as in lane 1 plus a 250-fold excess of unlabeled probe. Lane 3: as in lane 1 plus a 20-fold excess of a restriction fragment spanning the -250 to -500 region. Specific DNA/protein complexes are indicated by arrows on the left hand side of the figure. B: probe incubated with nuclear extracts from B-16 melanoma cells (M), L929 fibroblasts (L) and NIH 3T3 fibroblasts (T). Specific DNA/protein complexes are indicated by arrows on the left hand side of the figure.

3

4---

M

T

might be structurally related. The region upstream o f - 2 5 0 was examined for the binding of nuclear proteins using an E c o R I / E c o R V fragment extending from - 2 ~ to -500. One intense footprint and two fainter footprints were observed in the

-269 t o - 2 9 6 region (Fig. 2A). The more intense foot.. print is marked by a bold bracket and the fainter footprints by dotted brackets in Fig. 2A. It is presently uaclear whether the 31 bp region spanned by these footprints represents the binding site of a single large protein complex or whether the three footprints are induced by three separate smaller complexes. The region spannmg the -269 to -296 region was designated F8. No additional protected segments were observed in the -250 to -500 region (data not shown). The positions of the 4 footprinted segments in the 5" flank are outlined in the top diagram in Fig. 1. To confirm and further analyze binding of nuclear proteins to the F8 region, we used gel shift assays. For these assays, a 59 bp synthetic oligonucleotide spanning the -250 to -309 bp region was utilized. As shown in Fig. 3A, two complexes formed between nuclear proteins and oligonucleotide probe (Fig. 3A, lane 1). In the presence of a 200-fold excess of unlabeled oligonucleotide, complex formation was con~ple~ely abolished (lane 2). When a 20-fold molar excess of a cloned fragment spanning the -250 to -500 regior~ was used as a competitor, formation of the larger complex was severely reduced whereas the lower complex was relatively unaffected (lane 3). These results show that binding to the -250 to -309 segment is spe6~c and suggest that the factor generating the larger complex is present at lower amounts than the factor inducing the smaller complex.

Fig. 2. DNase-1 footprint analysis of the NGF promoter region. Restriction fragments spanning the '-8 t~ ,?.~0, and -250 to -500 regions of the mouse NGF genie were labeled at a single end with 32p and incubated with L929 nuclear extracts. Following partial DNAsc-1 digestion, the DNA was subjected to electmphoresis on 8% polyacrylamide gels containing 8 M urea and visu~!ized by autoradiography. A: footprint analysis using a BamHI-EcoRl fragment (+8 to-250), radiolabeled at the BamHl site. Lanes '--' and '+' represent the absence and presence of nuclear extracts in the analysis. Lane 'C' represents a competition experiment using 250-fold excess of an unlabeled oligonucleotide spanning the F2 region. B: footprint analysis using an EcoRI-EcoRV fragment (-250 to -500), radiolabeled at the EcoR site. Lanes '-' and '+' represent the absence and pre~nce of nuclear e.~__~O~.Lanes 'g+a" and 'c' represent Maxim and Gilbert sequencing reactions performed on the same restriction fragment.

260

Celhdar level of suppressor proteins correlate inversely with expression of the NGF gene The binding of nuclear proteins to a segment c~f the promoter region functionally defined as a suppressor region suggested that the bound segment may represent a cis regulatory element. This cis element could suppress NGF gene transcription not only in L929 cells, but could play a more general rc,le. If this were the case, one might expect an inverse relationship between levels of nuclear factors that bind to the suppressor element and basal NGF gene expression. To examine this possibility, we compared binding of proteins from L929 cells to those from a mouse melanoma cell line B-16 and NIH 3T3 cells. We have previously observed that NGF mRNA is undetectable in B-16 cells by Northern blot analysis, whereas NIH 3T3 cells express lower amounts of NGF mRNA as compared with L929 cells (unpublished observation). As shown in Fig. 3B, although the level of binding to the F8 region is similar in B-16 melanoma and NIH 3T3 cells it is much higher than the binding observed using L929 extracts. This result suggests that the level of the 'suppres~r' proteins may, in part, influence the level of NGF mRNA expressed. The similarity of the three binding patterns suggests th/it the same proteins bind to the F8 region in these three cell lines.

150 A O) C

tO0

T

E _= _c 0

so

pYGH5

pYGH~-sup

Fig. 4. Effect of the -250 to -750 region of the NGF gene on transcriptional activity of the mouse metallothionein promoter. The -250 to -750 region of the mouse NGF gene was inserted upstream of the mouse metallothionein gene promoter in pYGH5, a plasmid containing the gGH gene under the control of the metallothionein promoter. The resulting plasmid pYGH5-sup and pYGH5 were expressed in L929 cells and amount of hGH secreted in the culture medium measured 3 days after transfection.

suppressor fragment from the NGF and metaUothionein promoters differed by more than 250 bps in these constructs. This difference in position may also account for the lack of suppression. These results indicate that F8 is not a cis element with the characteristics of enhancers/ dehancers which genes'ally function independently of orientation, position and promoter context.

The transcr~otional suppressor region containing F8 may be promoter specific

A downstream AP-1 element binds nuclear proteins and enhances transcriptional activity of the NGF promoter

The F8 region contains the octamer A'ITITCAT, a sequence known as octarner motif and shown to be essential for the expression of immunoglobulin genes in B cells ~6. This sequence is present in several other genes, and proteins that bind to this element are found in a variety of cell types including L cells 16. It has been proposed that the octamer motif may be bound by repressor proteins except in B cells which actively transcribe immunoglobulin genes 16. In these cells the repressor protein may be modified by B cell-specific factors resulting in a conversion of the octamer motif from a suppressor to an activator of imrnu~globulin gepe transcription. In view of this proposed mechanism, it is possible that the F8 segment may be bound by a transcriptional suppressor which could act not only on the NGF but also on other promoters. 1"o test this possibility, the -750 to -250 fragment was inserted upstream of the mouse metallothionein gene promoter at -500. This promoter mediates 50-fold higher rates of basal transcription than the NGF gene promoter in L929 cells 31. Insertion of F8 upstream fror,l the MT promoter did not significantly affect transcription (Fig. 4), suggesting that F8 is specific for the NGF promoter. However, the distance of the

These experiments mapped several regions which affect basal transcription and bind nuclear factors to positions upstream from the promoter. We next extended these analyses to the region immediately downstream from the promoter. A major impetus for extending these studies to the downstream region was the presence of an AP-1 consensus sequence TGAGTCA. The AP-I consensus sequence is located at position +36 at the junction of exon 1 and intron 1. Cis regulatory regions containing AP-1 consensus sequences have been shown to play a role in basal transcription of several genes 3"2°. We have shown that nuclear proteins from un~timulated L929

catggaggaggggctaaataeagtoaggaagcotattaaagaagatakgtgot©cag

©a©gge

*10 420 &G&GAQeGCCTGG~GCCGGI~GGG~L~GC~t

+40 ct

Fig. 5. Position of the AP-I element within the mouse NGF geae. The nucleotide sequence of the promoter region of the mouse gene is shown. The nucleotide sequence of exon 1 is given in boldface, and the 5" flank and intron 1 in lower case letters. The promoter sequence is underlined. The transcription start site is numbered +1. The region protected by DNAse-1 digestion in footprinting experiments is circled. The AP-1 consensus sequence within the footprinted region is boxed.

261 -T50 .A!?-1 -T50 .MOT-

'fGAG'fCA ----C'I'G

NGF (-750 tO 4~65)

C v

4S

-

40

-

35

-

E _a

3° -

.l:

20

-

O

~s

-

5

-

[..ATG.'~..

as -

.1::

-750

.~

i

-750

.NO'J?

Fig. 6. Transcriptional assay of plasmids containing the NGF AP-1 element in the wild type and mutant form. NGF/hGH fusion genes containing the -750 to +65 region of the mouse NGF gene with the AP-1 consensus sequence in the wild type (-750.AP-I) and mutant (-750.MUT) form were transiently expressed in L929 cells. Levels of hGH in the medium were assayed 3 days after transfection. Top: diagrammatic representation of the constructs. Black boxes represent exons 2-5 of the hGH gene and the open box ¢xon 1 of the NGF gene. The lines are 5" flank, introns, and 3" flank. The positions and sequences of the wild type and mutant AP-1 element within intron 1 of the NGF gene are indicated. In the nucleotide sequences ~ s h e s indicate identical nucleotides. The position of the translation initiation codon and codons for the six N-terminal amino acids of hGH which were regenerated by insertion of synthetic oligonucleotides into exon 2 of the hGH gene is indicated in parentheses (...ATG...). Bottom: histogram of concentrations of hGH in medium of L929 cells transiently transfected with p-750.AP- and p-250.MUT.

six hGH codons were reconstituted by inserting an oligonucleotide encoding them into exon 2. The plasmid containing this fusion gene was designated p-750.AP-1 (Fig. 6, top diagram). When transiently transfected into L929 cells, p-750.AP-1 was expressed at levels approximately 5-fold higher than the -750 deletion construct shown in Fig. 1 which lacked the region containing the AP-1 consensus sequence (data not shown). This result suggested that the AP-1 consensus sequence may represent the cis regulatory element within the +8 to +65 region which mediated the observed increase in basal NGF gene transcription. To confirm this possibility, the AP-1 sequence within p-750.AP-1 was mutated from TGAGTCA to TGAGctg using syntheti~ oligonucleotides and the polymerase chain reaction (PCR). We mutated these nucleotides because these mutations have previously been shown to be critical for the binding of nuclear factors to the AP-1 consensus sequence ~ and they do not alter the splice donor sequence. To confirm that the introduced mutations affected binding of nuclear proteins to the NGF AP-1 consensus sequence, oligonucleotides containing

1

2

3

4

!,

1 •

'

2"

o-

fibroblasts bind to the same AP-1 sequence 6~ A role of the AP-1 consensus sequence in NGF gene expression is also supported by its conservation in mouse, rat, and human NGF genes. The region bound by nuclear proteins a~d the AP-I consensus sequence within it are outline~J !,'~~?~. ~..~ "~o examine ~:aether it may play a role in basal tran~ripfion of the NGF gene, we transiently expressed in L929 cells NGFtIIGH fusion genes containing the AP-1 consensus seque~¢e. These fusion genes were constructed by extending the NGF fragment from +8 to +65. The AP-1 consensus sequence is located in intron I of the NGF gene and ~s part of the splice donor site (Fig. 5). To preserve this location, and to provide a downstream splice ~cceptor site, we fused the NGF gene sequences to intron 1 of the hGH gene. However, the initiation codon of hGH as :~ell as the first six amino acids are located within exon i. Since this exon was eliminated by our fusion, the translation initiation codon and the first

~

~.

.~

Fig. 7. Gel retardation assay of L929 nuclear extracts using oligonucleotides containing wild type and mutant AP-1 element sequences. Radiolabeled probes representing the 'wild-type' and mutant AP-I elements were prepared as described in Materials and Methods. Radiolabeled probes were incubated with nuclear extracts, the mixtures subjected to electrophoresis on a 5% polyacryiamide gel and DNA/protein complexes visualized by autoradiography. Lane 1: nuclear extract plus 'wild-type' probe. Lane 2: as in lane 1 plus a 200-fold excess of unlabeled probe. Lai~e 3: nuclear extract plus mutant probe. Lane 4: as lane 3 plus a 200-fold excess of unlabeled mutar, t probe. Specific DNA/protein complexes are indicated by arrows on the left hand side of the figure.

262 the 'wild type' and 'mutant' AP-1 sequence were utilized in gel shift assays. Fig. 7 shows the results of these experiments. Two protein-DNA complexes were observed. In the presence of a 250-fold excess of unlabeled probe, formation of the larger complex was inhibited whereas the lower complex was unaffected, suggesting that the lower complex was due to non-specific binding of pro~oins to the probe. In contrast to the results obtained with the wild type AP-1 sequence, the larger complex did not form between extracts from L929 cells and the mutant AP-1 sequence (Fig. 7). This result shows that the mutations introduced into the AP-1 sequence prevent the binding of nuclear proteins. The effective inhibition of binding of nuclear proteins suggested that this mutation may also inhibit transcription. Indeed, transient transfection assays of the construct p-750.MUT (illustrated in Fig. 6, top panel) containing the mutated AP-1 sequence showed that it was expressed at approximately 6-fold lower levels than p-750.AP-1 (Fig. 6, bottom panel). This result shows that the AP-1 consensus sequence is a regulatory element which increases the basal activity of the NGF gene promoter. Hengerer et al. ~ recently proposed that the same AP-1 element we examined here may play a role in modulation of NGF gene transcription. Our present results show that in addition to modulation of NGF mRNA levels, the AP-1 element contributes to regulation of basal NGF gene transcription. The s',imulation of basal transcription by the AP-1 element raised the question whether its presence may alter the relative activities of ihe upstream suppressor and stimulator regions defined by the 5" deletion analyses. To address this question, additional deletion constructs containing 500 and 250 bp of 5" flank were generated from p-750.AP-1 and transiently expressed in L929 cells. The constn~cts p-750.AP-1 and p-500.AP-1 were expressed at similar levels whereas p-250.AP-1 was expressed at approximately 4-fold higher levels (data not shown). Thus, the upstream suppressor is active in the presence of the downstream AP-1 element. DISCUSSION

The results here delineate three regions surrounding the NGF gene promoter which affect basal NGF gene transcription. These include an upstream suppressor and activator, and a downstream stimulatory region which contains an AP-1 element. Two additional regions situated between the major upstream suppressor and stimulatory regions have relatively moderate effects on basal promoter activity. Within these functionally defined regions are segments that bind L929 nuclear proteins. It is likely that the binding segments represent cis regulatory

elements within the larger, functionally defined regions. This supposition is supported by the sequence conservation of these segments in the mouse and rat NGF genes 3~. In addition, our data using mutational analysis provide clear evidence for a cis regulatory role of the AP-1 element within the downstream stimulatory region. Similar mutation and transient transfection analyses of the putative upstream elements will confirm their role in basal NGF gene transcription. Both the F2 and F4 footprints which lie within the two upstream activating regions contain a G G A G G G G sequence suggesting that they may be bound by common factors. Several growth factor genes including the FGF, TGF-a, TGF-fl, IGF-1 and 2, and PDGF genes contain a GGAGGG sequence within their promot,~" regions. The GGAGGG sequence in the PDGF 2 gen~ is required for basal transcription as well as induction by the phorbol ester TPA 2~. A nuclear factor induced by TPA was shown to bind to an oligonucleotide that contains the GGAGGG sequence. It remains to be determined whether the proteins that bind to the G G A G G G sequences of the PDGF 2 and NGF genes are related. Our experiments define a transcriptional suppressor in the -250 to -500 region. The F8 region within this segment is bound by L929 nuclear proteins and contains an octamer that shares sequence homology with an element found to be important for cell-specific expression of immunoglobulin genes in B-cells ~6. It has been proposed that the octanucleotide within the immunoglobulin gene may serve as a general suppressor element. The inability of the NGF suppressor region to affect a heterologous promoter argues against this possibility. It is conceivable, however, that the octanucleotide is bound by a ubiquitous suppressor protein whose action becomes ~romoter-specific by interaction with NGF gene-specific factors that bind to the F8 region. Consistent with this possibility is the observation that the region upstream of the octanucleotide is bound by nuclear proteins (Fig. 3). The sup~ressor region may be a major determinant of the low levels of NGF mRNA in neuronal targets. This possibility is supported by our finding that L929 cells in which NGF mRNA is relatively abundant contain low amounts of nuclear factors that bind to the suppressor region whereas NIH 3T3 cells which express low levels of NGF mRNA and B-16 melanoma cells which express no detectable NGF mRNA, contain much higher levels of F8 binding proteins. Thus, F8 may play a role as a cell-specific regulatory, element in addition to regulating basal transcription ir~ L929 cells. Transfection experiments using other cell types will help clarify this possibility. The downstream AP-1 element serves as a transcriptional stimulator in L929 cells. This conclusion is sup-

263 ported by the effects of mutations which prevent binding of nuclear proteins and sharply reduce transcriptional activit~ of the N G F gene promoter. A role of the AP-1 element in basal transcription is also supported by recent analyses of expression of the NGF/gGH fusion gene carried by transgenic mice 6a. The N G F sequences in this NGF/hGH fusion gene terminate at +8. This fusion gene therefore represents a 3" deletion mutant with respect to the AP-I element which is located at position +36. We recently found that expression of this fusion gene cannot be detected by Northern blot hybridization in fibroblasts cultured from the kidneys of the transgenic mice although these cells express the endogenous N G F gene ~ . This finding is consistent with a role of the AP-1 element in basal N G F gene expression in fibroblasts. In contrast to its importance in fibroblasts, our previously reported analyses of these transgenic mice carrying the N G F / h G H fusion gene 1 imply that the AP-1 element is not required for high levels of N G F gene transcription in the SMG. The cis element(s) responsible for the high level of N G R gene transcription in the mot,se SMG therefore presumably lie upstream of +8, the 3" limit of the N G F sequences in the transgene. In view of our present results, it it tempting to invoke a mechanism for the high levels of N G F gene transcription which involves the upstream regions shown to be transcriptionally active in the deletion analyses. Accordingly, the SMG may produce large amounts of activator factors that bind the F2 and F4 regions, or it may lack the suppressor proteins that bind to the F6 and F8 regions,

REFERENCES 1 Alexander, J.M., Hsu, D., Penchuk, L. and Heinrich, G., Cellspecific and developmental regulation of a nerve growth factorhuman growth hormone fusion gene in transgenic mice, Neuron, 3 (1989) 133-139. 2 Aloe, L. and Levi-Montalcini, R., Comparative studies on testosterone and L-thyroxine effects on the synthesis of nerve growth factor in mouse submaxillary salivary glands, Exp. Cell Res., 125 (1980) 15-22. 3 Angel, E, Imagawa, M., Chin, R., Stein, B., Imbra, J.l., Rahmsdorf, HJ., Jonat, C., Herrlich, E and Karin, M., Phorbol ester-inducible genes contain a common cis element recognized by a TPA-modulated trans-ac~iag factor, Cell, 49 (1987) 729-739. 4 Bandtlow, C.E., Heumann, R., Schwab, M.E. and Thoenen, H., Cellular localization of nerve growth factor synthesis by in situ hybridization, EMBO 3., 6 (1987) 891-899. 5 Bradford, M.M., A rapid and sensitive method for the quanfitation of microgram quantifies of protein utilizing the principle of protein-dye binding, Anal. Biochem., 72 (1976) 248-254. 6 D'Mello, S.R. and Heinrich, G., Induction of nerve growth factor gene expression by 12-O-tetradecanoyl phorboi 12-acetate, J. Neurochem., 55 (1990) 718-721. 6a D'Meilo, S.R. and Heinrich, G., Nerve growth factor gene expression: involvement of a downstream, AP-1 element in basal and modulated transcription, Mol. Cell. Neurosci., 2 (1991) 157-167.

or both. On the other hand, it is also conceivable that an as yet unidentified element functions specifically in the mouse SMG to increase N G F gene transcription. A number of genes containing AP-1 elements are regulated by the phorbol ester TPA 3. Thus, besides its role in basal transcription in fibroblasts, the AP-1 element may also be involved in the regulation of N G F gene expression by the phorbol ester TPA. We he,ve recently provided evidence for such a role 6a. It is possible that additional elements are involved in the regulation by TPA, particularly the G G A G G G G sequence situated in the F2 and F4 activator regions. This sequence was shown to be required for the TPA response of the P D G F 2 gene 2~. Elements located in the suppressor region which contains F8 may also participate in the TPA response, conceivably by TPA-induced downregulation of the amounts or activities, or both, of suppressors which act on the F8 segment. Factors that bind to the stimulatory regions could be upregulated concomit~mtly. Related mechanisms may mediate the increases in NGF gene expression observed after sciatic nerve injury 11"12 and limbic seizures 9. Thus, basal and regulated NGF gene transcription may involve interrelated cis elements and transacting factors.

Acknowledgements. We thank Min Zheng for major contributions, Jun-Wei Wu and Mark Cartwright for expert technical help, and Stella C. Martin for stimulating discussions. This work was supported by NIH grant NS22422, the American Health Assistance Foundation and the Evans Medical Foundation.

7 Dignam, .I.D., Martin, P.L., Shastry, B.S. and Roeder, R.G., E~zkaryotic gent transcription with ?urificd components, Methods Enzymol., 101 (1983) 582-598. 8 Ebenda l, T., Larhammar, D. and Persson, H., Structure and expression of the chicken beta-nerve growth factor gene, E.MBO J., 5 (1986) 1483-1487. 9 Gall, M.C. and Isackson, P.J., Limbic seizures increase neuronal production of messenger RNA for nerve growth factor, Science, 245 (1989) 758-761. 10 Harper, G.P., Barde, Y.A., Burnstock, G., Carsteins, J.R., Dennison, M.E., Suda, K. and Vernon, C.A., Guinea pii', prostate is a rich source of nerve growth factor, Nature, 279 (1979) 160-162. 11 Hengerer, B., Lindholm, D., Heumann, R. Rfither, U., Wagner, E.W. and "Ihoenen, H., Lesion-induced increase in nerve growth factor mRNA is mediated by c-los, Proc. Natl. Acad. Sci. U.S.A., 87 (1990) 3899-3902. 12 Heumann, R., Korsching, S., Bandtlow, C. and Thoenen, H., Changes of nerve growth factor synthesis in non-neuronal cells in response to sciatic nerve transection, J. Cell Biol., 104 (1987) 1623-1631. 13 Jones, K.A., Yamamoto, K.R. and Tjian, R., Two distinct transcription factors bind to the HSV thymidine kinase promoter in vitro, Cell, 42 (1985) 559-572. 14 Knfisel, B., Winslow, J.W., Rosenthai, A., Burton, L.E., Seid, D.E, Nikolics, k. and Hefti, F., Promotion of central cholinergic and dopaminergic neuron differentiation by brain-derived neurotrophic factor but not neurotrophin-3, Proc. Natl. Acc,i.

264 Sci. U.S.A., 88 (1991) 961-965. 15 Leibrock, L, Lottspeich, F., Hohn, A., Hofer, M., Hengerer, B., Masiakowski, P., Thoenen, H. and Barde, Y.-A., Molecular cloning and expression of brain-derived neurotrophic factor, Nature, 341 (1989) 149-152. 16 Maeda, H., Araki, K., Kitamura, D., Wang, J. and Watanabe, T., Nuclear factors binding to the human immuno~iobulin heavy-chain gene enhancer, Nucl. Acid. Res., 15 (1987) 28512869. 17 Maisonpierre, P.C., Belluscio, L., Squinto, S., lp, N.Y., Furth, M.E., Lindsay, R.M. and Yancopoulos, G.D., Neurotrophin 3: a neurotrophic factor related to NGF and BDNF, Science, 247 (1.'~0) 1446-1451. 18 Meier, R., Becker-Andr6, M., G6tz, R., Heumann, R., Shaw, A. and Thoenen, H., Mt~lecular cloning of bovine and chick nerve growth factor (NGF): delineation of conserved and unconserved domains ~ad their relationship to the biological activity and antigenic~ty of NGF, EMBO J., 5 (1986) 1489-~.493. 19 Murphy, R.A., Watson, A.Y., McCarthy, M., Papastavro~, M., Neutra, M. anal Forssmann, W.¢3~ Submandi~alar gl~,,,ds in mice with muscular dystrophy: studies with nerve grc~wttt factoe, Anat. Rec., 200 (1981) 177-194. 20 Oshima, R.G., Abrams, L. and Kulesh, D., Activation of an intron enhancer within the keratin 18 gene by expression of c-fos and c-jun in undifferentiated F9 emhryona! carcinoma cells, Genes Dev., 4 (1990) 835-848. 21 Pech, M., Roa, C.D., Robbins, K.C. and Aaronson, S.A., Functional identification of regulatory elements within the promoter of platelet-derived growth factor 2, Mol. Cell. Biol., 9 (1989) 396-405. 22 Rennert, P.D. and Heinrich, G., Nerve growth factor mRNA in brain: localization by in situ hybridization, Biochem. Biophys. Res. Commun., 138 (1986)813-519. 23 Risse, G., Jooss, K., Neuberg, M., Bz~iiler, H.-J. and Mfiiler, R., Asymmetrical recognition of the palindromic AP-1 binding

site (TRE) by Fos protein complexes, Embo Z, 8 (1989) 38253832. 24 Selden, R.F., Howie, K.B., Rowe, M.E., Goodman, H.M. and Moore, D.D., Human growth hormone as a reporter gene in regulation studies employing transient gene expression, 114o1. Cell. Biol., 6 (1986) 3173-3179. 25 Sen, R. and Baltimore, D., Multiple nuclear factors interact with the immunoglobulin enhancer sequences, Cell, 46 (1986) 705-716. 26 Shelton, D.L. and Reichardt, L.F., Expression of the betanerve growth factor (NGF) gene ¢orrelates with density of sympathetic innervation in effector organs, Proc. Natl. Acad. Sci. U.S.A., 81 (1984) 7951-7955. 27 Siminoski, K., Rennert, ED., Murphy, R.A. and Heinrich, G., Cortisone, aldosterone, and tes~Jstcrone reduce levels of the messenger RNA coding for r~e~x,e growth factor in cultured L929 fibroblasts, Endocrir~ology, 121 (1987) 1432-1437. 28 Walker, P., Weichsel, M.E. Jr., Hoath, S.B., Poland, R.E. and Fisher, D.A., Effect of thyroxine, testosterone, and corticosterone on nerve growth factor (NGF) and epidermal growth factor (EGF) concentrations in adult female mouse submaxillary gland: dissociation of NGF and EGF responses, Endocrinology, 109 (1981) 582-587. 29 Whittemore, S.R., Ebendal, T., Larksfors, L., Olson, L., Seiger, ,~., Strombet'g, I. and Persson, H., Developmental and regional expression of beta-nerve growth factor messenger RNA and protein in the rat central nervous system, Proc. Natl. Acad. Sci. U.S.A., 83 (1986) 817-821. 30 Wion, D., Barrand, E, Dicou, E., Scott, J. and Brachet, P., Serum and thyroid hormones T3 and T4 regulate nerve growth factor mRNA levels in mouse L cells, FEBS Lea., 189 (1985) 37--41. 31 Zheng, M. and Heinrich, G., Structural and functional analysis of the promoter region of h'le nerve growth factor gene, Mol. Brain Res., 3 (1988) 133-140.

Structural and functional identification of regulatory regions and cis elements surrounding the nerve growth factor gene promoter.

The transcriptional mechanisms which contribute to the regulation of nerve growth factor (NGF) production are still largely unknown. We previously exp...
2MB Sizes 0 Downloads 0 Views