The Granulocyte-Macrophage Colony-Stimulating Factor Promoter ...

MOLECULAR AND CELLULAR BIOLOGY, Dec. 1993, p. 7399-7407

Vol. 13, No. 12

0270-7306/93/127399-09$02.00/0 Copyright © 1993, American Society for Microbiology

The Granulocyte-Macrophage Colony-Stimulating Factor Promoter cis-Acting Element CLEO Mediates Induction Signals in T Cells and Is Recognized by Factors Related to AP1 and NFAT ESTEBAN S. MASUDA,l HIROSHI TOKUMITSU,l AKIO TSUBOI,1 JOSEPH SHLOMAI,2 PEGGY HUNG,' KEN-ICHI ARAI,3 AND NAOKO ARAI'* Department of Molecular Biology, DNAX Research Institute of Molecular and Cellular Biology, 901 California Avenue, Palo Alto, California 94304-11041; Department of Cellular Biochemistry, The Hebrew University ofJerusalem, Hadassah Medical School, Jerusalem 91010, Israel2; and Department of Molecular and Developmental Biology, The Institute of Medical Science, University of Tokyo, Tokyo 108, Japan3 Received 14 May 1993/Returned for modification 24 July 1993/Accepted 8 September 1993

Expression of the granulocyte-macrophage colony-stimulating factor (GM-CSF) gene in T cells is activated by the combination of phorbol ester (phorbol myristate acetate) and calcium ionophore (A23187), which mimic antigen stimulation through the T-cell receptor. We have previously shown that a fragment containing bp -95 to +27 of the mouse GM-CSF promoter can confer inducibility to reporter genes in the human Jurkat T-cell line. Here we use an in vitro transcription system to demonstrate that a cis-acting element (positions -54 to -40), referred to as CLEO, is a target for the induction signals. We observed induction with templates containing intact CLEO but not with templates with deleted or mutated CLEO. We also observed that two distinct signals were required for the stimulation through CLEO, since only extracts from cells treated with both phorbol myristate acetate and A23187 supported optimal induction. Stimulation probably was mediated by CLEO-binding proteins because depletion of these proteins specifically reduced GM-CSF transcription. One of the binding factors possessed biochemical and immunological features identical to those of the transcription factor APi. Another factor resembled the T-cell-specific factor NFAT. The characteristics of these two factors are consistent with their involvement in GM-CSF induction. The presence of CLEO-like elements in the promoters of interleukin-3 (IL-3), IL-4, IL.5, GM-CSF, and NFAT sites in the IL-2 promoter suggests that the factors we detected, or related factors that recognize these sites, may account for the coordinate induction of these genes during T-cell activation.

phorbol esters (i.e., phorbol 12-myristate 13-acetate [PMA]) and calcium ionophores (i.e., A23187) (40). Addition of both of these agents causes maximal induction of most cytokine genes (4, 9). Yet the absolute requirement for the action of both PMA and A23187 varies for different cytokines, indicating that one pathway sometimes dominates over the other (4). We have focused on the regulation of granulocyte-macrophage colony-stimulating factor (GM-CSF) gene expression in T cells. Both PMA and A23187 are required to induce expression of GM-CSF; CsA or FK506 can completely inhibit this induction (4). After T-cell stimulation, GM-CSF mRNA can be detected in 2 to 4 h. The presence of this mRNA is largely due to transcription initiation (5, 19, 25) and requires the synthesis of specific factors, since cycloheximide (CHX), a protein synthesis inhibitor, blocks GM-CSF expression at the early stages of stimulation (27, 29). We have been characterizing presumed regulatory elements in the 5' flanking region of the mouse GM-CSF gene. This 5' region is highly conserved between murine and human clones; e.g., there is approximately 85% identity over the region extending about 360 bp upstream the transcription initiation site (18). Using transient transfection assays of 5' deletion constructs in the human T-cell leukemia Jurkat cell line, we found that the region from positions -95 to +27 of the mouse GM-CSF promoter region was capable of mediating induction in response to PMA plus A23187 (19). Further analysis identified two essential elements within this sequence, referred to as GM-KB/GC and CLEO elements.

When T lymphocytes encounter antigen in the proper context, they are activated and produce an array of protein mediators that regulate a variety of immune responses and inflammation (3, 9). These mediators are collectively called T-cell-derived cytokines and include a variety of interleukins, hemopoietic growth factors, and inflammation factors. The mechanisms that lead to coordinate production of these cytokines are being actively investigated. For most cytokines, it appears that their expression is regulated predominantly by initiation of transcription of their genes (3, 5, 9). Comparisons of the regions upstream of cytokine gene promoters have revealed a number of conserved lymphokine elements: CLE1 (CK1), CLE2 (CK2), and CLEO (3, 20, 32). In addition, the fact that immunosuppressants such as cyclosporin A (CsA) reduce or totally inhibit expression of cytokines in T cells suggests that there exists a common link in their regulation (35). The coupling of activation signals received at the cell surface to the initiation of transcription seems to involve many components, each of which may be regulated during the process of cell activation. Antigen recognition in the context of an antigen-presenting cell by the T-cell receptor rapidly switches on several signaling events, including protein phosphorylation, calcium mobilization, and inositol phospholipid hydrolysis (15, 40). With regards to transcriptional activation, these signals can be bypassed by the use of

*

Corresponding author. 7399

7400

MASUDA ET AL.

GM-KB/GC is a 23-bp cis-acting region (positions -95 to -73) that consists of two binding sites: GM-KB (previously called GM2) and a GC box. GM-KB defines a binding site for a protein induced by PMA whose characteristics are identical to those of the transcription factor NF-KB (30, 39). The GC box defines a site for constitutively bound proteins (36). The other element, CLEO (positions -54 to -40), lies near the TATA box; its sequence is similar to those of elements present in the regulatory regions for the interleukin-4 (IL-4) and IL-5 genes (20). Here, we report that CLEO is essential for induction of transcription at the GM-CSF promoter. These experiments were performed in vitro with a transcription system prepared from T cells. We found that CLEO is recognized by two factors, one with the characteristics of transcription factor AP1, and the other with features similar to those of NFAT. Finally, the implication of AP1, NFAT, or related factors in the induction of other T-cell-derived cytokines through CLEO-like elements is discussed.

MATERIALS AND METHODS Oligonucleotides and immunoreagents. The oligonucleotides used for competition and mobility shift assays contained the following sequences (only one strand is shown; sequence overhangs are presented in lowercase): CLEO element (20), 5'-gatcGTCACCATTAATCATT1CCTCTAA CTGT-3'; AP1 site (22), 5'-tcgaGCTATGACTCATCCG-3'; gm-mul,2 (36), 5'-gatcAGGTAGTTCCCCCGCCCCCC-3'; IgKB (21), 5'-gatcTCAACAGAGGGGACTTlCCGAGAG GCC-3'; and NFAT site (11), 5'-gatcGGAGGAAAAACT GTFlllCATACAGAAGGCGT-3'. GM47 and GM43,47 correspond to the CLEO oligonucleotide with a single substitution at position -47 and substitutions at both positions -43 and -47, respectively, as shown in Fig. 7. Rabbit polyclonal antisera raised against the N-terminal half (aFos-alu) or the basic region (aoFos-WB2.3) of c-Fos were kindly provided by T. Kerppola and T. Curran (Roche Institute of Molecular Biology, Nutley, N.J.). Other immunoglobulin G antibodies used were from Santa Cruz Biotechnology. Plasmids. All plasmids used as templates for in vitro transcription were digested with Fnu4HI, except for the adenovirus major late (AdML) promoter template, which was prepared by NdeI digestion of plasmid pDNAdML (8). Plasmids pGM6OLuc, pGM32Luc, pGM19Luc, and pGM+3Luc were constructed by replacing the chloramphenicol acetyltransferase (CAT) gene from the corresponding CAT constructs (19) with the luciferase gene as described previously (13). The constructions of pGM97Luc (also referred to as pKC1), pKC30, and pIL2Luc-1 will be reported elsewhere (38). Plasmid pKC30 is analogous to p43-48 (20) but with the luciferase reporter instead of the CAT gene. Plasmid pIL2Lucl, which contains positions -567 to +47 of the IL-2 gene, was derived from pIL2CAT-1 (obtained from G. Crabtree, Stanford University). Plasmid pGM6OLucl,2 (36) is analogous to pGM97Luc but has a reduced linker space between the GM-KB/GC and CLEO elements (39), giving slightly higher transcriptional activity. In vitro transcription. Jurkat T cells were cultured as described by Miyatake et al. (20). Unless otherwise mentioned, cells were stimulated for 2 h with PMA at 50 ng/ml and A23187 at 1.0 ,uM. Nuclear extracts were prepared by the method of Dignam et al. (10), using 0.2% Nonidet P-40 to disrupt the cells as described previously (36). Runoff transcription assays were carried as described by Miyatake et al. (20), using 40 to 50 ,ug of protein of nuclear extract, 100 ng of

MOL. CELL. BIOL.

template DNA, and a final KCl concentration of 100 mM. Radioactivity in runoff transcripts was quantitated by using the Phosphorlmager system (Molecular Dynamics). Depletion of CLEO-binding proteins was performed by using Dynabeads M-280 streptavidin (Dynal AIS) coupled to biotinylated multimers of the CLEO element as described by Gabrielsen et al. (12). The depletion was done on nuclear extracts from PMA-A23187-stimulated cells that were adjusted to 100 mM KCl in the presence of 10 ,ug of poly(dI-dC) per ml. Mobility shift assays. Nuclear extracts used for runoff transcription assays were used for gel shift assays. Alternatively, nuclear miniextracts were prepared according to the method of Schreiber et al. (31), modified as follows. After the appropriate treatments, 2 x 107 to 5 x 107 Jurkat cells (at 2 x 106 cells/ml) were collected and washed once with phosphate-buffered saline and then with buffer A (10 mM N2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid [HEPES; pH 7.6], 15 mM KCl, 2 mM MgCl2, 0.1 mM EDTA, 1 mM dithiothreitol [DTT], 0.5 mM phenylmethylsulfonyl fluoride, 10 mg of leupeptin per ml). The cell pellet was then resuspended on ice with buffer A containing 0.2% Nonidet P-40 and passed once through a syringe with a 27.5-gauge needle. Nuclei were gently centrifuged (-800 x g), resuspended in buffer A containing 0.25 M sucrose, and again pelleted by centrifugation. Nuclear proteins were extracted with constant shaking for 30 min at 4°C with 40 to 100 ml of a buffer containing 50 mM HEPES (pH 7.9), 400 mM KCl, 10% glycerol, 0.1 mM EDTA, 1 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, and 10 mg of leupeptin per ml. The extract was centrifuged, and the supernatant was aliquoted and stored at -80°C. The DNA-binding reactions were performed at room temperature for 30 min in 10 mM HEPES (pH 7.9)-10% glycerol-i mM EDTA-1 mM DTT2-50 ,ug of poly(dI-dC) per ml-250 ,ug of bovine serum albumin per ml-KCl adjusted to 100 mM--0.5 ng of labeled probe (-50,000 cpm)-2 to 10 ,g of nuclear extract. Samples were loaded onto 4% native acrylamide gel (Tris-glycine-EDTA buffer), run at 120 V, dried, and autoradiographed for 1 to 6 h at room temperature. Preparation of AP1 from Jurkat T cells. Nuclear extracts from Jurkat cells stimulated for 2 to 3 h with PMA-A23187 were prepared as described above. Fractions containing purified AP1 were obtained as described by Lee et al. (16), with slight modifications. Proteins were precipitated with 45% instead of 53% (NH4)2SO4, and a Superose 6 (Pharmacia) instead of a Sephacryl S-300 gel filtration column was used. Fractions eluted with 0.5 M NaCl from AP1 site affinity chromatography were shown to contain AP1 activity by mobility shift assays and Western blotting (immunoblotting). The fractions were collected into one pool, aliquoted, and stored at -80°C until further use. The aliquots had a protein concentration of .10 ng/,ul. Similarly obtained AP1 from HeLa cells were kindly provided by M. McMahon (DNAX Research Institute). RESULTS Analysis of the GM-CSF promoter by in vitro transcription. We have previously shown by promoter truncation studies that a fragment containing positions -95 to +27 of the mouse GM-CSF promoter region can confer inducibility to reporter plasmids in Jurkat T cells in response to the combination of the phorbol ester PMA and the calcium ionophore A23187 (13, 19). Here we report an analysis of the GM-CSF promoter activity in vitro by runoff transcription

VOL. 13, 1993

TRANSCRIPTIONAL REGULATION OF GM-CSF IN T CELLS

GM-KB/GC

A

CLEO

7401

TATA

pGM97Luc

pGM6OLuc

pGM32Luc pGM19Luc pGM+3Luc pKC30

v

B

C

N

0

2

3

C0

template fold stimulation

3.5x - + 10

Apw

U 0

1.1x

3.2x -

C0

c CIO

C

+

-

+

In

1.3x :. + -_

-

:1078 872 603

420nt 420nt 310 281 271

1, ,,,.

234

1

2

3

4

5

6

2

3

4

5

6

7

8

FIG. 1. Analysis of the GM-CSF promoter by in vitro transcription assays and identification of CLEO as a target of induction signals. (A) Schematic representation of the DNA templates. The mouse GM-CSF promoter sequences containing bp -95 to +27 fused to the reporter gene encoding luciferase (LUC) is shown at the top. Open rectangles indicate functional cis-acting elements. The arrow shows the transcription initiation site, and the TATA box is labeled. The lines below indicate the extent of the 5' deletions in the plasmids. Plasmid pKC30 contains a mutated CLEO element as described in the text. These plasmids were digested with Fnu4HI for use as templates for transcription assays. (B) Runoff transcription assays on truncated templates were performed with 53 jig of protein per assay of nuclear extracts from stimulated Jurkat T cells. The runoff transcript is 420 nucleotides (nt) long. The size marker on the right lane (M) is HaeIII-digested +X174 phage DNA that had been radioactively end labeled. (C) Runoff transcription assays using nuclear extracts (50 Fxg per assay) from nonstimulated (-) and stimulated (+) cells. Radioactivity in the transcripts was quantitated by using a PhosphorImager system (model 400S; Molecular Dynamics) and normalized to values obtained for AdML promoter templates. The fold stimulation shows the ratio between the transcription activity elicited by extracts from stimulated versus nonstimulated cells. The results are representative of at least three experiments using different preparations of nuclear extracts.

assays. For these assays, we used nuclear extracts prepared from Jurkat cells that had been treated for 2 h with a combination of PMA and A23187. As templates, we used DNA fragments containing the promoter that had been truncated at different distances relative to the transcription initiation site (Fig. 1A). The results (Fig. 1B) demonstrate a gradual decrease of the transcription signal as the size of the deletion increased. We tentatively concluded that both elements, GM-KB/GC and CLEO, and the TATA box played important roles in the extent of transcription initiation at the GM-CSF promoter. These findings are in general agreement with the data obtained in prior transfection studies (19, 20, 25). The CLEO element confers inducibility to the GM-CSF promoter. We next investigated the effects of the cis-acting elements on inducibility by comparing in vitro transcription activities in nuclear extracts prepared from cells that had been either mock stimulated or treated with PMA-A23187. We found that transcription of both the GM97Luc and GM6OLuc templates was strongly inducible, whereas induction was not observed on the GM32Luc template (Fig. 1C).

In fact, the slight induction seen with the GM32Luc template comparable to that observed for the unrelated AdML promoter template (data not shown; see also Fig. 2). The GM-CSF transcripts were quantified and normalized to the AdML signal (Fig. 1C). We could see that templates truncated up to position -60 still retained inducibility, which implied that at least one of the determinants for inducibility was located between positions -60 and -32, which contains the CLEO element. Next, we checked the requirement for CLEO in the presence of an intact GM-KB/GC element. We examined the KC30 template, which is analogous to GM97Luc but has a mutated CLEO element. The mutation, which abolished induction as determined by transient transfection assays, is a substitution of the residues between positions -48 and -43 (5'-CATITC-3') for a Spel site (5'-ACTAGT-3') (20). The results (Fig. 1C, lanes 7 and 8) demonstrate that the CLEO mutant template allowed only slight induction of transcription, pinpointing the CLEO element as an essential target sequence required for induction of transcription of the GM-CSF gene. Induction through CLEO requires activation of signals by

was

7402

MOL. CELL. BIOL.

MASUDA ET AL. GM97

1r--

Ir M

420nt-

N

A

P

P/A N

GM60

A

P

IL2

I IP/A N

A

AdML

rP

P/A N

I P/A

P

A

template

-~~~~~~440nt -4

* dIW.

_ * i... .w..; A. ffi

250nt

1.2 12.1 115.6 11.0 11.0 11.7 1 1.5| told normalized 11.2 J 1.2 10.4 FIG. 2. The stimulation in vitro of GM-CSF promoter templates requires the activation of two signals. In vitro transcription assays were carried out with nuclear extracts (50 fig per assay) from cells stimulated under different conditions: nonstimulated (N) or stimulated with calcium ionophore A23187 at 1.0 pLM (A), PMA at 50 ng/ml (P), or a combination of PMA and A23187 (P/A). Runoff transcripts lengths are 420 nucleotides (nt) for GM95Luc and GM6OLuc, 440 nucleotides for IL2Luc, and 250 nucleotides for AdML templates. For calculation of fold induction, the bands were quantitated and compared with the signal with nonstimulated extracts; the results are shown immediately below the corresponding lanes. These values, normalized to the values obtained for the AdML template, are presented below the values for fold induction. | 10 0.9 1.5 |1.0 |0.9 |0 9

4.6

|t3.1

1.0 1.0

1.0 12.5 9.0 1.0 I 1.5_ 6.0

both PMA and Ca21 ionophore. One of the characteristics of induction of GM-CSF gene expression in T cells, which is shared with IL-2 gene expression, is the requirement for the activation of pathways by treatment with both PMA and Ca2+ ionophore (4). We examined this dual requirement by measuring transcription of the GM-CSF templates in vitro. Nuclear extracts from Jurkat cells mock treated, treated with PMA or A23187, or treated with the combination of the two were prepared, and the transcription activity on different templates was assayed (Fig. 2). The double-stimulation requirements seen with the IL2Luc template were also observed for both the GM97Luc and the GM6OLuc templates, while the activity on the AdML template was not altered significantly. Again, the results with the GM6OLuc template suggested that pathways activated by both PMA and Ca2+ ionophore exert their effects through the CLEO element. Multiple CLEO-binding factors mediate transcriptional activation. To clarify the role of the CLEO element in the stimulation of transcription, we carried out transcription assays using nuclear extracts from stimulated cells that had been depleted of CLEO-binding proteins (see Materials and Methods). We found that depletion of >80% of the CLEObinding proteins (data not shown) reduced transcription through the GM-CSF promoter but had minimal effects on the AdML promoter template (Fig. 3A). We believe that the reduction was not due to the depletion of a particular basic transcription factor because addition of twice the amount of extract doubled the AdML signal but could not reconstitute the GM-CSF transcription to its original signal (Fig. 3A). This result supports the view that the presence of CLEObinding factors is necessary for transcriptional activation of the GM-CSF promoter. We examined the protein factors that bound the CLEO element in a mobility shift assay. The nuclear extracts used were from stimulated cells; a labeled oligonucleotide containing the CLEO element was used as the target. Previously, we and others reported two specific band shifts that reflected protein binding to the CLEO element (20, 23). We found here that by increasing the amount of extract, yet another specific band can be observed. We refer to the band shifts here as a, IB, and -y (Fig. 3B). A nonspecific band shift and could be observed with any oligonucleotide probe (see below). A factor related or identical to the transcription factor AP1

1.0 1.0

binds at the 5' end of CLEO element. During our characterization of CLEO-binding factors, we discovered that an unlabeled oligonucleotide probe containing a strong AP1 site (22) could efficiently compete for the binding of I to the CLEO element (Fig. 4, lanes 1 to 3). This effect was specific, as the oligonucleotide gm-mul,2, which contains the GMKB/GC element and is able to bind the factors NF-KdB and Spl (39), could not compete for 13 binding to CLEO (Fig. 4,

B

A GM

template:

AdML

extract

II

-P

-a

O

o

extract

v35

420nt

- -om

-6

35

m

c

35

3

ns

40

6

9 12

-w

C

70

itugl

-s

35

m

70 Ug

-25?0nt

1

2

3

4

5

6 free -x 1 2 3 4

FIG. 3. Multiple CLEO-binding factors mediate GM-CSF transcriptional activation. (A) In vitro transcription of GM-CSF and AdML templates using nuclear extracts from stimulated cells (lanes 1 and 4) or the same extracts that had been depleted of CLEObinding factors (lanes 2, 3, 5, and 6). The amount of extract used is expressed in micrograms of protein. The GM-CSF template used contained both the GM-KB/GC and CLEO elements and was prepared by Fnu4HI digestion of plasmid pGM60Luc1,2. The expected length of the transcript is 420 nucleotides (nt). (B) Mobility shift assays using the CLEO probe and increasing amounts of nuclear extract from stimulated cells. Bands a, A, and -y are indicated. ns, nonspecific band; free, unbound probe.

TRANSCRIPTIONAL REGULATION OF GM-CSF IN T CELLS

VOL. 13, 1993 probe:

CLEO AP1 IgicB --- r--- 1 -I I~~~~~~~~~~~ antisera AP1 protein competitor