shorter promoter fragments of appropriate size by PCR amplification using pMHFY .... NF-E2 oligonucleotide or cold PCR fragment was used as cold competi-.
Vol. 269,No.32,Issue of August 12,pp. 20281-20288,
THEJOURNALOF BIOLOGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.
1994
Printed in U.S.A.
Mouse Ferritin H Subunit Gene FUNCTIONAL ANALYSIS O F THE PROMOTER AND IDENTIFICATION OF AN UPSTREAM REGULATORY ELEMENT ACTIVE IN ERYTHROID CELLS* (Received for publication, December 30, 1993, and in revised form, May 16, 1994)
Carole BeaumontSP,Attila SeyhanS, Abdel-Kader YachouS, Bernard GrandchampS, and Richard Jones1 From the SGdnetique et Pathologie Moleculaires de l’He‘matopoi&e, ZNSERM Unite 409, Faculte‘ Xavier Bichat, 75018 Paris, France and the IZnstitute of Molecular Medicine, John Radcliffe Hospital, Oxford OX3 SOU, United Kingdom
change inits requirement for iron. An increase in the L subunit Ferritin synthesis is regulated at the translational level by iron, butit is likely that transcriptional regula- is associated with the needto increase iron stores, whereas the tion of H and L genes is responsible for tissue-specific H subunit is more abundant in circumstances where iron is distribution of H and L mRNAs. In order to define the neededfor cellular metabolism (for review, see Ref. 4). For regions important for transcriptional regulation of the example, induction of heme synthesis or stimulation of cell mouse ferritin H gene, we have linked the promoter, proliferation has been shown to increase the proportion of H including the transcription start site, and 5 kilobases of subunit (5-7). Other stimuli, like thyroid-stimulating hormone upstream sequence to a reporter gene (human growth or tumor necrosis factor, also increase the production of the H hormone).This construct anda seriesof 5’ deletion mu- subunit (8, 9), although the reason for these changes is still (K562, poorly understood. The regulationof ferritin H and L synthesis tantshavebeenusedtotransfecterythroid mouse erythroleukemia (MEL)) and hepatoma (HepG,) is complex and seems to operate at the levels of gene transcripcell lines. Measurement of growth hormone in the cultion and mRNA translation. Changes in mRNA stability may ture medium and analysis of ferritin-growth hormone transcripts by a ribonuclease protection assay revealedalso contribute to the control of ferritin production in some that a 140-base pair minimal promoteris sufficient to circumstances (10). confer a high level of expression to the reporter gene At inthe translational level, ferritin synthesis is regulated by iron. Under conditions of iron starvation, synthesis is repressed both cell types. In addition, a 180-base pair fragment, by interaction between an “iron-responsive element” (IRE),’ lying 4.5 kilobases upstream of the ferritin transcription start site, functions like an inducible enhancer during located in the 5”untranslatedregion of all ferritin mRNAs (11), N,N”hexamethylene-bis-acetamide-induceddifferentia- and a cytoplasmic RNA-binding protein(12). Iron entry into the tion of MEL cells.Aperfectmatch toa consensus binding cell induces conformational changes in the IRE-binding proNF-E2 is pres- tein, which leads to its release and subsequent translation of motif to the erythroid transcription factor ent in this regulatory element, but the mutant NF-E2 the mRNA(for review, see Ref. 13). This regulation operateson enhancer retains the inducible activity in stably trans- both H and L mRNA and is highly conserved throughout evofected MEL cells, and the results from gel retardation lution and between different cell types (14). Iron has also been assays suggest that protein-DNA complexes form that in shown to act at the transcriptionallevel to modulateexpression vitro between the ferritin enhancer and MEL nuclear of the L gene in HeLacells and hepatocytes withoutaffecting extracts do not contain NF-E2. Thus, nuclear factors expression of the H gene (15, 16). that mediate inducibility of the ferritin enhacer remain Very little is known about the tissue-specific regulation of to be identified. ferritin gene transcription. Induction of differentiation in HL60 or in mouse erythroleukemia (MEL) cells brings about an increase in theH/L mRNA ratio (17, 181, and we have been able Ferritin is the major ironstorageprotein of highereuto show that, in thecase of MEL, this is the consequence of an karyotes. Varying proportions of H and L subunits in the fer- increase in transcription of the ferritin H gene (19). ritin protein shell correlate with changes its in ability toincorWe (20) and others (21) have recently cloned and sequenced (1,2). Thisislargelyduetothe porateandreleaseiron the mouse ferritin H gene. Analysis of the promoter sequence ferroxidase activity of the H subunit, which catalyzes the oxi- reveals the presence of DNA motifs (GC-rich box, CAAT box) dation of Fe(I1) into Fe(III), and facilitates its incorporation that are frequentlyassociated with eukaryotic gene promoters into the iron core in the ferritin cavity (3). The L subunit is and areknown to bind ubiquitous transcriptionfactors (23,23). devoid of this ferroxidase activity. By modifylng the proportions However, computer search analysis of the first 1 kb of the of the H and L subunits in ferritin, the cell adapts to any promoter region did notreveal any homology with known binding sites for erythroid-specific transcription factors. The ferri* This work was supported by a grant from the Association pour la tin H gene is an interesting exampleof a gene that is ubiquiRecherche sur le Cancer. The costs of publication of this article were tously expressed but also up-regulatedduringerythroid defrayed in part by the payment of page charges. This article must differentiation. A similar pattern of expression has been detherefore be herebymarked “aduertisement” in accordance with 18 scribed for a number of other genes, such as the heme pathway U.S.C. Section 1734 solely to indicate this fact. The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTMIEMBL Data Bank with accession number(s) M85257. 8 Recipient of a INSERIWMedical Research Council fellowship. To whom correspondence should be addressed: GBnetique et Pathologie Moleculaires de YHematopoiese, INSERMU409, Faculte Xavier Bichat, 16, rue Henn Huchard, 75018 Pans, France. Tel.: 33-1-44-85-63-80; Fax: 33-1-42-26-46-24.
The abbreviations used are: IRE, iron-responsive element; MEL, mouse erythroleukemia; kb, kilobase(s); bp,base pair(s); GH, growth hormone; hGH, human growth hormone; PCR,polymerase chain reaction; HMBA, N,N”hexamethylene-bis-acetamide; F-GH, fenitingrowth hormone.
20281
20282
Mouse Ferritin H Gene Upstream Regulatory Region
enzymes (24, 25) or the membrane-bound glycophorin C, but the up-regulation of each of these genes presents distinctfeatures. 5-Aminolevulinate synthetase, the first enzyme of the heme pathway, is encoded by two different genes (26). One is ubiquitously expressed, whereas the otheris expressed only in erythroid cells (27). Porphobilinogen deaminase is encoded by a single gene with two promoters (28). Expression of the downstream promoter is restricted to cells of the erythroid lineage, whereas the upstreampromoter is active in all cell types (29). The housekeeping glycophorin C gene has a unique promoter that is regulatedby the binding of a combination of ubiquitous and erythroid-specific factors (30). Here we report that ferritinH gene expression is controlled by a strong promoter, is ubiquitously active,and, inaddition an upstream regulatory element, responsible is for increased transcription during induction of erythropoieticdifferentiation. This element maps to a 180-bp fragment, located 4.5 kb upstream of the transcription start site, but although it contains a perfect match t o a consensus binding site for the erythroidspecific transcription factor NF-E2, the data suggest that inducibility of this elementdoes not relyon the bindingof NF-E2.
amounts of plasmid were transfected, calculated to have 25 pg of the shortest construct. After electroporation, the cells were resuspended in medium and grown in culturefor a n additional 16-24-h period. Stable Zkansfections-MEL cells (2x lo7)were mixed with a total of 20 pg of linearized plasmid DNA in a final volume of 0.8 ml and electroporated using a Bio-Rad gene pulser apparatusat settings of 240 V and 960 microfarads. The 20 pg of plasmid consisted of the plasmid of interest andof a plasmid containinga selectable marker(pMC1Neo) at a molar ratio of 5:l. Transfected cells were divided into four plates, grown in nonselective mediumfor 48 h, and then transferred into G418 (0.8 mg/ml) containing medium for 2 weeks. Each plate is considered as an independant transfection and then grown in normal medium or induced with HMBA for 72 h as described above. G H a n dp-Galactosidase Assays-Cells transfected with constructs containing the hGH gene and with pIRV were grown for 24 h. The amount of hGH releasedby the cells was measured 100-p1 in aliquots of culture medium using the hGH Allegro radioimmunoassay kit (Nichols Institute, CA). The assay was performed in duplicate, and the total amount of hGH produced per plateover a 24-h period was calculated. The cells were also harvested, washedonce, resuspended in200 pl of 0.25 M Tris (pH 7.4), and subjectedt o three cycles of freeze thawing, andp-palactosidase was assayedin the resulting lysate using D-nitrophenyl 6-Dgalactopyranoside as substrate. /%Galactosidase (Sigma) was used as standard, and activities were calculated as milliunitdcell pellets. The amount of hGH accumulated in the culture medium was corrected for variations in transfection efficiency by dividing it by the P-galactosidase EXPERIMENTAL PROCEDURES activity of the corresponding cell pellet. Resultsare given as a percentage Construction of Ferritin-Growth Hormone Plasmids-All plasmids of the activity of the full-length promoter construct (-940 pF-GH). were constructed using standard techniques. Cloning ofpMHFY,, a Ribonuclease Protection&say-For RNA analysis, transfected cells pGEM7Zf(+) recombinant plasmid containing a 6-kb EcoRI fragment were grown for 16 h, harvested, and washed twice with phosphatewith the mouse ferritin H gene and flanking sequences, has been de- buffered saline. Total RNA was extracted using RNAzole B (CinnaBiotecx Laboratories), as previously described(20). RNA was analyzed scribed previously (20). The human growth hormone (hGH) gene has been excised from pOGH as a HindIII-NsiI fragment and inserted into by RNAse protection usingthe SP6 system (33). The hybrid ferritin-GH pMHFY,, between the NaeI site of the ferritin 5'-untranslated region mRNA, the ferritinmRNA transcribed from the transfected mouseferand the NsiI site of the pGEM7Zf(+) polylinker, in place of the 3' endof ritin H gene (pMHFY,), and the endogenous mouse ferritin H mRNA the ferritin gene. The resulting construct (designated -940 pF-GH) were all measured using a single probe. This was a derivative of -940 pF-GH madeby deletion between theEcolO9 site on the firstexon of the contained 940 bp of ferritin upstream region and 90bp of 5"untranslated region linked to the GH gene. The 90 bp downstream of the GH gene and the NsiI site at the 3' end of the gene. The resulting transcription start site retains the iron-responsive element. The5' de- plasmid was linearized using the BssHII site (at -210 in the ferritin promoter) and antisenseRNA probe generated using SP6 polymerase. letional mutants were obtained by using pre-existing restriction sites, This probe is protected by a 145-bp fragment from the F-GH hybrid and the following plasmids were generated: -300 pF-GH (KpnI), -210 mRNA or by a 85-bp fragmentfrom the first exon of the mouse ferritin pF-GH (BssHII), -60 pF-GH (Ball). The remaining constructs (-140 pF-GH, -90 pF-GH, and -70 pF-GH) were obtained by generating the H gene. The5' mouse a globin probe, pSP65a97,is protected by a 97-bp shorter promoter fragments of appropriate size by PCR amplification fragment from the first exon of the mouse or,-globin gene. RNA-RNA hybridization was performed in 80% formamide hybridusing pMHFY, as a template. The amplified fragments were then inserted inplace of the full-length promoter. The plasmid pC17F-GH was ization buffer a t 50 "C overnight. RNAse A + T1 digestion was carried out at room temperature for 30 min, and the protected fragments were constructed by replacing the1-kbEcoRI-NaeIfragment from -940 pF-GH by a 6-kb EcoRI-NaeI fragment excised from a genomic clone run on a denaturing 8% polyacrylamide gel. End labeled HpaII-cut pBR322 was used as a molecular weight marker. Autoradiography was overlapping with pMHFY, and containingmore 5"flanking sequences. Shorter enhanced constructs (numberedI to X) were generated by ex- done at -70 "C using intensifymg screen. Radioactive bands were quancising various fragmentsfrom pC1,F-GH using pre-existing restriction tified by cutting them out, along with appropriate backgrounds, and sites. These fragments were subsequently inserted into -940 pF-GH counting the radioactivity with an LKB-Wallac P-plate counter (34). Nuclear Extract Preparation and Gel Retardation Assay-Nuclear (between the EcoRI site of the polylinker and the KpnI site at -300) in extracts were preparedas previously described(38)from mouse eryththe ferritin promoter. All of these constructs retain 300 bp of ferritin promoter and 90 bpof 5"untranslated region linked to the hGH gene. roleukemia cells either exponentially growing or induced to differentiof 5 mM HMBA. The 154-bp probes for use A G to T mutation was introduced the in NF-E2 consensus sequence in ate for 3 days in the presence in binding assays, corresponding to the wild type and mutant enhancer, the 1-kb EcoRI-KpnI from NpF-GH construct to generate mNF-GH, using the Transformer Site-Directed Mutagenesis kit (Clontech). Pres-were generatedby PCR using the32P-endlabeled Pyr oligonucleotide as forward primer and either the NF-E2 or the mutant NF-E2 oligonucleence of the mutation was confirmed by sequencing. Plasmid pSPNaNAEco, used as an internal control for transfection efficiency, otide as reverse primer. Nonlabeledoligonucleotides were used to amconsists of a 3.74-kb BglII-EcoRI fragment containing the a1globin gene plify the fragments usedas cold competitor. A shorter110-bp fragment, and flanking sequences. pIRV, a gift from J. Morgensten, contains a n which did not contain the NF-E2 binding site, was also generated by PCR using Pyr and Enhl3oligonucleotides as amplimers and testedas actin-promoted P-galactosidase gene. a possible competitor in the binding assay. Cell Line Culturesand Zkansient Zkansfections-K562 (derived from The following primers were used: Pyr, + 8 to +28 from the SacI-BglII a human erythroleukemia) and MEL 707 were grown in RPMI 1640 enhancer fragment, CATCCTTCCTTCCCCTCCAGC; NF-E2, from medium, and HepG, (derived from a human hepatoblastoma) were +138 to +162, AGCAGAATGCTGAGTCACGGTGGM,mutant NF-E2, grown in minimum Eagle's medium medium; all were supplemented with 10% fetal calf serum. MEL cells were induced by the additionof 5 from +138to +162, AGCAGAATTCTGAGTCACGGTGGM, Enhl3, mM HMBA to the culture mediumfor 72 h (31). Cells were transfected from + 87 t o 111, GCCCTACCCCCTCCATGACAGCA. For binding assays, 20,000 cpm of end labeled PCR fragment was by electroporation in conditions previously described (32). Briefly, 5 x of poly(d1-dC)and 3 pg of lo7 cells wereharvested (by trypsinization,inthecase of HepG,) added last toa 15-pl reaction containing 2 pg nuclear proteins and incubated on ice for 15 min.Double-stranded washed once, and resuspended in 1 ml of phosphate-buffered saline containing 100 pgof supercoiled DNA. The cells were electroporated at NF-E2 oligonucleotide or cold PCR fragment was usedas cold competi325 V for HepG,, 400 V for K562, and 460 V for MEL cells. The DNA tors in some reactions and added before the probe. To assess for the consisted of 25 pg of the F-GH constructof interest, 25 pgof a control presence of NF-E2 in some of the protein-DNA complexes, gel retardaplasmidtocorrect for variationsintransfection efficiency (pIRV, tion assays were also carried out in the presence of 1 pl of antiserum prepared against recombinant p45 NF-E2 (35). Free and bound DNAs pSPNaNAEco or pMHFY,), and 50 pg of pSP65 as carrier DNA. To were separatedby electrophoresis in a nondenaturing5%polyacrylamcompare the activity between plasmids of different sizes, equimolar
20283
Mouse Ferritin H Gene Upstream Regulatory Region B
A
F O
-940
, !
Relative GH activity *
140
FERRITIN
GH
-300
"""""""
. -- .
-210
""""""""
"_"_""""""
_" " " "
__"_"_" " " "
-140
-80
Spl bindinp sib CCAAT box
-70
I
120
IRE
- -------
7
100 80
80 40
20 0 -940
-300 -140 -210
-90
-70
-60
corrected for transfection efflciency
-80
FIG.1.Deletional analysisof the mouse ferritin H promoter. A, schematic representation of the 5' deletion mutants used for the functional analysis of the ferritin H promoter. To generate the initial construct (-940 pF-GH),940 bp of ferritin promoter and90 bp of 5'-unstranslated region, containing the IRE were linked to the hGH gene. Positions of CCAAT box (*) and SP1 binding site (W are indicated.E GH accumulation in the pg of a F-GH construct, 25 pg of pIRV, a plasmid containing an medium of transfected K562 and HepG,. Cells were transfected with 25 actin-promoted P-galactosidase gene, and 25 pg of pSP65. The cells weregrownfor 24 h, and growth hormone in the culture medium and P-galactosidasein the cell pellet were measured in duplicate. Resultsare expressed in arbitrary units and represent the ratio of the hGH to the out of three typical experiments. P-galactosidase, made100 for the longest construct. The figure shows one ide gel in 0.5 x TBE (Tris borate, EDTA). Gels were run at 150 V for 3 h at room temperature, dried, andexposed to film.
RESULTS In Vivo Analysis of Ferritin H Promoter-The human growth a reporter gene t o test thefunctional hormone genewas used as activity of the ferritin H promoter. A DNA segment from the ferritin H gene consistingof 940 bp upstream from the cap site and 90 bp of 5"untranslated region containing the IRE was linked to the growth hormonegene. A series of unidirectional 5' deletion mutants was constructed (Fig. lA)and used to transfect erythroleukemia (K562) and hepatoma(HepG,) cell lines. A construct containing the actin promoter linked to the P-galactosidase gene was co-transfected to correct for variation in transfection efficiency. The transiently transfected cells were grown for a 24-h period, at the endof which accumulation of GH in the medium and P-galactosidase activity in the cell pellets was measured. The results (Fig. 1 B ) show that differences in expression between the various constructs are very similar in these two cell types. However, the absolute levels of expression cannot be compared since transfectionefflciencies vary between cell types and the GH assay relies on the ability of a given cell type to secrete hormone into themedium. For example, measurement of GH remaining in the cells at the end of the 24-h period clearly showed that the rate of secretion is rate limiting in K562, whereas in HepG, there is no accumulation of intracellular GH (not shown). This may be responsible for the greater variability observed in the amounts of GH in the culture medium of K562, especially in the case of more actively transcribed plasmids. Deletion from -940 to -140 had little or no effect on GH accumulation. Deletion from -140 to -90, which removes the most upstream SP1 binding site, led to a 5-fold drop in the amount of GH accumulated in themedium, whereas removal of the second SP1 motif, between -90 and -70, had no effect. Finally, removal of a CCAAT boxreduced GH production to barelydetectable levels. Since all of these constructs contain the IRE, it remained possible that the amount of GH produced did not reflect the actual amount of mRNA transcribed from these plasmids and that translationalrepression mediated by the IRE affected GH production in a different way in the two cell types. To test this possibility, we measured the hybrid ferritin-GH mRNAs from the transfected cells. mRNA Synthesis Correlates with GH Accumulation in Both Erythroid and Hepatoma Cell Lines-The hybrid ferritin-GH
constructs were co-transfected with pMHFY2, a plasmid containing the intact ferritin H gene with 1 kb of promoter sequence. We have previously shown that thisclone is functional when transfected into HeLa cells and is transcribed from a unique transcription start site(20). A single ferritin-GH antisense RNA probe was used t o measure mRNA from both the F-GH constructsand pMHFY2 byribonuclease protectionassay (see "ExperimentalProcedures"). When hybridizedto the hybrid F-GH transcripts, thisprobe would be protected over 145 bp by correctly initiated transcripts but only over 85 bp by the pMHFY,-derived mRNA. As shown in Fig. 2.4, the F-GH gene products are correctly initiated and give the expected 145-bp protected fragments. Thedoublet that is constantly observed when analyzing the F-GH or the pMHFY2transcripts is probably an artifact specific for this probe since the intact ferritin gene gives a single band when hybridizedwith a ferritin probe derived from exon 1 (20). Quantitative evaluation of the ferritin-GH-protected fragments, relatedt o the expression of the co-transfected control plasmid (pMHFYS), are summarized inTable I. The accumulation of the hybrid ferritin-GH mRNA in the transfected cells correlates with the accumulation of GH in the medium. Considering that the hybrid F-GH mRNA contains the iron-responsive element, which has been shown to mediate translational repression of ferritin synthesis(111,it is interesting to note that variations in the amount of transcripts produced by transfected cells are followed by similar variations in the amount of GH synthesis. A miminal promoter containing the TATAbox and 28 bp of upstream sequence (Fig. 2, -60 construct) gave a barely detectablesignal, althougha protected fragment showing correctinitiation could be detected on a longer exposure (not shown). Addition of a CAAT box (Fig. 2, -70 construct) brought about a 10-fold increase in the level of expression, maximum activity being obtained with the further addition of 70 bp of promoter sequence containing two SP1 binding sites (Fig. 2, -140 construct). The addition of sequences activity. in From up t o -940 did not lead to any further increase these results,it can be seen that thepromoter of the ferritinH gene is very active since as little as5 1.18of RNA gave a strong signal, comparable with thatobtained with atransiently transfected a-globin gene. The mRNA derived from the co-transfected ferritin H gene gave a signal of similar intensity when corrected for the difference in size of the protected sequence, suggesting that intragenic sequences are notnecessary for the basal activity of the 1-kb promoter in either cell type. The
Mouse Ferritin H Gene Upstream Regulatory
20284
Region
A
K 562
Hep,G2
"
FIG.2. RNAse protection assay of H Fer-GH the ferritin-GH constructs transiently transfected into K562 and HepG,. Cells were transfected with 25pg of a F-GH construct, 25 pgof p M H N 2 , a plasmid containing the entire ferritin H gene, and with 50 pg of pSP65. The cells were grown for 16 h, and the transient Mouse expression of the constructs was analyzed H f e r r i t i n by a ribonuclease protection assay a s de( p MHFY2) scribed under "Experimental Procedures"; a schematic representation of the probe used and of the expected protected fragments is shown in the lower part of the figure. Each lane represents hybridization of 5 pg of total RNA with loficpm B of probe, and the autoradiography is 5-h a exposure.
-145b-
- 85b-
pCL7F-QH and 5' deletions
Eeo 109
145b
i
4
.
4
-
Probe 355b
1 1
Protected trwrnenta
Forritin H germ (DMHFY2)
TABLE I Quantitative evaluation of the protected fragments fromFig. 2 Signal from construct 60 was too low to be accurately quantitated in eithercell type. HepG2
K562 C17"
940"
300" 90" 140" 210"
2420
3580
70"
C17"
940"
300"
210"
140"
90"
665 1.9
1187 607 1.95
330 718 0.45
cpm
Ferritin-GH mRNA 8625 pMHFY2 1765 1709 1542 1781 1085 1850 Ratio 4.6 (Ferritin-GWpMHFY2)
2.2
2
3630 2.3
70"
cpm
3500 1267 1592 1048 1763 960 1101 2 1.6
0.60 2.8
1467 0.65
631
659
662 2.4
555 503 1.10
" Construct. apparent difference between the two cell types in the amount of mRNA produced by the various constructs could easily be accounted for by the difference in transfection efficiency. These results show that 140 bp of promoter sequence is sufficient to sustain a high level of reporter gene transcription. Furthermore, thedecrease in activity with progressive deletion of this 140-bp promoter follows a similar pattern inboth erythroid and hepatoma cell types (Table I), suggesting that cisacting sequences responsible for tissue-specific variations in ferritin H gene expression must lie further away from the transcription start site. In order to test this hypothesis, we made anotherconstruct containing6 kb of upstream sequence (pC17F-GH)and used this to transfect both cell types. We found that inK562 this constructproduced a 2-3-fold greater amount of mRNA than the shorter constructs, whereas in HepG,, these additional sequences had almost no effect (Fig. 2 and Table I). To explore the possibility that the upstreamregion contains an erythroid-specific transcriptional regulator, we followed the expression of these two constructs during erythroid differentiation of mouse erythroleukemia cells. A Regulatory Element, Inducible with Erythroid Differentiation, Lies Upstream of the Ferritin H Gene-After 3 days of culture in thepresence of a chemical inducer (HMBA), mouse erythroleukemia cells are committed to terminal erythroid dif-
ferentiation, and the differentiated phenotype is maintained, even after removal of the inducer. We transfected uninduced and 72-h-induced MEL cells with the constructcontaining 6 kb of upstream sequence (pC1,F-GH) and with the shorter construct containing only 1 kb of promoter (-940 pF-GH). Cells were maintained in culture for 24 h after transfection in the absence of the inducer, and RNA was analyzed usingthe F-GH probe. This probe cross-hybridizes with the endogenous mouse ferritin H mRNA to give the same85-bp protected fragments as was seen in human cells transfected with the mouse ferritin gene (pMHFY,) (see Fig. 2). Accumulation of endogenous mouse ferritin mRNA increased by5-10-fold during HMBA-induced differentiation of MEL cells (Fig. 3), in agreement with our previous observations (19). Expression of the ferritin-promoted GH gene containing 6kb of upstream sequence was also increased to a similar extent in induced MEL, whereas the construct with only 1 kb of upstream sequence has the same transcriptionalactivity in both uninduced and induced cells (Fig. 3). Regulatory sequences responsible for the inducibility of the ferritinH gene, therefore, appear tobe located between -1 and -6 kb upstreamof the cap site in pCI,F-GH. Enhancer Activity Resides on a 180-bp Fragment 4.5 kb Upstream of the Danscription Start Site-To map the regulatory
Mouse Ferritin H Gene Upstream Regulatory Hlndlll I
;:
pCL7F-OH
-940F-OH
Region
20285
cisely to a 180-bp SacI-BglII fragment, which retained theability to up-regulate the reportergene when tested in either orientation (only the natural orientation is shown in Fig. 4B, lane 9). These results demonstrate that the upstream regulatory region from the ferritin H gene contains an element that confers to the ferritinpromoter an enhanced transcriptional activity ininduced MEL cells but does not bring aboutany increase in transcription eitherin noninduced MEL cells (see Fig. 3) or in the non-erythroid cell line HepG, (not shown). Inducibility of Ferritin Enhancer Does Not Correlate with NF-E2 Binding-Sequence of the 180-bp enhancer fragment
A
B Constucts
H Fer-GH
H
Mouse Ferritin
Mouse H ferritin
FIG.3. An inducible enhancerlies upstream of the ferritin H gene. Two F-GH constructs, containing either 1 kb (-940 pF-GH) or 6 kb (pCL7F-GH)of upstream sequences were transfected intoMEL cells, either noninduced (-1 or induced (+) by HMBA for 72 h prior to transfection. Transient expression of the transfected plasmids was analyzed a s described in Fig. 2. The lower band represents the accumulation of the endogenous ferritin H mRNA. elements responsible for inducibility of the ferritin gene in MEL, the 5-kb HindIII-EcoRI fragment from pC1,F-GH was divided into smaller fragments. These weresubcloned between the EcoRI and KpnI sites of -940 pF-GH, thus replacing the most 5' 640 bp of the promoter region. Removal of this region does not affect expression of the reportergene (Fig. 2). Equimolar amounts of these constructs were used to transfect 72-hinduced MEL cells, which were then maintained in culturefor another 16 h in the absence of the inducer. Expression of the transfected plasmids and of the endogenous H ferritin gene was analyzed by ribonuclease protection assay using the F-GH probe (Fig. 4). A plasmid containing the human cq globin gene under thecontrol of its own promoter was usedas a control for transfection efficiency, and a-globin transcripts were analyzed separately. Quantitative evaluation of the protected fragments are reported in Table 11. The shortest construct(-300 pF-GH) wasused as a noninducible control (Fig. 4B, lane 1), and pC1,F-GH was used as an inducible control (Fig. 4B, lane 2).A first series of constructs (Fig. 4B, lanes 3-6) mapped the sequence responsible for induction to a 1-kb EcoRI-KpnI fragment (constructIV).This fragment was thendivided into a set of smaller fragments thatwere inserted upstream of the minimally promoted reporter constructs and transientlytransfected into induced MEL cells (Fig. 4B, lanes 7-9). By these further experiments, the inducible enhancer was located more pre-
Lanes
d globin
C SKI GAGCTCTCATCC~TCC~TCCCCTCCA~CC~TCCTGTCCACC~CCC~AAC CACAMACCACAGCCCTCCA(C~TCCAGACTGCCCTACCCC~CCATGA
e TAAAGCAC~~TTGCA(~CCCAACCCCTCCAAAGGA~AGAATGCTGAGTCAC CGTGCAACAACAAT~CCAACAAGATCT
NFEZ
Bo12
FIG.4. Finemapping of the ferritin H enhancer. The 5-kb HindIII-EcoRI fragmentfrom pCL7F-GH was divided into a set of overlapping fragments, numberedI-V, and fragment N was subsequently divided into smaller fragments using preexisting Sac1 and BglII sites. The individual DNA fragments were linked at the KpnI site of -300 pF-GH, and the resulting constructs were tested for their enhancing activity by transienttransfectioninto 72-h-induced MEL cells and mRNA quantification a s described in Fig. 2. A, schematic representation of the various ferritin-growth hormone constructs.B , RNAse protection assay of the transiently transfected MEL cells. Construct -300 pF-GH was used as a noninducible control, and pCI,F-GH was used a s a n inducible control. To correct for transfection efficiency, a plasmid containing the entire a,globin gene was co-transfected and hybridized separately. C , nucleotide sequence of the minimal 180-bp Sad-BglII fragment with enhanceractivity. The sequenceof the consensus NF-E2 been motif is underlined and asterisk indicates the G residue that has mutated to generatemIVpF-GH.
Mouse Ferritin H Gene Upstream Regulatory Region
20286
TABLE I1 Mauuine o f the enhancer actiuitv Construct
I
c17
300
111
VI11V
IV
X
VI1
cpm
Ferritin-GH mRNA a mRNA Ratio (Ferritin-GWn) Enhanced expression
A-
+
-
1500 309 4.8
205 177 1.16 -
+
-
362 312 1.16 -
+
+
-
1202 240 5
+
1470 131 11.2
-
+
B - +
+
643 155 4.14
184 116 1.6 -
250 156 1.6
-
+
3
4
-
293 154 1.9
-
+
+
-
+
6
7
8
Mouse
H Ferritin
1
2
3 EcoRl-
4
5
6
7
8
Kpnl
IVpFGH
1
2
5
-r
