Portal de Periódicos da CAPES

Expression of a Human β-Globin Transgene in Mice with the CACC Motif and Upstream Sequences Deleted from the Promoter Still Depends on Erythroid Krüppel-like Factor

2000; Elsevier BV; Volume: 275; Issue: 5 Linguagem: Inglês

10.1074/jbc.275.5.3675

1083-351X

Louis-Georges Guy, Nathalie Delvoye, Lee Wall,

Epigenetics and DNA Methylation

Mice in which the erythroid Krüppel-like Factor (EKLF) gene is inactivated die in fetal life due to down-regulation of the β-globin gene. Results have suggested that EKLF functions through the proximal CACC motif of the β-globin promoter. For example, natural mutations of this element that fail to bind EKLF give reduced gene expression and the ability of EKLF to activate reporter genes in co-transfection assays is dependent on an intact CACC. Here, removal of the CACC motif and upstream promoter sequences from the β-globin gene resulted in reduced expression in transgenic mice. However, breeding onto an EKLF−/− background demonstrated that a CACC-less β-globin transgene remains highly dependent on EKLF. Hence, although the β-globin gene partly depends on the proximal CACC motif for expression, it is unlikely that the major mechanism of gene activation by EKLF is through this element. We also show that a lacZ reporter gene linked to the β-globin promoter, with or without the CACC box present, is actually expressed higher in EKLF−/− fetuses than in wild type animals, suggesting that EKLF may be able to act as an inhibitor of transcription with certain transgene configurations. Mice in which the erythroid Krüppel-like Factor (EKLF) gene is inactivated die in fetal life due to down-regulation of the β-globin gene. Results have suggested that EKLF functions through the proximal CACC motif of the β-globin promoter. For example, natural mutations of this element that fail to bind EKLF give reduced gene expression and the ability of EKLF to activate reporter genes in co-transfection assays is dependent on an intact CACC. Here, removal of the CACC motif and upstream promoter sequences from the β-globin gene resulted in reduced expression in transgenic mice. However, breeding onto an EKLF−/− background demonstrated that a CACC-less β-globin transgene remains highly dependent on EKLF. Hence, although the β-globin gene partly depends on the proximal CACC motif for expression, it is unlikely that the major mechanism of gene activation by EKLF is through this element. We also show that a lacZ reporter gene linked to the β-globin promoter, with or without the CACC box present, is actually expressed higher in EKLF−/− fetuses than in wild type animals, suggesting that EKLF may be able to act as an inhibitor of transcription with certain transgene configurations. locus control region erythroid Krüppel-like factor polymerase chain reaction murine erythroleukemia base pair(s) kilobase pair(s) The human β-globin locus contains five active genes (Fig. 1) that are each expressed only at a specific time during development; the ε-globin gene is transcribed during early embryonic life, the Gγ and Aγ-globin genes during fetal development, and the δ (a minor contributor) and β-globin genes in late fetal stages and throughout adult life (reviewed in Ref. 1.Stamatoyannopoulos G. Neinhuis A.W. Stamatoyannopoulos G. Neinhuis A.W. Majerus P.W. Varmus H. The Molecular Basis of Blood Diseases. W. B. Saunders Co., Philadelphia1994: 107-136Google Scholar). Transcription at high levels of any of these genes requires the β-globin LCR,1 which is a strong, erythroid-specific enhancer located in the 5′ region of the locus (Fig.1) that has the special property of being able to protect globin transgenes from position effects in mice (reviewed in Refs. 2.Dillon N. Grosveld F. Trends Genet. 1993; 9: 134-137Abstract Full Text PDF PubMed Scopus (177) Google Scholar, 3.Felsenfeld G. Gene (Amst.). 1993; 135: 119-124Crossref PubMed Scopus (34) Google Scholar, 4.Higgs D.R. Cell. 1998; 95: 299-302Abstract Full Text Full Text PDF PubMed Scopus (91) Google Scholar). Studies have shown that the developmental regulation of the β-globin locus is controlled, in part, by the fact that the genes must compete with each other to be activated by the LCR (5.Behringer R.R. Ryan T.M. Palmiter R.D. Brinster R.L. Townes T.M. Genes Dev. 1990; 4: 380-389Crossref PubMed Scopus (224) Google Scholar, 6.Enver T. Raich N. Ebens A.J. Papayannopoulou T. Constantini F. Stamatoyannopoulos G. Nature. 1990; 344: 309-313Crossref PubMed Scopus (269) Google Scholar). Results have further suggested that a globin gene must directly interact with the LCR to be enhanced; however, the LCR can only interact with one gene at a time (7.Hancombe O. Whyatt D. Fraser P. Yannoutsos N. Greaves D. Dillon N. Grosveld F. Genes Dev. 1991; 5: 1387-1394Crossref PubMed Scopus (236) Google Scholar, 8.Peterson K.R. Stamatoyannopoulos G. Mol. Cell. Biol. 1993; 13: 4836-4843Crossref PubMed Google Scholar). Therefore, changes in the trans-acting environment of the red cell compartment must take place during development to alter the relative affinity each globin gene has for the LCR and dictate which gene will be expressed (reviewed in Refs. 2.Dillon N. Grosveld F. Trends Genet. 1993; 9: 134-137Abstract Full Text PDF PubMed Scopus (177) Google Scholar, 3.Felsenfeld G. Gene (Amst.). 1993; 135: 119-124Crossref PubMed Scopus (34) Google Scholar, 4.Higgs D.R. Cell. 1998; 95: 299-302Abstract Full Text Full Text PDF PubMed Scopus (91) Google Scholar). One transcription factor that appears to be involved in this developmental competition process is EKLF. Mice in which the EKLF gene has been crippled by targeted mutation die at approximately 15 dayspost coitus from a severe anemia that is caused by a 10–20-fold decrease in expression of the β-globin gene (9.Nuez B. Michalovich D. Bygrave A. Plaemacher R. Grosveld F. Nature. 1995; 375: 316-318Crossref PubMed Scopus (492) Google Scholar, 10.Perkins A.C. Sharpe A.H. Orkin S.H. Nature. 1995; 375: 318-322Crossref PubMed Scopus (540) Google Scholar). The other mouse globin genes are not affected, and transgenic mice carrying human globin genes express the γ-globin genes at higher levels in EKLF−/− animals up until the time the fetuses perish (11.Perkins A.C. Gaensler K.M.L. Orkin S.H. Proc. Natl. Acad. Sci. 1996; 93: 12267-12271Crossref PubMed Scopus (131) Google Scholar,12.Wijgerde M. Gribnau J. Trimborn T. Nuez B. Philipsen S. Grosveld F. Fraser P. Genes Dev. 1996; 10: 2894-2902Crossref PubMed Scopus (184) Google Scholar). Thus, EKLF may play an important role in determining the affinity between the β-globin gene and the LCR and in establishing the ability of the β-globin gene to inhibit expression of the γ-globin genes in later development by competing out their interaction with the LCR. The precise mechanism of EKLF function is not known. However, EKLF binds to the proximal CACC motif of the β-globin gene promoter with high affinity and can activate transcription through this site (13.Miller I.J. Bieker J.J. Mol. Cell. Biol. 1993; 13: 2776-2786Crossref PubMed Scopus (667) Google Scholar, 14.Feng W.C. Southwood C.M. Bieker J.J. J. Biol. Chem. 1994; 269: 1493-1500Abstract Full Text PDF PubMed Google Scholar, 15.Donze D. Townes T.M. Bieker J.J. J. Biol. Chem. 1995; 270: 1955-1959Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar, 16.Donze D. Jeancake P.H. Townes T.M. Blood. 1996; 88: 4051-4057Crossref PubMed Google Scholar). Moreover, natural point mutations of this CACC element, which result in decreased β-globin synthesis in red blood cells, fail to bind EKLF (14.Feng W.C. Southwood C.M. Bieker J.J. J. Biol. Chem. 1994; 269: 1493-1500Abstract Full Text PDF PubMed Google Scholar, 17.Hartzog G.A. Myers R.M. Mol. Cell. Biol. 1993; 13: 44-56Crossref PubMed Google Scholar). On the other hand, EKLF binds only very weakly to the CACC regions in the promoters of the γ-globin genes (15.Donze D. Townes T.M. Bieker J.J. J. Biol. Chem. 1995; 270: 1955-1959Abstract Full Text Full Text PDF PubMed Scopus (185) Google Scholar), a fact that could account for the specificity of β-globin gene down-regulation in EKLF−/− mice. Therefore, it has been widely accepted that EKLF acts through the CACC box of the β-globin gene promoter. Interestingly, at least four other proteins homologous to EKLF have now been discovered and they too bind to CACC sequences (18.Anderson K.P. Kern C.B. Lingrel J.B Mol. Cell. Biol. 1995; 15: 5957-5965Crossref PubMed Scopus (231) Google Scholar, 19.Crossley M. Whitelaw E. Perkins A. Williams G. Fujiwara Y. Orkin S.H. Mol. Cell Biol. 1996; 16: 1695-1705Crossref PubMed Scopus (209) Google Scholar, 20.Matsumoto N. Laub F. Aldabe R. Zhang W. Ramirez F. Yoshida T. Terada M. J. Biol. Chem. 1998; 273: 28229-28237Abstract Full Text Full Text PDF PubMed Scopus (109) Google Scholar, 21.Shields J.M. Yang V.W. Nucleic Acids Res. 1998; 26: 796-802Crossref PubMed Scopus (157) Google Scholar). Moreover, very recently, another EKLF related factor (embyronic/fetal β-like gene-activating krüppel-like factor) that is expressed in red cells and may be a preferential activator of the embryonic and fetal globin genes has been described (22.Asano H. Li X.S. Stamatoyannopoulos G. Mol. Cell. Biol. 1999; 19: 3571-3579Crossref PubMed Google Scholar). Thus, understanding where and how EKLF acts in regulating the β-globin gene should provide insights into how a whole family of transcription factors is implicated in controlling gene expression and in regulating developmental expression of globin genes. Although EKLF can bind to and stimulate transcription from the β-globin CACC box, the β-globin gene does not appear to depend on the CACC box to the same extent that it depends on EKLF. For example, mutations that prevent binding of EKLF to the CACC element only decrease β-globin gene expression 3–5-fold (23.Orkin S.H. Kazazian Jr., H.H. Antonarakis S.E. Geoff S.C. Boehm C.D. Sexton J.P. Waber P.G. Giardina P.J. Nature. 1982; 196: 627-631Crossref Scopus (720) Google Scholar, 24.Orkin S.H. Antonarakis S.E. Kazazian Jr., H.H. J. Biol. Chem. 1984; 259: 8679-8681Abstract Full Text PDF PubMed Google Scholar, 25.Kulozik A.E. Bellan-Koch A. Bail S. Kohne E. Kleihauer E. Blood. 1991; 77: 2054-2058Crossref PubMed Google Scholar). This contrasts with the 10–20-fold decrease in β-globin gene transcription seen in EKLF gene knockout mice (9.Nuez B. Michalovich D. Bygrave A. Plaemacher R. Grosveld F. Nature. 1995; 375: 316-318Crossref PubMed Scopus (492) Google Scholar, 10.Perkins A.C. Sharpe A.H. Orkin S.H. Nature. 1995; 375: 318-322Crossref PubMed Scopus (540) Google Scholar). Moreover, EKLF was unable to stimulate transcription from a γ-globin gene promoter in which the CACC box element in this promoter had been replaced with that from the β-globin gene (26.Asano H. Stamatoyannopoulos G. Mol. Cell. Biol. 1998; 18: 102-109Crossref PubMed Google Scholar). This suggests that the β-globin CACC element might not be the most important site of action of EKLF or, at least, may not be the only element required for EKLF function. Thus, we investigated the relationship between EKLF function in β-globin gene expression and the proximal CACC box element of this gene. The βA and βB constructs, which contain the β-globin gene promoter (up to +32) linked to thelacZ gene followed by SV40 poly(A) addition sequences and cloned into the μ-locus plasmid, which contains a reduced, but fully active form of the LCR, have been described previously (27.Guy L.-G. Kothary R. DeRepentigny Y. Delvoye N. Ellis J. Wall L. EMBO J. 1996; 15: 3713-3721Crossref PubMed Scopus (58) Google Scholar). The deletions of the β-globin promoter in the βG and βH transgenes were made by using PCR to synthesize the appropriate promoter fragments (−103 to +32 wild type sequences for βG and −103 to +32 with mutant 5 sequences (28.Delvoye N.L. Destroismaisons N.M. Wall L.A. Mol. Cell. Biol. 1993; 13: 6969-6983Crossref PubMed Google Scholar) introduced into the CAAT box region for βH), sequencing the fragments to ensure no undesired mutations occurred, and then exchanging them with the promoter fragment in the βB transgene. To construct the βJ transgene, which contains the complete β-globin structural gene, a marked β-globin gene containing the sequence gcaagcttgtccagacaccatg in the 5′ untranslated region, where the underlined sequence is a HindIII site, was made via PCR. A promoter fragment without the CACC box and upstream sequences was synthesized using PCR and was exchanged with the complete promoter in the marked gene using the introduced HindIII site. The modified gene was then cloned into the poly-linker of the μ-locus plasmid. All transgene constructs to be injected into fertilized mouse eggs were cut with SacII, and the appropriate fragment was purified from vector sequences and the tkNeo gene that are part of the μ-locus plasmid. For cell transfection experiments, the constructs were made linear with ScaI. For bred transgenic lines, F1 tail DNA was used to determine transgene copy number and transgene integrity. For transgenic founders, tail DNA was used to test for transgenic status and transgene integrity, while fetal liver DNA was used to determine copy number. Ten μg of genomic DNA were digested with SacI and analyzed by Southern blot, hybridizing with a 2-kb fragment from the 5′ end of the LCR in the μ-locus and then with a 1.3-kbSacI fragment from the dystonia musculorum locus. The EKLF genotype of adult mice was determined by digesting tail DNA withPvuII and hybridizing Southern blots with a 1.3-kb fragment from the Neo gene. The presence of a 1.3-kb hybridization signal indicated the mice were EKLF+/− as the Neo gene is incorporated into the inactivated EKLF locus (10.Perkins A.C. Sharpe A.H. Orkin S.H. Nature. 1995; 375: 318-322Crossref PubMed Scopus (540) Google Scholar). The EKLF genotype of fetuses isolated at 14.5 days of gestation was established as described previously (29.Guy L.-G. Mei Q. Perkins A.C. Orkin S.H. Wall L. Blood. 1998; 91: 2259-2263Crossref PubMed Google Scholar). The original EKLF knockout mice were from Dr. Stuart Orkin's laboratory (10.Perkins A.C. Sharpe A.H. Orkin S.H. Nature. 1995; 375: 318-322Crossref PubMed Scopus (540) Google Scholar), and these as well as all other transgenic lines were maintained on a B6C3F1 background for several generations. Total RNA was isolated from half of a fetal liver using 1 ml of Trizol reagent (Life Technologies, Inc.) as described by the manufacturer. The final pellet was dissolved in 25 μl of H2O, and 1–3 μl was used for each RNase protection assay. The assays were performed as described previously (27.Guy L.-G. Kothary R. DeRepentigny Y. Delvoye N. Ellis J. Wall L. EMBO J. 1996; 15: 3713-3721Crossref PubMed Scopus (58) Google Scholar). The antisense RNA probes used represented a Sau3A toHgaI (−15 to +60 relative to the Cap site) from the mouse β-globin gene, +1 to +323 of the mouse GATA-1 cDNA (28.Delvoye N.L. Destroismaisons N.M. Wall L.A. Mol. Cell. Biol. 1993; 13: 6969-6983Crossref PubMed Google Scholar), and a 199-bp HaeIII fragment (−69 to +131 relative to the Cap site) from the human β-globin gene with a HindIII site introduced into the 5′-untranslated region (see above). The human and mouse β-globin probes were synthesized using [32P]UTP with a specific activity of 100 Ci/mmol, while the GATA-1 probe was synthesized with a specific activity of 400 Ci/mmol. Gels were exposed to film for different periods of time (several hours to several days) and the signals on the autoradiograms were quantified using a Alphaimager 2000 (Alpha Innotech Corp.). MEL cells were transfected with 5 μg of linear plasmid by the lipofection method, and transfected cell populations were selected in medium containing 800 μg/ml G418 as described previously (28.Delvoye N.L. Destroismaisons N.M. Wall L.A. Mol. Cell. Biol. 1993; 13: 6969-6983Crossref PubMed Google Scholar). Each construct was transfected in triplicate. Duplicate samples of each selected population were induced for 3 days in medium containing 2% dimethyl sulfoxide, and freeze-thaw extracts of the induced cells were used to determine β-galactosidase activity as described previously (27.Guy L.-G. Kothary R. DeRepentigny Y. Delvoye N. Ellis J. Wall L. EMBO J. 1996; 15: 3713-3721Crossref PubMed Scopus (58) Google Scholar), while DNA extracts were used to determine transgene copy number as outlined above. Previously, we used transgenic mice to study expression of a lacZ reporter gene linked to the human β-globin gene promoter and in cis with the LCR (27.Guy L.-G. Kothary R. DeRepentigny Y. Delvoye N. Ellis J. Wall L. EMBO J. 1996; 15: 3713-3721Crossref PubMed Scopus (58) Google Scholar). In that investigation, we had looked at a construct in which the promoter extended to −800 as well as one in which the proximal CACC box and all sequences upstream of it had been deleted from the promoter (βA and βB constructs, see Fig. 1). In the present study, we created transgenic mouse lines carrying two additional lacZ reporter constructs (Fig. 1). In the first transgene (βG), the β-globin promoter was deleted upstream from the proximal CACC sequence, while the second transgene (βH) had, in addition to the same promoter deletion, a mutation in the CAAT box that renders this cis-acting element non-functional (28.Delvoye N.L. Destroismaisons N.M. Wall L.A. Mol. Cell. Biol. 1993; 13: 6969-6983Crossref PubMed Google Scholar). To determine whether the transgenes that contained an intact CACC box (βB, βG, and βH) might depend on EKLF for full expression, while the transgene without the CACC (βA) might be independent of EKLF, we crossed one or two high expressing lines obtained for each construct onto an EKLF−/− background. Surprisingly, all four transgenes were actually expressed at higher levels in EKLF−/− animals than in wild type animals (Table I). Note that six of the seven lines studied demonstrated a greater than 2-fold increase in expression on an EKLF−/− background, while one of the lines, βG6, gave only a modest increase. Hence, the effect was not specific to any of the transgenes. Thus, it appeared that EKLF actually represses expression of the lacZ reporter gene linked to the β-globin promoter, and this occurs whether or not the CACC box is present.Table Iβ-Galactosidase expression in EKLF−/− miceLineEKLF−/−EKLF+/−EKLF+/+milliunits/mgβB41900950780βG6260300220βG81802628βH31402824βH41000100140βA226090120βA4890550370β-Galactosidase activity was measured in extracts from 14.5-day fetal livers isolated from animals of the EKLF genotype indicated. The activity was not corrected for transgene copy number. Open table in a new tab β-Galactosidase activity was measured in extracts from 14.5-day fetal livers isolated from animals of the EKLF genotype indicated. The activity was not corrected for transgene copy number. The above observations compelled us to determine whether or not expression of the lacZ transgenes actually depended on the CACC box. Unfortunately, lacZ transgenes demonstrate very strong position effects in transgenic mice, with expression per transgene copy varying greatly (several hundredfold) between different animals carrying the same transgene and with some lines giving no detectable expression at all. We have shown this previously for the βA and βB constructs (27.Guy L.-G. Kothary R. DeRepentigny Y. Delvoye N. Ellis J. Wall L. EMBO J. 1996; 15: 3713-3721Crossref PubMed Scopus (58) Google Scholar), and the same phenomenon occurred with the βG and βH transgenes (data not shown). Therefore, it was not possible to use transgenic mice to make a quantitative comparison of expression levels between the four constructs. However, we have shown that expression of the lacZ reporter gene linked to the β-globin promoter in the μ-locus plasmid gives much less variation in expression per copy in stable transfected MEL cells (27.Guy L.-G. Kothary R. DeRepentigny Y. Delvoye N. Ellis J. Wall L. EMBO J. 1996; 15: 3713-3721Crossref PubMed Scopus (58) Google Scholar). Hence, each of the lacZ constructs was transfected into MEL cells and three distinct cell populations for each construct were assessed for β-galactosidase activity per transgene copy. The results presented in Table II showed that within the experimental variation all four constructs gave approximately the same average activity per copy. Thus, it would appear that, when linked to the β-globin promoter and the μ-locus version of the LCR, thelacZ reporter gene does not depend to a significant extent on the proximal CACC box of the β-globin promoter. Nor does it appear to be highly dependent on the CAAT box or further upstream sequences.Table IIβ-Galactosidase expression in MEL cellsTransgenePopulationβ-GalaEach population was assayed in duplicate. β-Gal, β-galactosidase.Copy no.bThe population with the lowest copy number was assigned a value of 1, and the copy numbers for the other populations were measured relative to it.β-Gal/copymilliunits/mgA166628βAB58168C62152Average (± av. var.):cav. var., average variation.49 (±14)A652793βBB921.355C3401034Average (± av. var.):61 (±22)A4931241βHB336656C270390Average (± av. var.):62 (±18)A3251.4232βGB631.157C190448Average (± av. var.):112 (±80)a Each population was assayed in duplicate. β-Gal, β-galactosidase.b The population with the lowest copy number was assigned a value of 1, and the copy numbers for the other populations were measured relative to it.c av. var., average variation. Open table in a new tab The observation that the lacZ transgenes do not rely on EKLF for expression obviously made it impossible to use such transgenes to assess the relation of the CACC box to EKLF's role in β-globin expression. Therefore, we produced transgenic mice carrying the β-globin gene itself in cis with the LCR and with the proximal CACC box and upstream sequences deleted from the promoter (βJ construct in Fig. 1). In the first instance, transgenic founders were isolated at the 14.5-day fetal stage of development. Of 11 fetuses that were deemed to be transgenic based on Southern blot analysis of body DNA, 3 that were high copy were severely anemic and did not contain sufficient fetal liver tissue for further experimentation. Such an anemia occurs when β-globin transgenes are present in multiple copies, and it has been suggested that a β-globin to α-globin chain imbalance may be the cause (30.Grosveld F. Bloom van Assendelft G. Greaves D.R. Kollias G. Cell. 1987; 51: 975-985Abstract Full Text PDF PubMed Scopus (1540) Google Scholar). However, it has not been ruled out that the cause may be competition by the transgenes themselves for transcription factors that are required for normal erythroid homeostasis. In fact, that βJ transgenes are expressed at a much lower level than the full-length β-globin gene (see below) points to this latter explanation. For the eight non-anemic, βJ transgenic fetuses, transgene copy number was determined by Southern blotting of DNA isolated from fetal liver (Fig. 2 a). The exact copy number was assessed by comparing to μD transgenic mice, which are from a line that carries a single copy of the complete β-globin gene with the promoter extending to −1,300 (31.Ellis J. Tan-Un K.C. Harper A. Michalovich D. Yannoutsos N. Philipsen S. Grosveld F. EMBO J. 1996; 15: 562-568Crossref PubMed Scopus (203) Google Scholar). Expression of the transgenes was measured via RNase protection using the endogenous mouse β-globin mRNA as a loading control (Fig.3). Of the eight βJ transgenic animals analyzed, seven expressed the transgene. The one animal that did not give detectable expression (βJ7) was found to have a deletion in the 3′ end of the transgene (Fig. 2 b), so it was not considered further. Among the other seven animals, the average level of expression per transgene copy was 24% of the endogenous mouse β-globin gene (Table III). In contrast, it has been shown by others that, in the context of the μ-locus, the β-globin gene with an extended promoter is expressed at approximately 100% the level of the endogenous mouse β-major gene in transgenic mice (31.Ellis J. Tan-Un K.C. Harper A. Michalovich D. Yannoutsos N. Philipsen S. Grosveld F. EMBO J. 1996; 15: 562-568Crossref PubMed Scopus (203) Google Scholar,32.Talbot D. Collis P. Antoniou M. Vidal M. Grosveld F. Greaves D. Nature. 1989; 338: 352-355Crossref PubMed Scopus (263) Google Scholar). For example, the single copy μD line that was used as a qualitative control here was expressed at 80% the level of the mouse β-major gene (Table II). Thus, the results demonstrated that the human β-globin gene is approximately 4-fold dependent on the proximal CACC box and upstream sequences of its promoter. We could not assess whether this 4-fold effect was due to deletion of the CACC box and/or further upstream sequences. However, the fact that natural point mutations of the β-globin CACC motif decrease gene expression 3–5-fold (23.Orkin S.H. Kazazian Jr., H.H. Antonarakis S.E. Geoff S.C. Boehm C.D. Sexton J.P. Waber P.G. Giardina P.J. Nature. 1982; 196: 627-631Crossref Scopus (720) Google Scholar, 24.Orkin S.H. Antonarakis S.E. Kazazian Jr., H.H. J. Biol. Chem. 1984; 259: 8679-8681Abstract Full Text PDF PubMed Google Scholar, 25.Kulozik A.E. Bellan-Koch A. Bail S. Kohne E. Kleihauer E. Blood. 1991; 77: 2054-2058Crossref PubMed Google Scholar) and the demonstration that β-globin promoter sequences upstream of the CACC box have no effect on expression when the LCR is present (33.Antoniou M. Grosveld F. Genes Dev. 1990; 4: 1007-1013Crossref PubMed Scopus (80) Google Scholar), suggest that the major effect may be through the CACC box.Figure 3RNase protection analysis of βJ transgenic founders. RNA isolated from 14.5-day fetal liver of the animal indicated on the bottomof each lane was subjected to RNase protection analysis simultaneously with two probes. The mouse β-globin probe detects a 60-bp fragment from the 5′ end of the endogenous mouse β-globin mRNA. The human β-globin probe detects a 131-bp fragment from the 5′ end of the βJ transgene and a 89-bp fragment near the 5′ end of the μD transgene as indicated to the right of the figure. Note, the βJ and μD transgenes give different size fragments in this assay since the βJ transgene was altered to contain a HindIII site in its 5′-untranslated region and the probe was homologous to this transgene. In the case of the μD transgene, the probe is not homologous at theHindIII site; hence, the probe is further cut into two smaller fragments, one of 89 bp, which is present in the figure, and another of 35 bp, which is not seen.View Large Image Figure ViewerDownload Hi-res image Download (PPT)Table IIIExpression of the βJ construct in transgenic miceTransgenic sampleExpressionaExpression was determined by RNase protection (see Fig. 3). The signal for the human transgene was determined as a percentage of mouse β-globin mRNA after correction for U content. ND, not detectable.Copy no.bCopy number was determined by Southern blot (see Fig. 2). Transgene signals were corrected to the dt gene, and absolute copy numbers were determined relative to the single-copy μD line.ExpressioncExpression per copy was corrected for the fact that there are two copies of the endogenous mouse gene. per copy% mouse β-globin%βJ1150.933βJ2281.929βJ31007.826βJ416910.233βJ5708.417βJ6101.612βJ7ND0.6NDβJ8333.917Average:24(±8)μD401.080a Expression was determined by RNase protection (see Fig. 3). The signal for the human transgene was determined as a percentage of mouse β-globin mRNA after correction for U content. ND, not detectable.b Copy number was determined by Southern blot (see Fig. 2). Transgene signals were corrected to the dt gene, and absolute copy numbers were determined relative to the single-copy μD line.c Expression per copy was corrected for the fact that there are two copies of the endogenous mouse gene. Open table in a new tab It was also interesting to note that among the seven expressing βJ transgenic fetuses the level of transgene expression per copy varied less than 3-fold with a coefficient of variation of only 33%. This suggested that removal of the proximal CACC box and upstream sequences from the β-globin promoter did not adversely affect the ability of the LCR to confer position-independent, copy number-dependent expression onto the β-globin gene. Born mice carrying the βJ transgene were obtained only at a low frequency, and several founder transgenic animals failed to transfer the transgene to their offspring. We suspect that this was because the βJ transgene was causing a fatal anemia in higher copy number mice, as was described above. Such an anemia may be less pronounced in founders if they are mosaics. Two established, low copy βJ transgenic lines were crossed onto an EKLF null background. The transgenic status of fetuses isolated at 14.5 days post coitus and their EKLF genotype were determined by Southern blot (data not shown), and fetal liver RNA was assessed for expression of the transgene by RNase protection (Fig. 4). Expression of the transgene and the endogenous mouse β-globin gene were measured relative to GATA-1 mRNA, which is not affected by EKLF inactivation (10.Perkins A.C. Sharpe A.H. Orkin S.H. Nature. 1995; 375: 318-322Crossref PubMed Scopus (540) Google Scholar), and then expression of the transgene in EKLF +/+ and −/− fetuses was determined as a percentage of mouse β-globin gene expression in wild type litter mates. As a qualitative control, the human β-globin transgene in the μD line containing a promoter extending to −1300 was found to be expressed 11-fold lower in EKLF−/− mice (Table III), which is a similar value to what we and others have described previously (29.Guy L.-G. Mei Q. Perkins A.C. Orkin S.H. Wall L. Blood. 1998; 91: 2259-2263Crossref PubMed Google Scholar, 34.Tewari R. Gillemans N. Wijgerde M. Nuez B. von Lindern M. Grosveld F. Philipsen S. EMBO J. 1998; 17: 2334-2341Crossref PubMed Scopus (71) Google Scholar). Expression of the βJ transgene was decreased 13-fold in a two-copy line and 5-fold in a single copy line in EKLF−/− animals (lines 1and 2 in Table IV and Fig. 4, respectively). Thus, the βJ transgene, in which the proximal CACC box and upstream sequences have been deleted from the β-globin promoter, was still highly dependent on EKLF for expression. Therefore, we conclude that the major role of EKLF in β-globin gene expression probably does not involve the proximal CACC box of the β-globin gene promoter.Table IVDependence of the βJ transgene on EKLFTransgeneExpression in EKLF+/+ miceaValues were not corrected for copy number. Three or four EKLF−/− animals of each line were assayed.Expression in EKLF−/− miceaValues were not corrected for copy number. Three or four EKLF−/− animals of each line were assayed.Change +/+ / −/−% mouse β-globin% mouse β-globin-FoldμD423.711 (±4)βJ-line 1332.613 (±2)βJ-line 2193.75 (±1)a Values were not corrected for copy number. Three or four EKLF−/− animals of each line were assayed. Open table in a new tab Although there has been an accumulation of studies to suggest that the most important role for EKLF in β-globin gene expression is played through the CACC box (see Introduction), Asano and Stamatoyannopoulos (26.Asano H. Stamatoyannopoulos G. Mol. Cell. Biol. 1998; 18: 102-109Crossref PubMed Google Scholar) first presented evidence that the CACC box alone is not sufficient for activation by EKLF. These authors demonstrated that, even when the γ-globin promoter was altered to contain the identical CACC sequence to the β-globin promoter and in the same position relative to the Cap site, EKLF was still unable to activate transcription through the γ-globin promoter. They concluded that it might be the overall structure of the β-globin promoter rather than just the CACC box that determines the specificity of activation by EKLF. We have now extended these results by showing that, even when the CACC box is deleted from the promoter, the β-globin gene still remains highly dependent on EKLF. Thus, it appears likely that the major role of EKLF does not involve the CACC element. It should be realized, however, that our results cannot rule out that EKLF has an essential role at the CACC motif. For example, because of the limited number of transgenic lines available for analysis (Table IV), we cannot say whether the βJ construct is as dependent on EKLF as is the μD construct with an extended promoter. Moreover, in previous studies, no expression of the human β-globin gene could be detected in EKLF −/− mice that were transgenic for the complete human β-globin locus (12.Wijgerde M. Gribnau J. Trimborn T. Nuez B. Philipsen S. Grosveld F. Fraser P. Genes Dev. 1996; 10: 2894-2902Crossref PubMed Scopus (184) Google Scholar). On the other hand, when the human β-globin gene is alone with the LCR, as is the case in the μD and βJ transgenic lines studied here, expression is very low, but is still easily detectable in EKLF−/− fetuses (see Fig. 4). Thus, the function of EKLF may be more profound in the context of the entire locus compared with when the β-globin gene is linked by itself to the LCR. Any additional role that EKLF may play, which might be related to competition for LCR interactions from the other globin genes in the locus (11.Perkins A.C. Gaensler K.M.L. Orkin S.H. Proc. Natl. Acad. Sci. 1996; 93: 12267-12271Crossref PubMed Scopus (131) Google Scholar, 12.Wijgerde M. Gribnau J. Trimborn T. Nuez B. Philipsen S. Grosveld F. Fraser P. Genes Dev. 1996; 10: 2894-2902Crossref PubMed Scopus (184) Google Scholar), could be dependent on the β-globin CACC motif. To investigate this possibility would require mutation of the CACC box directly in the locus. In any case, that the β-globin gene with the CACC motif and further downstream sequences removed from the promoter remains highly dependent on EKLF, even in the absence of other competing genes, shows that a major role played by EKLF is independent of the CACC motif. There are several CACC like sequences in the HSS regions of the β-globin LCR and it has been demonstrated that EKLF can act to increase transcription through at least one of these CACC elements in the HSS3 region of the LCR (35.Gillemans N. Tewari R. Lindeboom F. Rottier R. deWit T. Wijgerde M. Grosveld F. Philipsen S. Genes Dev. 1998; 12: 2863-2873Crossref PubMed Scopus (62) Google Scholar). Hence, it is plausible that the major function of EKLF is done at the level of the LCR. This idea is somewhat dampened by the fact that all the globin genes in the locus depend on the LCR for high level expression (36.Epner E. Reik A. Cimbora D. Telling A. Bender M.A. Fiering S. Enver T. Martin D.I.K. Kennedy M. Keller G. Groudine M. Mol. Cell. 1998; 2: 447-455Abstract Full Text Full Text PDF PubMed Scopus (166) Google Scholar), but only the β-globin gene is down-regulated by the absence of EKLF. However, some reports have suggested that different regions of the LCR may have developmental and/or globin gene specificity (37.Fraser P. Pruzina S. Antoniou M. Grosveld F. Genes Dev. 1993; 7: 106-113Crossref PubMed Scopus (201) Google Scholar, 38.Bungert J. Davé U. Lim K.-C. Lieuw K.H. Shavit J.A. Liu Q. Engel J.D. Genes Dev. 1995; 9: 3083-3096Crossref PubMed Scopus (178) Google Scholar). Thus, it may be that EKLF acts in regions of the LCR that are implicated in enhancing expression of the β-globin gene only. In this context, the recent discovery of an EKLF homologue that is expressed in red cells and may have some specificity for fetal globin genes (22.Asano H. Li X.S. Stamatoyannopoulos G. Mol. Cell. Biol. 1999; 19: 3571-3579Crossref PubMed Google Scholar) could have important implications if this factor can also act at the level of the LCR. Another possibility is that EKLF acts somewhere else within the β-globin gene itself. This possibility is highlighted by the fact that when β-globin sequences downstream of the TATA box are replaced by the lacZ gene and SV40 sequences the transgenes no longer show any requirement for EKLF. In fact, they actually appear to be inhibited by EKLF (Table I). Thus, it can be hypothesized that an element within the β-globin structural gene must participate, either directly or indirectly, in the action of EKLF in stimulating β-globin gene transcription. We are presently searching for this element. That EKLF appeared to inhibit expression of the lacZ reporter gene linked to the β-globin promoter and LCR was a surprising result. This had not been observed previously for the lacZ gene linked to a heat shock promoter and in cis with the LCR (34.Tewari R. Gillemans N. Wijgerde M. Nuez B. von Lindern M. Grosveld F. Philipsen S. EMBO J. 1998; 17: 2334-2341Crossref PubMed Scopus (71) Google Scholar), for example. It was also unexpected that expression of the lacZ transgene did not depend on the upstream sequences of the β-globin promoter, including the CACC and CAAT box (Table II). Previously, it had been shown that a H2Kk reporter gene linked to this promoter and in cis with the LCR is highly dependent on both the CACC and CAAT motifs for expression in MEL cells (33.Antoniou M. Grosveld F. Genes Dev. 1990; 4: 1007-1013Crossref PubMed Scopus (80) Google Scholar), as is the β-globin gene itself. Although we can only speculate on the explanation for these results, they suggest that the specific context of a reporter gene will dictate its relationship with regulatory sequences. For instance, we have shown that the lacZ reporter gene linked to the β-globin promoter is not properly protected from position effects by the LCR. On the other hand, the complete β-globin gene with the LCR gives position-independent expression (27.Guy L.-G. Kothary R. DeRepentigny Y. Delvoye N. Ellis J. Wall L. EMBO J. 1996; 15: 3713-3721Crossref PubMed Scopus (58) Google Scholar, 31.Ellis J. Tan-Un K.C. Harper A. Michalovich D. Yannoutsos N. Philipsen S. Grosveld F. EMBO J. 1996; 15: 562-568Crossref PubMed Scopus (203) Google Scholar, 32.Talbot D. Collis P. Antoniou M. Vidal M. Grosveld F. Greaves D. Nature. 1989; 338: 352-355Crossref PubMed Scopus (263) Google Scholar). Hence, the ability of the LCR to modulate a downstream gene is influenced by the structure of the gene itself. The improper interaction between lacZ transgenes and the LCR could relate to the fact that the transgene is inhibited by EKLF and is not dependent on the upstream β-globin promoter. In this respect, it has been shown that, although alacZ transgene linked to the heat shock promoter is not influenced by EKLF when the complete LCR is present, the transgene is highly dependent on EKLF when only HSS3 of the LCR is present (34.Tewari R. Gillemans N. Wijgerde M. Nuez B. von Lindern M. Grosveld F. Philipsen S. EMBO J. 1998; 17: 2334-2341Crossref PubMed Scopus (71) Google Scholar). In our case, it may be that the structure the LCR assumes when EKLF is present is more apathetic to stimulating the β-globin promoter linked to the lacZ gene than the LCR structure formed in the absence of EKLF. In any case, these differences that occur between different reporter genes show that one must be very careful in interpreting data with regards to gene-regulatory sequence interactions when one replaces the natural gene with a reporter gene. We thank Yves DeRepentigny and Michel Ste-Marie for producing the transgenic mice.