Non-erythroid Genes Inserted on Either Side of Human HS-40 Impair the Activation of Its Natural α-Globin Gene Targets without Being Themselves Preferentially Activated*

The human α-globin gene complex includes three functional globin genes (5′-ζ2-α2-α1–3′) regulated by a common positive regulatory element named HS-40 displaying strong erythroid-specific enhancer activity. How this enhancer activity can be shared between different promoters present at different positions in the same complex is poorly understood. To address this question, we used homologous recombination to target the insertion of marker genes driven by cytomegalovirus or long terminal repeat promoters in both possible orientations either upstream or downstream from the HS-40 region into the single human α-globin gene locus present in hybrid mouse erythroleukemia cells. We also used CRE recombinase-mediated cassette exchange to target the insertion of a tagged α-globin gene at the same position downstream from HS-40. All these insertions led to a similar decrease in the HS-40-dependent transcription of downstream human α-globin genes in differentiated cells. Interestingly, this decrease is associated with the strong activation of the proximal newly inserted α-globin gene, whereas in marked contrast, the transcription of the non-erythroid marker genes remains insensitive to HS-40. Taken together, these results indicate that the enhancer activity of HS-40 can be trapped by non-erythroid promoters in both upstream and downstream directions without necessarily leading to their own activation.

Human ␣-globin genes are clustered on a single complex located in the telomeric region of the short arm of chromosome 16. This complex includes three functional genes, the embryonic 2 gene and the two fetal/adult ␣2 and ␣1 genes, which are arranged in the order 5Ј-2-␣2-␣1-3Ј of their expression during development (1). Although the ␣-globin genes are transcribed exclusively in erythroid cells, the whole complex is located in a GC-rich isochore, within an early replicating and constitutively DNase I-sensitive chromatin domain in both erythroid and non-erythroid cells (2)(3)(4). The human ␣-globin gene complex is also surrounded by several widely expressed genes, including the Ϫ14 gene of unknown function, which is located in the opposite transcriptional orientation, 14 kb 1 upstream from the 2 gene (5,6). Several studies have shown that the erythroidspecific transcriptional activation of all ␣-globin genes present in the locus is controlled by a single positive regulatory element, named HS-40, which corresponds to a DNase I-hypersensitive site located 40 kb upstream from the 2 gene (7,8). HS-40 is characterized by a high density of DNA-binding sites for ubiquitous and erythroid-specific transcription factors (8 -10). It is a strong erythroid-specific transcriptional enhancer in cell lines (7,(11)(12)(13)(14), and it confers erythroid lineage-specific, autonomous, and appropriate developmental patterns of expression of either the 2 or ␣-globin promoter in transgenic mice (14 -17). Despite this strong enhancer activity, the expression level of globin genes linked to the HS-40 element in transgenic mice remains sensitive to position effects, is not copy number-dependent, and tends to decrease in adults. Natural or targeted deletions of HS-40 lead to a complete loss of globin gene transcriptional activation in erythroid cells, but do not affect the DNase I sensitivity or the replication timing of the whole complex (18 -21). Furthermore, the expression level of the Ϫ14 gene is independent of HS-40 despite the location of HS-40 in its fifth intron (18). The HS-40 regulatory element thus appears to be involved in the erythroid-specific and selective transcriptional activation of all globin genes belonging to the complex, but the mechanisms responsible for this selective activation are still poorly understood.
One way to approach an understanding of these mechanisms is to investigate how this HS-40-mediated transcriptional activation of human ␣-globin genes can be affected by the insertion of new genes into the complex. In this study, we addressed these questions by using homologous recombination and CRE recombinase-mediated cassette exchange to target the insertion of either an extra ␣-globin gene or a marker gene driven by the non-erythroid promoter immediately upstream or downstream from HS-40 into the single chromosome 16 present in hybrid mouse erythroleukemia cells. We found that all of these insertions led to a similar drastic reduction of the HS-40-dependent transcription of resident ␣-globin genes, regardless of the position, the orientation, or the identity of the newly inserted gene. Although this down-regulation is associated with the strong activation of the newly inserted ␣-globin gene, this is not the case for newly inserted non-erythroid marker genes, the transcription of which appears to be independent of HS-40.
Taken together, these results suggest that the enhancer activity of HS-40 spreads in both upstream and downstream directions and can be trapped by non-erythroid promoters without necessarily leading to their own activation.

Cell Culture
All experiments were performed in the mouse erythroleukemia (MEL) hybrid cell line LT585P3, which contains a single copy of normal human chromosome 16 (18,22). Cells were cultured as already described (18). Erythroid terminal differentiation of the cells was induced by the addition of 5 mM hexamethylenebisacetamide (HMBA; Sigma) to the culture medium for 4 days.

Plasmid Constructions
pLTR-neo Targeting Plasmids-All four plasmids used to target the insertion of the LTR-neo gene are derived from a single starting plasmid, pHS-40. This pHS-40 plasmid is based on pUC18 in which 9.2 kb of isogenic genomic DNA overlapping the HS-40 regulatory region has been cloned. This isogenic DNA was cloned as two cassettes: a 4.1-kb HindIII genomic fragment including HS-40 was directly cloned in pUC18, and the downstream adjacent 5.1-kb HindIII genomic fragment was cloned as a SalI-XhoI fragment to leave a unique SalI site for inserting the LTR-neo gene between the two isogenic cassettes. In addition, pHS-40 contains a herpes simplex virus thymidine kinase XhoI-HindIII gene cassette, derived from a pIC19R/MCI-tk plasmid (23), which was cloned immediately downstream from the 3Ј-homology arm. The LTR-neo gene derives from the pGEM-I-FLP-neo plasmid (24). It is driven by the enhancer/promoter of the Friend retrovirus long terminal repeat and is flanked on both sides by FLP recombinase targets (24). The pHS-neoS targeting plasmid was thus obtained by subcloning the LTR-neo gene (taken as a SalI-XhoI fragment) into the single SalI site of pHS-40 and in the same transcriptional orientation as that of resident ␣-globin genes. The pHS-neoAS targeting plasmid was obtained similarly by subcloning the LTR-neo gene cassette in reverse orientation. The other two plasmids used to target the insertion of the LTR-neo gene upstream from HS-40 were obtained from the same pHS-40 starting plasmid in which a 1.1-kb HpaI fragment including the HS-40 site was first removed and replaced by the LTR-neo gene (taken as a HincII fragment) in either orientation. The HpaI fragment containing HS-40 was then reinserted in the correct orientation into the single SalI site remaining downstream from the LTR-neo gene. The pneoS-HS targeting plasmid thus contains the LTR-neo gene in the same transcriptional orientation as that of the resident ␣-globin genes, whereas the pneoAS-HS targeting plasmid contains the LTR-neo gene in reverse orientation.
pCMV-hygroTK Targeting Plasmid-The CMV-hygroTK gene, which is a hygroTK gene fusion driven by the cytomegalovirus promoter, was first isolated as a XhoI fragment from a plasmid kindly provided by Dr. P. Greenberg (Fred Hutchinson Research Center) and cloned using BamHI linkers between two inverted Lox sequences L1 and 1L (25,26). The resulting L1-CMV-hygroTK-1L cassette isolated as a XhoI-PvuII fragment was then subcloned into the single SalI site of plasmid pHS-40 in the same transcriptional orientation as that of resident ␣-globin genes, thus leading to the targeting plasmid pCMV-hygroTK. All targeting plasmids were linearized at the unique ScaI site present in the pUC18 sequence before transfection.
p␣ T -globin Gene Exchange Plasmid-The human ␣ T -globin gene was previously cloned from the genomic DNA of an ␣ ϩ -thalassemic patient homozygous for the rightward 3.7-kb deletion that generates an ␣2/␣1 fusion gene (27). This ␣ T -globin gene carries a two-nucleotide deletion at positions Ϫ2 and Ϫ3 preceding the ATG initiation codon, which is responsible for reduced translation efficiency, but which does not affect the transcription of the gene (28). The ␣ T -globin gene cassette was taken as a 1.5-kb PstI fragment and cloned using BamHI linkers between two inverted Lox sequences L1 and 1L, thus generating the p␣ T gene exchange plasmid.

Isolation of Homologous Recombinant Clones
Hybrid MEL cells (10 7 ) were transfected by electroporation using 20 g of each linearized targeting plasmid as described previously (18). Twenty-four hours after electroporation, surviving cells were plated in selective medium containing 0.6 mg/ml G418 (Life Technologies, Inc.) and 10 M ganciclovir (Syntex Research Co.) for cells transfected with the pLTR-neo targeting plasmids or containing 1 mg/ml hygromycin (Life Technologies, Inc.) for cells transfected with the pCMV-hygroTK targeting plasmid. After 15 days, individual resistant clones were analyzed by Southern blotting using an HS-40 probe to identify homologous recombinant clones.

Deletion of the Selectable Marker Gene by FLP Recombinase
A yeast FLP recombinase expression vector (pHook-3-FLP) was obtained by subcloning a XbaI fragment containing the FLP coding sequence driven by a cytomegalovirus promoter and derived from vector pCFIZ (24) in the polylinker of the pHook-3 plasmid (Invitrogen), which itself contains a gene encoding Zeocin resistance. Cells (10 7 ) from clones harboring insertion of the LTR-neo gene were transfected by electroporation using 30 g of pHook-3-FLP DNA. After 24 h, cells were placed in selective medium (300 g/ml Zeocin; CAYLA). Among the Zeocinresistant clones, G418-sensitive clones were analyzed by Southern blotting to verify excision of the LTR-neo gene.

Recombinase-mediated Cassette Inversion or Exchange
Recombinase-mediated inversion of the CMV-hygroTK gene was obtained through the transient expression of CRE recombinase. For this purpose, 10 6 cells of a clone harboring targeted insertion of the L1-CMV-hygroTK-1L cassette were cotransfected using DAC-30 (Eurogentec), 1 g of expression plasmid DNA encoding the CRE recombinase (pCMV-CRE), and 1 g of expression plasmid DNA encoding green fluorescent protein (pSV40-GFP). Forty-eight hours following transfection, cells expressing green fluorescent protein were purified by fluorescence-activated cell sorting and recloned in the presence of hygromycin. Isolated clones were then amplified and analyzed by Southern blotting. Recombinase-mediated exchange of the hygroTK gene by the ␣ T -globin gene was obtained similarly by cotransfecting 10 6 cells of a clone harboring targeted insertion of the CMV-hygroTK gene using 1 g of pCMV-CRE, 1 g of pSV40-GFP, and 2 g of p␣ T plasmid carrying the tagged human ␣-globin gene flanked by two inverted Lox sequences. Purified green fluorescent protein-positive cells were then cloned in medium containing 10 M ganciclovir. Individual ganciclovir-resistant clones were amplified and analyzed by Southern blotting to verify the insertion of the ␣ T -globin gene.

RNase Protection Assays
Total cellular RNA was prepared using RNA-plus TM (Quantum Biotechnologies) according to the manufacturer's instructions. RNase protection assays were performed as described previously (18) using 8 g of total RNA and the following labeled antisense RNA probes: (i) a human riboprobe (18) that is expected to produce a single protected fragment of 133 nucleotides with normal ␣-globin mRNA and two protected fragments of 97 and 34 nucleotides with ␣ T -globin mRNA carrying a deletion of two nucleotides at positions Ϫ2 and Ϫ3 preceding the AUG initiation codon; (ii) a mouse ␣-globin riboprobe that includes 180 nucleotides complementary to the 3Ј-end of the first exon of the mouse ␣-globin gene and that gives a protected fragment of 75 nucleotides with mouse ␣-globin mRNA; (iii) a neo riboprobe (pT3TKN) that is expected to produce a protected fragment of 260 nucleotides with transcripts of the LTR-neo gene (24); (iv) a hygromycin riboprobe (pT7HY) that contains the BamHI-EcoRI fragment including the hygromycin coding sequence from the L1-CMV-hygroTK-1L gene and that is expected to produce a protected fragment of 258 nucleotides with transcripts of the CMV-hygroTK gene. Radioactive signals corresponding to each specific protected fragment were quantified using a GS-525 Molecular Imager (Bio-Rad) and Molecular Analyst software (Bio-Rad).

Nuclear Run-on Assays
Cells (10 8 ) from a 2-day culture in the presence of 5 mM HMBA were harvested by centrifugation, washed with phosphate-buffered saline, and lysed for 5 min on ice in buffer A (10 mM Tris, pH 7.5, 10 mM NaCl, and 2.5 mM MgCl 2 ) containing 0.3% Nonidet P-40. Nuclei were pelleted by centrifugation through a 30% sucrose cushion made in buffer A, resuspended in glycerol buffer (50 mM Tris, pH 7.9, 75 mM NaCl, 0.1 mM EDTA, 50% glycerol, 0.5 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride), frozen, and stored in liquid nitrogen in 100-l aliquots containing 5 ϫ 10 7 nuclei. Nuclei were thawed on ice by adding an equal volume of transcription buffer (10 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 10 mM MnCl 2 , and 300 mM KCl) supplemented with 5 mM each ATP, GTP and CTP; 100 Ci of [␣-32 P]UTP (3000 Ci/mM; Amersham Pharmacia Biotech), and 100 units/ml RNasin (Promega). Transcription reactions were carried out at 30°C for 15 min and terminated by centrifugation for 30 s. Labeled nuclei were resuspended in 500 l of 10 mM Tris, pH 7.5, 0.5 M NaCl, and 10 mM MgCl 2 containing 40 units of RNase-free DNase (Roche Molecular Biochemicals) and incubated at 30°C for 15 min. Deproteinization was performed for 30 min at 37°C after the addition of proteinase K (500 g/ml) and SDS (0.5%). Labeled RNA was extracted with phenol/chloroform, ethanol-precipitated in the presence of 2 M ammonium acetate, and incubated for 15 min at 37°C in DNase buffer (10 mM Tris, pH 7.5, and 10 mM MgCl 2 ) containing 10 units of RNase-free DNase. RNA was purified again by phenol/chloroform extraction and two rounds of ethanol precipitation, resuspended in water, and hybridized to membranes (Hybond-C Extra, Amersham Pharmacia Biotech) loaded with unlabeled DNA probes. Membranes were loaded using a slot-blot apparatus with a 5 g of DNA/slot concentrations of the following denatured DNA probes: the pMC1neo plasmid (23), mouse ␤-major globin 5-kb EcoRI fragment, and pGEMT plasmid. Membranes were prehybridized for 4 h at 42°C in 50% formamide, 6ϫ saline/sodium phosphate/EDTA, 5ϫ Denhardt's solution, 0.1% SDS, and 20 g/ml yeast tRNA and hybridized overnight with labeled RNA under the same conditions. Membranes were washed for 15 min in 1ϫ saline/sodium phosphate/EDTA and 0.1% SDS at room temperature and for 5 min in 0.1ϫ saline/sodium phosphate/EDTA and 0.1% SDS at 65°C. Hybridization signals were revealed by autoradiography and quantified using the Molecular Imager and Molecular Analyst software.

Targeted Insertion of an LTR-neo Gene Near the HS-40 Regulatory Region of the Human ␣-Globin
Complex-Four different DNA constructs were designed to target the insertion of an LTR-neo gene on either side of HS-40 and in both possible orientations by homologous recombination (Fig. 1A). Each of these DNA constructs was introduced by electroporation into hybrid MEL cells carrying a single copy of human chromosome 16. G418-and ganciclovir-resistant clones obtained were analyzed individually by Southern blotting using a human HS-40 hybridization probe. As schematically presented in Fig. 1A, the targeted integration of the LTR-neo gene was expected to lead to the conversion of the normal 20-kb BamHI fragment to shorter BamHI fragments of 11.3, 12.3, 9, or 14.7 kb depending on the transfected targeting DNA construct. Five correctly targeted clones were identified among 301 clones obtained after transfection with the pHS-neoS construct (Fig. 1B, lanes 3-7), and one correctly targeted clone was identified among 300, 150, or 225 clones obtained after transfection with the pneoS-HS (lane 8), pHS-neoAS (lane 9), or pneoAS-HS (lane 10) construct. Further Southern blot analyses using other restriction endonucleases and either an HS-40 or a human ␣-globin probe confirmed that all eight clones displayed the expected targeted insertions (data not shown).
Targeted Integration of the LTR-neo Gene Near HS-40 Leads to a Decrease in the HS-40-dependent Transcription of Downstream ␣-Globin Genes-The eight targeted clones described above as well as parental cells and the previously described clone containing a targeted replacement of HS-40 by the same LTR-neo gene (18) were grown for 4 days in the presence or absence of HMBA, a chemical inducer of differentiation. Equal amounts of total cellular RNA prepared from induced and uninduced cells were then analyzed by RNase protection assay using a mixture of probes allowing the specific detection of human and mouse ␣-globin and neo gene transcripts. Typical results obtained from three different experiments are shown in Fig. 2A.
These data indicate that the eight clones harboring a targeted insertion of the LTR-neo gene in the vicinity of HS-40 displayed a marked reduction of the HMBA-induced increase in the expression of human ␣-globin genes ( Fig. 2A, compare lanes 13-20 with lane 12). Quantitative analysis revealed that the levels of human ␣-globin mRNA in uninduced cells did not significantly differ between parental cells and cells harboring insertions of the LTR-neo gene either adjacent to or in place of HS-40 (Fig. 2B). In contrast, the levels of human ␣-globin mRNA in induced cells were markedly reduced in all cells harboring insertions of the LTR-neo gene compared with pa-rental cells (Fig. 2C). Interestingly, the levels of human ␣-globin mRNA in the eight clones harboring insertions of the LTRneo gene in the vicinity of HS-40 still remained 5-25-fold above the background level observed in clone ⌬HS-neo (Fig. 2C, compare bars 2-9 with bar 10), in which the LTR-neo gene has been inserted in place of HS-40 (18). These levels correspond to a 3-8-fold reduction of the human ␣-globin mRNA level observed in induced parental cells compared with a Ͼ60-fold reduction in clone ⌬HS-neo.
Since only one example of clones neoS-HS, HS-neoAS, and neoAS-HS could be analyzed, the possibility remained that the reduction of the levels of human ␣-globin mRNA in these clones was due to clonal variation rather than the direct effect of the inserted LTR-neo gene. To exclude this possibility, we verified that excision of the LTR-neo gene in these three clones was indeed able to rescue a level of human ␣-globin mRNA in induced cells similar to that in parental cells. For this purpose, each clone was transfected with an expression vector encoding FLP recombinase (24); and in each case, one G418-sensitive clone, potentially lacking the LTR-neo gene, was then selected for further analyses. Southern blot analysis using an HS-40 probe revealed the presence of a 20-kb BamHI fragment identical to the fragment present in parental cells (Fig. 3A), thus demonstrating the excision of the LTR-neo gene. As expected, RNase protection analysis using human and mouse ␣-globin probes performed in induced cells revealed that all three clones that lost the LTR-neo gene recovered a level of human ␣-globin mRNA similar to that observed in parental cells (Fig. 3, B and C).
Taken together, these data establish that insertions of an LTR-neo gene near the HS-40 regulatory region of the human ␣-globin gene complex markedly reduce, but do not abolish, the HS-40-mediated transcriptional activation of the downstream human ␣-globin genes in hybrid MEL cells. This negative effect on the transcription of downstream ␣-globin genes appears to occur whether the LTR-neo gene is upstream or downstream from HS-40 and regardless of the chromosomal orientation of the inserted LTR-neo gene.
Transcriptional Activity of the LTR-neo Gene Inserted into the Human ␣-Globin Gene Locus Is Independent of HS-40 -One possible explanation for the above observations could be that HS-40 preferentially activates the LTR-neo gene inserted into the human ␣-globin locus at the expense of downstream human ␣-globin genes. According to this hypothesis, transcriptional activity of the LTR-neo gene should be higher in induced cells harboring the inserted LTR-neo gene in the vicinity of HS-40 than in induced cells harboring the same LTR-neo gene in place of HS-40. Unexpectedly, both types of induced cells displayed similar levels of neo gene transcripts (Fig. 2E, compare bars 2-9 with bar 10). However, these levels of neo gene transcripts might not reflect the real transcriptional activities of the genes due to the eventual saturation of the degradation process, which is known to affect selectively non-erythroid gene transcripts during the terminal differentiation of MEL cells (29). We therefore decided to use a nuclear run-on assay to compare more directly the transcriptional activities of the LTRneo gene in induced cells from clone ⌬neo-HS as well as from one example of clones harboring the LTR-neo gene in the four different positions and orientations with respect to HS-40. Induced parental cells were used as a negative control. Briefly, nuclei were prepared from each type of cell and incubated in vitro in the presence of labeled UTP, and labeled nuclear transcripts were hybridized to membranes loaded with neo, mouse ␤-globin, and empty vector DNA probes (Fig. 4A). Hybridization signals were quantified, and the neo gene signals were standardized to the ␤-globin gene signals to allow the direct comparison of the transcriptional activities of the LTR-neo gene between the different types of cells (Fig. 4B). As estimated by this assay, the maximum difference in the transcriptional activities of the LTR-neo gene between the five analyzed clones was 3-fold. The transcriptional activities of the LTR-neo gene in the four clones harboring insertions near HS-40 were alternatively higher, as in clones HS-neoS and neoAS-HS (Fig. 4B, bars 2 and 4), or lower, as in clones neoS-HS and HS-neoAS (Fig. 4B, bars 3 and 5), than that in the clone harboring the insertion in place of HS-40 (bar 6). Furthermore, the variations in the transcriptional activities of the neo gene estimated in the different clones harboring the insertions near HS-40 did not correlate with the variations in the reduction of human ␣-globin gene expression. Taken together, these data indicate that the transcriptional activity of the LTR-neo gene inserted into the human ␣-globin gene is not affected by induction of differentiation and therefore does not apparently benefit from HS-40-mediated activation. This unexpected result led us to investigate the effect of the insertion of a new ␣-globin gene, instead of the LTR-neo gene, at the same downstream position near HS-40.
Targeted Insertion of a Tagged Human ␣-Globin Gene Near HS-40 Using Recombinase-mediated Gene Cassette Exchange-In a previous study, we identified a two-nucleotide deletion at positions Ϫ2 and Ϫ3 preceding the ATG initiation codon in one human ␣ ϩ -thalassemic gene (27). We have shown that this deletion is responsible for a 2-fold reduction of mRNA translation efficiency, but does not affect the transcriptional activity of the gene (28). We therefore decided to use this human ␣ T -globin gene as a marked ␣-globin gene because its transcript can be easily distinguished from the transcripts of normal human ␣-globin genes using the appropriate antisense RNA probe. To introduce this tagged ␣ T -globin gene downstream from HS-40, we used the strategy recently described as recombinase-mediated gene cassette exchange with inverted Lox sites (25,26). The first step of this strategy consists of using classical homologous recombination to introduce into the human ␣-globin gene locus a fusion gene cassette (CMV-hy-groTK) encoding hygromycin resistance and ganciclovir sensitivity and flanked by two Lox sequences in inverted orientations (L1 and 1L). The second step consists of realizing the exchange of the inserted L1-CMV-hygroTK-1L cassette by the tagged ␣ T -globin gene using CRE recombinase-mediated recombination and ganciclovir selection (25).
Among 140 hygromycin-resistant clones analyzed in the first step, two clones, HYTK-S1 and HYTK-S2, showed the targeted replacement of a normal 20-kb BamHI fragment with a 9.5-kb fragment as expected for targeted integration of the CMV-hygroTK gene immediately downstream from HS-40 (Fig. 5, A  and B, lanes 2 and 3). Further Southern blot analyses using other restriction endonucleases confirmed the correct integration of the CMV-hygroTK gene in these two clones (data not shown). Clone HYTK-S1 was further used to derive new clones harboring the CMV-hygroTK gene in the opposite orientation through the transient expression of CRE recombinase (see "Experimental Procedures"). Forty-five hygromycin-resistant clones were then analyzed individually by Southern blotting using an HS-40 probe and BglII digestion to check the orientation of the CMV-hygroTK gene (Fig. 5C). Four of them (Fig.  5C, lanes 3-6) were found to display the expected 8.3-kb BglII fragment instead of the initial 10. In the second step, cells from clone HYTK-S1 were cotransfected by plasmids p␣ T and pCMV-CRE to exchange the CMV-hygroTK gene with the tagged ␣ T -globin gene through the transient expression of the CRE recombinase (see "Experimental Procedures"). Ganciclovir-resistant clones were analyzed individually by Southern blotting after BglII digestion using HS-40 or ␣-globin probes to identify those clones harboring insertion of the ␣ T -globin gene and to check its orientation (Fig.  6A). Among 24 ganciclovir-resistant clones analyzed, five clones gave the expected pattern indicating that the ␣ T -globin gene has been correctly inserted (Fig. 6B). Two of them gave the expected 9.4-kb BglII fragment revealed by both the HS-40 and ␣-globin probes (Fig. 6B, lanes 3 and 7), thus indicating the insertion of the ␣ T -globin gene in the same orientation as that of downstream human ␣-globin genes. The three other clones gave the expected 7.9-and 2-kb BglII fragments revealed by HS-40 and ␣-globin probes, respectively, thus proving the insertion of the ␣ T -globin gene in the opposite orientation (Fig.  6B, lanes 4 -6). Further experiments revealed that the remaining ganciclovir-resistant clones analyzed had lost the human chromosome 16 (data not shown).
Targeted Insertion of the CMV-hygroTK Gene Near HS-40 Leads to a Decrease in the HS-40-mediated Transcriptional Activation of Downstream ␣-Globin Genes-Typical results obtained by RNase protection assays of the CMV-hygroTK, human ␣-globin, and mouse ␣-globin gene transcripts expressed in several clones harboring insertions of the CMV-hygroTK gene are shown in Fig. 7A. As described above, in clones har- ; lane 2, initial clone HYTK-S1; lanes 3-6, four independent clones derived from the initial clone HYTK-S1 following its transfection with an expression vector encoding CRE recombinase. boring insertions of the LTR-neo gene, the most striking observation was a marked decrease in the human ␣-globin mRNA levels in differentiated cells harboring insertions of the CMV-hygroTK gene compared with those in differentiated parental cells (Fig. 7A, compare lane 14 with lanes 2, 4, 6, 8, 10, and 12). Furthermore, the levels of hygromycin mRNA remained very low and poorly affected by the differentiation state of the cells compared with the highly inducible and very high levels of human ␣-globin mRNA in differentiated parental cells. Quantitative analyses of these data (Fig. 7B) revealed that the decrease in the human ␣-globin mRNA levels in differentiated cells corresponded to a 5-20-fold reduction of the levels in differentiated parental cells, with no significant difference depending on the orientation of the inserted CMV-hygroTK gene. Taken together, these data indicate that insertions of the CMV-hygroTK gene immediately downstream from HS-40 lead to an orientation-independent decrease in the HS-40-mediated transcriptional activation of downstream human ␣-globin genes similar to that induced by the insertion of the LTR-neo gene at the same position.
Decrease in the Transcriptional Activation of Downstream ␣-Globin Genes Is Associated with the Strong Transcriptional Activation of the Tagged Globin Gene Inserted Near HS-40 -Results obtained by RNase protection assays of human and mouse ␣-globin gene transcripts in five clones harboring insertions of the ␣ T -globin gene are shown in Fig. 8A. Due to the two-nucleotide deletion in the tagged human ␣ T -globin gene, ␣ T -globin mRNA could be identified by two protected fragments of 97 and 34 instead of the single 133-nucleotide protected fragment corresponding to normal human ␣-globin mRNA. As described above, differentiated cells from all clones harboring insertions of the tagged human ␣ T -globin genes were characterized by a marked 5-10-fold reduction of the normal human ␣-globin mRNA levels compared with those in differentiated parental cells (Fig. 8A, compare lanes 4, 6, 8, 10, 12 with lane 2). However, in marked contrast to observations with the LTRneo and CMV-hygroTK genes inserted at the same position, the inserted human ␣ T -globin gene was characterized by highly inducible and very high levels of expression in differentiated cells. These high levels of ␣ T -globin mRNA in differentiated cells appear to be independent of the orientation of the inserted ␣ T -globin gene (Fig. 8B, compare bars 2 and 3 with bars 4 -6) and ranged from 30 to 80% of the levels of human ␣-globin mRNA in differentiated parental cells. DISCUSSION We have analyzed the expression of human ␣-globin genes in several clones of hybrid MEL cells carrying a single human chromosome 16 in which we have inserted an LTR-neo gene, a CMV-HYTK gene, or a new ␣-globin gene in the vicinity of HS-40. Our results show that, compared with parental cells, all these clones displayed a similar and drastic reduction of the HMBA-induced transcription of the human ␣-globin genes (Fig.  9). More important, this drastic reduction of human ␣-globin gene transcription was undoubtedly induced by the newly inserted genes and cannot be explained by clonal variations in the extent of cell differentiation. Indeed, the reduction of human ␣-globin gene transcription was invariably observed in all of the independent clones carrying insertions and was completely reversed after excision of the newly inserted gene. Furthermore, the reduction of human ␣-globin gene transcription was observed only in HMBA-treated cells, but not in untreated cells. Since we have previously shown that the increased transcription of human ␣-globin genes in HMBA-treated cells is strictly dependent on HS-40 (18), we conclude that all the insertions specifically impair the HS-40-mediated transcriptional activation of human ␣-globin genes.
This study thus confirms a common phenomenon already observed in numerous examples of other loci in which insertions of new genes have been shown to reduce or even suppress the enhancer-dependent transcriptional activation of resident genes (reviewed in Ref. 30). Our finding that the same negative effect is observed regardless of the transcriptional orientation of the inserted gene excludes the possibility that this negative effect is due to a simple transcriptional interference mecha- FIG. 7. Analysis of human and mouse ␣-globin and CMV-hy-groTK gene expression in clones harboring targeted insertions of the CMV-hygroTK gene. Parental hybrid MEL cells (wild-type (WT)) and each of the indicated clones were grown for 4 days in the presence or absence of HMBA. Total RNA was prepared, and equal amounts from each lot of cells were analyzed by RNase protection assay using a mixture of mouse and human ␣-globin and hygroTK antisense RNA probes. Protected fragments were separated by electrophoresis on denaturing polyacrylamide gel, visualized by autoradiography, and quantified using a Molecular Imager. A, autoradiogram of the gel. The positions and lengths of specific protected fragments are indicated on the left. B, quantitative analysis of the data presented in A for differentiated cells. Results are expressed as the ratios of human to mouse ␣-globin signals (black bars) and hygromycin to mouse ␣-globin signals (hatched bars).
nism. Most important, this study presents the first demonstration that such an enhancer-blocking effect can be induced by proximal genes located on either side of HS-40 and independently on their own transcriptional activation.
One of the most trivial explanations of our data could be that the HS-40 enhancer activity is unspecifically affected by the insertion of any transcription unit in its immediate proximity. However, we do not favor this possibility given the strong activation of the newly inserted ␣-globin gene, which clearly indicates that most (if not all) of the enhancer activity of HS-40 is conserved even in the presence of the inserted gene. The strong activation of the newly inserted ␣-globin gene suggests that its negative effect on the transcription of downstream ␣-globin genes might result from its preferential activation due to its closer position to HS-40, as has been shown in similar experiments performed in the ␤-globin gene complex (31)(32)(33)(34)(35)(36)(37). However, although certain sequences of the ␤-LCR could also function as enhancer(s), the latter element definitely possesses chromatin-remodeling functions not shared by HS-40, and for that reason, direct comparison of data collected on both type of complexes must be considered with caution.
A previous study performed in the mouse ␣-globin gene complex has shown that a PGK-driven neo gene induces a stronger down-regulation of resident ␣-globin genes when it is inserted closer 3Ј to the HS-40 mouse homologue than when it is inserted farther (38). We have shown that the insertion of another non-erythroid gene downstream from resident ␣-globin genes did not affect their HS-40-mediated transcription, although in this case, the inserted non-erythroid gene was itself activated by HS-40 (22). Taken together with those results reported in this study, all these data tend to suggest that the reduction of the HS-40-mediated transcription of resident ␣-globin genes is dependent on the distance of the newly inserted gene from HS-40. Unfortunately, the expression levels of the PGK-driven neo gene inserted at two different positions into the mouse ␣-globin gene complex have not been compared, and it therefore remains unknown whether the different effects observed on the expression of the resident mouse ␣-globin genes are associated with different expression levels of the inserted gene (38). However, we found in this study that either non-erythroid genes or a new ␣-globin gene inserted at the FIG. 8. Analysis of mouse ␣-globin, endogenous human ␣-globin, and human ␣ T -globin gene expression in clones harboring targeted insertion of the tagged ␣ T -globin gene. Parental hybrid MEL cells (wild-type (WT)) and each of the indicated clones were grown for 4 days in the presence or absence of HMBA. Total RNA was prepared, and equal amounts from each lot of cells were analyzed by RNase protection assay using a mixture of mouse and human ␣-globin antisense RNA probes. Protected fragments were separated by electrophoresis on denaturing polyacrylamide gel, visualized by autoradiography, and quantified using a Molecular Imager. A, autoradiogram of the gel. The positions and lengths of specific protected fragments are indicated on the right. B, quantitative analysis of the data presented in A for differentiated cells. Results are expressed as the ratios of endogenous human to mouse ␣-globin signals (black bars) and human ␣ Tglobin to mouse ␣-globin signals (hatched bars). nt, nucleotides. The mean expression levels of ␣-globin genes corresponding to each insertion are given as the percentage of the level determined in parental cells. Contrary to the strong activation of the newly inserted ␣-globin gene, the transcription of the newly inserted LTR-neo and CMV-hy-groTK genes is independent of HS-40. same position downstream from HS-40 induces the same negative effect of the HS-40-mediated transcription of downstream resident ␣-globin genes whatever their own activation by HS-40. Thus, one of the new findings of this study is that this reduction can be clearly uncoupled from the transcriptional activation of the inserted gene by HS-40. Intriguingly, these data seem to contradict the current knowledge that HS-40 regulates the transcription of the humanand ␣-globin genes in an autonomous manner. However, recent studies have shown that HS-40 binds a different set of transcription factors during development (17,39). Furthermore, a single-point mutation modifying the nature of transcription factors bound to HS-40 has been shown to induce the derepression of -globin gene transcription in an adult erythroid cell context (17,39). This strongly suggests that humanand ␣-globin gene transcription, while being activated by the same cis-HS-40 element, is dependent on the binding of two different sets of transcription factors to HS-40. This, in turn, does not contradict the result of this study showing a functional interference between transcription units located in the same ␣-globin complex in the same cellular context and thus with the same set of factors bound to HS-40. All these data further suggest that the functional interference between HS-40 and promoters located in the complex is mainly dependent on the combination of transcription factors loaded on HS-40 and the promoters. The precise underlying mechanism responsible for such functional interference still remains to be established. Among models already suggested by others (40 -45), one attractive possibility could be that physical interactions occur between transcription factor complexes loaded on HS-40 and promoters through DNA looping. According to this hypothesis, the negative effect of nonerythroid genes evidenced in this study could be explained by competitive but sterile interactions of their promoters with HS-40. This model is compatible with the observation that the same negative effect of non-erythroid genes can be induced even when inserted upstream from HS-40.
Whatever the underlying mechanism, the interesting perspective suggested by our data is that it should be possible to identify different promoter sequences that are involved in trapping the enhancer activity of HS-40 and that are necessary to achieve efficient transcriptional activation by HS-40. We believe that the cellular clones and recombinase-mediated cassette exchange strategy described here should be useful tools to identify these two types of sequences for a better understanding of how HS-40 works.