Dynamic posttranscriptional regulation of ϵ-globin gene expression in vivo

Human β-like globins are encoded by 5 homologous genes (5′-ϵ-^Gγ-^Aγ-δ-β-3′) arranged in the order of their developmental expression. More than 200 gene mutations are known to adversely affect either the level or the function of the principal adult (β) globin, resulting in significant morbidity and mortality worldwide.¹  For example, individuals with severe deficits in β-globin expression (β thalassemia) exhibit profound anemia, hepatosplenomegaly, and delays in growth and development,^2,3  whereas homozygotes for a single-nucleotide mutation resulting in a Glu→Val substitution at β-globin codon 6 (sickle cell disease) display chronic hemolytic anemia and severe multiorgan damage.⁴  Although many of the phenotypical consequences of these gene defects can be mitigated by periodic erythrocyte transfusions or allogeneic bone marrow transplantation, neither therapy is universally available and both are attended by significant risk.^5-7  Another therapeutic strategy pharmacologically derepresses developmentally silenced β-like globin genes in terminally differentiating adult erythroid cells using hydroxyurea or any of several short-chain fatty acid derivatives. This approach can sometimes reactivate fetal γ-globin expression to levels that are sufficiently high to ameliorate the sickle phenotype^8-10  but are typically too low to benefit individuals with β thalassemia.¹¹  Importantly, the teratogenic, carcinogenic, and developmental risks of long-term therapy with these agents are still largely undefined. Despite important progress, then, safe and effective therapies for β-globin gene disorders are still urgently needed.

The β-globin gene cluster contains another developmentally silenced gene, ϵ-globin, whose expression is ordinarily limited to primitive nucleated erythroblasts in the blood islands of the embryonic yolk sac.^2,3  Like the fetal γ-globin genes, the embryonic ϵ-globin gene remains physically intact in definitive erythroid progenitors and is therefore available for therapeutic derepression. The potential value of this approach for β thalassemia and sickle cell disease has been demonstrated by proof-of-principle studies carried out in vivo in disease-specific mouse models. In one study, enforced expression of human ϵ-globin protein in definitive erythrocytes restored viability to transgenic animals with homozygous, embryonic-lethal inactivation of their endogenous adult β-globin genes, demonstrating the physiologic neutrality of an ϵ-for-β substitution in heterotetrameric hemoglobin.¹²  A subsequent study demonstrated that coexpressed human ϵ-globin significantly improved the phenotypes of mouse models of sickle cell disease, suggesting a second therapeutic application for its targeted reactivation.¹³  Although demonstrating the therapeutic potential of embryonic ϵ-globin, neither study directly addressed fundamental questions concerning regulatory processes that may affect the expression of the native ϵ-globin gene in stage-discordant erythroid cells.

There is general consensus that ϵ-globin gene silencing in definitive erythroid cells is a consequence of transcriptional arrest. Studies conducted both in vitro^14,15  and in transgenic mice^16-18  have identified a number of positive and negative transcriptional regulatory motifs within the approximate 200-bp ϵ-globin gene promoter that bind erythroid cell-restricted and -ubiquitous factors. Site-specific mutations in several of these motifs can reactivate ϵ-globin gene expression, indicating the importance that transcriptional controls play in regulating ϵ-globin gene expression and, in addition, illustrating the principle that developmental silencing of the structurally intact ϵ-globin gene is not irreversible.

The value of transcriptionally derepressed ϵ-globin genes in adults with defects in β-globin gene expression, however, may be limited by poorly understood regulatory mechanisms that act on the mature ϵ-globin mRNA. For example, the benefits of even the most efficient transcriptional reactivation method will be negated in translationally active definitive erythroid cells if the encoded ϵ-globin mRNA is unstable. The current report investigates aspects of the posttranscriptional regulation of ϵ-globin gene expression that have direct relevance to its potential therapeutic application. First, a novel method is established that permits the relative stabilities of embryonic ϵ- and adult β-globin mRNAs to be directly compared in erythroid-phenotype mouse erythroleukemia (MEL) and nonerythroid HeLa cells. Subsequent analyses demonstrate that the stability of ϵ-globin mRNA in undifferentiated definitive erythroid cells in vivo is sufficient to guarantee its high-level accumulation in translationally active cells at later stages of differentiation. The possibility that the human ϵ- and β-globin mRNAs are coregulated by a common posttranscriptional mechanism is also investigated to determine whether the efficiency of ϵ-globin gene derepression might be enhanced in individuals with certain β-thalassemic gene defects. A potential structural basis for this effect is then addressed in experiments carried out in cultured cells and in transgenic mice. In sum, these experiments indicate that the stability of ϵ-globin mRNA does not limit the value of ϵ-globin gene-reactivation strategies, and further suggest that relevant posttranscriptional mechanisms may provide a substantial therapeutic advantage to individuals with β thalassemia in whom ϵ-globin gene expression is successfully derepressed. These data complement previous reports demonstrating that the biochemical properties of heterotetramers containing human ϵ-globin subunits are physiologically useful in adults,^12,19  providing assurances that posttranscriptional events do not pose a significant barrier, and may even enhance, the therapeutic value of reactivated embryonic ϵ-globin.

The construction and validation of a human (h) β-globin gene linked to a tetracycline-conditional regulatory element (TRE) is described elsewhere.²⁰  A TRE-linked hϵ-globin gene was constructed by sequential ligation of 3 DNA fragments, encompassing the full 1.6-kb transcribed region and 0.2 kb of contiguous 3′-flanking region, into the SacII-EcoRV site of pTRE-2 (BD Biosciences, San Jose, CA). Derivative TRE-linked hϵ^3′β and hβ^3′ϵ genes were constructed by substituting polymerase chain reaction (PCR)–generated hβ- and hϵ-globin 3′UTRs and 1.6 or 0.2 kb of contiguous 3′-flanking regions, respectively, for the corresponding TRE-hϵ and TRE-hβ sequences. The integrity of each DNA construct was verified by restriction digest analysis and automated dideoxy sequencing.

Transgenes that encode hϵ- and hβ-globin mRNAs are described elsewhere^12,21 ; each is linked in its native orientation to a 6.5-kb DNA fragment derived from DNase I-hypersensitive sites 1-4 of the hβ-globin locus control region (μβLCR).²²  The μβLCR-linked transgenes encoding hϵ^3′β and hβ^3′ϵ mRNAs are identical to the TRE-linked genes, except that both transgenes are flanked by identical β-globin 5′ and 3′ flanking-region DNA.^12,21 

All animal studies were approved by the University of Pennsylvania Institutional Animal Care and Use Committee. The generation and validation of mice with germline integration of μβLCR-hϵ and μβLCR-hβ transgenes has previously been described.^12,21 ClaI-EcoRV DNA fragments containing the μβLCR-hϵ^3′β and μβLCR-hβ^3′ϵ DNAs were purified over an Elutip filtration column (Schleicher & Schuell, Keene, NH) and provided to the University of Pennsylvania Transgenic and Chimeric Mouse Facility for injection into B6SJLF1/J × B6SJLF1/J fertilized oocytes.²³  Founder mice identified by PCR analysis of tail DNA were mated with CD-1 females or C57BL6 males (Jackson Laboratories, Bar Harbor, ME) to generate F1 progeny with germline transgene integration. Mice heterozygous for targeted knockout of adjacent β^Maj and β^Min genes (Hbb^th-3 heterozygotes) were generously provided by O. Smithies (University of North Carolina, Chapel Hill, NC).²⁴  Globin phenotypes of transgenic animals were established by Triton-acid-urea gel electrophoresis of hemolysates,^25,26  which, in combination with known pedigrees, permitted the globin genotypes of complex transgenic-knockout mice to be deduced.

As previously described,^21,27,28  bone marrow hematopoietic cells and peripheral blood reticulocytes are harvested from individual transgenic mice, and total RNA purified from each tissue using TRIzol reagent (Invitrogen, Carlsbad, CA). Levels of transgenic and control endogenous mouse (m) α-globin mRNAs are determined by RNase protection using corresponding [³²P]-labeled antisense RNA probes (see “RNase protection analysis”). Protected probe fragments are resolved on a denaturing acrylamide/urea gel and band densities quantitated by PhosphorImager analysis. All assays are carried out under conditions of probe excess.

MEL cells expressing the tetracycline-regulated transactivator (tTA) protein were provided by S. A. Liebhaber (University of Pennsylvania, Philadelphia, PA).²⁹  tTA-expressing HeLa cells were maintained in FBS-supplemented DMEM media as recommended by the manufacturer (BD Biosciences). The capacities of the derivative tTA-expressing MEL and HeLa cell lines to support doxycycline-conditional silencing of TRE-linked genes have been previously demonstrated.^20,29  Transfections were carried out using 5 × 10⁵ cells and 5 μg supercoiled DNA using Superfect reagent as recommended by the manufacturer (Qiagen, Valencia, CA). Doxycycline was added to a final concentration of 1 μg/mL when required.

RNAs from mouse bone marrow and peripheral blood were purified as described^12,21 ; RNAs from cultured cells were prepared using TRIzol reagent as recommended by the manufacturer (Gibco-BRL). The [³²P]-labeled hβ- and hϵ-globin probes were transcribed in vitro from PCR-generated DNA templates using SP6 RNA polymerase (Ambion, Austin, TX). The 287-nucleotide (nt) β-globin probe protects a 199-nt sequence of hβ-globin mRNA exon 2, whereas the 317-nt ϵ-globin probe protects a 221-nt fragment of human ϵ-globin mRNA exon 2. A 313-nt probe protects 160-nt exonic fragment of hβ-actin mRNA.²⁰  A DNA template encoding an mα-globin mRNA probe has previously been described.²¹  Band intensities were quantitated from PhosphorImager files using ImageQuant software (Amersham Biosciences, Piscataway, NJ). The RNase protection analyses (RPAs) were carried out using previously described linearity controls.²¹ 

Unfractionated PBS-washed cells from heparin-anticoagulated whole blood were resuspended in 15 μL PBS containing 2 mg/mL dextrose and supplemented with 1.5 μL [³⁵S]methionine (Amersham; 15 mCi/mL [555 MBq/mL], >1000 Ci/mmol [37 TBq/mmol]), and incubated and for 30 minutes at 37°C. Cells were washed in excess PBS, osmotically lysed in excess ddH₂O, and clarified hemolysates resolved by Triton-acid-urea gel electrophoresis.^25,26  PhosphorImager exposures were quantitated using ImageQuant software. Pulse-chase analyses used a 5-minute metabolic-labeling period, at which time cells were washed in PBS and resuspended in excess heparin-anticoagulated plasma from a nontransgenic donor mouse for defined intervals.

The possibility that a transcriptionally derepressed hϵ-globin gene can produce therapeutically beneficial levels of hϵ-globin protein is critically dependent on the stability of its encoded mRNA in definitive erythroid cells. To address the extent to which successful reactivation of ϵ-globin gene transcription might be limited by posttranscriptional regulatory mechanisms, a novel transcriptional chase strategy was designed to establish the stabilities of globin mRNAs under a variety of informative experimental conditions. The approach capitalizes on the properties of derivative MEL cells that constitutively express the hybrid tTA transactivator. Although tTA constitutively activates the transcription of genes linked to a recombinant TRE, its activity is rapidly and efficiently inhibited in the presence of tetracycline or doxycycline. Consequently, the stabilities of mRNAs encoded by TRE-linked genes can be established by interval analysis of their levels in doxycycline-exposed tTA-expressing cells.²⁰  For the current studies, 2 TRE-linked genes encoding full-length hϵ- and hβ-globin mRNAs were constructed and structurally verified (Figure 1A). These genes were cotransfected into fully characterized tTA-expressing MEL cells,^29,30  and the relative levels of the encoded hϵ- and hβ-globin mRNAs established at defined intervals after doxycycline exposure using an RNase protection method. Triplicate analyses demonstrated a reproducible 30% reduction in hϵ-globin mRNA, relative to hβ-globin mRNA, following an 8-hour period of transcriptional silencing (Figure 1B-C). Parallel experiments in previously validated doxycycline-exposed nonerythroid tTA-expressing HeLa cells revealed a similar reduction in hϵ-globin mRNA levels (Figure 1D-E). These results indicate potentially important differences in the posttranscriptional fates of hϵ- and hβ-globin mRNAs in definitive erythroid cells and additionally suggest that the relevant globin mRNA-stabilizing mechanisms are not erythroid cell-type restricted.

Although hϵ- and hβ-globin mRNAs are unequally stable in proerythroblast MEL cells, the effect of this difference on the accumulation of hϵ-globin mRNA in translationally active cells at later stages of terminal differentiation is not known. Consequently, we elected to study the extent to which the constitutive stability of hϵ-globin mRNA affects its accumulation over the full course of terminal differentiation. The experimental approach uses mice with hϵ- and hβ-globin transgenes linked to transcriptional control elements that ensure their high-level transcription in adult (definitive) erythroid cells¹²  (Figure 2A). A previously established method was adapted to assess the stability of hϵ-globin mRNA (Figure 2B), which is defined as the proportion of hϵ-globin mRNA originally present in murine marrow erythroid cells (largely proerythroblasts, basophilic erythroblasts, and polychromatophilic erythroblasts) that survives in circulating reticulocytes from the same animal.^21,23  Analyses of 2 or more animals from each of 4 hϵ-transgenic mouse lines demonstrated a stability for hϵ-globin mRNA equal to about 60% of the value previously established for hβ-globin mRNA under identical experimental conditions²¹  (Figure 2C). These studies indicate that the stability of an mRNA in early, transcriptionally active erythroid cells poorly predicts its overall accumulation in later stages of terminal differentiation, because the modest instability of hϵ-globin mRNA in MEL cells seems to have little practical impact on its accumulation in cells corresponding to later stages of differentiation. Consequently, the constitutive stability of the hϵ-globin mRNA does not appear to limit the utility of therapeutic approaches to hϵ-globin gene reactivation in disorders of β-globin gene expression.

The observed high-level stabilities of the evolutionarily related hϵ- and hβ-globin mRNAs suggested the possibility that the 2 mRNAs might be coregulated by a shared posttranscriptional mechanism. For example, the hϵ- and hβ-globin mRNAs might compete for an essential mRNA-stabilizing mechanism that is easily saturable. Because this possibility would have critical implications vis-à-vis the efficiency of ϵ-globin gene reactivation strategies—particularly in thalassemics with deficiencies in β-globin mRNA—a functional screen for relevant regulatory relationships was developed. In vivo analyses were designed to assess whether the expression of transgenic hϵ-globin is affected by the level of β-globin gene expression in mature study animals. hϵ-transgenic mice containing none, one, or 2 copies of a mβ knockout allele (genotypes mβ^+/+/hϵ, mβ^+/−/hϵ, and mβ^−/−/hϵ, respectively) were generated through iterative mating of hϵ transgenics with mice containing heterozygous knockout of their endogenous adult mβ-globin genes (Hbb^th-3).²⁴  Metabolic-labeling studies of anticoagulated whole blood from these animals revealed striking 3- and 9-fold increases in hϵ expression in mβ^+/− and mβ^−/− mice, respectively, compared to its expression in mβ^+/+ animals, consistent with a predicted coregulatory activity affecting the expression of the 2 related genes (Figure 3A-B). This effect was observed in 2 different hϵ-transgenic lines, suggesting that the relevant mechanism is independent of transgene integration-site effects. In addition, levels of [³⁵S]methionine pulse-labeled hϵ-globin remained stable in reticulocytes from mβ^+/− and mβ^−/− animals following a prolonged translational chase interval (Figure 3C-D), indicating that hϵ-globin augmentation in thalassemic animals is not due to an increase in the stability of the hϵ-globin protein. The results of these studies strongly suggest that coregulation is likely to affect the stability of the intact hϵ-globin mRNA.

Evidence indicating that β-globin expression alters neither the transcription of the hϵ transgene nor the stability of its encoded protein suggested the existence of a posttranscriptional process that coregulates the stabilities of the hϵ- and hβ-globin mRNAs. To address this possibility, mRNA stability studies were carried out in complex transgenic-knockout mβ^+/+/hϵ, mβ^+/−/hϵ, and mβ^−/−/hϵ mice. The results of these experiments revealed a clear relationship between hϵ-mRNA survival and the number of intact mβ-globin alleles; the stability of hϵ mRNA increased 2- and 6-fold in mβ^+/− and mβ^−/− animals, respectively, relative to its value in mβ^+/+ animals (Figure 4A-B). A corresponding prediction, that β-globin gene overexpression would reduce hϵ-globin mRNA stability, was tested by similar studies of hϵ-transgenic mice that did or did not coexpress an independent hβ-globin transgene. Metabolic-labeling experiments demonstrated a 3-fold decrease in hϵ-globin synthesis in reticulocytes from hβ-expressing transgenics (not shown), whereas marrow-reticulocyte mRNA analyses demonstrated a 2-fold reduction in the stability of hϵ mRNA in 7 hβ-expressing mβ^+/− animals (Figure 4C-D). These results substantiate the hypothesis that the stabilities of the hβ- and hϵ-globin mRNAs are likely to be coregulated through a shared mechanism. From a practical perspective, these data indicate that the physiologic effect of hϵ-globin gene derepression would likely be accentuated in β-thalassemic individuals who expressed low levels of functional β-globin mRNA.

We speculated that the stabilities of both the hϵ- and hβ-globin mRNAs might be mediated by related cis-acting determinants, including either of 2 structural elements that have been identified within the β-globin 3′UTR.^20,21  Sequence alignments, however, failed to reveal any clear similarities in the primary structures of the hϵ- and hβ-globin 3′UTRs (data not shown). Because mRNA-stability determinants can be highly degenerate, and therefore difficult to identify with certainty,^27,31,32  a functional assay was designed to assess whether the stability of the hϵ-globin mRNA is dictated by elements within its 3′UTR. TRE-linked hϵ- and hβ-globin genes were constructed that contained reciprocal exchanges of their 3′UTRs (Figure 5A); these genes were then cotransfected into tTA-expressing MEL cells and the relative stabilities of the encoded mRNAs assessed by doxycycline chase (Figure 5B). In contrast to the unequal stabilities of the parental hϵ- and hβ-globin mRNAs, the 2 chimeric mRNAs were equally stable in erythroid MEL cells (Figure 5C). Analyses in nonerythroid HeLa cells produced nearly identical results, validating earlier conclusions that this aspect of hϵ- and hβ-globin gene expression is not restricted to erythroid cells (Figure 5C,E). The physiologic importance of this effect was subsequently tested in mice expressing transgenes encoding the chimeric hϵ^3′β-globin and hβ^3′ϵ-globin mRNAs (not illustrated). The mRNA stability analyses confirmed the importance of 3′UTR identity to globin mRNA accumulation during terminal differentiation; substitution of an ϵ-globin 3′UTR reduced hβ-globin mRNA stability by more than 4-fold whereas, conversely, a substituted β-globin 3′UTR augmented hϵ-globin mRNA survival nearly to the level of full-length hβ-globin mRNA (Figure 5F). These studies indicate the importance of the 3′UTR to the constitutive stabilities of the hϵ- and hβ-globin mRNAs and, in the setting of their common evolutionary heritage, suggest that the relevant structural determinants may be distantly related.

Detailed biochemical studies conclude that ϵ-globin subunits can incorporate into heterotetrameric hemoglobins exhibiting physiologically valid, therapeutically important characteristics. In vitro studies using hemoglobins expressed in yeast³³  and in mammalian erythrocytes¹⁹  observe little practical difference between the O₂-binding characteristics of Hb Gower-2 (α₂ϵ₂) and Hb A (α₂β₂), including highly similar P₅₀ values, Hill coefficients, Bohr properties, and 2,3-BPG–binding affinities. Parallel in vivo analyses demonstrate that expression of hϵ-globin improves the phenotype in transgenic mouse models of thalassemia¹²  and sickle cell disease,¹³  confirming the therapeutic utility of hϵ-globin expression in both disorders. These studies provide a clear rationale for aggressive investigation into the possibility that the developmentally silenced hϵ-globin gene can be transcriptionally derepressed in disorders of β-globin gene expression.

Generic gene-reactivation strategies for β thalassemia and sickle cell disease, though, are predicated on the expectation that derepressed embryonic and fetal globin mRNAs will exhibit stabilities similar to those of adult β-globin mRNA. This assumption is likely to be valid for fetal γ-globin mRNA, which can be reactivated to high-level expression using several pharmacologic agents.^8-10,34  The possibility that embryonic-stage hϵ-globin mRNA will exhibit the necessary stability, however, is substantially more difficult to predict because there are no known human conditions, either constitutive or acquired, in which the hϵ-globin mRNA is transcribed at physiologically relevant levels. The importance of assessing hϵ-globin mRNA function is not trivial; the transcription of mRNAs that are either highly unstable or that encode dysfunctional globin subunits would negate many of the anticipated benefits of hϵ-globin gene reactivation. Our analyses investigate this critical property of hϵ-globin mRNA to ensure that it is compatible with the therapeutic targeting of its encoding gene for transcriptional reactivation.

The current studies demonstrate that hϵ-globin mRNA is sufficiently stable to permit high-level expression of hϵ globin in definitive erythroid cells. Differentiating definitive erythroid cells remain translationally active for several days after they are transcriptionally silenced, favoring the expression of genes, like those encoding hβ and hγ globin, that transcribe highly stable mRNAs. In contrast, the expression of hϵ globin is normally restricted to transcriptionally active primitive erythroblasts³⁵  in which the evolutionary benefits of high mRNA stability are less pronounced. Consequently, there is significant reason to query whether the stability of hϵ-globin mRNA is sufficient to permit its high-level accumulation in developmental stage-discordant erythroid progenitors. These concerns are justified by studies demonstrating the relative instability of a related embryonic-globin mRNA (ζ-globin mRNA) when it is expressed in definitive mouse erythrocytes.^23,27  Using a tetracycline-conditional expression system (Figure 1A) we demonstrate that hϵ-globin mRNA is less stable than β-globin mRNA in cultured MEL cells (Figure 1B-C). Although murine in origin, these erythropoietic cells are similar to human proerythroblasts in several respects and are consequently used to model early-stage definitive human erythroid progenitors.³⁶  Unexpectedly, though, the impact of the stability difference between the 2 mRNAs over the full course of terminal differentiation is relatively small (Figure 2C). The basis for this paradox is not clear, although it seems reasonable to speculate that the stabilities of individual globin mRNAs are not fixed, but rather change as erythroid cells pass through successive stages of maturity. This possibility would account for the wide range of half-life values for adult α- and β-globin mRNAs that have emerged from studies using experimental systems corresponding to different periods of terminal differentiation.^37-40  From a practical perspective, the observed (relative) instability of hϵ-globin mRNA in immature erythroid cells has surprisingly little impact on its overall accumulation in more differentiated cells.

The accompanying observation that the hϵ- and hβ-globin mRNAs accumulate to nearly equal levels during the course of terminal differentiation may have important implications vis-à-vis the efficiency of gene reactivation approaches. Other globin mRNAs appear to be coregulated through related or identical posttranscriptional processes,^20,21,27  providing a strong precedent for the hypothesis that β-like mRNAs may share vestiges of a posttranscriptional regulatory mechanism that predates evolutionary divergence of the embryonic and adult β-like globin genes. Our data indicate that this is likely to be the case, demonstrating that hϵ-globin is expressed at significantly higher levels in β-thalassemic animals than in nonthalassemic transgenic mice (Figure 3A-B). Two observations argue that this effect is not a consequence of transcriptional up-regulation. First, up-regulation is independent of transgene integration site and does not require physical proximity to the endogenous mouse β-globin gene locus (data not shown). Second, levels of hϵ-globin mRNA in transcriptionally active marrow erythroid cells (normalized for control mα-globin mRNA) are not materially increased in β-thalassemic animals, seemingly inconsistent with a mechanism involving an increase in hϵ-globin gene transcription (Figure 4A). Likewise, pulse-chase analyses indicate that compensatory up-regulation is not a consequence of alterations in the stability of hϵ-globin subunits (Figure 3C-D). Additional studies did, however, demonstrate an unequivocal increase in the stability of hϵ-globin mRNA in β-thalassemic animals (Figure 4A-B), consistent with the hypothesis that the hϵ- and hβ-globin mRNAs are posttranscriptionally coregulated. These data suggest that the consequences of hϵ-globin gene derepression will be highly leveraged in β thalassemias that are characterized by reduced levels of β-globin mRNA. A mechanistic arrangement of this type would also have an impact on the fundamental design of globin transgenes for human therapy to avoid competitive destabilization of their encoded mRNAs.

The nature of the elements that define the high stabilities of the hϵ- and hβ-globin mRNAs remain undefined, but are anticipated to be structurally related. Although strict homologies between the 3′UTRs of the 2 mRNAs were not observed, it is worth noting that hϵ-globin mRNA contains a pyrimidine-rich track similar to one that is thought to stabilize the β-globin mRNA.²¹  This track can be highly polymorphic, as evidenced by different forms in the 3′UTRs of the human²⁷  and murine⁴¹  globin mRNAs, as well as several nonglobin mRNAs.^31,32  As might be predicted, reciprocal exchange of this region results in a decrease in the stability of the hβ-globin mRNA and a corresponding increase in the stability of the hϵ-globin mRNA in both cultured cells and in transgenic mice (Figure 5). One future challenge will be to identify the nature of the site-specific mRNA stability elements in the hϵ-globin 3′UTR and to determine how nucleotide differences affect the binding of trans-acting effector factors, as well as the overall stability of hϵ-globin mRNA.

The authors thank Dr Jia Yu for technical assistance, Dr Oliver Smithies for providing mβ-knockout animals, and Dr Stephen A. Liebhaber for sharing tTA-expressing MEL cells.

This work was supported in part by National Institutes of Health grants HL-R01-061399 and HL-U54-070596.