A 3′UTR mutation affects β-globin expression without altering the stability of its fully processed mRNA


J. Eric Russell, MD, Abramson Research Building, Room 316F, The Children's Hospital of Philadelphia, 34th Street and Civic Center Boulevard, Philadelphia, PA 19104, USA. E-mail: jeruss@mail.med.upenn.edu


Summary. Determinants of mRNA stability are frequently positioned in the 3′UTR where they are not subject to disruption by actively translating ribosomes. Two related individuals with β thalassaemia who carry a β-globin gene containing a 13 nt deletion in its 3′UTR have recently been described. Its position within the 3′UTR, as well as its relative distance from other known functionally important elements, suggested that the deletion might overlay previously unrecognized determinants of β-globin mRNA stability. We studied the impact of the Δ13 mutation on β-globin gene expression in vitro and in vivo. The adverse effect of the Δ13 mutation on β-globin expression was confirmed in studies utilizing reticulocytes from a βΔ13 heterozygote, which indicated a sixfold reduction in the relative level of the mutant mRNA. Additional in vitro analysis indicated that the deletion did not affect the capacity of the βΔ13 mRNA to assemble an mRNA-stabilizing mRNP ‘β-complex’. Unexpectedly, functional tests in both primary erythroid cells and in a transgenic mouse model demonstrated that the βΔ13 mRNA was fully stable, suggesting that the Δ13 mutation affects accumulation of the fully processed mRNA at an earlier step. Consistent with this, there was a relative excess of unprocessed βΔ13 mRNA in erythroid progenitors from a βΔ13 heterozygote. Taken together, these results define a new thalassaemic determinant, which acts to decrease β-globin mRNA levels by inhibiting the efficiency of nuclear processing events, and suggest a previously unanticipated complexity to the role of the 3′UTR elements in the regulation of β-globin gene expression.

The normal expression of nearly all eukaryotic genes is dependent upon a series of defined post-transcriptional events that occur both in the nucleus and in the cytoplasm. Newly transcribed mRNAs are initially restructured by nuclear processing reactions and then transported into the cytoplasm, where their stabilities and translational efficiencies dictate the expression of their encoded protein products. The importance of these processes to regulated gene expression is illustrated by the severe phenotypical consequences resulting from naturally occurring mutations that interfere with any of these steps (Liebhaber & Kan, 1981; Hunt et al, 1982; Orkin et al, 1985; Jankovic et al, 1990; Russell & Liebhaber, 1993).

The primary sequence of a nascent mRNA transcript contains all of the information necessary for its appropriate nuclear processing and efficient nuclear-to-cytoplasmic transport. Embedded RNA signals serve as a nidus for the assembly of specific mRNP effector complexes and/or direct their activities to appropriate RNA target sequences. For example, accurate and efficient intronic splicing requires 5′ donor and 3′ acceptor elements and an adenosine-containing branchpoint sequence (Hastings & Krainer, 2000; Reed, 2000), and may benefit from defined enhancer and silencer determinants positioned within flanking exons (Del Gatto-Konczak et al, 1999; Mayeda et al, 1999). These RNA signals recruit more than a dozen trans-acting factors that assemble into a structurally complex mRNP spliceosome that efficiently deletes the targeted intron with single-nucleotide accuracy (Hastings & Krainer, 2000; Reed, 2000). In a separate set of reactions, the AAUAAA hexanucleotide and a short GU-rich motif direct 3′ cleavage and polyadenylation of the nascent mRNA (Colgan & Manley, 1997). Trans-acting factors that are recruited to these elements, including cleavage/polyadenylation specificity factor, cleavage stimulation factor, and poly(A) polymerase, ultimately direct site-specific processing. The mRNA signals that mediate transport of mature mRNAs into the cytoplasm are less well defined, although the process appears to be facilitated both by cis elements (such as the 5′ cap) (Hamm & Mattaj, 1990) and functionally distinct trans-acting factors (e.g. shuttling and non-shuttling hnRNP proteins) (Nakielny et al, 1997).

In the cytoplasm, fully processed eukaryotic mRNAs retain signals that specify both their translational efficiencies as well as their stabilities. Protein synthesis initiates at a highly conserved consensus sequence [(A/G)NNAUGG] (Kozak, 1987) with an efficiency that can be modified by secondary structures that slow or stall 40S scanning (Hentze et al, 1987; Kozak, 1994). Other sequences within the transcribed region mediate accelerated clearance of mutant mRNAs that translate truncated protein products (nonsense-mediated decay) (Thermann et al, 1998). The nature of their interaction with the translational apparatus requires that signals for translational initiation or nonsense-mediated decay are located either in the 5′UTR or coding region. In contrast, determinants of mRNA stability are, with few exceptions (Kabnick & Housman, 1988; Shyu et al, 1989; Wisdom & Lee, 1991), confined to the 3′UTR where their interactions with trans-acting factors are not subjected to sterical disruption by actively translating ribosomes. A broad array of linear AU-rich (Shaw & Kamen, 1986) and CU-rich motifs (Weiss & Liebhaber, 1995), as well as several structurally dissimilar stem–loop structures (Pandey & Marzluff, 1987; Klausner et al, 1993; Scheper et al, 1995), are known to direct mRNA stability from this region. Naturally occurring (De Jong et al, 1975; Hanash et al, 1977; Weiss & Liebhaber, 1994) and synthetic mutations (Casey et al, 1989; Weiss & Liebhaber, 1995; Holcik & Liebhaber, 1997) that disrupt the interaction of specific trans-acting factors with these cis elements, either directly or indirectly, can dramatically alter expression of the cognate gene. The clinical relevance of many such mutated genes emphasizes the importance of mRNA stability to normal gene expression.

The structures mediating the high stabilities of human (h) globin mRNAs have been particularly well defined using in vitro (Wang et al, 1995; Holcik & Liebhaber, 1997), cell culture (Weiss & Liebhaber, 1994, 1995), and transgenic approaches (Morales et al, 1997; Russell et al, 1998). The unusually long half-life of hα-globin mRNA, estimated between 24 and 60 h (Lodish & Small, 1976; Volloch & Housman, 1981; Ross & Pizarro, 1983; Ross & Sullivan, 1985), requires the presence of a defined pyrimidine-rich track within its 3′UTR (Weiss & Liebhaber, 1995; Holcik & Liebhaber, 1997). A 39-kDa RNA-binding factor, αCP, binds to this region, resulting in the assembly of an mRNP ‘α complex’ (Kiledjian et al, 1995; Wang et al, 1995). The mRNA-stabilizing function of the α complex has been confirmed by experiments demonstrating that mutations affecting its assembly in vitro have a parallel effect on α-globin mRNA stability in cultured cells (Kiledjian et al, 1995; Wang et al, 1995; Weiss & Liebhaber, 1995). Related or identically functioning mRNP complexes are also known to assemble on the 3′UTRs of other mRNAs that encode factors essential to erythroid cell development (Ostareck-Lederer et al, 1994; Czyzyk-Krzeska & Beresh, 1996; Holcik & Liebhaber, 1997; Stefanovic et al, 1997).

In contrast with α-globin mRNA, relatively little is known about the structures or mechanisms through which the hβ-globin mRNA is stabilized. Although single-nucleotide polymorphisms have been described within the β-globin 3′UTR (Jankovic et al, 1991; Cai et al, 1992), naturally occurring mutations affecting β-globin mRNA stability have not been identified. Genes containing anti-termination mutations permitting ribosomes to read 31 nt into the 3′UTR are expressed at near-normal levels (Flatz et al, 1971; Bunn et al, 1975; Delanoe-Garin et al, 1988), indicating that cis-acting stability elements are unlikely to reside in the proximal 3′UTR. These observations have been confirmed by experiments in cultured cells demonstrating the presence of a stability determinant(s) distal to this region (Russell & Liebhaber, 1996). More recent work in MEL cells indicates that an mRNP α complex assembles on the β-globin 3′UTR, and that targeted deletion or substitution of a 14-nt pyrimidine-rich track in this region substantially reduces the stability of hβ-globin mRNAs expressed in transgenic mice (Yu & Russell, 2001). These studies do not address the possibility that additional 3′UTR elements might contribute to β-globin mRNA stability, an arrangement that has been suggested but never formally studied (Russell & Liebhaber, 1996; Yu & Russell, 2001).

Recently, an individual was described who possessed a well described β thalassaemia determinant (IVS-II-745) as well as a β-globin gene carrying a novel 13-nt deletion within its 3′UTR (Basak et al, 1993). The functional impact of the Δ13 mutation on β-globin expression was suggested by the severity of the clinical course as well as the phenotype of the propositus' mother, an uncomplicated βΔ13 heterozygote (genotype βΔ13A). Its position in the 3′UTR fuelled speculation that the deletion affected the function of an undefined mRNA stability determinant. The current work confirms the adverse effect of the Δ13 mutation on β-globin mRNA accumulation in primary erythroid cells from a βΔ13A heterozygote, and assesses the impact of the Δ13 mutation on β-globin mRNA stability in vitro as well as in vivo in humans and in a transgenic mouse model. The results of these analyses indicate that the Δ13 mutation reduces accumulation of cytoplasmic β-globin mRNA without significantly impairing its stability.

Finally, we show that unprocessedβΔ13 mRNAs are over-represented in βΔ13A erythroid cells. Although unanticipated, these data were consistent with a model in which the Δ13 mutation leads to accumulation of unspliced β-globin mRNAs by interfering with the efficiency of its nuclear processing. These results also identified the 3′UTR as an important repository for an array of mRNA signals that direct functionally diverse post-transcriptional processes in both the nucleus and the cytoplasm.

Materials and methods

Preparation of RNA from human and mouse tissues. Informed consent was obtained from all individuals before venepuncture or bone marrow aspirate, which were subsequently carried out using standard clinical method. Heparin-anticoagulated peripheral blood and marrow cells were washed in phosphate-buffered saline, pelleted and stored in aliquots at −80°C. Approximately 50 µl of each tissue was resuspended in 400 µl of TRIzol reagent and processed as recommended by the manufacturer (GibcoBRL, Gaithersburg, MD, USA), except that haemolysis was facilitated by incubating the cell suspension at 80°C for 10 min. Total RNA was prepared from the peripheral blood and bone marrow of adult mice as described previously (Russell & Liebhaber, 1996; Morales et al, 1997; Yu & Russell, 2001).

Preparation of antisense RNA probes.  DNA templates for in vitro transcription of antisense RNA probes were generated by polymerase chain reaction (PCR) amplification of the cloned hβ-globin gene (Yu & Russell, 2001) using primers annealing to exon 3 (β4814–4833: 5′-TATCAGAAAGTGGTGGCTGG-3′) and to its 5′ flanking region (β5060–5042: 5′-ATACGATTTAGGTGACACTATAGAACTCGCACTGACCTCCCACATTC-3′). The reverse primer encodes an SP6 promoter element (boldface). Then, 50-µl reactions were set up, which contained 100 pmol each primer, 1·5 mmol/l MgCl2, 500 µmol/l each dNTP, and 5 U Taq polymerase (Applied Biosystems, Indianapolis, IN, USA). Reactions were amplified in a Perkin Elmer Cetus thermal cycler for one cycle (94°C × 3 min, 58°C × 15 s, 72°C × 30 s), followed by an additional 30 cycles (92°C × 1 min, 61°C × 15 s, 72°C × 30 s), and a 3-min extension at 72°C. In vitro transcriptions were carried out with ≈40 ng DNA template and [α-32P]-CTP (370 MBq/ml, 14·8 TBq/mmol, Amersham, Arlington Heights, IL, USA) using a Maxiscript SP6 kit as recommended by the manufacturer (Ambion, Austin, TX, USA). RNAs were de-salted over a G50 minispin column and probe integrity was verified by denaturing acrylamide/urea gel electrophoresis. The 247-nt β-globin probe generates 186 and 144 nt fragments when incubated with βA and βΔ13 mRNAs respectively. Probes for hα-globin and mouse (m)α-globin mRNAs have been described previously (Russell & Liebhaber, 1996; Morales et al, 1997; Yu & Russell, 2001).

RNase protection assay (RPA).  The RPA method using purified reticulocyte (≈50–100 ng) or marrow RNA (≈1000–1500 ng) has been described in detail elsewhere (Russell & Liebhaber, 1996; Morales et al, 1997; Yu & Russell, 2001). Samples containing excess levels of mRNA were analysed in parallel to assure probe excess and assay linearity. In some cases, the specific activities of individual probes was adjusted to facilitate visual comparison of band intensities on single autoradiographs.

RNA electrophoretic mobility-shift assays (EMSA).  DNA templates for in vitro transcription of 32P-labelled hα- and hβ-globin 3′UTRs were prepared by PCR amplification of the cognate cloned full-length genes, and RNAs were transcribed and purified as detailed previously (Yu & Russell, 2001). EMSA analyses were carried out in MEL cell cytoplasmic extract using a standard method (Yu & Russell, 2001).

Transgenic mice.  All procedures were carried out under protocols approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania. A splice-overlap-extension PCR method was used to generate a 0·9-kb EcoRI–EcoNI fragment of the hβ-globin gene containing a 13-nt deletion in its 3′UTR (Yu & Russell, 2001). The modified fragment was substituted for the cognate region of a previously constructed hβ-globin transgene (Yu & Russell, 2001) linked in its native orientation to a micro β Locus Control Region (Talbot et al, 1989). The structurally verified 10·6-kb µLCR-transgene fragment was subsequently provided to the Transgenic and Chimeric Mouse Facility at the University of Pennsylvania for pronuclear injection. Positive founders identified by Southern blot analysis were used to generate two independent, sustainable lines with germline transgene integration (Liebhaber et al, 1996; Russell et al, 1998; Yu & Russell, 2001).

PCR amplification of pre-RNA.  A reaction containing 400 ng of purified RNA and 70 pmol of primer (preβ-1: 5′-ATACGATTTAGGTGACACTATAGAACTCGCAATGAAAA- TAAATGTTTTTTATTA-3′) was reverse transcribed following a standard method provided by the manufacturer (Promega, Madison, WI, USA). After a 1-h incubation at 42°C, 1 µl of product was incorporated into a 25-µl reaction containing 2·5 mmol/l MgCl2, 40 µmol/l each dNTP and 1·5 U Taq polymerase in 1× buffer supplied by the manufacturer (Applied Biosystems). The reaction was supplemented with 20 pmol forward primer annealing to IVS-2 (preβ-2: 5′-CTCTTATCTTCCTCCCACAG-3′), 5 pmol reverse primer (preβ-3: 5′-TTTAGGTGACACTATAGAACTC-3′), and 10 pmol primer preβ-3 that had been 32P-end-labelled using polynucleotide kinase under conditions recommended by the manufacturer (New England Biolabs, Beverly, MA, USA). Reactions were amplified for one cycle (94°C × 2 min, 58°C × 15 s, 72°C × 45 s) followed by 30 or 35 additional cycles (92°C × 30 s, 60°C × 15 s, 72°C × 45 s) and a 6-min terminal extension at 72°C. Reaction products were resolved on a denaturing acrylamide/urea gel and band intensities quantified from exposed PhosphoImager screens.


The level of βΔ13 mRNA is reduced in a βΔ13A heterozygote

The possibility that a site-specific 13-nt deletion might compromise β-globin gene expression was initially indicated by the unanticipated clinical severity of β thalassaemia in a βΔ13IVS–II−745 double heterozygote (Basak et al, 1993). As the presence of alternately spliced βA mRNAs had the potential to interfere with the analysis, we elected to study βΔ13 mRNA expression in the propositus' mother, a βΔ13 heterozygote carrying a normal βA gene in trans. To validate the anticipated effect of the Δ13 mutation on β-globin gene expression, the relative levels of βΔ13 and βA mRNAs were determined by RPA using 32P-labelled α- and β-globin probes (Fig 1A). Compared with control hα-globin mRNA, βA mRNA was present at high levels in the mother (but not in the propositus), whereas the βΔ13 mRNA was barely detectable (Fig 1B). To better assess the relative expression of the βΔ13 and βA mRNAs, an RPA was carried out using only the β-globin probe (Fig 1C). In four separate determinations the level of βΔ13 mRNA was reduced by an average of sixfold relative to βA mRNA levels. This data was consistent with independent analysis of maternal RNA using a semiquantitative reverse transcription (RT)-PCR method (unpublished observations) indicating that expression of the βΔ13 gene was significantly compromised by a process resulting in reduced levels of its RNA.

Figure 1.

βΔ13 mRNA is expressed at low levels in a βΔ13A heterozygote. (A) An antisense probe distinguishes βΔ13 and βA mRNAs. A DNA template corresponding to a contiguous region of the β-globin exon 3 and 3′ flanking region transcribes a 247-nt reverse complement RNA that protects 186 and 144 nt fragments of human βA and βΔ13 mRNAs respectively. The sizes and positions of the full-length probes and protected probe fragments are indicated. (B) RNase protection analysis of βΔ13 expression. Total reticulocyte RNA prepared from a βΔ13A heterozygote (M, mother), or a βΔ13IVS–II−745 double heterozygote (P, propositus) was analysed by RPA using in vitro transcribed 32P-labelled α- or β-globin RNA probes. Fragments of each probe that are protected by the human α-, βA, and βΔ13 mRNAs are indicated to the right. Control lanes indicate migration of the intact and fully digested α- and β-globin probes (lanes 1–4), as well as analyses of RNA prepared from a normal (βAA) volunteer (C, lanes 5 and 6). The specific activity of the β-globin probe was fivefold greater than that of the α-globin probe. (C) Single-probe RNase protection analysis of βΔ13 expression. Total reticulocyte RNA prepared from a βΔ13A heterozygote (M) was analysed by RPA using 32P-labelled β-globin probe (lane 5). Reticulocyte RNA from βA and βΔ13 transgenic mice (lanes 2 and 3) and from a normal human βAA control (C, lane 4) were analysed in parallel to verify the identities of the protected probe fragments (described in Fig 4 and accompanying text). Lane 1 contains fully digested 32P-labelled β-globin probe.

The βΔ13 mutation does not affect assembly of an mRNA-stabilizing mRNP complex

The observed decrease in the accumulation of βΔ13 mRNA suggested that the 13-nt deletion might adversely affect the high stability of βA mRNA. Examination of the 135-nt β-globin 3′UTR (Fig 2A) indicated that the region of interest is more than 40 nt downstream of a cis-element that is believed to stabilize β-globin mRNA through assembly of a characteristic nRNP ‘β complex’ (Yu & Russell, 2001). To test whether the Δ13 mutation interferes with β-complex assembly, in vitro transcribed 32P-labelled βΔ13 3′UTRs (Fig 2B) were analysed for a shift in their electrophoretic gel mobility after incubation in MEL cell cytoplasmic extract (Fig 2C). The βΔ13 3′UTRs assembled an mRNP complex that comigrated with well characterized α and β complexes assembled on control α- and βA-globin 3′UTRs (Fig 2C, lanes 2, 4 and 6). Moreover, the assembly of all three mRNP complexes was efficiently blocked by the addition of poly(dC) (lanes 3, 5 and 7), a characteristic of the α and β complexes that verified their identities (Wang et al, 1995; Yu & Russell, 2001). Hence, the Δ13 mutation reduces β-globin mRNA levels through a mechanism that is unrelated to β-complex specified mRNA stabilization.

Figure 2.

The Δ13 mutation does not affect assembly of an mRNP β complex in vitro. (A) Map of the 135-nt β-globin mRNA 3′UTR. The UAA translation termination codon and AAUAAA polyadenylation signal are underlined. The pyrimidine-rich element effecting high-level β-globin mRNA stability is indicated in boldface, whereas nucleotides deleted by the Δ13 mutation are indicated in lower case boldface. (B) An mRNP β complex assembles on the βΔ13 3′UTR: left, autoradiograph verifying integrity of in vitro transcribed 32P-labelled α-, βΑ- and βΔ13-globin 3′UTRs; right, RNA electrophoretic gel mobility-shift analysis (EMSA) of 32P-labelled probes incubated in MEL cell cytoplasmic extract were carried out in the absence (lanes 2, 4 and 6) or presence of competitor oligo(dC) (lanes 3, 5 and 7). A control reaction (lane 1) demonstrates full digestion of input RNA in the absence of cytoplasmic extract.

Human βΔ13 mRNA is highly stable in vivo

Although the βΔ13 mRNA assembles an mRNP β complex, in vitro EMSA analysis does not rule out the possibility that the Δ13 mutation adversely affects an independent, previously unrecognized determinant of β-globin mRNA stability. Consequently, we tested the relative stabilities of the βΔ13 and βA mRNAs using an assay that was specifically designed for this purpose (Russell et al, 1998; Yu & Russell, 2001). In this assay, the relative levels of the βΔ13 and βA mRNAs are determined in erythroid progenitors representing early (marrow) and late (reticulocyte) stages of terminal erythroid differentiation. Because erythroid cells in both tissues are largely post transcriptional, the interval change in the relative levels of the two mRNAs reflects a difference in their stabilities. Using an in vitro transcribed 32P-labelled β-globin probe, the ratio of βΔ13 to βA mRNA was measured in marrow and reticulocyte cells from the βΔ13A mother (Fig 3). The relative stabilities of βΔ13A mRNAs were measured at 0·75 and 0·83 in two independent analyses. These values were consistent with normal or near-normal stability of the βΔ13 mRNA. In comparison, mutations that block α-complex assembly reduced the stability of hα-globin mRNA more than fourfold in mice (Morales et al, 1997) and by an even larger margin in humans (Liebhaber & Kan, 1981). These data indicated that the deficit in βΔ13 mRNA levels was likely to reflect mechanisms that do not act via destabilization of the fully processed cytoplasmic mRNAs.

Figure 3.

Human βΔ13 mRNA is highly stable in vivo. Total RNA prepared from bone marrow aspirate (B) and peripheral blood cells (P) from a βΔ13A heterozygote was subjected to RNase protection analysis using an in vitro transcribed 32P-labelled β-globin RNA probe. Reactions containing undigested (lane 1) and fully digested probes (lane 2), as well as human reticulocyte RNA from a βAA homozygote (lane 3), were run in parallel as controls. RNA fragments protected by βA and βΔ13 mRNAs are indicated on the right.

Human βΔ13 mRNA is highly stable in transgenic mice

As the preceding results were unexpected and could not be independently verified (no other genotypically similar individuals have been identified), we elected to extend our study of human βΔ13 mRNA to a transgenic mouse model. Two independent murine lines expressing high levels of the βΔ13 mRNA were generated and subsequently verified (Fig 4A and data not shown). Previously generated mice expressing the transgenic hβΑ-globin mRNA were studied in parallel as controls (Yu & Russell, 2001).

Figure 4.

Transgenic human βΔ13 mRNA is stable in intact mice. (A) Construction of a transgene encoding human βΔ13 mRNA. The βΔ13 and βA transgenes are identical throughout their intronic and exonic sequences (black lines and boxes respectively) except for the defining 13-nt deletion (indicated in exon 3). Tick marks denote the translational initiation and termination codons. Both transgenes are linked to identical µLCR (LCR) and β-globin promoter (P) and enhancer (E) sequences. (B) Representative analysis of transgenic human βΔ13 mRNA stability in normal and β-thalassaemic mice. RNase protection assays were carried out on RNA purified from the bone marrow (B) and peripheral blood (P) of transgenic mice that were normal (mβ+/+, left) or nullizygous (mβ–/–, right) for the endogenous mβ-globin genes. Left, reticulocyte RNA from a non-transgenic control (C) positively identifies the mα probe fragment (lane 1), and demonstrates the specificity of the hβ probe for human β-globin mRNA. Right, samples containing no RNA (lane 1) or RNA from a non-transgenic control (C) (lane 2) were analysed in parallel as controls. The composition of each reaction is shown above the respective lane, and the position of protected βA, βΔ13 and mα probe fragments to the right of each autoradiograph. (C) Stability of human βΔ13 mRNAs in normal and β-thalassaemic mice. Left, the stabilities of βΔ13 mRNAs from five individual mβ+/+ mice, relative to the stability of endogenous mα globin mRNA, are plotted (•), as is their average (bar). The average stability of βA mRNA (Yu & Russell, 2001), is shown for comparison. Right, the stabilities of βΔ13 and βA mRNAs from individual mβ–/– transgenics, each nullizygous for the endogenous mβ-globin genes, are plotted (•), as are their averages (bars).

The design of experiments for analysing the stability of βΔ13 mRNA in transgenic mice was dictated by the observed difference in βΔ13 expression in the propositus and his mother (Fig 1B, lanes 8 and 7), who displayed severe and mild β thalassaemia phenotypes respectively. This inverse correlation between β-globin genotype and βΔ13 mRNA phenotype indicated that the overall level of β-globin mRNA in erythroid progenitors might affect the stability of βΔ13 mRNA, a possibility that has been previously suggested by studies of transgenic hγ-globin mRNA stability in normal and severely β-thalassaemic mice (Yu & Russell, 2000). Consequently, we studied the stability of βΔ13 mRNA in normal mice (mβ+/+) as well as in mice that contained homozygous knockout of their endogenous mβ-globin genes (mβ–/–), using a variation of the two-probe RPA described above. Representative studies indicated that βΔ13 mRNAs and endogenous mα-globin mRNAs were equally stable in both mβ+/+ mice (Fig 4B, left) as well as in mβ–/– animals (Fig 4B, right). When normalized to levels of endogenous mα-globin mRNA, the average stability of βΔ13 in six independent mβ+/+ mice was nearly identical to the stability of βA mRNA previously determined by the same method (Fig 4C, left). Likewise, βΔ13 and βA mRNAs were equally stable in mβ–/– animals (Fig 4C, right). These data indicate that the Δ13 deletion does not materially destabilize the β-globin mRNA in either normal cells or in cells that are severely thalassaemic.

βΔ13 pre-mRNAs accumulate to high levels in marrow erythroid progenitors

The above data indicated that the reduction in the level of fully processed βΔ13 mRNA was not related to a change in its stability. An alternate possibility, that a defect in mRNA processing and/or transport might affect βΔ13 expression, was addressed by quantitative analysis of unprocessed RNA present in erythroid progenitors. Total RNA purified from the bone marrow of the heterozygous βΔ13A mother was subjected to RT-PCR amplification using an intron-specific DNA primer (Fig 5A). The nascent (unprocessed) βΔ13 and βA mRNAs were predicted to generate 291 and 304-bp products, respectively, whereas the fully processed mRNAs, lacking the targeted intron, should not be amplified. A single oligomer pair was used to amplify both mRNAs to eliminate potential differences in their amplification efficiencies, an approach that was validated by noting that the βΔ13A product ratio did not vary in aliquots amplified for either 30 or 35 cycles. With these controls in place, the ratio of unprocessed βΔ13A mRNAs was determined to be ≈1·5 in each of two independent experiments (Fig 5B). This ratio was nearly 10-fold greater than the ratio of the two fully processed mRNAs (Fig 1B and C), suggesting that the expression of βΔ13 mRNA was impeded at a point prior to its appearance in the cytoplasm. A model for this mechanism (Fig 5C) proposes that the Δ13 mutation directly blocks processing of the βΔ13 pre-RNA, resulting in nuclear retention of the unprocessed form with a consequent reduction in the cytoplasmic levels of its fully processed, translationally capable form. It is also possible that the mutation inhibits nuclear-to-cytoplasmic transport of the fully processed mRNA with secondary effects on the efficiency of mRNA processing (not shown).

Figure 5.

Accumulation of unprocessed βΔ13 pre-RNA in human erythroid cells. (A) RT-PCR distinguishes nascent βΔ13 and βA mRNAs. DNase I-treated RNAs were reverse transcribed and PCR amplified using oligomers annealing to IVS-2 and exon 3 (depicted as arrows). The position of the 13-nt deletion is indicated (Δ). (B) Unprocessed βΔ13 pre-RNA accumulates in marrow erythrocytes from a βΔ13A heterozygote. RNA purified from the bone marrow of a βΔ13A heterozygote (M, mother) was reverse transcribed and PCR amplified (lane 2) using the oligomers illustrated in (A). The indicated positions of the amplified PCR products are verified by RT-PCR analysis of transgenic mice expressing the two human β-globins (βΔ13 and βA; lanes 3 and 4). As a control, maternal RNA was also subjected to PCR amplification without initial reverse transcription (lane 1). (C) Model for dysregulated nuclear processing of βΔ13 pre-RNA. The sequence of nuclear events affecting the expression of the βA and βΔ13 mRNAs is illustrated. The exon affected by the Δ13 mutation is indicated in grey, with the other exons indicated in black. Poly(A) tails (AAAAA) are also depicted. The βA and βΔ13 mRNAs are transcribed with equal efficiency, but the Δ13 mutation retards processing of the nascent βΔ13mRNA (indicated by a thin arrow). Fully processed mRNAs of both types are efficiently transported, leading to a deficit of βΔ13 mRNA in the cytoplasm that complements the excess of its unprocessed form in the nucleus. Although not illustrated, a defect in βΔ13 nuclear-to-cytoplasmic transport would be anticipated to slow processing of the βΔ13 pre-mRNA as the fully processed form accumulated in the nucleus.


Normal gene expression is regulated by processes that impact on both the level of the encoded mRNA as well as its ability to efficiently translate its intended protein product. These events are regulated by effector trans-acting factors that interact with specific mRNA signals. The relevant mRNA sequences are not randomly distributed but are rather compartmentalized within functionally distinct regions of the mRNA. For example, independent groups of signals act locally to direct splicing of individual introns (Hastings & Krainer, 2000; Reed, 2000). The positioning of these elements within introns insures that they are eliminated by the processes that they mediate. Likewise, translation initiates at a consensus sequence that operationally defines the border of the 5′UTR and coding regions and encodes the N-terminal methionine (Kozak, 1999). Other sequences impacting translational efficiency are similarly positioned at sites that afford trans-acting factors direct access to ribosomal subunits. The locations of signals that specify mRNA stability are likewise determined by functional considerations. With few exceptions (Kabnick & Housman, 1988; Shyu et al, 1989; Wisdom & Lee, 1991), stability elements are positioned in the 3′UTR where they are protected from sterical disruption by actively translating ribosomes. The 3′UTR is not known to harbour sequences important to mRNA splicing, and its potential role in mRNA transport remains vague (Eckner et al, 1991; Jarmolowski et al, 1994; Nakielny et al, 1997). In this context, it was widely anticipated that 3′UTR mutations impacting β-globin mRNA levels were likely to do so by adversely affecting its stability.

Contrary to all expectations, our studies indicated that a 13-nt deletion in its 3′UTR downregulated β-globin gene expression without altering the cytoplasmic stability of its encoded, fully processed mRNA. RPA analyses indicated that the βΔ13 mRNA was expressed at approximately one-sixth the levels of the wild-type βA mRNA, consistent with the observed β-thalassaemic phenotypes in both the propositus and his mother (Fig 1). Although defects in gene transcription might account for the reduced mRNA levels, careful sequencing through the β-globin promoter region has not revealed any abnormalities (Basak et al, 1993). The possibility that distant mutations affecting the β-globin locus control region might manifest as deficient β-globin mRNA levels appears to be remote: functionally important mutations in this region are large deletions that fully ablate expression from all downstream β-cluster genes (Van der Ploog et al, 1980; Curtin et al, 1985; Driscoll et al, 1989). Consequently, our investigation has focused on the possibility that the Δ13 mutation affects the post-transcriptional regulation of β-globin mRNAs. Structural analysis indicated that the Δ13 mutation does not affect assembly of an mRNP β complex, which is thought to mediate high-level β-globin mRNA stability in vivo (Fig 2). Parallel, functional assessments indicated that the βΔ13 and βA. mRNAs are equally stable in a βΔ13 heterozygote (Fig 3) as well as in normal and β-thalassaemic mice expressing the βΔ13 and βA mRNAs (Fig 4). Careful analysis did not reveal defects in 3′-cleavage of the βΔ13 mRNA (data not shown). Hence, in vitro studies, as well as in vivo studies in both humans and transgenic mice, suggest that the Δ13 mutation reduces the level of β-globin mRNA without impacting on its cytoplasmic stability. These results implicate a nuclear processing or transport defect as the basis for the reduced cytoplasmic levels of βΔ13 mRNA.

The observation that the βΔ13 mRNA was stable in the cytoplasm of terminally differentiating erythroid progenitors indicates that the mutation adversely impacts an intranuclear event(s). We determined that the relative levels of unprocessed βΔ13 mRNA (Fig 5) were 50% higher than levels of βA pre-RNA, a finding that is consistent with this hypothesis. This observation is even more striking in light of the deficit of fully processed βΔ13 mRNAs in these same cells noted here (Fig 1) and elsewhere (unpublished observations). Our data is consistent with a block in processing or transport of the βΔ13 mRNA, with subsequent nuclear accumulation of the unprocessed mRNA transcripts. This data implies that the β-globin 3′UTR encompasses RNA signals that are crucial to processes occurring in the nucleus, in addition to signals specifying cytoplasmic stability. In fact, the possibility that 3′UTRs may contain elements facilitating nuclear-to-cytoplasmic transport has been previously suggested (Eckner et al, 1991; Jarmolowski et al, 1994; Nakielny et al, 1997). In this context, our data may suggest reconsideration of the proposed mechanisms through which other known 3′UTR mutations affect β-globin gene expression. As previously mentioned, the mechanism through which the AAUAAA mutations lead to decreased levels of β-globin mRNA has never been directly evaluated (Orkin et al, 1985; Jankovic et al, 1990). The possibility that these mRNAs are reduced because of a defect in their nuclear processing, rather than destabilization in the cytoplasm per se, must be considered in light of the current data. The presence of three or more functionally diverse elements identifies the 3′UTR as a surprisingly vital participant in the regulation of β-globin gene expression.

An interesting facet of our studies was that the level of βΔ13 mRNA expression was several-fold higher in the severely thalassaemic propositus than in the mildly thalassaemic mother (Fig 1). Expression from the βΔ13 transgene was similarly elevated in β-thalassaemic mice (nullizygous for the mβ-globin genes) than in mice with intact mβ-globin genes (Fig 4A and B). In mice, the increased expression does not reflect a significant change in the stability of the βΔ13 mRNA, suggesting that compensatory upregulation is due to changes in gene transcription or in the efficiency of nuclear processing and transport events. The effect of these changes is to moderate the phenotype of severely thalassaemic individuals. Whether a compensatory mechanism acting at this level is mediated through a specific mRNA signal may merit attention in the future.


The authors thank Dr N. Akar for providing patient samples, J. Yu for performing the EMSAs, and Z. He for technical assistance. This work was supported in part by the Bogazici University Research Fund #101B-105 (A.N.B.), the Bogazici University Foundation (O.B.), and NIH/NHLBI #HL61399 (J.E.R.).