INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Eukaryotic genes exhibiting complex patterns of expression may be regulated through an array of controls that act both transcriptionally and posttranscriptionally. Globin genes, which are expressed in a well-characterized developmental stage-specific manner, serve as an informative model for such regulatory complexity. The -globin gene cluster in humans and mice contains three functional genes: , 2, and 1. These genes undergo a switch in expression at the transition between embryonic and fetal development, from coexpression of the - and -globin genes, to exclusive expression of the two fetal/ adult -globin genes. This event, which occurs at gestational weeks 6 to 8 in humans and postcoital days 8.5 to 10.5 in mice, is known as -globin gene silencing. The switch is remarkably efficient: although -globin protein is expressed at high levels in primitive erythroblasts in the embryonic yolk sac, it cannot be detected in definitive erythroid cells from normal term fetuses or adults. Recent evidence indicates that -globin gene silencing is a complex event requiring both transcriptional downregulation (23, 29, 36, 37), as well as other, less well-defined mechanisms (23). Specifically, the -globin transcribed region was noted to play an important and previously unanticipated role in this process (23). In experiments utilizing transgenic mice, levels of -globin mRNAs transcribed from full-length human -globin (h-globin) transgenes were appropriately silenced during the embryonic-to-fetal transition. In contrast, chimeric -globin transgenes into which the -globin transcribed region was substituted (thereby preserving -globin promoter elements) were incompletely silenced, with significant residual mRNA expressed in adult erythroid cells. While confirming the importance of promoter elements to regulation of h-globin gene expression (29, 36, 37), these transgenic studies also indicated that promoter elements alone are insufficient to effect its full developmental silencing.

The high-level expression of h- and h-globins in adult erythrocytes is dependent on the unusually long half-lives of their fully-processed mRNAs in erythrocyte progenitors. These half-lives have been estimated to be between 16 and 72 h in a broad range of cell culture and in vivo experiments (2, 5, 18, 21, 24, 31, 32). The high stability of - and -globin mRNAs facilitates their accumulation in transcriptionally active erythroid precursors and ensures that they remain at high levels for continued translation in subsequent, transcriptionally silent stages of terminal erythroid differentiation. A molecular basis for the stability of h-globin mRNA has recently been established with both cultured cells and transgenic mice (19, 43, 45, 46). The stability of -globin mRNA in vivo is paralleled by the ability of its 3' untranslated region (3'UTR) to assemble a messenger ribonucleoprotein (mRNP) complex (the -complex) in vitro (42). Mutations that disrupt -complex assembly in vitro decrease -globin mRNA stability in vivo. The cis elements crucial to -globin mRNA stability and -complex assembly map to a defined pyrimidine-rich region that is highly conserved between human and mouse -globin 3'UTRs (15, 43, 44). These data, combined with the observation that the stabilizing activity of the -globin mRNA 3'UTR can be successfully transferred to a reporter mRNA (34), suggest that -globin mRNA stability is determined by a mechanism that acts locally within the 3'UTR. Although the nature of this mechanism is not yet clear, recent studies indicate that mutations that destabilize -globin mRNA by interfering with -complex assembly promote accelerated shortening of the poly(A) tail (26). These studies demonstrate that assembly of the sequence-specific 3'UTR RNP -complex is crucial to -globin mRNA stability, and is consequently an important determinant of -globin gene expression.

The studies mentioned above illustrate the importance of mRNA stability to -globin gene expression. The possibility that these controls might also be relevant to the observed posttranscriptional component to silencing of the evolutionarily related -globin gene has not been previously investigated. In the current work, we establish a transgenic mouse model system that permits direct in vivo comparison of the stabilities of h- and h-globin mRNAs. We demonstrate that h-globin mRNA is significantly less stable than h-globin mRNA in definitive erythroid cells, that the h-globin 3'UTR encompasses major structural determinants of this instability, and that the relative stabilities of the h- and h-globin mRNAs in vivo parallel their capacities to assemble a specific 3'UTR RNP complex in vitro. A specific difference in the sequences of the h- and -globin mRNA 3'UTRs which contributes to their distinctly different affinities for the mRNP stability complex is identified. Finally, we demonstrate that mRNA instability encoded by the h-globin 3'UTR is accompanied by a shortened poly(A) tail, suggesting a mechanistic link between -globin mRNA destabilization and deadenylation. Collectively, these data provide a detailed account of a posttranscriptional mechanism that in conjunction with transcriptional controls, is essential for the full developmental silencing of embryonic -globin gene expression.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Transgene construction. The construction of plasmids containing the full-length h- and h-globin genes has been previously described (23). The high-level transcription of h-globin mRNA in adults was facilitated in some lines by deleting a 3' flanking region silencer element (23, 44a). The h-globin gene is carried as the 1.5-kb PstI human genomic DNA fragment inserted into the EcoRI site of pSP72, and the h-globin gene is carried as the 2.3-kb EcoRI-BamHI human genomic DNA fragment, also in the EcoRI site of pSP72 (23). Chimeric 3' and 3' transgenes were constructed by a splice overlap extension-PCR strategy utilizing either of the two full-length genes as templates (23, 26). First-stage reaction mixtures (Tables 1 and 2) were assembled from 10 ng of template DNA, 200 pmol (each) forward and reverse primer, 4 µl of deoxynucleoside triphosphates (2 mM each), and 2 U of Vent polymerase in 100 µl of 1× reaction buffer provided by the manufacturer (New England Biolabs, Beverly, Mass.). Reaction mixtures were initially denatured at 95°C (5 min), annealed at 57°C (15 s), and extended at 73°C (25 s), and then were cycled an additional 28 times at 92°C (1 min), 57°C (15 s), and 73°C (25 s), with a terminal extension at 73°C (25 s). Second-stage reactions were identical to first-stage reactions, except that they comprised 1 µl each of the two related first-stage reactions (3'-A and -B, and 3'-A and -B) along with oligomer sp72poly and either oligomer 589-607 or 1431-1448. Second-stage reactions were amplified as described above, except that the extension times were prolonged to 30 s. The amplified DNAs were digested with BstEII and SphI and directionally cloned into the corresponding sites of the original h-globin plasmid to create 3', or into the h-globin plasmid to create 3'. Ampicillin-resistant transformants were verified by restriction digest analysis and by sequencing both strands of PCR-amplified sequences. EcoRI fragments containing the 1.6-kb chimeric genes were ligated into the unique pSP72 EcoRI site adjacent to a previously inserted 6.5-kb DNA fragment containing core elements of the -locus control region (-LCR) (µLCR [40]) as described previously (23). Insert orientations were verified by restriction analysis. DNA fragments containing the 8.0-kb µLCR linked to the transgene were released by digestion with EcoRV and SalI and purified for microinjection as previously described (23). The construction of the ^P transgene, containing the -globin promoter linked to the -globin transcribed region, has been previously described (23). The h mRNAs transcribed from the h and h promoters of the ^P and h transgenes, respectively, differ only in the sizes of their 5'UTRs (23).

Generation and characterization of transgenic mice. Mice transgenic for the h-globin transgene have been previously characterized (23). Mice transgenic for chimeric globin genes were generated by the Transgenic Mouse Core Facility at the University of Pennsylvania as previously described (23). Founder and F₁ transgenic mice were screened by dot blot analysis of tail DNA (26) and verified by Southern analysis. Tail DNA (5 µg) from transgenic (F₁ or subsequent generations) and control mice was digested with EcoRI (h- and h3' transgenes) or BamHI (h- and h3' transgenes), resolved on an 0.8% agarose gel in 1× TEA buffer, transferred to Zetabind (CUNO, Meriden Conn.), and probed with ³²P-labeled DNA fragments complementary to regions of the h- or h-globin promoter regions and mouse -globin (m-globin) 3' flanking region (X-region [26]). These probes detect a 1.6-kb EcoRI fragment of the h and h3' globin transgenes or a 2.3-kb BamHI fragment of the h- and h3'-globin transgenes, and either a 4.1-kb (EcoRI) or 1.3-kb (BamHI) fragment of genomic DNA originating from the m-globin 3' flanking region. For all lines, transgene copy number was established as previously described (23).

Assay of transgenic globin mRNA stability. Adult transgenic mice were pretreated with three intraperitoneal injections of acetyl-2-phenylhydrazine (Sigma) as described previously (26), resulting in an initial peripheral hemolysis followed by a compensatory erythropoietic response characterized by an increase in both the marrow erythroid/myeloid ratio and the peripheral reticulocyte count. RNA was purified from unfractionated marrow cells and peripheral reticulocytes as described previously (26). Internally ³²P-labeled antisense-oriented RNA probes for RNase protection assays were transcribed in vitro from template DNA by using SP6 RNA polymerase (SP6 Maxiscript kit; Ambion). RNA samples were desiccated and resuspended in 20 µl of Berk buffer {80% formamide, 40 mM PIPES [piperazine-N,N'-bis(2-ethanesulfonic acid)] (pH 6.4), 400 mM NaCl, 1 mM EDTA} supplemented with both transgene-specific and murine -globin probes, heat denatured for 10 min at 80°C, and incubated overnight at 52°C. Digestion buffer (200 µl of 10 mM Tris [pH 7.5], 300 mM NaCl, 5 mM EDTA, 20 mg of RNase A per ml, 1 mg of RNase T₁ per ml) was added, and the samples were digested at room temperature for 25 min. To terminate the reaction, 17 µl of a 4:1 mixture of 10% sodium dodecyl sulfate-10-mg/ml proteinase K was added, and the samples were incubated at 37°C for an additional 20 min. The samples were extracted with phenol-chloroform-isoamyl alcohol, ethanol precipitated, and analyzed on an 8 M urea-6% acrylamide gel. This method permits calculation of normalized stabilities for mRNAs which are highly reproducible (26). The specific activities of individual probes was adjusted to facilitate visual comparison of band intensities on single autoradiographs; consequently, observed band intensities do not reflect actual transgene expression levels.

Gel mobility shift analysis. h- and h-globin RNA 3'UTR probes were transcribed in vitro from PCR-generated cDNA templates according to the manufacturer's instructions (T7 Maxiscript kit). Murine erythroleukemia (MEL) cells (2.5 × 10⁸) washed twice in ice-cold phosphate-buffered saline were resuspended in 3.5 ml of buffer A (10 mM KCl, 1.5 mM MgCl, 10 mM Tris HCl [pH 7.4], 0.5 mM dithiothreitol) supplemented with pepstatin (0.2 µg/ml), leupeptin (0.2 µg/ml), and aprotinin (10 µg/ml). Cells were lysed by 1 pass through a 23-gauge needle and four additional passes through a 25-gauge needle. KCl (1 M, 0.49 ml) was added, and the mixture was centrifuged in an HS-4 rotor (Beckman) at 3,200 rpm for 10 min at 4°C. The supernatant was transferred into 5-ml Beckman tubes and spun at 32,500 rpm for 60 min at 4°C in an SW41 Ti rotor. Glycerol (0.1 volume) was added to the supernatant, which was subsequently stored in 50-µl aliquots at 80°C. Quick-thawed aliquots were preincubated with 1 µl of RNasin (5 Prime3 Prime, Boulder, Colo.). and 1 µl of -mercaptoethanol for 30 min at room temperature. ³²P-labeled probe (5 × 10⁴ cpm/ml) and specific cold competitor (100 to 200 µg) were added to a total volume of 15 µl, and the incubation continued for an additional 30 min. The reaction was terminated by addition of RNase T₁ (20 U/reaction) for 10 min at room temperature. Heparin (1 µl of a 50-mg/ml stock) and loading dye were added, and the reaction was resolved on a 60:1 acrylamide-bis-acrylamide gel prerun at 110 V for 1 h in 1× TBE. -Complex assembly on 42, 42G, and 42 DNA oligomers was assessed by a similar method which omitted digestion with RNase (14). Pilot experiments (not shown) indicated that the -complex assembles with equal affinity on sequence-identical RNA and DNA oligomers.

Estimation of apparent K_d. Cytoplasmic extract (5 µg) was incubated for 30 min with ³²P-labeled - or -globin 3'UTR RNA probe in concentrations ranging from 0.05 to 10 nM ( 3'UTR) or 0.1 to 24 nM ( 3'UTR), and the resulting RNP complexes were resolved on a 5% native polyacrylamide gel. Quantitation of bound and free RNA probe was performed by PhosphorImager densitometry (Molecular Dynamics) with ImageQuant software. The values of apparent equilibrium dissociation constant (K_d) were calculated by linear regression in a double-reciprocal plot for the one-binding-site hyperbola (10). Each curve comprises data points from two independent experiments.

RNase H mapping. Purified RNA (0.1 to 1.0 µg) and specific oligomer (300 ng) were heat denatured and renatured in a 10-µl reaction mixture (20 mM Tris [pH 7.5], 1 mM EDTA, 50 mM NaCl). Details of analysis of the -globin mRNA poly(A) tail have been previously described (26); a 21-nucleotide (nt) oligomer which anneals 15 nt 5' to the TGA termination site was used for analysis of the -globin mRNA poly(A) tail (5'TCAGGACAGAGGATACGACCG). The samples were supplemented with 10 µl of a mixture of 20 mM Tris (pH 7.5), 20 mM MgCl₂, 100 mM NaCl, 2 mM dithiothreitol, 60 mg of bovine serum albumin per ml, and 0.1 U of RNase H and incubated at 30°C for 60 min. The reaction was terminated by the addition of 130 ml of stop mix (5 µg of tRNA, 20 mM EDTA, 300 mM NaOAc), and the products were phenol extracted, ethanol precipitated, and resuspended in 10 µl of 95% formamide-5 mM EDTA loading dye. The products were resolved on a 6% acrylamide-8 M gel and then were electrotransferred to a Nytran membrane (Schleicher and Schuell, Keene, N.H.) as directed by the manufacturer (Hoefer Scientific, San Francisco, Calif.). The membrane was hybridized with random-primer ³²P-labeled probe corresponding to a region immediately 3' to the oligomer annealing site and then washed according to a standard Northern transfer protocol (26). The probe for the -globin 3' fragment was generated with forward and reverse primers that anneal within its 3'UTR (5'CAGGACAGGCTGCGGC3' and 5'ATTGGTTTATTGGCGC3', respectively) as previously described (26).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

In vivo determination of h- and h-globin mRNA stabilities. An assay was designed to compare the relative stabilities of transgenic - and -globin mRNAs in the intact mouse. This assay exploits the non-steady-state conditions resulting from the generalized transcriptional silencing that occurs midway through terminal erythroid differentiation in all mammalian species. Transcriptionally active erythroid progenitors in the bone marrow of adult animals mature into transcriptionally silent intermediates and finally into anucleate reticulocytes that migrate into the peripheral circulation (Fig. 1A). Thus, bone marrow erythroid cells, which are mostly posttranscriptional, and peripheral blood reticulocytes, which are entirely posttranscriptional, represent early and late time points, respectively, in a physiologic transcriptional chase experiment. The uncertain length of the transcriptionally silent interval precludes calculation of absolute mRNA half-lives. Therefore, the stabilities of a variety of transgenic globin mRNAs are determined by assessing the extent to which their levels decline in the interval between these two points in time relative to the level of a standard control globin mRNA. In each tissue, transgenic globin mRNA levels are normalized to the level of stable, endogenous m-globin mRNA (Fig. 1B). The relative stability of a transgenic mRNA in terminally differentiating erythroid cells is calculated as the ratio of its normalized level in reticulocytes to that in the marrow (the "normalized stability"). As defined, the normalized stability of a transgenic globin mRNA would be 1.0 if it were as stable as m-globin mRNA, while lower values would indicate relative instability of the transgenic mRNA (26).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   In vivo determination of globin mRNA stabilities in transgenic mice. (A) Bone marrow erythroid cells and circulating peripheral reticulocytes represent two sequential stages in terminal erythroid differentiation. The sequence of events during terminal differentiation of erythroid cells in the bone marrow (left) and peripheral circulation (right) is illustrated. In the bone marrow, transcriptionally active erythroid progenitors (top) undergo global transcriptional silencing, nuclear condensation, and nuclear extrusion (middle), resulting in transcriptionally silent, anucleate marrow reticulocytes (bottom). Marrow reticulocytes, which contain substantial levels of actively translating globin mRNA, are subsequently released into the bloodstream as peripheral reticulocytes. Reticulocyte mRNA degrades as the cell matures into an erythrocyte over the ensuing 2 to 3 days (not illustrated). (B) Assessment of relative human globin mRNA stabilities in transgenic mice. Bone marrow hematopoietic cells and peripheral blood reticulocytes were harvested from individual transgenic mice previously rendered anemic by treatment with phenylhydrazine. Total RNA was purified from each tissue, and the levels of transgenic (- or -) globin mRNA and control endogenous m-globin mRNA were determined by RNase protection with corresponding 32P-labeled antisense RNA probes. Protected probe fragments were resolved on a denaturing acrylamide-urea gel, and band densities were quantitated by PhosphorImager analysis. All assays were carried out under conditions of probe excess. B, bone marrow; R, peripheral blood reticulocytes (Retics).

h-Globin mRNA is less stable than h-globin mRNA in adult-stage erythroid cells. Mouse lines containing h- and h-globin transgenes have been previously generated and characterized for transgene copy number and overall level of transgene expression by standard assays (23, 26). The stabilities of h- and h-globin mRNAs were determined relative to that of -globin mRNA in individual transgenic mice by the RNase protection assay (RPA [Fig. 1B]). Representative analyses of mice from each of three independent lines transgenic for the h-globin gene demonstrate that h-globin mRNA is as stable as endogenous m-globin mRNA in terminally differentiating adult erythroid cells (Fig. 2A). In contrast, analyses of mice from each of three independent lines transgenic for the h-globin gene demonstrate a rapid loss of the h-globin mRNA relative to endogenous m-globin mRNA (Fig. 2B). The normalized stability of h-globin mRNA was determined in two or more mice from each of 16 independent transgenic lines; these data are plotted in Fig. 2C. The mean stability of 0.28 for the h-globin mRNAs is significantly lower than the mean stability of 0.85 for h-globin mRNA (P < 0.005).

View larger version (44K): [in this window] [in a new window]   FIG. 2.   h-globin mRNA is less stable than h-globin mRNA in adult murine erythroid cells. (A) h-Globin mRNA is stable in adult erythroid cells. Levels of transgenic -globin mRNA and endogenous m-globin mRNA were determined by RPA of bone marrow erythrocytes (B) and peripheral blood reticulocytes (R) from individual mice transgenic for the h-globin gene. Autoradiographs of three representative lines are shown. The positions of the protected h- and m-globin mRNA fragments are indicated. Transgenic line designations are shown at the top of respective lanes. In this experiment and others, the specific activities of individual probes were adjusted to ensure that the intensities of the experimental and control bands were both within the linear range of autoradiographic detection. (B) h-Globin mRNA is unstable in adult erythroid cells. RPA analyses of h-globin mRNA stability in mice from three representative transgenic lines is shown. The positions of the protected h- and m-globin mRNA fragments are indicated. (C) -Globin mRNA is less stable than -globin mRNA in terminally differentiating adult erythroid cells. The normalized stability values of h-globin mRNA determined in each of 16 independent transgenic lines () are plotted. A minimum of two mice from each transgenic line were studied. The normalized stability of endogenous m-globin mRNA (defined as 1.0) is indicated by a dashed horizontal line. The mean ± 2 standard errors for h-globin mRNA is indicated by an open circle with error bars. The mean ± 2 standard errors for h-globin mRNA is reproduced from a similar analysis of 10 independent -globin transgenic lines (26). The difference between the mean stabilities of h- and h-globin mRNA is significant at P < 0.005. (D) Direct comparison of h- and h-globin mRNA stabilities in a mouse coexpressing both transgenes. A mouse was generated which was transgenic for both the human - and -globin genes, and the levels of expressed h- and h-globin mRNA in its bone marrow (B) and peripheral blood reticulocytes (R) was determined by RPA. An autoradiograph from the study is illustrated, with the positions of the protected h- and h-globin mRNA fragments indicated to the left. A linear increase in the amount of probe protected by 5- and 10-fold amounts of reticulocyte mRNA confirmed that the analysis was performed under conditions of probe excess. Parallel assay of reticulocyte mRNA from a wild-type (wt) mouse illustrates the specificity of the probes for their transgenic h- and h-globin mRNA targets. The above results were confirmed by the study of an additional doubly hemizygous mouse (not shown). (E) Apparent differences in h- and h-globin mRNA stabilities do not reflect differences in promoter function. The stability of an h-globin mRNA transcribed from an h-globin promoter was studied in transgenic mice with the described stability assay. The normalized stability values of h-globin mRNA in P mice was determined for mice from each of three independent transgenic lines (); the mean ± 2 standard errors is indicated by an open circle with error bars. This value does not differ significantly from the mean stability ± 2 standard errors for h-globin mRNA transcribed from the -globin promoter, which has been reproduced from Fig. 2C. The normalized stability of endogenous m-globin mRNA (defined as 1.0) is indicated by a dashed horizontal line.

Differences in the stabilities of h- and h-globin mRNAs are encoded by elements within their 3'UTRs. We have previously linked the stability of h-globin mRNA to the structure of its 3'UTR (26, 43, 45, 46). To test whether the relative instability of h-globin mRNA is related to structural divergence in this region, we created chimeric h- and h-globin transgenes with exchanged 3'UTRs (Fig. 3A). The chimeric 3' gene encodes an mRNA comprising the h-globin 5'UTR and coding region and the h-globin 3'UTR. The reciprocal transgene, 3', encodes an mRNA comprising the h-globin 5'UTR and coding region, and the h-globin 3'UTR. As in the case of the h- and h-globin transgenes, high-level expression of the chimeric transgenes was achieved by linking each to core elements of the -LCR (23, 41). Transgenic lines were established from each of the chimeric globin genes. In each case, transgene copy numbers (between one and five per mouse genome in all but one line [data not shown]) were similar to those previously determined for h- and h-globin transgenic lines (23).

View larger version (17K): [in this window] [in a new window]   FIG. 3.   Elements within the -globin 3'UTR mediate mRNA destabilization. (A) Construction of chimeric h- and h-globin transgenes with exchanged 3'UTRs. Two chimeric transgenes were constructed from segments of h- and h-globin genes (black and gray, respectively). The 3' gene comprises the full length -globin gene in which the -globin 3'UTR and 3' flanking region have been substituted. The 3' transgene contains the full-length h-globin gene in which the -globin 3'UTR and 3' flanking region have been substituted. The positions of the normal translation initiation and termination sites are indicated. (B) Distinct stability phenotypes are determined by h- and h-globin mRNA 3'UTRs. Transgenic mouse lines were generated from each chimeric globin transgene illustrated in panel A, and the stabilities of their encoded mRNAs were determined by RPA (Fig. 1). The normalized stability value for transgenic mRNAs from each individual line is plotted (). The mean ± 2 standard errors from all 3' lines is shown to the left, along with the average stability of transgenic h-globin mRNA reproduced from Fig. 2C. The mean ± 2 standard from all 3' lines is shown to the right, along with the average stability of transgenic human -globin mRNAs reproduced from Fig. 2C. Average stabilities which differ significantly from either h- or h-globin mRNA at the P < 0.05 level are indicated by asterisks (*).

The -globin mRNA 3'UTR assembles a stability-determining -complex with reduced affinity. The stability of h-globin mRNA in vivo has been linked to the assembly of an mRNP complex (the -complex) on a pyrimidine-rich tract within its 3'UTR (15, 19, 43). The -complex has a characteristic mobility on nondenaturing polyacrylamide gels, and its assembly is specifically competed by poly(C). When incubated under standard conditions in cytosolic extract from MEL cells, the h-globin 3'UTR assembles an mRNP complex with the same migration (Fig. 4A, lanes 3 and 10) and competition profile (lanes 11 to 14) as the authentic -complex (lanes 4 to 7). These data suggest that the h-globin mRNA 3'UTR assembles a complex that is identical to or closely related to the authentic -complex.

View larger version (25K): [in this window] [in a new window]   FIG. 4.   h- and h-globin mRNA 3'UTRs differ in their abilities to assemble an RNP -complex. (A) The - and -globin mRNAs assemble similar RNP complexes on their respective 3'UTRs. 32P-labeled h- and h-globin mRNA 3'UTRs were incubated in cytosolic (S-100) extract prepared from MEL cells, and RNA not protected by RNP complexes was subsequently digested by addition of RNase T1. Protected RNP complexes were resolved on a nondenaturing 5% polyacrylamide gel. Assays were carried out in the absence (lanes 3 and 10) or presence (lanes 4 to 7 and 11 to 14) of excess homoribopolymer competitors [poly(C), poly(G), poly(A), or poly(U)]. Control lanes (1, 2, 8, and 9) verify the integrity of the probes and the efficiency of RNase digestion. The positions of the -complex and free RNA probe are indicated to the left of the autoradiograph. (B) Determination of the Kd of the RNP complexes which assemble on the - and -globin mRNA 3'UTRs. Defined quantities of 32P-labeled h or h 3'UTR transcripts were incubated in MEL cell cytosolic extract (5 µg) and resolved on nondenaturing 5% polyacrylamide gels. For each sample, the quantity of probe incorporated into -complexes (bracketed) was determined by PhosphorImager analysis. The position of free h and h 3'UTR probes are indicated. h-3'UTRs are nearly completely incorporated into -complexes when present in concentrations of <1 nM, consistent with the extremely low Kd of this interaction. Free probe can be visualized upon prolonged autoradiograph exposure and in samples which contain probe in concentrations of >2.5 nM (not shown). (C) The -complex assembles on the - and -globin mRNA 3'UTRs with distinct affinities. The affinity of -complex interaction with the - and -globin 3'UTRs (black and gray, respectively) was estimated as the apparent Kd, calculated from linear regression analysis of a double-reciprocal plot (inset). Curves comprise points from results illustrated in panel B and from additional independent experiments (not shown).

10

9

A CG substitution in the major polypyrimidine track within the -globin 3'UTR reduces the efficiency of -complex assembly. An analysis of the structural basis for the difference in -complex assembly and mRNA stability was initiated by inspection of the - and -globin 3'UTR nucleotide sequences (Fig. 5A). The -globin 3'UTR was aligned to maximize homology to three pyrimidine-rich elements of the -globin 3'UTR, which have been directly implicated in both -globin mRNA stability and -complex assembly (14, 43, 45). All three of the pyrimidine-rich tracts within the -globin 3'UTR are conserved within the -globin 3'UTR. However, the central tract is shifted 5' a distance of 7 nt in the -globin 3'UTR relative to the -globin 3'UTR, and the longest of the three polypyrimidine tracts within the -globin 3'UTR is interrupted by a single purine (G) transversion. Either (or both) of these structural differences might account for the decreased affinity of the -complex for the -globin 3'UTR and for the consequent decrease in -globin mRNA stability. Evolutionary comparisons and structural mapping studies suggest, however, that the major polypyrimidine tract is the most important feature determining -globin mRNA stability and -complex assembly (14, 44). Thus, the effect of the single-base transversion on -complex assembly was studied. The -complex readily assembled on a 42-nt oligomer (42) corresponding to the segment of the native -3'UTR previously demonstrated to be fully sufficient for -complex assembly (14). A single CG mutation in the major polypyrimidine track of 42 (42G) markedly reduced -complex assembly (Fig. 5C). Moreover, 42G and a 42-nt oligomer corresponding to the native 3'UTR (42), which contains a similarly positioned G, both compete poorly for -complex assembly relative to native 42. These data indicate that a naturally occurring G which interrupts the major polypyrimidine tract in the native -globin mRNA 3'UTR decreases its capacity to assembly a stabilizing -complex.

View larger version (25K): [in this window] [in a new window]   FIG. 5.   Assembly of an mRNP stability complex on the -globin 3'UTR is inhibited by a CG transversion within the major polypyrimidine tract. (A) Alignment of h- and h-3'UTR minimal sequences. The h- and h-globin mRNA 3'UTRs are aligned to maximize homology; the lengths of 3'UTR sequences on either end of the aligned segments are indicated parenthetically. The termination codons (UAA or UGA) and poly(A) hexanucleotides (AAUAAA) are underlined. The polypyrimidine tracts implicated in -globin mRNA stability are boxed, as are corresponding polypyrimidine-rich regions of the -globin mRNA 3'UTR. The 5' displacement of the central -globin polypyrimidine tract, relative to the related -globin sequence, is emphasized by a narrow arrow; interruption of the major polypyrimidine tract by a purine (G) is indicated by a thick vertical arrow. (B) Oligomers used to assess the importance of an uninterrupted pyrimidine tract to high-efficiency -complex assembly. 42 corresponds to the 42-nt region of the -globin mRNA 3'UTR protected from RNase digestion by a fully assembled -complex. 42G differs from the wild-type 42 by the substitution of a G within the 3' terminal of the major polypyrimidine tract. 42 is the wild-type -globin 3'UTR sequence corresponding to the 42 segment. The substituted G in 42G is in boldface, and both it and the native G which interrupts the major polypyrimidine tract in the -3'UTR sequence (42) are indicated by arrows. (C) Interruption of the major polypyrimidine tract of the -3'UTR by a purine G inhibits -complex assembly. 32P-labeled 42 and 42G were incubated in MEL cell S-100 extract in the absence () or presence of unlabeled competitors present in 100-fold excess, and resolved on a native polyacrylamide gel as described in the legend to Fig. 4. The positions of the -complex, the previously defined alternate complex (43), and free probe are indicated to the left of the autoradiograph.

Destabilization of h-globin mRNA is linked to its accelerated deadenylation. The observation that destabilization of the h-globin mRNA in vivo parallels inefficient assembly of the -complex on its 3'UTR in vitro suggests that the two processes are related. To begin to characterize the molecular steps that accompany accelerated h-globin mRNA decay, we assessed the length of poly(A) tails on the h-, h-, and h3'-globin mRNAs (Fig. 3A). RNA was isolated from reticulocytes of mice carrying each of the three transgenes (Fig. 6A). The transgenic RNAs were subjected to site-specific RNase H cleavage, resolved on a denaturing polyacrylamide gel, and electroblotted to a nylon membrane, and the 3'-terminal fragments containing the poly(A) tails were visualized with  or  3'UTR sequence-specific probes. The poly(A) tail on the h-globin 3'UTR demonstrated discrete peaks of activity that occurred at intervals of approximately 20 nt (Fig. 6B). The major peak corresponded to a poly(A) tail length of approximately 60 nt (A₆₀), while large proportions of shorter (A₄₀) and longer (A₈₀) poly(A) tails were also observed. In contrast, both mRNAs with -3'UTRs (the  and 3' mRNAs) contained poly(A) tails almost entirely composed of the shorter A₄₀ variety. Hence, two characteristics of -globin mRNAits relative instability and a short poly(A) tailare encoded by elements contained within the 105-nt  3'UTR.

View larger version (18K): [in this window] [in a new window]   FIG. 6.   The h-globin mRNA 3'UTR contains a short poly(A) tail. (A) Measurement of poly(A) tail length (An) on transgenic globin mRNAs. Site-specific RNase H cleavage of globin mRNAs was targeted by an antisense DNA oligomer (horizontal bar) complementary to a sequence located approximately 100 nt 5' to its poly(A) addition site. The RNA fragments were resolved on a denaturing acrylamide gel and electrotransferred to a nylon membrane. The 3'-terminal fragment, which contains the poly(A) tail, was visualized by hybridization with a site-specific 32P-labeled 3'UTR DNA probe, the position of which is indicated (*). The length of the fully deadenylated 3'-terminal RNA fragment was established by addition of excess oligo(dT) to the hybridization reaction prior to the RNase H digestion. (B) The h and 3' mRNAs have shortened poly(A) tails. The poly(A) tail lengths of h-, h-, and chimeric 3'-globin mRNAs from transgenic reticulocytes were determined with the RNase H assay. The positions of the fully deadenylated 3'-terminal fragments are aligned, and the positions equivalent to a poly(A) tail length of 40 residues (determined by DNA size markers electrophoresed in parallel for each study) are indicated by an arrow. The results for each mRNA were confirmed in experiments with reticulocytes from mice representing two or more independent transgenic lines. (C) Densitometric analyses of the poly(A) tails from panel B. Signal intensities from h-, h-, and chimeric 3'-globin poly(A) tails from panel B were quantitated by scanning densitometry, and the adjusted scans were aligned to a standard scale to facilitate direct comparison.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The induction of -globin gene expression and reciprocal silencing of -globin gene expression in erythroid cells between the 6th and 8th weeks of human gestation define the end of the embryonic and the initiation of the fetal stages of erythroid development. This process is paralleled in the -globin gene cluster by a switch from embryonic - to fetal -globin gene expression (reviewed by Russell and Liebhaber [33]). It is hypothesized that coordination of globin gene switching between the two clusters provides a mechanism through which hemoglobin heterotetramers with specific O₂ affinities can be expressed during different developmental stages; such an arrangement might facilitate transplacental diffusion of O₂ from maternal to embryonic or fetal erythrocytes. A potential hazard of such an arrangement is that failure to fully silence expression of embryonic globin genes might result in the inappropriate expression of embryonic hemoglobin heterotetramers with high O₂ affinities in fetal and adult erythrocytes (14). This situation might be detrimental to the fetus by altering sensitive transplacental O₂ gradients and to adults by triggering a secondary polycythemia (11). These possibilities emphasize the potential importance of full -globin gene silencing at the appropriate gestational age.

The contribution of transcriptional controls to globin gene switching has been intensively studied, and the specific importance of this mechanism to -globin silencing has been clearly documented (23, 29, 36, 37). However, independent lines of evidence from experiments utilizing primary human erythroid tissue suggest that additional mechanisms which act posttranscriptionally on both - and -globin gene products are also crucial to this process (1, 49). Nuclear run-on experiments utilizing progenitor erythroid cells from normal adult bone marrow yield surprisingly high levels of -globin transcripts (49); however, -globin mRNA can be detected at only trace levels in near fully differentiated adult reticulocytes (1). This apparent paradox is resolved if one postulates a posttranscriptional defect in -globin mRNA accumulation due to instability of the -globin mRNA. This hypothesis is supported by our recent studies of transgenic mice, which demonstrate that h-globin transcriptional control elements are only partially effective in -globin gene silencing and that differences within the - and -globin transcribed regions are important determinants of their expression (23). Collectively, these data support a model in which posttranscriptional eventsspecifically, instability of -globin mRNAplay a necessary role in the full silencing of -globin gene expression.

The general importance of mRNA stability to the control of eukaryotic gene expression is well established. The half-lives of eukaryotic mRNAs vary more than 1,000-fold, from as short as several minutes to as long as several days (30). The short half-lives of labile mRNAs such as c-myc, c-fos, and other cytokine and lymphokine mRNAs, permit changes in transcription to be rapidly reflected in the level of cytoplasmic mRNA (6, 30). In comparison, "bulk" proteins, such as collagens (13, 25) or crystallins (22), are typically encoded by highly stable mRNAs, providing the cell with a substantial savings in the energy required for their transcription, processing, and transport. In some cases, mRNA stability is dynamic, as occurs with transferrin receptor (TfR) mRNA, whose half-life varies in response to physiologically relevant changes in cellular iron levels (20). Hence, mRNA stability plays an important regulatory role in the expression of a wide variety of genes central to cellular differentiation and function.

In the current work, we test the hypothesis that posttranscriptional controls of -globin gene expression act through destabilization of -globin mRNA in fetal and adult erythroid cells. A mouse model system was established in which the stabilities of transgenic globin mRNAs could be measured in vivo (Fig. 1). We have previously verified the utility of the transgenic approach for studies of globin mRNA stability by comparing the stability of h^wt mRNA to an h-globin mRNA with a known destabilizing mutation (^CS) (26). This system closely models the physiology of normal human erythropoiesis and precisely models the compensatory erythropoietic response to a variety of anemias, including those (as typified by the thalassemias) that result from ineffective erythropoiesis and accelerated peripheral hemolysis. The current approach also permits long-term monitoring of highly stable globin mRNAs under non-steady-state conditions that occur in cells undergoing rapid changes in morphology, function, and gene expression. The "physiologic transcriptional chase" has obvious advantages over the use of global, nonspecific inhibitors of transcription, which may impart artifactual effects in studies of mRNA stability (27, 39, 48). Moreover, cell culture-based model systems which permit interval assessment of mRNA levels following a transcriptional pulse (such as the fos promoter response to serum induction) are poorly suited for the study of highly stable mRNAs, because it is difficult to maintain cell viability during the prohibitively long serum-free preparatory interval necessary to establish sufficiently low background levels of the test mRNA (45). Studies summarized in Fig. 2 demonstrated that h-globin mRNA is significantly less stable than h-globin mRNA in erythroid cells from transgenic mice. This effect complements, but is independent of, transcriptional downregulation of the -globin gene occurring at the embryonic-to-fetal transition. The mRNA sequences which determine the difference in stability map to the - and -globin mRNA 3'UTRs (Fig. 3). These results confirm the previously established importance of the 3'UTR to -globin mRNA stability (45, 46) and demonstrate that the -globin mRNA 3'UTR lacks an equivalent stability-determining function. These results also confirm in a whole animal system earlier observations in cultured cells that the stabilizing function accompanies physical transfer of the -globin 3'UTR (34).

Although exchange of the h- and h-globin mRNA 3'UTRs significantly alters the stability of the resultant chimeric mRNAs in the anticipated direction, this effect is incomplete. Compared to the stabilities of the native h- and h-globin mRNAs, the intermediate values for the chimeric mRNAs suggest that additional, functionally distinct regions outside the 3'UTR of either (or both) the - and -globin mRNAs may also contribute to their stability. A similar arrangement of spatially and functionally distinct elements mediates accelerated decay of several short-lived mRNAs, including those of c-myc, c-fos, and beta interferon (17, 40, 47, 48), and may mediate high-level stability of the human -globin mRNA in adult erythroid cells (34). The evolutionary forces that favor a multiplicity of stability-determining features remain obscure, although it seems likely that such an arrangement functions in part to ensure the stability of -globin mRNA, which is crucial to the generation of a fully functional erythrocyte.

An unexplained aspect of the current data is the variation in the normalized stabilities of transgenic mRNAs among lines generated from identical transgene constructs. This effect is best appreciated by the scatter in the stabilities of h-globin mRNA illustrated in Fig. 2; a similar scatter has also been observed in measurements of the stability of transgenic h-globin mRNAs (26). The stability of a transgenic mRNA determined in multiple mice from the same line is highly reproducible (~12% variation), as are serial determinations from an individual mouse. This high reproducibility suggests that biologic factors contribute to the scatter in mRNA stabilities observed among independent lines. An analysis of possible variables that could result in line-to-line variation in mRNA stability failed to demonstrate linkage to transgene structure, copy number, level of expression, or the sex of the adult mouse. Additional experiments (Fig. 2E) indicated that the stability of transcribed -globin mRNAs is independent of promoter identity. This raises the possibility that the observed differences might reflect local influences at each transgene insertion site. Such random integration site-dependent effects might explain some of the variation that has previously been observed in studies of transgenic -globin gene expression (38). Although of potential interest, the observed line-to-line variation in mRNA stability does not detract from the highly significant, threefold difference in the stability of the human - and -globin mRNAs in the large number of independent transgenic lines studied.

A molecular basis for the difference in the stabilities of the - and -globin mRNAs was suggested by RNA gel mobility shift analysis (Fig. 4). The high stability of -globin mRNA is believed to result from assembly of an RNP complex (the  complex) on the -globin 3'UTR (14, 19, 43). It was thus surprising to find that despite its relatively low stability, the -globin mRNA 3'UTR assembles an mRNP complex which, as judged by its gel migration and its susceptibility to competition by poly(C), appears similar, if not identical, to an authentic -complex. However, the affinity of the -complex for the -globin 3'UTR was sixfold lower than its affinity for the -globin 3'UTR (Fig. 4), thus linking the relative instability of -globin mRNA to the relatively high K_d of -complex formation on its 3'UTR. The structural basis for this difference in the efficiency of -complex assembly was suggested by alignment of the - and -globin mRNA 3'UTR sequences (Fig. 5A). Both 3'UTRs contain a series of three pyrimidine-rich tracts which have been largely conserved in the 350 to 400 million years since separation of the h- and h-globin genes (12). These pyrimidine-rich tracts appear to be essential for assembly of an -complex on the -globin 3'UTR (14, 43, 44). We have shown that interruption of the major pyrimidine-rich tract by a single purine negatively affects the efficiency of -complex assembly (Fig. 5). Still, the fact that these pyrimidine tracts have been largely conserved within the context of otherwise poorly conserved 3'UTRs suggests that they may serve an additional unrecognized function(s). For example, the  3'UTR might play a limited role in enhancing -globin mRNA stability in primitive (embryonic) erythroid cells. To date, our attempts to test this possibility have been frustrated by technical barriers that limit direct measurement of the stability of globin mRNAs in this tissue.

An analysis of the poly(A) tails of the - and -globin mRNAs indicated a potential mechanism through which -globin mRNA decay is accelerated (Fig. 6). Although mRNA degradation may proceed via a number of overlapping pathways in both yeast and higher eukaryotes, a common pathway involves initial deadenylation followed by cleavage of the 5' ^7mG(5') ppp(5')N mRNA cap (8, 9, 16, 28, 40). The demonstration that functional characteristics, such as mRNA stability, and structural attributes, such as poly(A) tail length, both accompany physical transfer of the -globin 3'UTR to the -globin mRNA emphasizes the importance of this region to normal -globin gene expression. The shortened poly(A) tails may result from inefficient polyadenylation of the nascent mRNA in the nucleus and/or from accelerated shortening of a normally polyadenylated mRNA in the cytoplasm. Attempts to resolve this issue by analysis of poly(A) tail length in early-stage (marrow) erythroid progenitors have been frustrated by the limited amount of tissue and the low-level expression of the h transgene. It seems likely, however, that accelerated deadenylation related to degradation occurs in the cytoplasm, because we have observed that unusually short poly(A) tail lengths accompany mRNAs destabilized by either of two mechanisms: the loss of cis elements in the -globin mRNA 3'UTR (this study) or through physical disruption of the -complex by ribosomal read-through into the 3'UTR (26). A causal link between accelerated mRNA decay and a short poly(A) tail has been demonstrated in yeast, but has not yet been firmly established in higher eukaryotes (8, 9).

The data from the current study support a model for -globin gene silencing which incorporates both transcriptional downregulation and posttranscriptional destabilization of -globin mRNA (Fig. 7). In this model, high-affinity cis elements within the -globin mRNA 3'UTR assemble -complexes efficiently, resulting in stabilization of -globin mRNA. In contrast, the accelerated deadenylation and decay of human -globin mRNA in adult erythroid cells reflect relatively inefficient assembly of a stabilizing -complex on low-affinity cis elements within its 3'UTR. This relatively inefficient assembly of the -complex is likely to be of only minimal importance to -globin expression in circulating primitive (embryonic) erythroblasts, because these cells contain low levels of competitor -globin mRNA. Moreover, -globin mRNA is continually replenished in circulating primitive erythroblasts, which maintain a transcriptionally active nucleus. In contrast, the relative instability of -globin mRNA is a crucial determinant of full -globin gene silencing in definitive (fetal or adult) erythrocytes. Reciprocal induction of the -globin genes and downregulation of -globin gene transcription result in a large increase in the ratio of  to  mRNAs. According to the model, the -globin mRNAs assemble stabilizing -complexes on their high-affinity 3'UTRs from limiting quantities of essential components. -globin mRNAs, which compete poorly for these limiting factors, are left unprotected and are subject to accelerated decay. These molecular events are paralleled by a cellular switch from transcriptionally active, nucleated primitive erythrocytes to transcriptionally silent, anucleate definitive erythrocytes. The shortened half-life of -globin mRNA results in an exponential fall in its levels relative to -globin mRNA in the 3- to 5-day transcriptionally silent (but translationally active) period of terminal differentiation. Hence, the temporal confluence of these two eventsthe molecular and cellular switchesresults in a situation in which direct molecular communication or "cross-talk" between the - and -globin 3'UTRs assumes a prominent role in -globin gene silencing. This model would predict that naturally occurring mutations which downregulate -globin gene transcription would result in increased levels of -globin mRNA. In fact, this pattern is observed in individuals with some forms of deletional -thalassemia who express especially low levels of -globin mRNA (7, 42). A detailed knowledge of these mechanisms would permit formulation of therapeutic strategies designed to alter the balance between - and -globin gene expression through modulation of these posttranscriptional controls. In addition, the identification of additional mRNAs whose functions depend on assembly of RNP complexes which share components with the -complex (15) will suggest a wider role for direct mRNA-mRNA interaction in the control of gene expression.

View larger version (14K): [in this window] [in a new window]   FIG. 7.   Model in which instability of -globin mRNA reflects its reduced affinity for the -complex and is initiated through accelerated decay of the poly(A) tail. In terminally differentiating adult erythroid cells, limiting quantities of one or more components of the -complex (gray objects) assemble preferentially on high-affinity sites within the -globin mRNA 3'UTR. Lower-affinity sites in the -globin 3'UTR compete poorly for -complex assembly, subjecting the unprotected -globin mRNA to accelerated degradation. In this model, assembly of the poly(A) tail by PABP multimers results in discontinuous deadenylation, observed experimentally as phasing of poly(A) tail lengths (26). The - and -globin mRNA sequences are indicated in black and gray, respectively; the mRNA cap () and translation initiation and termination sites (tick marks) are noted, as are PABP monomers (open ovals).