Three mouse models of human thalassemia

Hemoglobin α in the blood vessel wall

Hemoglobin has been studied and well characterized in red blood cells for over 100 years. However, new work has indicated that the hemoglobin α subunit (Hbα) is also found within the blood vessel wall, where it appears to localize at the myoendothelial junction (MEJ) and plays a role in regulating nitric oxide (NO) signaling between endothelium and smooth muscle. This discovery has created a new paradigm for the control of endothelial nitric oxide synthase activity, nitric oxide diffusion, and, ultimately, vascular tone and blood pressure. This review discusses the current knowledge of hemoglobin׳s properties as a gas exchange molecule in the bloodstream and extrapolates the properties of Hbα biology to the MEJ signaling domain. Specifically, we propose that Hbα is present at the MEJ to regulate NO release and diffusion in a restricted physical space, which would have powerful implications for the regulation of blood flow in peripheral resistance arteries.

Differential Regulatory and Compensatory Responses in Hematopoiesis/Erythropoiesis in α- and β-Globin Hemizygous Mice

Characterization of hematopoiesis/erythropoiesis in thalassemias from multipotent primitive cells to mature erythrocytes is of fundamental importance and clinical relevance. We investigated this process in alpha- and beta-globin hemizygous mice, lacking the two adult tandemly organized genes from either the alpha- or beta-globin locus. Although both mice backcrossed on a homogeneous background exhibited similar reduced red blood cell (RBC) survival, beta-globin hemizygous mice had less severe reticulocyte loss and globin chain imbalance, suggesting an apparently milder thalassemia than for alpha-globin hemizygous mice. In contrast, however, beta-globin hemizygous mice displayed a more marked perturbation of hematologic parameters. Quantification of erythroid precursor subpopulations in marrow and spleen of beta-globin hemizygous mice showed more severely impaired maturation from the basophilic to orthochromatophilic erythroblasts and substantial loss of these late precursors probably as a consequence of a greater susceptibility to an excess of free alpha-chain than beta-chain. Hence, only erythroid precursors exhibiting stochastically moderate chain imbalance would escape death and mature to reticulocyte/RBC stage, leading to survival and minimal loss of reticulocytes in the beta-globin hemizygous mice. Furthermore, in response to the ineffective erythropoiesis in beta-globin hemizygous mice, a dynamic compensatory hematopoiesis was observed at earlier differentiation stage as evidenced by a significant increase of erythroid progenitors (erythroid colony-forming units approximately 100-fold) as well as of multipotent primitive cells (day 12 spleen colony-forming units approximately 7-fold). This early compensatory mechanism was less pronounced in alpha-globin hemizygous mice. The expansion of multipotent primitive and potentially stem cell populations, taken together with ineffective erythropoiesis and increased reticulocyte/RBC destruction could confer major cumulative advantage for gene targeting/bone marrow transplantation. Therefore, this study not only corroborated the strong potential effectiveness of transplantation for thalassemic hematopoietic therapy but also demonstrated the existence of a differential regulatory response for alpha- and beta-thalassemia.

Hemoglobinopathies

The outlook for patients with sickle cell disease has improved steadily during the last two decades. In spite of these improvements, curative therapies are currently available only to a small minority of patients. The main theme of this chapter is to describe new therapeutic options that are at different stages of development that might result in further improvements in the outlook for patients with these disorders.

Dr. Joseph DeSimone and his colleagues had previously made the important observation that the hypomethylating agent 5-azacytidine can reverse the switch from adult to fetal hemoglobin in adult baboons. Although similar activity was demonstrated in patients with sickle cell disease and β-thalassemia, concern about the toxicity of 5-azacytidine prevented its widespread use in these disorders. In Section I, Dr. DeSimone discusses the role of DNA methylation in globin gene regulation and describe recent clinical experience with decitabine (an analogue of 5-azacytidine) in patients with sickle cell disease. These encouraging studies demonstrate significant fetal hemoglobin inducing activity of decitabine in patients who fail to respond to hydroxyurea.

In Section II, Dr. George Atweh continues the same theme by describing recent progress in the study of butyrate, another inducer of fetal hemoglobin, in patients with sickle cell disease and β-thalassemia. The main focus of his section is on the use of a combination of butyrate and hydroxyurea to achieve higher levels of fetal hemoglobin that might be necessary for complete amelioration of the clinical manifestations of these disorders. Dr. Atweh also describes novel laboratory studies that shed new light on the mechanisms of fetal hemoglobin induction by butyrate.

In Section III, Dr. Ronald Nagel discusses the different available transgenic sickle mice as experimental models for human sickle cell disease. These experimental models have already had a significant impact on our understanding of the pathophysiology of sickle cell disease. Dr. Nagel describes more recent studies in which transgenic sickle mice provide the first proof of principle that globin gene transfer into hematopoietic stem cells inhibits in vivo sickling and ameliorates the severity of the disease.

Although stroke in adult patients with sickle cell disease is not as common as in children, adult hematologists, like their pediatric colleagues, need to make management decisions in adult patients with a stroke or a history of stroke. Dr. Robert Adams has led several large clinical studies that investigated the role of transfusions in the prevention of stroke in children with sickle cell disease. Much less is known, however, about the prevention of first or subsequent strokes in adult patients with sickle cell disease. In Section IV, Dr. Adams provides some general guidelines for the management of adult patients with stroke while carefully distinguishing between recommendations that are evidence-based and those that are anecdotal in nature.

Mutagenesis and human genetic disease: an introduction

This special issue attempts to provide a fresh perspective on the importance of germ-cell mutagenesis studies and restate the questions and challenges inherent in efforts to minimize the incidence of human genetic diseases. We are working in a time when rapidly advancing molecular technologies provide the tools that permit a more detailed understanding of germ-cell mutagenesis and genetic disease. Meanwhile, discoveries of new genetic disease phenomena challenge our abilities to conceive and develop research models for their study. It is hoped that the collection of articles in this issue will serve to stimulate interest in scientists of varied disciplines and help focus those interests on the issues surrounding the relationship between environmental mutagens and human genetic disease.

Intestinal iron absorption studies in mouse models of iron-overload

Three mouse strains have been evaluated as suitable models for investigations into the pathogenesis of iron-overload syndromes. Mice with hereditary heterozygous alpha-thalassaemia had moderately raised reticulocyte counts, but were not anaemic and showed little, if any, iron loading. In contrast, mice with homozygous beta-thalassaemia showed microcytic anaemia, reticulocytosis and splenomegaly. Iron-loading was marked, progressive with age and mainly confined to the spleen. Liver iron-loading increased until the age of 7-8 weeks, with no further increase over successive weeks. Although intestinal iron absorption was modestly increased due to enhanced mucosal uptake, the majority of the 'excess' liver and spleen iron could be accounted for by re-distribution of iron from the erythrocytic compartment. Homozygous hypotransferrinaemic mice, with approximately 1-2% of normal plasma transferrin levels, were markedly anaemic with hypochromic microcytic erythrocytes. Intestinal iron absorption increased 3-4-fold (predominantly due to changes in mucosal transfer), as compared to wild-type controls and heterozygotes, and was ascertained to be a major factor causing the marked hepatic iron overload. Heterozygous hypotransferrinamic mice, with over half normal plasma transferrin levels and a mild degree of hepatic iron loading, showed very similar characteristics to wild-type controls. Thus, of the three models, hpx/hpx mice showed the greatest enhancement in intestinal iron absorption and net iron-loading and provides a suitable animal model of spontaneous iron-overload. Comparison of iron absorption values between the models suggests that reticulocytes cannot account for the enhanced absorption seen in the hpx/hpx mice.

High expression of human beta S- and alpha-globins in transgenic mice: hemoglobin composition and hematological consequences

A line of transgenic mice (alpha H beta S-11; where alpha H is human alpha-globin) was created in which the human beta S and human alpha 2 globin genes, each linked to the beta-globin locus control region, were cointegrated into the mouse genome. On a normal genetic background, the transgenic mice produced 36% human beta S-globin chains with an alpha H/beta S ratio of 1.3. Higher levels of beta S were achieved by breeding the transgenic mice with mutant mice carrying a mouse beta major-globin gene deletion. Mice heterozygous for the beta major deletion (alpha H beta S[beta MD]; MD, mouse deletion) had 54% beta S with an alpha H/beta S ratio of 1.0; mice homozygous for the beta major deletion (alpha H beta S[beta MDD]) had 72.5% beta S and an alpha H/beta S ratio of 0.73. Because mouse alpha chains inhibit hemoglobin (Hb) S polymerization, we bred the mice to heterozygosity for a mouse alpha-globin deletion. These mice (alpha H beta S[alpha MD beta MDD]) had an increased alpha H/beta S ratio of 0.89 but expressed 65% beta S. Expression of the human genes cured the thalassemic phenotype associated with the murine beta major deletion. Transgenic alpha H beta S[beta MDD] mice had normal hematocrit and Hb and somewhat elevated reticulocytes (6% vs. 3% for control), whereas the mice carrying the alpha-globin deletion (alpha H beta S[alpha MD beta MDD]) had a normal hematocrit and Hb and more elevated reticulocytes (10.3 +/- 7.6% vs. 3.4 +/- 1.0%). Expression of the transgene restored a normal distribution of erythrocyte densities when compared to thalassemic mice; however, the average mean corpuscular Hb concentration of alpha H beta S[beta MDD] mice increased to 35.7 g/dl (vs. control 33.7 g/dl) whereas that of alpha H beta S[alpha MD beta MDD] mice was further elevated to 36.3 g/dl. The intrinsic oxygen affinity was increased in transgenic mouse erythrocytes at 280 milliosmolal, and the PO2 at midsaturation of alpha H beta S[alpha MD beta MDD] erythrocytes was higher than that of alpha H beta S[beta MDD] cells (37.4 +/- 2 vs. 33.5 +/- 1 mmHg). The higher values of the mean corpuscular Hb concentration and intrinsic PO2 at midsaturation, which favor in vivo sickling, may explain the slightly more severe hematological picture in alpha H beta S[alpha MD beta MDD] mice. We conclude that the transgenic mouse with high Hb S expression does not exhibit adult anemia but does have abnormal hematological features: increased erythrocyte density, high oxygen affinity, and reticulocytosis with increased stress reticulocytes.

A transgenic mouse model of sickle cell disorder

A single base-pair mutation (beta s) in codon 6 of the human beta-globin gene, causing a single amino-acid substitution, is the cause of sickle cell anaemia. The mutant haemoglobin molecule, HbS, polymerizes when deoxygenated and causes deformation of the erythrocytes to a characteristic 'sickled' shape. Sickling of cells in small vessels causes painful crises and other life-threatening complications. Although the molecular basis for sickle cell anaemia has been known for 30 years, no definitive treatment is available. An animal model of sickle cell anaemia would not only allow a detailed analysis of the factors that initiate erythrocyte sickling in vivo and of the pathophysiology of the disease, but would also permit the development of novel approaches to the treatment of the disease. By using the dominant control region sequences from the human beta-globin locus, together with human alpha- and beta s-globin genes, we have obtained three transgenic mice with HbS levels ranging from 10 to 80% of total haemoglobin in their red cells. As observed in homozygous and heterozygous Hbs patients, the erythrocytes of this mouse sickle readily on deoxygenation. Irreversibly sickled cells, which are characteristic of sickle-cell patients homozygous for beta s, are also observed in the peripheral blood of the mouse with high levels of HbS.

High-level, erythroid-specific expression of the human alpha-globin gene in transgenic mice and the production of human hemoglobin in murine erythrocytes

Using the dominant control region (DCR) sequences that flank the beta-globin gene locus, we have been able to achieve high-level expression of the human alpha-globin gene in transgenic mice. Expression in fetal liver and blood is copy number dependent and at levels comparable to that of the endogenous mouse alpha-globin genes. Transgenic fetuses with high-copy numbers of the transgene suffer severe anemia and die before birth. Using a construct with both the human alpha- and beta-globin genes and the beta-globin DCR, live mice with low-copy numbers were obtained. Both human globin genes are expressed at high levels in adult red cells to give human hemoglobin HbA in amounts equal to or greater than endogenous mouse hemoglobin. Expression of HbA in murine red cells is not accompanied by any increase in mean corpuscular volume (MCV) or mean corpuscular hemoglobin concentration (MCHC). However, these transgenic mice tend to have an increased number of reticulocytes in peripheral blood; consistent with some degree of hemolysis. Metabolic labeling experiments showed balanced mouse globin synthesis, but imbalanced human globin synthesis, with an alpha/beta biosynthetic ratio of approximately 0.6. Thus, these mice have mild anemia. These results are discussed with relation to the coordinate regulation of alpha- and beta-globin synthesis in erythroid tissues.

High-level erythroid expression of human alpha-globin genes in transgenic mice

The human alpha 1-globin gene was fused downstream of two erythroid-specific DNase I super-hypersensitive sites that are normally located upstream of the human beta-globin locus. This construct was injected into fertilized mouse eggs, and expression was analyzed in 16-day fetal livers and brains. All 11 fetuses that contained intact copies of the transgene expressed correctly initiated human alpha-globin mRNA in the erythroid fetal liver but not in brain. Levels of expression ranged from 4% to 337% of endogenous mouse beta-globin mRNA. A human alpha-globin construct that did not contain super-hypersensitive sites was not expressed. These results demonstrate that human beta-globin locus activation sequences can stimulate high levels of human alpha-globin gene expression in erythroid tissue of transgenic mice. The results also provide a foundation for experiments designed to coexpress human alpha- and beta-globin genes in transgenic mice and suggest a feasible approach for production of a mouse model for human sickle cell disease.

Characterization of mouse reticulocyte free globin mRNP

Globin messenger ribonucleoprotein particles (free and polysomal) from mouse reticulocyte lysates were characterized for their mRNA composition, translational activity as well as the proteins in direct contact with them. In contrast to the homogeneous single-peak distribution of rabbit and duck reticulocyte free mRNPs, mouse free mRNP particles were heterogeneously dispersed on the sucrose density gradient into two major domains called region I and region II. Region I appeared enriched with alpha-globin mRNP and region II with beta-globin mRNP. mRNP from both regions was translationally active. Examination of lysates prepared from beta-thalassemic mice revealed a reduction of translatable beta minor mRNP within region I, supporting the hypothesis of a compensatory recruitment of beta minor free mRNP into polysomes in beta-thalassemic mice.

Introduction and expression of the human Bs-globin gene in transgenic mice

Owing to the episodic and unpredictable nature of the sickling crisis, many aspects of the disease sickle cell anemia have resisted in vivo analysis. The lack of an animal model has hindered the pathophysiological investigation of this disease, as well as deterred the development of pharmacological therapies. The transgenic mouse system offers a new means for creating animals that make a specified mutant gene product, and we have used this system to create a series of mice that contain the human beta s-globin gene. These animals express this gene in the appropriate tissues and at the same point in development as the adult mouse globin genes are expressed. We have crossed the human beta s-containing transgenic mice with a beta-thalassemic mouse line and examined the hemoglobins produced by these mice. Their red cells contain 10% mouse alpha/human beta s hybrid hemoglobin, which partially corrects the thalassemic phenotype of the homozygous beta-thalassemic animals. Though the red cells do not sickle, other properties of the human beta s gene in these mice indicate the potential for the eventual development of a transgenic animal model for sickle cell anemia.

Independent expression of the two mouse adult beta-globin genes

Dot blot hybridization was used to determine the relative amounts of the beta-major and beta-minor globin RNAs present in reticulocytes of mice at 14.5 and 17.5 days of gestation, newborns, and adults of the Hbab/Hbab;Hbbs2/Hbbs2 globin genotype. RNAs isolated from embryonic yolk sac, fetal liver, and adult reticulocytes were hybridized with the following labeled DNA probes: alpha-1, beta-minor specific, and beta-major specific. The level of beta-sminor RNA in reticulocytes at 14.5 and 17.5 days of gestation is nearly the same as in induced reticulocytes of adult mice. In contrast, the level of beta-s2major RNA in reticulocytes at 14.5 days of gestation is 0.23 X and at 17.5 days of gestation is 0.66 X the amount found in reticulocytes of adult mice. These results correlate well with earlier observations that the beta-sminor globin gene approaches its normal adult level of expression by 14.5 days of gestation, whereas the beta-s2major globin gene expression increases between 14.5 days of gestation and 6 days postnatally. They indicate that the differential expression of beta-sminor and beta-s2major globins during development is regulated at the level of transcription. Expression of the beta-minor globin gene in reticulocytes of adult normal mice is not maximal, however, because the levels of the beta-minor globin and its RNA are increased further in reticulocytes of thalassemic mice.

Red cell volume distribution curves and intracellular globin chain precipitation in the alpha-thalassaemic mouse, Hbath-J

Red cell volume distribution curves were studied in alpha-thalassaemic mice (Hbath-J/+ mice) and normal mice (+/+ mice) of various ages. Individual Hbath-J/+ mice could not be reliably distinguished from their +/+ littermates on the basis of modal cell volume either at birth or during the first 3 weeks of life. However, between the ages of 4 and 30 weeks Hbath-J/+ mice displayed a degree of microcytosis that enabled them to be readily distinguished from their normal littermates using the criterion of modal red cell volume. Preliminary studies of alpha:beta globin chain synthesis ratios given by blood reticulocytes of Hbath-J/+ and +/+ mice after incubation with 3H-leucine for 5 min and 2 h suggest that there is little or no proteolysis of excess beta-chains in the alpha-thalassaemic mouse. Electron microscope studies revealed that the erythroblasts, marrow reticulocytes and circulating red cells of Hbath-J/+ but not +/+ mice contain stellate and branching intracytoplasmic inclusions, presumed to consist of precipitated beta-chains. These inclusions were ultrastructurally similar to the inclusions which have been previously reported in the erythroblasts and marrow reticulocytes of people with various alpha-thalassaemia syndromes. The proportion of erythropoietic cell profiles with inclusions was higher in Hbath-J/+ mice (in which two of the four alpha-globin genes are deleted) than in Thai patients with HbH disease (in whom there is usually a deletion of three of the four alpha-globin genes); this finding is probably related to a relatively low proteolytic capacity in the more mature mouse erythroid cells when compared with human cells. The presence of inclusion-containing red cells (mainly reticulocytes) in the peripheral blood of unsplenectomized Hbath-J/+ animals contrasts with the absence of such cells in unsplenectomized patients with alpha-thalassaemia I trait and HbH disease; this difference seems to be at least partly due to a poorly-developed pitting function in the mouse spleen.

Gene transfer into mouse embryos

Gene transfer into the murine genome was accomplished nearly a decade ago by use of chimeras and teratocarcinomas; however, the low frequencies of transfer into the germ line and other difficulties stemming from mosaicism and karyotypic abnormalities in chimeric mice have limited the general usefulness of this procedure in achieving transformation in mammalian embryos. The introduction of cloned genes into teratocarcinoma cells, selection for a mutant phenotype, and transfer of those cells into mouse embryos holds some promise as a technique to employ mouse chimeras for gene transfer into mice. Infection with animal viruses and retroviral vectors provides another way to introduce exogenous DNA into mouse embryos. Infection with Mo-MuLV has been utilized to characterize the relationship between sites of integration and gene function in developing and adult mice. Gene transfer by microinjection of cloned recombinant DNA has been used by many laboratories for the transfer of DNAs into mouse embryos. The factors affecting transformation frequencies and sites of integration are unknown at present, although it seems that integration is not strictly mediated by homology-dependent events. Many genes have been introduced into mouse embryos by these procedures and many of these are expressed at high levels in appropriate tissues. No realistic possibility exists at the present time for the utilization of embryo gene transfer in the medical field for the correction of genetic defects for several reasons. First, in order to effectively provide "gene therapy" it would be necessary to determine the genotype of each recipient egg, a technical impossibility. The genetic diseases that would be amenable to germ line intervention are recessive diseases and there would be only a 25% chance of any one embryo derived from heterozygous parents being a homozygous recessive. Moreover, it would be impossible to distinguish the normal from abnormal embryos. Second, the frequencies of transformation are so low as to exclude work on human beings on ethical grounds. Third, the parameters effecting chromosomal integration sites and gene expression have not been fully characterized. Until it becomes experimentally possible to target the newly introduced DNA into expressable chromosomal sites and actively replace or supplement defective genes, the possibility of gene therapy through manipulation of embryos is remote. Yet, efforts to provide gene therapy in somatic tissues have been promising, leading to expression of a modified phenotype (Anderson, 1984). In contrast to embryo gene therapy, gene therapy in somatic tissues would not lead to germ line propagation of the manipulated genotype.(ABSTRACT TRUNCATED AT 400 WORDS)

Hematology of a murine beta-thalassemia: a longitudinal study

Mice homozygous for a spontaneous mutation, in which the beta-major globin gene is deleted, have clinical symptoms of beta-thalassemia. These mice have a hypocellular, hypochromic, microcytic anemia that becomes more severe with increasing age. The defective red cell morphology, decreased osmotic fragility of erythrocytes and shortened red cell life span found in beta-thalassemic mice are similar to those observed in human beta-thalassemia. Synthesis of beta-globin is depressed but not as much as might be expected because the expression of the beta-minor globin gene is enhanced to encode two to three times more globin than in normal mice. Splenomegaly, an enlarged pool of stem cells for erythropoiesis, and iron overloading occur in older mice. The fact that these mice remain moderately healthy makes them a very suitable animal model in which to develop and test alternative techniques of gene therapy that could be successfully applied to the treatment of human thalassemia. Homozygous beta-thalassemic mice have large deposits of iron in their tissues, which might make these mice also useful for in vivo tests of the effectiveness and possible long-term side effects of newly developed iron chelators.

Animal model of human disease: thalassemia: alpha-thalassemia in laboratory mice

Images

Analysis of a mouse alpha-globin gene mutation induced by ethylnitrosourea

A DBA/2 mouse treated with ethylnitrosourea sired an offspring whose hemoglobin showed an extra band following starch gel electrophoresis. The variant hemoglobin migrated to a more cathodal position in starch gel. Isoelectric focusing indicated that chain 5 of the mutant hemoglobin migrated to a more cathodal position than the normal chain 5 from DBA/2 mice and that the other alpha-globin, chain 1, was not affected. On focusing gels the phenotype of the mutant allele, Hbay9, was expressed without dominance to normal chain 5, and Hbay9/Hbay9 homozygotes were fully viable in the laboratory. The molecular basis for the germinal mutation was investigated by analyzing the amino acid sequence of chain 5y9, the mutant form of alpha-chain 5. A single amino acid substitution (His leads to Leu) at position 89 was found in chain 5y9. We propose that ethylnitrosourea induced an A leads to T transversion in the histidine codon at position 89 (CAC leads to CTC). This mutation has apparently not been observed previously in humans, mice or other mammals, and its novel occurrence may be indicative of other unusual mutational events that do not ordinarily occur in the absence of specific mutagen exposure.

Deletions in the alpha-globin gene complex in alpha-thalassemic mice

Three induced, heritable mutations in the mouse cause alpha-thalassemias. The adult alpha-globin genes on each mutant chromosome are no longer expressed. Embryos heterozygous for one normal and any of the three mutant chromosomes also seem to be deficient in embryonic alpha-globin-like x-globin, suggesting that the x-globin gene is nearby and also inactivated. A normal genetic polymorphism for a specific EcoRI site in or around the mouse alpha-globin gene complex has been used here to show that each of the three mutated chromosomes has a deletion that includes the segment of a 12-kilobase EcoRI band which normally carries one of the two adult alpha-globin genes. The deletion of the comparable part of the second alpha-globin gene site is also inferred. Nonetheless, a 4.7-kilobase EcoRI segment which carries a characterized alpha-globin-like pseudogene is still present in each mutant. These mutations were recovered after triethylenemelamine or x-ray treatments.

Mouse alpha-globin genes and alpha-globin-like pseudogenes are not syntenic

A genetic polymorphism for a Bgl I endonuclease site near the alpha-globin-like pseudogene alpha-4 of C57BL/6 and C3H/HeN mice was used to show that alpha-4 was not affected by three independent mutations in which the adult globin genes alpha-1 and alpha-2 were deleted. These results indicated that alpha-4 might not be located adjacent to the adult alpha-globin genes on chromosome 11. Restriction endonuclease analysis of DNA of a primary clone of a Chinese hamster--mouse somatic cell hybrid that had lost mouse chromosomes 11 and 18 showed that this clone lacked the adult murine globin genes alpha-1 and alpha-2 but it did contain the alpha-globin-like pseudogenes alpha-3 and alpha-4. These results indicated that the adult alpha-globin genes and alpha-globin-like pseudogenes are not located on the same chromosome. Similar analyses of several other Chinese hamster--mouse somatic cell hybrids that had segregated other mouse chromosomes indicated that the alpha-globin-like pseudogenes alpha-3 and alpha-4 are located on mouse chromosomes 15 and 17, respectively. These data explain why alpha-3 and alpha-4 were not affected by the three independently induced deletion-type mutations that cause alpha-thalassemia in the mouse.