Cell. Vol. 63, 665-672,
November
16, 1990, Copyright
0 1990 by Cell Press
G lobin Gene Regulation and Switching: Circa 1990 Stuart H. Orkin Division of Hematology/Oncology Children’s Hospital and Dana Farber-Cancer Department of Pediatrics Howard Hughes Medical Institute Harvard Medical School Boston. Massachusetts 02115
Institute
Hemoglobins have assumed a special place in the hearts and minds of devotees of protein structure, gene structure and regulation, cell differentiation, and human genetics. Over forty years ago the association between an amino acid mutation in the 8-globin chain and a disease, sickle cell anemia, was first established. More than ten years ago the human globin genes were among the first mammalian genes to be cloned in bacteriophage and analyzed at the nucleotide level. Identification of mutations in globin genes in inherited disorders of hemoglobin synthesis (thalassemia syndromes) provided the first near-complete description of a disease at the molecular level and the experimental basis for DNA-based prenatal diagnosis. Notwithstanding these achievements, major questions pertaining to the control of globin gene expression and red blood cell development remained unanswered. How do hematopoietic stem cells, which are capable of differentiating into all blood lineages, commit to erythroid development? How is cell-specific gene expression activated and maintained in erythroid progenitors? How is the orderly transition from embryonic to fetal to adult globin synthesis programmed during human ontogeny? What regulatory factors participate in establishment of cellspecific gene expression and gene switching? Finally, can insights gleaned from basic molecular studies be applied in design of improved strategies for clinical management, and perhaps cure, of patients afflicted with the major disorders of the 8-globin gene? Since 1978 a diverse group of investigators interested in these problems have convened at a biannual meeting, jointly organized since its inception by Arthur Nienhuis (National Institute of Health) and George Stamatoyannopoulos (University of Washington). Held for the seventh time September 8-11 at Airlie House, Virginia, the conference has served as the major forum for the field of erythroid cell gene regulation. After traversing a difficult period during which many dispaired of learning how globin genes are normally regulated, the tenor of the field has changed. Once again, the study of hemoglobin is ripe to provide novel insights into gene regulation and cellular differentiation in eukaryotes. Here findings presented at the meeting are reviewed in the context of emerging concepts in the field. C&Elements Globin Gene Regulation: A Miss Is As Good As a Mile Vertebrate globins are encoded in small a- and B-like gene
Meeting Review
families that are expressed in a developmental sequence. This “switching” of globins is accomplished at the transcriptional level. As development proceeds, the site of erythropoiesis shifts from the yolk sac in the early embryo, to the liver in fetal life, and finally to the bone marrow in the adult. The human a-globin cluster, which resides near the telomere of the short arm of chromosome 16, includes an embryonic gene (6) and two expressed adult genes (al and a2) encompassing approximately 30 kb. The 8-globin cluster on the short arm of chromosome 11 spans roughly 60 kb and is comprised of an embryonic gene (E), duplicated fetal (y) genes, and the adult 8 gene, in addition to a minor adult gene (6). As proper assembly of globin tetramers into hemoglobins requires the balanced synthesis of a- and p-like globins, it is presumed that common mechanisms exist to ensure coordinated expression of the unlinked gene clusters. The organization of globin genes in the a- and p-like clusters of other vertebrates differs in detail, but general aspects of gene regulation appear to be otherwise highly conserved. Early efforts to define the cis-elements responsible for specifying globin gene expression in erythroid progenitors concentrated on DNA sequences just upstream or downstream of globin genes. In the assessment of putative regulatory sequences, transgenic mice have proven invaluable. While the human y- and 8-globin genes themselves contain sequences that direct developmentally appropriate, erythroid-specific expression in transgenic mice, the level of expression observed is very low, not dependent on gene copy number, and particularly sensitive to the site of integration in the host genome. Therefore, local sequences are not sufficient to establish normal globin gene regulation. Locus Control Region (LCR): Far, Far, Away . . . The observations that rare deletions upstream of the 8-globin gene inactivate its expression and that erythroidspecific DNAase I hypersensitive sites are present in a broad region of more than 20 kb 5’ of the .s-globin gene (Tuan et al., 1985) led to the hypothesis that important control sequences might reside at a distance from the 6 gene. Demonstration that the DNAase I hypersensitivity region confers high level, position-independent expression on linked globin, or heterologous, genes in transgenic mice (Grosveld et al., 1987) established this principle and has provided an experimental basis for subsequent studies of globin gene regulation aid switching. As these control sequences have been termed either the dominant control region (DCR) or locus-activating region (LAR) and numbered differently on the two sides of the Atlantic, participants at the conference were asked to ratify new nomenclature presented in Figure 1. Henceforth, sequences that confer position-independent expression of linked genes in transgenic mice will be known as locus control regions (LCRs). Studies reported at the conference focused on the structure of the 8-LCR, the identification of the elusive LCR for the a-globin complex, and
Cell 666
FiHS- 4 3
2 1
3’HS-1 Gy Ay
P
Table 1. Expression
of Globin Gene-LCR Expression
Construct
LCR
Gene(s) Figure 1. Erythroid DNAase f3-globin Gene Complex
I Hypersensitive
Sites in the Human
In the new nomenclature sites are numbered proceeding in the Sand 3’ direction from the 6 gene. 5’ HS-1 through HS-4 correspond to sites I-IV (-6.1, -10.9, -14.7, -16 kb) of the LAR (Forrester et al., 1966) and to sites 4-l of the DCR (Grosveld et al., 1967). The 6-LCR includes 5’ HS-1 through HS-4. The role of 3’ HS-1 in LCR function is uncertain.
effects of the LCR on chromatin organization and hemoglobin switching. The f3-LCR includes four prominent hypersensitive sites (HSs), designated 5’ HS-1 through HS-4, situated upstream of the E gene (Figure 1). Dissection of the LCR by several groups reveals that its full activity, as defined experimentally in transgenic mice, can be retained in “mini” or “micro” LCRs in which small segments (0.5-l kb) including each HS are joined. The Phenomenology of Switching: Two Heads Are Better Than One Attachment of the 5-LCR to single or linked genes has provided a powerful system for examining the regulation of globin genes in the context of a developing mouse. While endogenous 5 gene switching in the mouse proceeds from embryonic to adult genes without a true fetal period, available data tend to support the use of the transgenie mouse as an adequate (though, perhaps not perfect) model for human gene switching. As a reflection of differences between mouse and human ontogeny, it should be noted that y-globin transgenes are expressed in yolk sac rather than fetal life. An extraordinary (and sometimes bewildering) array of LCR-P-gene complex constructs were presented and discussed. Inclusion of slightly different fragments of the LCR and globin genes, as well as assay at different days of mouse development, prevents strict comparison among laboratories. At the risk of minimizing small but potentially significant differences, I have idealized the data to relate conclusions on which nearly all participants could agree. Table 1 summarizes the experimental observations and includes previously published results pertaining to expression of human p-like genes in the absence of the LCR. These studies reveal the following: l As previously reported (Grosveld et al., 1987), high level expression of globin genes in transgenic mice requires linkage of the LCR in cis. However, LCR-5 (as well as LCR-y) constructs are expressed at all developmentally stages. Thus, the LCR can override stage-specific developmental cues of individual globin genes. l As reported more recently (Behringer et al., 1990; Enver et al., 1990) seemingly appropriate switching from y-to f3-globin occurs during mouse development when the y and 5 genes are introduced together in cis to the LCR. Thus, these genes are in competition with each other for
LCR
E E A.prE Y Y P s YW
+ +
Y6 BY
+ +
+ + +
Yolk Sac (9-10 day) 4+ 4+ 1+ 4+ 4+ Y4+ BY4+ 84+
Constructs
in Transgenic
in Mice
Mice
Fetal Liver (12-14 day)
Adult
4+ 3+ l+ 4+ Yl+ P4+ Y4+ 8+
2+ l+ 4+ y
November
16, 1990, Copyright
0 1990 by Cell Press
G lobin Gene Regulation and Switching: Circa 1990 Stuart H. Orkin Division of Hematology/Oncology Children’s Hospital and Dana Farber-Cancer Department of Pediatrics Howard Hughes Medical Institute Harvard Medical School Boston. Massachusetts 02115
Institute
Hemoglobins have assumed a special place in the hearts and minds of devotees of protein structure, gene structure and regulation, cell differentiation, and human genetics. Over forty years ago the association between an amino acid mutation in the 8-globin chain and a disease, sickle cell anemia, was first established. More than ten years ago the human globin genes were among the first mammalian genes to be cloned in bacteriophage and analyzed at the nucleotide level. Identification of mutations in globin genes in inherited disorders of hemoglobin synthesis (thalassemia syndromes) provided the first near-complete description of a disease at the molecular level and the experimental basis for DNA-based prenatal diagnosis. Notwithstanding these achievements, major questions pertaining to the control of globin gene expression and red blood cell development remained unanswered. How do hematopoietic stem cells, which are capable of differentiating into all blood lineages, commit to erythroid development? How is cell-specific gene expression activated and maintained in erythroid progenitors? How is the orderly transition from embryonic to fetal to adult globin synthesis programmed during human ontogeny? What regulatory factors participate in establishment of cellspecific gene expression and gene switching? Finally, can insights gleaned from basic molecular studies be applied in design of improved strategies for clinical management, and perhaps cure, of patients afflicted with the major disorders of the 8-globin gene? Since 1978 a diverse group of investigators interested in these problems have convened at a biannual meeting, jointly organized since its inception by Arthur Nienhuis (National Institute of Health) and George Stamatoyannopoulos (University of Washington). Held for the seventh time September 8-11 at Airlie House, Virginia, the conference has served as the major forum for the field of erythroid cell gene regulation. After traversing a difficult period during which many dispaired of learning how globin genes are normally regulated, the tenor of the field has changed. Once again, the study of hemoglobin is ripe to provide novel insights into gene regulation and cellular differentiation in eukaryotes. Here findings presented at the meeting are reviewed in the context of emerging concepts in the field. C&Elements Globin Gene Regulation: A Miss Is As Good As a Mile Vertebrate globins are encoded in small a- and B-like gene
Meeting Review
families that are expressed in a developmental sequence. This “switching” of globins is accomplished at the transcriptional level. As development proceeds, the site of erythropoiesis shifts from the yolk sac in the early embryo, to the liver in fetal life, and finally to the bone marrow in the adult. The human a-globin cluster, which resides near the telomere of the short arm of chromosome 16, includes an embryonic gene (6) and two expressed adult genes (al and a2) encompassing approximately 30 kb. The 8-globin cluster on the short arm of chromosome 11 spans roughly 60 kb and is comprised of an embryonic gene (E), duplicated fetal (y) genes, and the adult 8 gene, in addition to a minor adult gene (6). As proper assembly of globin tetramers into hemoglobins requires the balanced synthesis of a- and p-like globins, it is presumed that common mechanisms exist to ensure coordinated expression of the unlinked gene clusters. The organization of globin genes in the a- and p-like clusters of other vertebrates differs in detail, but general aspects of gene regulation appear to be otherwise highly conserved. Early efforts to define the cis-elements responsible for specifying globin gene expression in erythroid progenitors concentrated on DNA sequences just upstream or downstream of globin genes. In the assessment of putative regulatory sequences, transgenic mice have proven invaluable. While the human y- and 8-globin genes themselves contain sequences that direct developmentally appropriate, erythroid-specific expression in transgenic mice, the level of expression observed is very low, not dependent on gene copy number, and particularly sensitive to the site of integration in the host genome. Therefore, local sequences are not sufficient to establish normal globin gene regulation. Locus Control Region (LCR): Far, Far, Away . . . The observations that rare deletions upstream of the 8-globin gene inactivate its expression and that erythroidspecific DNAase I hypersensitive sites are present in a broad region of more than 20 kb 5’ of the .s-globin gene (Tuan et al., 1985) led to the hypothesis that important control sequences might reside at a distance from the 6 gene. Demonstration that the DNAase I hypersensitivity region confers high level, position-independent expression on linked globin, or heterologous, genes in transgenic mice (Grosveld et al., 1987) established this principle and has provided an experimental basis for subsequent studies of globin gene regulation aid switching. As these control sequences have been termed either the dominant control region (DCR) or locus-activating region (LAR) and numbered differently on the two sides of the Atlantic, participants at the conference were asked to ratify new nomenclature presented in Figure 1. Henceforth, sequences that confer position-independent expression of linked genes in transgenic mice will be known as locus control regions (LCRs). Studies reported at the conference focused on the structure of the 8-LCR, the identification of the elusive LCR for the a-globin complex, and
Cell 666
FiHS- 4 3
2 1
3’HS-1 Gy Ay
P
Table 1. Expression
of Globin Gene-LCR Expression
Construct
LCR
Gene(s) Figure 1. Erythroid DNAase f3-globin Gene Complex
I Hypersensitive
Sites in the Human
In the new nomenclature sites are numbered proceeding in the Sand 3’ direction from the 6 gene. 5’ HS-1 through HS-4 correspond to sites I-IV (-6.1, -10.9, -14.7, -16 kb) of the LAR (Forrester et al., 1966) and to sites 4-l of the DCR (Grosveld et al., 1967). The 6-LCR includes 5’ HS-1 through HS-4. The role of 3’ HS-1 in LCR function is uncertain.
effects of the LCR on chromatin organization and hemoglobin switching. The f3-LCR includes four prominent hypersensitive sites (HSs), designated 5’ HS-1 through HS-4, situated upstream of the E gene (Figure 1). Dissection of the LCR by several groups reveals that its full activity, as defined experimentally in transgenic mice, can be retained in “mini” or “micro” LCRs in which small segments (0.5-l kb) including each HS are joined. The Phenomenology of Switching: Two Heads Are Better Than One Attachment of the 5-LCR to single or linked genes has provided a powerful system for examining the regulation of globin genes in the context of a developing mouse. While endogenous 5 gene switching in the mouse proceeds from embryonic to adult genes without a true fetal period, available data tend to support the use of the transgenie mouse as an adequate (though, perhaps not perfect) model for human gene switching. As a reflection of differences between mouse and human ontogeny, it should be noted that y-globin transgenes are expressed in yolk sac rather than fetal life. An extraordinary (and sometimes bewildering) array of LCR-P-gene complex constructs were presented and discussed. Inclusion of slightly different fragments of the LCR and globin genes, as well as assay at different days of mouse development, prevents strict comparison among laboratories. At the risk of minimizing small but potentially significant differences, I have idealized the data to relate conclusions on which nearly all participants could agree. Table 1 summarizes the experimental observations and includes previously published results pertaining to expression of human p-like genes in the absence of the LCR. These studies reveal the following: l As previously reported (Grosveld et al., 1987), high level expression of globin genes in transgenic mice requires linkage of the LCR in cis. However, LCR-5 (as well as LCR-y) constructs are expressed at all developmentally stages. Thus, the LCR can override stage-specific developmental cues of individual globin genes. l As reported more recently (Behringer et al., 1990; Enver et al., 1990) seemingly appropriate switching from y-to f3-globin occurs during mouse development when the y and 5 genes are introduced together in cis to the LCR. Thus, these genes are in competition with each other for
LCR
E E A.prE Y Y P s YW
+ +
Y6 BY
+ +
+ + +
Yolk Sac (9-10 day) 4+ 4+ 1+ 4+ 4+ Y4+ BY4+ 84+
Constructs
in Transgenic
in Mice
Mice
Fetal Liver (12-14 day)
Adult
4+ 3+ l+ 4+ Yl+ P4+ Y4+ 8+
2+ l+ 4+ y
