`
`R E V I E W A R T I C L E
`
`The human b-globin locus control region
`A center of attraction
`
`Padraic P. Levings and Jo¨ rg Bungert
`
`Department of Biochemistry and Molecular Biology, Gene Therapy Center, Center for Mammalian Genetics, College of Medicine,
`University of Florida, Gainesville, FL, USA
`
`The human b-globin gene locus is the subject of intense
`study, and over the past two decades a wealth of information
`has accumulated on how tissue-specific and stage-specific
`expression of its genes is achieved. The data are extensive and
`it would be di(cid:129)cult, if not impossible, to formulate a com-
`prehensive model integrating every aspect of what is cur-
`rently known. In this review, we introduce the fundamental
`characteristics of globin locus regulation as well as questions
`on which much of the current research is predicated. We then
`outline a hypothesis that encompasses more recent results,
`focusing on the modification of higher-order chromatin
`structure and recruitment of transcription complexes to the
`globin locus. The essence of this hypothesis is that the locus
`control region (LCR) is a genetic entity highly accessible to
`
`and capable of recruiting, with great e(cid:129)ciency, chromatin-
`modifying, coactivator, and transcription complexes. These
`complexes are used to establish accessible chromatin
`domains, allowing basal factors to be loaded on to specific
`globin gene promoters in a developmental stage-specific
`manner. We conceptually divide this process into four steps:
`(a) generation of a highly accessible LCR holocomplex;
`(b) recruitment of transcription and chromatin-modifying
`complexes to the LCR; (c) establishment of chromatin
`domains permissive for transcription; (d) transfer of tran-
`scription complexes to globin gene promoters.
`
`Keywords: chromatin domains; globin genes;
`transcription; locus control region; transcription.
`
`intergenic
`
`O R G A N I Z A T I O N A N D S T R U C T U R E
`O F T H E H U M A N b-G L O B I N L O C U S
`The five genes of the human b-globin locus are arranged in a
`linear array on chromosome 11 and are expressed in a
`developmental stage-specific manner in erythroid cells
`(Fig. 1) [1]. The e-globin gene is transcribed in the embry-
`onic yolk sac and located at the 5¢ end. After the switch in
`the site of hematopoiesis from the yolk sac to the fetal liver,
`the e-gene is repressed and the two c-globin genes, located
`downstream of e, are activated. In a second switch,
`completed shortly after birth, the bone marrow becomes
`the major site of hematopoiesis, coincident with activation
`of the adult b-globin gene, while the c-globin genes become
`silenced. The d-globin gene is also activated in erythroid cells
`derived from bone marrow hematopoiesis but is only
`expressed at levels less than 5% of that of the b-globin gene.
`The complex program of
`transcriptional regulation
`leading to the differentiation and developmental stage-
`specific expression in the globin locus is mediated by DNA-
`regulatory sequences located both proximal and distal to the
`
`Correspondence to J. Bungert, Department of Biochemistry and
`Molecular Biology, Gene Therapy Center, Center for Mammalian
`Genetics, College of Medicine, University of Florida, 1600 SW Archer
`Road, Gainesville, FL 32610, USA. Fax: + 352 392 2953,
`Tel.: + 352 392 0121, E-mail: jbungert@college.med.ufl.edu
`Abbreviations: LCR, locus control region; HS, hypersensitive; EKLF,
`erythroid kru¨ ppel-like factor; MEL cells, murine erythroleukemia
`cells; ICD, interchromosomal domain; HLH, helix–loop–helix.
`(Received 15 November 2001, revised 16 January 2002, accepted
`21 January 2002)
`
`gene-coding regions. The most prominent distal regulatory
`element in the human b-globin locus is the locus control
`region (LCR), located from about 6 to 22 kb upstream of
`the e-globin gene [2–4]. The LCR is composed of several
`domains that exhibit extremely high sensitivity to DNase I
`in erythroid cells (called hypersensitive, or HS, sites), and is
`required for high-level globin gene expression at all develop-
`mental stages [5].
`The entire b-globin locus remains in an inactive DNase
`I-resistant chromatin conformation in cells in which the
`globin genes are not expressed. In erythroid cells, the entire
`locus shows a higher degree of sensitivity to DNase I,
`indicating that it is in a more open and accessible chromatin
`configuration [6]. Studies analyzing the human b-globin
`locus in transgenic mice have shown that sensitivity to
`DNase I in specific regions of the globin locus varies and
`depends on the developmental stage of erythropoiesis (yolk
`sac, fetal liver, adult spleen) [7]. The LCR remains sensitive
`to DNase I at all developmental stages, whereas sensitivity
`to DNase I in the region containing the e-globin and
`c-globin genes is higher in embryonic cells, and DNase I
`sensitivity in the region containing the d-globin and b-globin
`gene is higher in adult erythroid cells [7].
`This review focuses on the regulation of the human
`b-globin gene locus, and we would like to refer the reader to
`another recent review that compares the regulation of
`different complex gene loci [8].
`
`D E V E L O P M E N T A L S T A G E - S P E C I F I C
`E X P R E S S I O N O F T H E G L O B I N G E N E S
`
`The stage-specific activation and repression of the individual
`globin genes during development is regulated by various
`
`GENENTECH 2013
`GENZYME V. GENENTECH
`IPR2016-00383
`
`
`
`1590 P. P. Levings and J. Bungert (Eur. J. Biochem. 269)
`
`(cid:211) FEBS 2002
`
`e-promoter EKLF binding site leads to expression of the
`e-globin gene at the adult stage [15]. This observation
`indicates that repression of the e-globin gene at the definitive
`stage is in part due to proteins that interfere with the
`interaction of the transcriptional activator EKLF.
`There is also increasing evidence for the presence of stage-
`specific factors regulating the expression of the two c-globin
`genes. In particular, it has been shown that CACCC and
`CCAAT motifs are required for activation of the c-globin
`genes. The CACCC element is bound by members of the
`family of kru¨ ppel-like zinc finger (KLF) proteins [16].
`Potential candidates for proteins acting through this
`element are EKLF, FKLF, FKLF-2, and BKLF [17]. The
`CCAAT box interacts with the heterotrimeric protein NF-Y
`[18], which appears to play a role similar to EKLF and may
`recruit chromatin-remodeling activities to the c-globin gene
`promoters at the fetal stage.
`The combined data demonstrate that stage-specific
`factors interacting with individual globin gene promoters
`play important roles in the regulation of local chromatin
`structure and stage-specific gene expression.
`Another important parameter regulating the stage-speci-
`fic activity of the globin genes is the relative position of the
`genes with respect to the LCR [19,20]. Inverting the genes
`relative to the LCR leads to an inappropriate expression of
`the adult b-globin gene at the embryonic stage and the
`absence of e-globin gene expression at all stages [21].
`Although the mechanistic basis for the importance of gene
`order in the globin locus is not entirely clear, it is in
`agreement with the hypothesis that the genes in the globin
`locus are competitively regulated by the LCR [22,23] and
`suggests that repressors restrict the ability of the LCR to
`activate transcription of only one or two genes at specific
`developmental stages. These factors could either modulate
`the chromatin structure around the inactive genes [7] or
`interact with globin gene promoters to prevent the interac-
`tion of a gene with the LCR in a developmental stage-
`specific manner [15].
`
`S T R U C T U R E A N D F U N C T I O N
`O F T H E L C R
`
`The overall organization of the LCR is conserved among
`several vertebrate species. The conservation of individual
`factor-binding sites within the HS core elements implies that
`these sites are important for LCR function [24]. However,
`this by no means leads to the conclusion that transcription
`factor-binding sites that are not conserved are functionally
`irrelevant. Some of these nonconserved sites may mediate
`novel functions acquired during evolution. For example, the
`developmental pattern of globin gene expression in humans
`is quite different from that in mice (Fig. 1) [25].
`Whereas almost all studies agree that the human b-globin
`LCR is required for high-level transcription of all b-like
`globin genes, the question of whether the LCR also
`regulates the chromatin structure over the whole locus is a
`matter of debate. Deletion of the complete LCR from either
`the murine or human locus does not appear to change the
`overall general sensitivity to DNase I of the locus, indicating
`that the LCR is not required for unfolding of higher-order
`chromatin structure [26–28]. Our understanding of the
`structural basis for general DNase I sensitivity of chromatin
`is limited. Loci permissive for transcription are within
`
`Fig. 1. Diagrammatic representation of the human b-globin gene locus
`(not drawn to scale). The five genes of the human b-globin gene locus
`are arranged in linear order reflecting their expression during develop-
`ment. The LCR is represented as the sum of the five HS sites. It should
`be noted that additional HS sites were mapped 5¢ to HS5 [95], but it
`is currently not known whether these sites participate in globin gene
`regulation or whether they are associated with the regulation of
`genes located upstream of the globin locus. The HS core elements are
`200–400 bp in size and separated from each other by more than 2 kb.
`During normal human development, the e-globin gene is expressed in
`the first trimester in erythroid cells derived from yolk sac hematopoi-
`esis. The c-globin genes are expressed in erythroid cells generated in the
`fetal liver until around birth. The adult b-globin gene is expressed
`around birth predominantly in cells derived from bone marrow
`hematopoiesis. The expression pattern of the human globin genes is
`somewhat different when analyzed in the context of transgenic mice
`[96], where the e-globin and c-globin genes are coexpressed in the
`embryonic yolk sac and the b-globin gene is expressed at high levels in
`fetal liver and circulating erythroid cells from bone marrow.
`
`mechanisms. First, genetic information governing the stage-
`specificity for all b-like globin genes is located in gene
`proximal regions. These elements represent transcription
`factor-binding sites that recruit proteins or protein com-
`plexes in a stage-specific manner. Examples exist for the
`presence of both positive and negative acting factors that
`turn genes on or off at a specific developmental stage [1].
`The most extensively studied stage-specific activator is
`EKLF (erythroid kru¨ ppel like factor), which is crucial for
`human b-globin gene expression [9]. Gene-ablation studies
`in mice have shown that EKLF deficiency leads to a specific
`reduction in adult b-globin gene expression, with a
`concomitant increase in expression of the fetal genes [10–
`12]. Associated with the dramatic decrease in adult b-globin
`gene expression is a reduction in DNase I HS site formation
`in the b-globin gene promoter as well as in LCR element
`HS3 [13]. These results demonstrate that EKLF is critically
`required for the expression of the adult b-globin gene and
`suggest that EKLF may exert part of its function by
`changing chromatin structure. Indeed, Armstrong et al. [14]
`showed that EKLF recruits chromatin-remodeling factors
`to the adult b-globin promoter and that this remodeling
`activity was sufficient to activate b-globin gene expression in
`an erythroid-specific manner in vitro. EKLF acts in a
`sequence-specific context to activate transcription of the
`b-globin gene [15]. Although both the e-globin and b-globin
`gene promoters harbor binding sites for EKLF, only the
`b-globin gene is expressed at definitive stages of erythro-
`poiesis. Disruption of direct repeat elements flanking the
`
`
`
`(cid:211) FEBS 2002
`
`Multistep model for locus control region function (Eur. J. Biochem. 269) 1591
`
`domains of general DNase I sensitivity. However, the
`presence of a DNase I-sensitive domain does not indicate
`that all of the genes residing within the domain are
`transcribed or even that they are permissive for transcription
`[28]. In this respect the LCR could be involved in regulating
`chromatin structure beyond the formation of a general
`DNase I-sensitive domain, for example by regulating the
`modification of histone tails (methylation, acetylation,
`phosphorylation) [29].
`It is unquestionable that the LCR provides an open and
`accessible chromatin structure at ectopic sites in transgenic
`assays [5]. Whether this is true for all chromosomal
`positions is not known, because there are no data available
`that demonstrate LCRs function from within a defined
`heterochromatic
`environment. However, globin gene
`expression constructs reveal strong position-of-integration
`effects in transgenic assays in the absence of the LCR,
`suggesting that at most sites the LCR is able to confer an
`accessible chromatin structure. It is important to understand
`that any model describing globin gene regulation must
`address the LCR’s ability to open chromatin and enhance
`globin gene expression at ectopic sites.
`Current models propose that the individual HS core
`elements interact to form a higher-order structure, com-
`monly referred to as the LCR holocomplex [30,31].
`Evidence supporting the holocomplex model came from
`the genetic analysis of mutant LCRs in transgenic assays
`[31–34]. Deletion of individual LCR HS elements in single-
`copy YAC transgenes led to strong reductions in globin
`gene expression and also impaired the formation of
`DNase I HS sites associated with the LCR and the globin
`gene promoters. These data suggest that LCR HS site
`deletions render the LCR unable to protect from position-
`of-integration effects in transgenic studies [32]. In contrast
`with these findings, the consequence of deleting HS sites
`from the endogenous mouse locus on globin gene expression
`is much milder and does not appear to affect the formation
`of remaining HS sites [35–37]. The different results from
`studies of globin locus transgenes vs. endogenous loci could
`be explained in several ways [38]. First, the differences could
`solely be based on the observation that an incomplete LCR
`is not able to confer position-independent chromatin
`opening and gene expression in the globin locus at ectopic
`sites. Secondly, differences in the size of the deleted
`fragments could result in different phenotypes. The most
`severe effects on globin gene expression were observed in
`those transgenes in which only the 200–400-bp (cid:212)core(cid:213)
`enhancer elements were deleted. All the experiments in the
`endogenous murine globin locus removed the cores together
`with the flanking sequences. Finally, it is possible that the
`endogenous murine globin locus contains sequences in
`addition to the LCR that are able to provide an open
`chromatin configuration.
`Recently, Hardison and colleagues analyzed the function
`of LCR HS sites in the presence or absence of the HS core
`flanking sequences in murine erythroleukemia (MEL) cells
`using recombination mediated cassette exchange [39]. At
`several fixed positions,
`the inclusion of
`the flanking
`sequences leads to a synergistic enhancement of expression
`by the combination of HS units, whereas combining the
`core HS elements only additively enhanced reporter gene
`expression. Similarly, May et al.
`[40] showed that the
`combination of HS2, 3, and 4 led to therapeutic levels of
`
`b-globin gene expression in b-thalassemic mice only in the
`presence of sequences flanking the LCR HS cores. Taken
`together, the data suggest that the HS units interact with
`each other to generate an LCR holocomplex, formation of
`which is required for high-level b-globin gene expression.
`The flanking sequences could be important in positioning
`the HS core elements in ways that facilitate their interactions
`[39].
`
`L C R I N T E R A C T I N G P R O T E I N S
`
`Knowledge about the proteins that interact with the LCR
`in vivo is very limited. Here we will focus on more recent
`results describing the activities of specific proteins or protein
`complexes implicated in LCR function. For a more
`comprehensive summary of proteins interacting with regu-
`latory sequences throughout the globin locus, we would like
`to refer the reader to previous reviews [1,24].
`The DNA sequence motifs that are most conserved
`among different species are MARE (maf recognition
`element) and GATA sequences in HS2, 3 and 4, KLF-
`binding sites in HS2 and HS3, and an E-box motif in HS2
`[24]. MARE sequences are bound in vitro by a large number
`of different proteins that all heterodimerize with small maf
`proteins [41]. Individual members of
`this family are
`characterized by the presence of leucine zipper motifs, the
`founding member being NF-E2 (p45) [42]. Other members
`of this family also expressed in erythroid cells are Bach1,
`NRF1 and NRF2 (NF-E2 related factor 1 and 2) [43–45].
`A variety of data suggest a pivotal role for p45 in LCR
`function [42,46]. However, gene ablation studies have
`shown that erythropoiesis is not affected in mice lacking
`NF-E2 (p45), NRF1 or NRF2, suggesting functional
`redundancy among the NF-E2 family members in erythroid
`cells [47–49].
`It should be noted that, although the NF-E2-like proteins
`are all thought to interact with the same DNA-binding site,
`they are structurally different. Bach1 for example contains a
`BTB/Poz domain and forms oligomers while bound to
`DNA in vitro [50]. This observation prompted investigators
`to analyze whether Bach1/small maf heterodimers could
`simultaneously bind to HS2, 3, and 4 and mediate the
`interaction between the core elements [51]. Using atomic
`force microscopy,
`it was shown that Bach1-containing
`heterodimers could indeed cross-link HS sites in vitro,
`indicating that proteins exist that bind to the LCR and are
`able to mediate the interaction of HS sites. Importantly, this
`activity of Bach1 depends on the presence of the BTB/Poz
`domain.
`The CACCC sites in HS2 and HS3 are probably bound
`in vivo by EKLF. First, transgenic mice containing the
`human b-globin locus and lacking EKLF exhibit a reduc-
`tion in the formation of HS3 [13]. In addition, using the
`Pin-Point assay, Lee et al. [52] demonstrated that EKLF
`binds to both HS2 and HS3 in vivo. Interestingly, the
`binding of EKLF to HS3 is reduced in the absence of HS2,
`suggesting some form of communication between these two
`elements [52].
`The GATA sites are bound by either GATA-1 or
`GATA-2, the only two members of the GATA family of
`transcription factors known to be expressed in erythroid
`cells [53]. GATA-1 is one of the earliest markers in red cell
`differentiation and is detectable in progenitor cells that do
`
`
`
`1592 P. P. Levings and J. Bungert (Eur. J. Biochem. 269)
`
`(cid:211) FEBS 2002
`
`not yet express the globin genes [54]. Interestingly, LCR HS
`sites are already detectable in these undifferentiated precur-
`sor cells [55]. These results suggest that GATA-1 may be
`involved in the regulation of chromatin structure at an early
`stage of erythroid differentiation.
`The E-box in HS2 interacts with helix–loop–helix (HLH)
`proteins in vitro, and both USF and Tal1 were shown to
`interact with this element [56,57]. USF is a ubiquitously
`expressed member of the HLH family of proteins and binds
`to DNA as a heterodimer usually composed of USF1 and
`USF2. USF has been implicated in the regulation of many
`genes and normally acts as a transcriptional activator.
`However, it has also been reported to function through
`initiator elements, in which case it mediates the recruitment
`of Pol II transcription complexes [58,59]. Tal1 is hemato-
`poietic specific and appears to function at an early step
`during the specification of hematopoietic progenitor cells
`[60].
`Protein–protein interactions probably play important
`roles in LCR function. We have already discussed the
`multimerization of Bach/maf heterodimers. Other protein–
`protein interactions known to occur among LCR-binding
`proteins involve those between the GATA factors and
`between GATA factors and EKLF, LMO2/Tal1, and Sp1
`[61–63]. In addition, GATA-1, EKLF and NF-E2 (p45)
`were shown to interact with coactivators and acetyltrans-
`ferase activities [64,65]. EKLF has also been demonstrated
`to interact with members of the Swi/SNF family of
`chromatin-remodeling complexes [14]. These results show
`that most proteins binding to one LCR core element have
`the potential to interact with proteins binding to another
`LCR core HS site, which could initiate and stabilize an LCR
`holocomplex. In addition, the results also demonstrate that
`LCR-interacting proteins recruit macromolecular com-
`plexes involved in chromatin remodeling and histone
`acetylation.
`
`R E P L I C A T I O N A N D C H R O M A T I N
`S T R U C T U R E
`The human b-globin locus replicates early in erythroid cells
`and late in nonerythroid cells. Earlier studies suggested that
`the LCR regulates the timing and usage of an origin of
`replication located between the d-globin and b-globin gene
`[66]. This interpretation was based on the observation that a
`large deletion in the human b-globin locus, starting
`immediately upstream of HS1 and spanning about 30 kb,
`inactivates the entire globin locus [66]. The globin genes
`linked to this deletion are not transcribed, the locus becomes
`late replicating, and remains in a DNase I-resistant and
`inaccessible configuration. However, recent analysis of the
`consequence of a targeted deletion of the LCR demonstrates
`that the LCR regulates neither the timing of replication in
`the globin locus nor the usage of the replication origin [67].
`Thus, a putative element regulating replication timing in the
`human b-globin locus must be located 5¢ to the LCR.
`An important question that has to be addressed is
`whether activation of the globin locus and LCR function
`requires replication. During differentiation of erythroid
`cells, the locus undergoes various transitions, the first of
`which is the formation of DNase I HS sites in the LCR [55].
`Does the formation of HS sites at this early stage in
`differentiation require replication? In other words, do the
`
`proteins responsible for HS site formation require a window
`of opportunity after replication to bind and then prevent the
`generation of repressive chromatin structure or do these
`proteins recruit chromatin-remodeling activities that change
`the chromatin structure in a replication-independent man-
`ner? Experiments that indirectly addressed this issue were
`those in which investigators generated heterokaryons with
`MEL cells, which represent definitive erythroid cells that
`express the adult b-globin gene, and human K562 cells,
`which represent primitive erythroid cells that express the
`e-globin gene [68]. These studies showed that trans-acting
`factors in the MEL cells are able to activate transcription of
`the human b-globin gene. Interestingly,
`the onset of
`b-globin gene expression in these experiments occurred
`about 12 h after fusion. Because the globin locus replicates
`early in erythroid cells, these results could be interpreted to
`mean that replication is required for trans-activation of the
`human b-globin genes in the heterokaryons. On the other
`hand, this experiment could also lead to the interpretation
`that the human locus can be activated by transcription
`factors and accessory proteins already present in the adult
`(MEL) erythroid cells. This mode of regulation would be
`similar to the induction of genes by hormone receptors [69].
`However, differences in the two systems may exist, as the
`globin locus is a developmentally regulated locus, the
`expression of which changes as the cell differentiates. Genes
`regulated by hormone and orphan receptors are transcribed
`in mature cells and their expression is regulated by external
`stimuli, i.e. hormones. Obviously more studies are needed
`that examine the relationship between replication and
`chromatin structure in the globin locus. For example, it
`would be interesting to examine the binding of chromatin
`components and transcription factors during the cell cycle in
`erythroid cells.
`
`I N T E R G E N I C T R A N S C R I P T S
`I N T H E G L O B I N L O C U S
`
`In 1992, Tuan et al. [70] reported that long transcripts
`initiate within LCR HS2 and proceed in a unidirectional
`manner toward the globin genes. Further studies by the
`same group led to the startling observation that transcrip-
`tion always proceeds in the direction of a linked gene,
`independent from the orientation of HS2 [71]. This result
`suggests some form of communication between the promo-
`ter and LCR HS2 in these experiments. Subsequent studies
`in the laboratories of Proudfoot [72] and Fraser [7] identified
`noncoding transcripts over the entire LCR and in between
`the globin gene coding regions. Interestingly, the pattern of
`intergenic transcription during development appears to
`correlate with the pattern of general DNase I sensitivity [7].
`Mutations that delete the start site of the adult-specific
`intergenic transcripts lead to a decrease in general DNase I
`sensitivity and b-globin gene transcription, suggesting that
`intergenic transcription modulates the chromatin structure
`of globin locus subdomains. Intergenic transcripts appear to
`be generated in a cell-cycle-dependent manner, detectable
`during early S-phase but predominantly present in G1 [7].
`These results provide evidence for the hypothesis that
`intergenic transcription is transient. Recently Plant et al.
`[73] analyzed intergenic transcripts across the globin locus
`by nuclear run-on analysis and did not find any evidence for
`the stage-specific generation of
`these transcripts. The
`
`
`
`(cid:211) FEBS 2002
`
`Multistep model for locus control region function (Eur. J. Biochem. 269) 1593
`
`discrepancy between this study and that of Gribnau et al. [7]
`is not understood at the moment, but it is possible that at
`certain stages of the cell cycle, the entire locus is transcribed
`for a short period of time. A subsequent step could then shut
`off transcription in silenced domains, but reduced tran-
`scription could still be detectable by the more sensitive assay
`employed by Plant et al. [73].
`
`I N S U L A T O R S
`The chicken b-globin locus is flanked by insulator elements
`which mark clear boundaries between active and inactive
`chromatin [74,75]. No such sequences have been conclu-
`sively identified in the human or murine globin locus, and it
`appears that the DNase I-sensitive domain in these loci
`extend far 5¢ of the LCR and far 3¢ of the b-globin gene.
`Recent experiments distinguish between insulator sequences
`that block the action of an enhancer or silencer and that of
`boundary elements that separate open and closed chromatin
`domains [74]. The 5¢ most HS site of the chicken LCR, HS4,
`appears to harbor both activities [75]. In this sense it is quite
`possible that the human b-globin locus contains insulator
`elements that restrict the action of the LCR to within
`specific domains. Some evidence suggests that HS5 may
`harbor insulator activity. First, HS5 harbors a binding site
`for the protein CTCF, which is largely responsible for
`insulator function of chicken HS4 [76]. Secondly, inversion
`of the entire LCR with respect to the genes reduces globin
`gene expression to less than 30% of wild-type levels [21].
`Thirdly, an e-globin gene placed upstream of the LCR is not
`transcribed [21]. Finally, HS5 was shown to exhibit
`insulator activity in cell culture experiments [77].
`
`N U C L E A R L O C A L I Z A T I O N
`
`Recent data suggest that enhancer and other regulatory
`elements affect the position of genes within the nucleus
`[78,79]. For example, it was shown that in the absence of
`the b-globin gene is located close to
`an enhancer,
`centromeric heterochromatin, an environment within the
`nucleus that is incompatible with transcription [80]. In the
`presence of LCR element HS2, the b-globin gene localizes
`away from centromeric heterochromatin, suggesting that
`activities associated with HS2 are able to relocate the
`transgene to a transcriptionally permissive nuclear region
`[80]. This phenomenon has been most intensively analyzed
`in yeast, in which specific protein complexes appear to
`direct the location of genes into active or inactive regions
`of the nucleus [81]. However, Milot et al. [32] showed that
`a wild-type globin locus that integrated close to centro-
`meric heterochromatin was still active, suggesting that, in
`the presence of the LCR, the globin locus is active even
`when situated close to a defined heterochromatic envi-
`ronment.
`Recent advances in fluorescent labeling of chromatin as
`well as three-dimensional fluorescent microscopy indicate
`that chromosomes occupy distinct regions, or domains,
`within the cell nucleus [82]. These chromosome domains
`may be composed of up to 1 Mb of chromatin supported by
`the nuclear architecture and appear to contain loops of
`about 50–200 kb of DNA possessing one or several gene
`loci that may or may not be co-regulated. The spaces
`between these territories are believed to be occupied by a
`
`(cid:212)matrix(cid:213)-like structure, consisting of filamentous proteins,
`which is defined as the interchromosomal domain (ICD).
`Active gene loci are located at the surface of chromosomal
`domains in direct contact with the ICD, whereas inactive
`loci are located away from the ICD within chromosomal
`domains. It
`is proposed that macromolecular protein
`complexes involved in chromatin remodeling, transcription,
`and splicing are enriched in the ICD, whereas single proteins
`or smaller protein complexes can diffuse into regions of the
`chromatin domains that are not in contact with the ICD.
`The former ideas are based on indirect observations using
`microscopy and fluorescent labeling. We can therefore only
`describe the existence of chromosome territories and the
`ICD as speculative at best. However, it is safe to say that
`gene loci are located in specific regions of the nucleus and
`that the relative position of these loci changes on activation.
`If applied to the regulation of the globin genes, the ICD
`model could explain why deletion of the LCR in the
`endogenous human or murine globin loci silences globin
`gene expression without altering the establishment of
`DNase I and hyperacetylated chromatin. It is possible that
`transcription factors could gain access to the globin locus
`and change higher-order chromatin structure, but that the
`LCR is required to organize the globin locus in a way that it
`is located in close proximity to the ICD. The situation is
`similar in concept to mechanisms described for the regula-
`tion of gene loci during differentiation of B-lymphocytes.
`Fisher and colleagues [79] have shown that specific gene loci
`relocate to inactive regions in the nucleus of cycling B-cells.
`The relocation and inactivation is regulated by the DNA-
`binding protein Ikaros, which mediates the association of
`gene loci with centromeric heterochromatin.
`
`A M U L T I S T E P M O D E L F O R H U M A N
`b-G L O B I N G E N E R E G U L A T I O N
`
`Step 1: generation of a highly accessible
`LCR holocomplex
`
`We propose that the first step towards activation of the
`globin genes during differentiation is the partial unfolding of
`the chromatin structure containing the globin locus into a
`DNase I-sensitive domain (Fig. 2A). This step may or may
`not require replication. The initial unfolding of
`the
`chromatin structure is mediated by the diffusion of eryth-
`roid-specific proteins into chromosomal domains that are
`not permissive for transcription. These proteins bind to
`sequences throughout the globin locus leading to the partial
`unfolding and perhaps hyper-acetylation of the chromatin.
`If replication is required for globin locus activation, we
`propose that erythroid-specific proteins bind to the globin
`locus after DNA synthesis, prevent the formation of
`repressive chromatin, and mark the locus by modification
`of histone tails.
`GATA factors may be involved in the initial step of
`globin locus activation, as their binding sequences are
`located throughout the globin locus. In addition, GATA-1
`is one of the earliest markers of red cell differentiation [54]
`and is known to associate with proteins containing histone
`acetyltransferase activities. The partial unfolding into a
`DNase I-sensitive structure does not require activities
`associated with the LCR. This is shown by the fact that
`even in the absence of an intact LCR, the rest of the globin
`
`
`
`1594 P. P. Levings and J. Bungert (Eur. J. Biochem. 269)
`
`(cid:211) FEBS 2002
`
`Fig. 2. Multistep model for human b-globin gene regulation. The model depicts four steps proposed to be involved in the regulation of chromatin
`structure and gene expression in the human b-globin locus. The model focuses on the regulation of the human globin locus in the context of
`transgenic mice, but it is assumed that the same principal mechanisms govern the co