Synergistic regulation of human 13-g.lobin gene switching by locus control region elements HS3 and HS4 J6rg Bungert, Utpal Dav6, Kim-Chew Lim, Ken H. Lieuw, Jordan A. Shavit, Qinghui Liu, and James Douglas Engel 1 Department of Biochemistry, Molecular Biology, and Cell Biology, Northwestern University, Evanston, Illinois 60208-3500 USA Proper tissue- and developmental stage-specific transcriptional control over the five genes of the human ~-globin locus is elicited in part by the locus control region (LCR), but the molecular mechanisms that dictate this determined pattern of gene expression during human development are still controversial. By use of homologous recombination in yeast to generate mutations in the LCR within a yeast artificial chromosome (YAC) bearing the entire human 13-globin gene locus, followed by injection of each of the mutated YACs into murine ova, we addressed the function of LCR hypersensitive site (HS) elements 3 and 4 in human 13-globin gene switching. The experiments revealed a number of unexpected properties that are directly attributable to LCR function. First, deletion of either HS3 or HS4 core elements from an otherwise intact YAC results in catastrophic disruption of globin gene expression at all erythroid developmental stages, despite the presence of all other HS elements in the YAC transgenes. If HS3 is used to replace HS4, gene expression is normal at all developmental stages. Conversely, insertion of the HS4 element in place of HS3 results in significant expression changes at every developmental stage, indicating that individual LCR HS elements play distinct roles in stage-specific [~-type globin gene activation. Although the HS4 duplication leads to alteration in the levels of ¢- and ~/-globin mRNAs during embryonic erythropoiesis, total [3-type globin mRNA synthesis is balanced, thereby leading to the conclusion that all of the human [3-locus genes are competitively regulated. In summary, the human [3-globin HS elements appear to form a single, synergistic functional entity called the LCR, and HS3 and HS4 appear to be individually indispensable to the integrity of this macromolecular complex. [Key Words: LCR; competition; HS3; HS41 Received August 3, 1995; revised version accepted October 24, 1995. Each of the genes in the human [3-globin locus is sequen- tially activated during embryonic development: the 5' most (embryonic) e-globin gene is expressed during the first trimester of gestation, the two ~-globin genes during the second and third trimesters, and, shortly after birth, the 3' adult 8- (minor) and [3-globin proteins replace ~-globin chains in tetrameric hemoglobin (Stamatoyan- nopoulos and Neinhuis 1994). The [3-type globin genes are all regulated by the locus control region (LCR), posi- tioned far 5' of the structural genes themselves (Forrester et al. 1987; Grosveld et al. 1987). The LCR is composed of one constitutive and four tissue-specific DNaseI hy- persensitive sites (designated HS1, closest to ~-globin, to the constitutive element HS5, lying furthest away; see Orkin 1990, for nomenclature). The LCR (or smaller units containing the HSs, termed mini- or micro-LCRs) was originally studied as a single tCorresponding author. contiguous structure, where its effects on single genes or a subset of the genes within the [3-globin locus were as- sayed (e.g., Talbot et al. 1989). Individual LCR elements HS2, HS3, and HS4 were then shown to be indepen- dently capable of conferring high level, tissue-specific transcription to linked human 13-locus genes in trans- genic mice (Curtin et al. 1989; Ryan et al. 1989; Ney et al. 1990; Talbot et al. 1990; Pruzina et al. 1991; Talbot and Grosveld 1991; Lloyd et al. 1992; Morley et al. 1992). Subsequently, it was shown that the HS elements indi- vidually elicited markedly different developmental stage-specific activities in constructs harboring several [3-locus genes (Fraser et al. 1993). Thus LCR elements were shown to be more than passive amplifiers of globin gene transcription, and rather, were intrinsically capable of conferring a discriminating developmental response in transcriptional control over individual /3-locus genes. While the LCR is now generally acknowledged to be an active participant in this program of temporal gene acti- vation, gene-proximal regulatory sequences have also GENES & DEVELOPMENT 9:3083-3096 ~ 1995 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/95 $5.00 3083
`
`SKI Exhibit 2052
`Page 1 of 14
`
`

`

`Bungert et al. been known for some time to confer significant tissue- specific activity (Magram et al. 1985; Kollias et al. 1986; Behringer et al. 1987; Trudel et al. 1987; Antoniou et al. 1988). The underlying mechanism by which the LCR modulates the appropriate transcriptional activity of each B-type globin gene during development nonetheless remains obscure. The nomenclature used to describe the individual be- havior of different globin genes during development (he- moglobin switching) was originally applied to the chicken embryonic e- or adult [3-globin genes. Autono- mously regulated genes are both activated and sup- pressed in proper developmental time with only an en- hancer in cis, whereas competitive regulation requires other cis elements to be present in a particular construct for proper temporal gene activation and suppression (Choi and Engel 1988). Enver et al. (1990) first showed that competition was also applicable to describing gene regulation in the human B-globin locus. More recent ev- idence supporting the notion that B-globin locus genes are regulated by a competitive mechanism showed that LCR function is grossly disrupted if a strong transcrip- tion unit is placed between human HS1 and HS2 (Kim et al. 1992); however, deletion of HS2 alone from the en- dogenous mouse B-globin locus results in only a mild phenotype (Fiering et al. 1995). Transgenic experiments examining fragments of the LCR linked to human B-lo- cus genes led to the conclusions that ~-globin was auton- omously regulated (Raich et al. 1990; Shih et al. 1990), while adult f~-globin transcription was competitive (En- vet et al. 1990). Evidence for control of the ~/-globin genes was, however, contradictory: Two early reports sug- gested that ~/-globin transcription was regulated compet- itively, whereas subsequent observations provided com- pelling evidence for autonomy (Behringer et al. 1990; En- vet et al. 1990; Dillon and Grosveld 1991). The disparity in the results analyzing human "y-globin gene regulation could be attributed to differences in the arrangement and number of cis-regulatory elements sur- rounding the gene. Such contradictions arising from the analysis of slightly different DNA constructions could reflect a requirement for mutant transgenes to be exam- ined within a context where all known (as well as per- haps currently unidentified) regulatory elements are present. Thus transformation of mice with YACs con- taining large segments of contiguous genomic DNA came as a timely technical advance. In independent re- ports, YAC transformation of the murine germ line re- sulted in the recovery of tissue-specific control of the tyrosinase or cxl(I) collagen genes, where earlier studies had failed to demonstrate complete complementation with smaller (genomic ~ or cosmid DNA) clones {Schedl et al. 1993b; Strauss et al. 1993). In contemporary exper- iments, it was shown that YACs bearing the entire hu- man B-globin locus also resulted in appropriate develop- mental regulation of the human genes after introduction into the mouse germ line (Gaensler et al. 1993; Peterson et al. 1993). We therefore devised a strategy to create LCR mutations in a YAC bearing the whole human B-globin locus by use of homologous recombination in yeast. YAC DNAs were then isolated from pulse-field gels and injected into fertilized ova to generate trans- genic mice. Pups containing intact, unrearranged, single- copy YACs were monitored for the expression of each human B-locus gene at different stages of erythroid de- velopment by use of a reverse-transcription-PCR (RT- PCR) assay. In this report, we describe several explicit tests of a regulatory sequence competition hypothesis formulated for the human B-globin gene locus (Engel 1993). In es- sence, the model predicted that the stage-specific activa- tion of the genes within the human locus would be achieved by preferential binary association of individual HS sites with (known or theoretical) gene-proximal reg- ulatory elements at various times during erythroid de- velopment. DNA looping and direct juxtaposition of di- stal and proximal elements were proposed to mediate these interactions {Choi and Engel 1988; Gallarda et al. 1989). Each of the mutations revealed new insight into the complex roles that the HS3 and HS4 elements play in the generation of LCR function and in developmental stage-specific competition between human [3-globin lo- cus genes and LCR hypersensitive sites. The present experiments show that deletion of either HS3 or HS4 fundamentally impairs expression of all the B-type globin genes at all erythroid developmental stages, despite the fact that other powerful LCR ele- ments (for example in the case of the HS4 core element- deletion mutant, HS2 and HS3) remain in the YAC trans- gene. The replacement of HS4 by HS3 appears to fully complement HS4 function at every developmental stage. The converse, however, is not true; when HS4 is substi- tuted for HS3, the expression of every gene at all stages of erythropoiesis is altered, and thus HS4 cannot fully com- pensate for HS3 function. Moreover, when the wild-type and the HS4 substitution mutant transgenes are com- pared, alteration of the transcript level of human e- and ~/-globin mRNAs in the embryonic yolk sac is reciprocal. This suggests that HS3 and HS4 competitively and col- laboratively control the expression of these two genes at the embryonic stage of erythroid development. Taken together, these data reveal that the LCR behaves as a single cooperative unit and that, not only is this syner- gistic macromolecular complex comprised of the indi- vidual HS elements, but also the elements are not always uniquely specified for discrete functions in the complex. Results Generation of human [3-globin YA C mutants A201F4 is a 155-kb YAC that contains the entire human B-globin locus (Gaensler et al. 1991, 1993). The YAC initially segregated unstably in the parent yeast strain, and therefore, a clone which stably mitosed was isolated by mating and sporulation. The YAC DNA was then cloned into a bacteriophage ~, vector (Maniatis et al. 1982), and individual recombinant phage containing overlapping segments of the [3-globin locus were isolated (Fritsch et al. 1980). Recombinants specifying the 3084 GENES & DEVELOPMENT
`
`SKI Exhibit 2052
`Page 2 of 14
`
`

`

`Human ~-globin switching in YAC transgenic mice [3-globin LCR HS elements and the individual genes were further subcloned into plasmid vectors to facilitate mu- tagenesis. The minimal HS4 core element has been localized to a 280-bp SacI-AvaI segment of a 3.2-kb parental EcoRI- HindlII fragment (Pruzina et al. 1991 ), while the minimal HS3 core element was defined as a 225- bp HphI-Fnu4HI fragment within a larger 1.9-kb HindIII fragment (Phil- ipsen et al. 1990; Fig. 1). For the creation of the HS3 replacement mutant, the HS4 core element was ampli- fied by PCR from the parental HS4 subclone with prim- ers that incorporated unique XbaI and XhoI sites at the ends. The subcloned products of all PCR reactions were verified by sequencing. This 280-bp HS4 minimal ele- ment was then used to replace the corresponding HS3 core sequence within the 1.9-kb HS3 HindIII subclone, which generated the mutant hypersensitive site desig- nated HS343 (i.e., a 280-bp HS4 core element embedded in HS3-flanking DNA; Fig. 1). A plasmid incorporating the HS3 core element surrounded by HS4-flanking se- quences (HS4341 was generated by use of a conceptually identical strategy. Two other plasmids containing dele- tions of each of the two HS core elements within the bodies of the otherwise unmodified parental fragments were constructed at the same time (see Materials and methods). To facilitate the generation of mutant [3-locus YACs, two manipulations were performed. First, the URA3 se- lectable marker gene in the right arm of A201F4 was retrofitted with the LYS2 gene, thereby inactivating URA3 in the YAC by homologous recombination (Srivastava and Schlessinger 1991). At the same time, each of the four mutated HS plasmids (described above) was subcloned into the yeast integrative plasmid vector pRS306, containing the URA3 gene (Sikorski and Hieter 1989). Each pRS306 subclone was then used to transform the (LYS2-modified) wild-type human [3-globin YAC. For example, HS343 (Fig. 1) subcloned in pRS306 was di- gested with SauI and then used to transform yeast bear- ing the LYS2-retrofitted YAC clone (Fig. 2; see Materials and methods). The majority of the yeast colonies grow- ing on selective (trp-/lys -/ura - ) medium contained the targeting plasmid integrated at the homologous human [~-globin HS site within the YAC, thereby creating an intermediate with the LCR elements arranged (e.g., in replacing HS3 with HS4) in the order: 5'-HS4--[HS343] - URA3-HS3-HS2-3' (Fig. 2). Selective excision of the tar- geting plasmid was mediated by growth on medium con- taining uracil, followed by counterselection on 5-fluoro- orotic acid /FOA}-containing plates Ilethal to cells retaining the URA3 marker; see Materials and methods). Excision of pRS306 resulted in reversion to either the parental YAC structure or replacement of the parental HS site by the desired mutant element (Fig. 2; Winston et al. 1983). Thus the number of YACs to be tested for transgene expression characteristics in mice were five (Fig. 3): wild-type (the LYS2-retrofitted YAC, designated HS4321), an HS3 duplication (HS3321; incorporating HS4g41, an HS3 core element deletion (HS4021; incorpo- rating HS3°3), an HS4 duplication (HS4421; incorporat- ing HS343), and the HS4 core element deletion (HS0321; incorporating HS4°4). The structures of the four mutants and the parental wild-type YAC were verified by South- ern blotting on both conventional and pulse-field gels (Fig. 4). Transgenic mice Two methods have been used to introduce YACs into the mouse germ line. One involved microinjection into HS5 HS4 HS3 HS2 . , t \ .~:~ I -20 / ~ -10 HS4 ..... -~ HS3 (3.2 kbp) EcoRl rill] Hind'Ill [ \ Hi~dlII (1.9 kbp) / \ / \ / \ / \ HS4 core i [ [ I HS3 core (280 bp) Sacl AvaI HphI Fnu4Hl (225 bp) |-~. |~ ~'1 ~ "-'- ~ XbaI "~ -~ )O~oI ~ ~" ~ .~ .~ X]~I~ XhoI ... ~ ~ PCR 8s 34 ...... , EcoR1 BamHl HindlIl BstXI Saul Clal HSMa Figure 1. Diagram of the human [3-globin LCR and generation of the HS3 and HS4 re- placement mutants. The top line indicates the positions of the hypersensitive sites in the human [3-globin LCR ITalbot et al. 19891. The second line depicts the two subclones {HS3 and HS4) that were used to initiate the yeast targeting mutagenesis strategy (see Ma- terials and methods). These two HS element subclones were used as substrates for PCR amplification of the individual core elements and flanking sequences. The PCR amplifica- tion was carried out with primers that incor- porated new, unique restriction enzyme sites at a number of positions to aid in subsequent cloning steps (Table 1). Both the HS3 and HS4 core elements (line 3} were amplified from the respective parental plasmids using these primers to incorporate new sites into the cores (line 4), and finally the two core elements were exchanged between the two {appropriately PCR-adaptedl parent clones to yield the desired mutant elements HS343 and HS434 (line 5). GENES & DEVELOPMENT 3085
`
`SKI Exhibit 2052
`Page 3 of 14
`
`

`

`Bungert et al. Figure 2. Generation of LCR HS replace- ment and deletion mutants by use of homol- ogous recombination. In this example, the HS343 fragment (generated as outlined in Fig. 1) was subcloned into yeast shuttle vector pRS306 (Sikorski and Hieter 19891. After di- gestion with Saul {top line), it was then used to transform (LYS2-modified; see Materials and methods) yeast clone A201F4, harboring the 155-kb human ~-globin YAC (line 2; Gaensler et al. 1991 ). Colonies containing the targeting vector (i.e., growth on Lys , Trl0-, Ura- medium) were tested for homologous integration by Southern blot analysis (line 3). Individual clones that had undergone homol- I I __ _~ _-...~.~. ~ -- -- HS4321 HS4 HS3 HS2 (L YS2-retrofitted clone A201F4) Transform A201F4 with linearized HS343 HS4 HS343 HS3 HS2 Select pRS306 excision - -- ~ ~ ~,~ ...... H~4421 HS4 HS343 HS2 ogous recombination were replated on Ura + medium and then replica plated onto FOA plates {see Materials and methods). Clones growing on FOA were tested individually for the presence of the original YAC 1HS4321) vs. the replacement mutant structure (HS4421; line 4) by Southern blots (see Fig. 4). fertilized murine eggs (Schedl et al. 1993a, b), whereas another involved lipofection-mediated transfer of puri- fied YAC DNA into embryonic stem (ES) cells, followed by blastocyst injection to generate chimeric mice (Strauss et al. 1993). We chose to use the former method because microinjection is reported to result in more fre- quent integration of intact YAC transgene DNA {Gnirke et al. 1993), and because we planned to use these same mutant YACs for transgenic studies after further mu- tagenesis. Wild-type and mutated YACs were isolated after pulse-field gel electrophoresis (PFGE), purified, and mi- croinjected into fertilized murine ova (see Materials and methods; Schedl et al. 1993a, b; Peterson et al. 1993). After initial PCR identification and then Southern blot- ting of founder tail DNA to confirm the presence of the transgene, all positive animals were bred to ensure that the transgenes could be stably transmitted to progeny. At least two transgenic lines were analyzed for each of the YACs, arbitrarily referred to below as lines a and b (an additional line, c, was analyzed, which contained the HS4421 YAC mutant transgene). Southern blotting of each of the F~ or F 2 lines, with probes both flanking and internal to the locus IRI 1.8, RI 3.3, RK29, HS4, HS3, HS2, e-, y-, and f3-globin; Gaensler et al. 1991; Fig. 3) showed that each contained restriction fragments of the expected size. Fragmented YACs discovered through this analysis were excluded from further consideration. The a RI1.8 RI3.3 HS4 ~ T ? 8 ~ RK29 EEl i • • • Be • l II ~ Left YAC arm ~ "~ ~ Right YAC arm .H~ HS3 HS2 g G'y a,~ 6 ~~. HS4321~, • n [] , .- ," , , ,r-, T' ' , ='"' , -20 -10 0 10 20 30 40 HS3 HS3 HS2 n n [] - ~ ~ r-ll ~ HS3321 , ,- , , I , ' -20 -10 0 10 20 30 40 HS4 HS2 • [] . HI'ill am I ~ lr'-I n n HS4021 , ,_ , , , l , -20 -10 0 10 20 30 40 HS4 HS4 HS2 • • n . m n N 'l mn~m~m~lm i--IN am • H54421 , ,- u l , ' -20 -I0 0 10 20 30 40 HS3 HS2 ~i 0 . gill N m l ~ I" • iI n HS0321 , ,- u , , ' -20 -10 0 10 20 30 40 Figure 3. Diagrammatic representation of the LCR mutant YACs. Diagrammatic representations of the five individual human [3-globin YACs generated for this analysis are shown. The bottom five lines depict the structures of the parental and four mutant YACs studied here, whereas the top line depicts the approximate position of all these elements within the YAC borne by A201F4. In addition, other sequences flanking the gene and LCR elements were used as probes in Southern blot analysis to characterize the physical integrity of the locus before and after transgene integration into the germ line of mice (see Results; Fig. 4; data not shown; Gaensler et al. 1991}. 3086 GENES & DEVELOPMENT
`
`SKI Exhibit 2052
`Page 4 of 14
`
`

`

`Human [~-globin switching in YAC transgenic mice A, 1 2 3 4 5 6 7 8 9 i B@ C@ HS4 (10.6) 8(5.5) (3.7) 10 4~3 3~4 4~3 3~4 + A4 A3 + A4 A3 I I ! l + HS4 core* + HS3 core* A201F4 155 kbp) HS3 (10.6) ,:(1.6) RK29 (1.3} Figure 4. Characterization of the A201F4 YAC HS mutations. All four YAC mutants (HS3321, lanes 2, 7; HS0321, lanes 3,8; HS4421, lanes 4,9; and HS4021, lanes 5,10) generated by use of the strategy outlined in Fig. 2 and depicted diagrammatically in Fig. 3, were analyzed by Southern blotting. (A) Confirmation of targeted mutagenesis in each of the YACs was first investigated by PFGE. Examination of ethidium bro- mide-stained pulse-field gels of the parental and mu- tated human ~-globin YACs showed that the mutant YAC chromosomes are unaltered in size. (B) The pulse-field gel shown in A was transferred to a nylon membrane and then hybridized with radiolabeled HS4 (lanes 1-5) or HS3 (lanes 6-I0) core element probes. (C) DNA prepared from the wild-type and each of the mutant YACs was digested with EcoRI, transferred to nylon filters, and then hybridized to radiolabeled HS4, adult [3-globin and embryonic e-globin probes (lanes 1-5) or HS3, fetal y-globin and 3' flanking marker RK29 probes (lanes 6-10; Fig. 3, Gaensler et al. 1991). Abbreviations representing each of the YACs are ( + ) HS4321 (lanes 1,6; the wild- type YAC; Fig. 3); (4---*3)HS3321; (A4)HS0321; (3 --* 4) HS4421; and (A3} HS4021. line animals used for breeding and subsequent mRNA analysis were found to contain a single copy of the YAC left and right vector arms, except a single transgenic line in which HS3 had been deleted (HS4021a, Figs. 3 and 7, below), which contained neither (not shown). The b transgenic lines also contained all markers within the locus on contiguous restriction enzyme fragments of the expected size, and virtually identical band intensities. In summary, these data indicated that all of the animals subjected to detailed analysis here contained intact, sin- gle-copy YAC transgenes that were transmissible through the germ line. Human [3-globin multiplex PCR assay To determine the pattern, timing, and abundance of ex- pression of each of the human fS-globin locus genes in transgenic mice, RNA isolated from the yolk sacs, fetal livers, or adult spleens of transgenic embryos or animals (see Materials and methods)was analyzed by semiquan- titative RT-PCR (Foley and Engel 1992; Foley et al. 1993; Leonard et al. 1993). The level of ~-, ~-, and [3-globin mRNAs were compared to an internal control, mouse o~-globin mRNA, which remains relatively con- stant during murine gestation (Whitelaw et al. 1990). This RNA analysis method relied on the fact that unique primers for each of the human p-type globin genes could be defined (Fig. 5A; Table 1), that they would give rise to specific amplicons that differed from one another in size, and that these primers would not cross-react, either with one another or with cDNAs produced after reverse tran- scription of mouse [3-type globin mRNAs. We demon- strated that these expectations could be fulfilled in con- trol experiments (Fig. 5B), thus simplifying the analysis by allowing simultaneous (multiplex) assay for accumu- lation of all three human f~-globin transcripts (hut, huT, and hu~) and the endogenous ot-globin control (muc,) in each RNA sample at every stage of murine erythroid development. The assay is only semi-quantitative, how- ever, because primer sets for each gene may differ subtly in annealing efficiency and could therefore differ in the amount of isotope-labeled deoxynucleotide incorporated (the products are of different length and G + C content); thus the absolute abundance of one transcript compared to another at a different stage of development is not quantitative, but the relative abundance of any individ- ual PCR product at specific developmental stages can be directly compared after normalization to the intensity of the mu~ globin internal control, the length and G+ C content of the PCR products, and the transgene copy number in comparison to the endogenous mouse ~-globin genes (see Mateials and methods; Table 2). RNAs were prepared by use of standard procedures (Chomczynski and Sacchi 1987), from transgenic F2 or F 3 embryos at 9.5 days post-coitus (dpc)(yolk sac) and 14.5 dpc (liver) or from the spleens of 2- to 6-month-old ane- mic animals, representing embryonic, fetal or adult de- GENES & DEVELOPMENT 3087
`
`SKI Exhibit 2052
`Page 5 of 14
`
`

`

`Bungert et al. Figure 5. Developing a multiplex RT-PCR as- say for human ~3-type globin gene expression in transgenic mice. (A) The human f3-globin gene locus DNA sequences (EMBL data base] were compared to one another and to the mouse ~3-globin locus sequences to define primers (Ta- ble 1) that would both be specific for each hu- man globin gene and additionally would not cross-react with mouse ~-type globins; the ap- proximate positions and anticipated product lengths as cDNA are shown. When possible, we chose primer positions spanning introns to en- sure that products arise originally from mRNA and not contaminating genomic DNA. (B) Con- trol PCR reactions were performed to determine the temperature optima and substrate specificity for each set of human ~3-globin and mouse o-globin gene primers. In every reaction shown here, all four globin gene-specific primer sets (Table 1) were included; in addition, each lane represents the inclusion of different individual cloned substrate globin cDNAs {30 pg each, shown in the table below the diagraml in the PCR reactions (see Materials and methodsl. A41 hu[} I+ [/,./I ~'~'t I, Ii~ I muo:l ~ B liiTiiiiii!iiiiii~iiiiiiiillilil ~,,,,,,G,, , , - (amplicon = l~bp) (amplicon .. 165 b p) (ampIicon = ~2bp) (amplicon ,, l~bp) Iii[iiiiiiiiiii!1 V///A I velopmental stages and hematopoietic sites, respectively (Strouboulis et al. 1992; Gaensler et al. 1993). Stage-specific expression of [3-type globin mRNA in LCR HS mutant transgenic mice After qualitative and quantitative analysis of the four HS3/HS4 LCR mutants (Fig. 6; Table 2), numerous al- terations in the expression patterns of the individual globin genes emerged. The first unanticipated observa- tion was that the substitution of core element HS3 for the HS4 core (mutant HS3321; Fig. 3) resulted in no sig- nificant quantitative change in stage-specific expression of any of the human ~3-1ocus genes in comparison to the wild-type YAC, indicating that HS3 can functionally complement HS4 activity at every stage of erythroid de- velopment. The converse duplication of HS4 (mutant HS4421 ), on the other hand, resulted in drastic alteration in expression of all the genes at all stages, showing that HS4 fails to functionally complement HS3 expression during erythropoiesis. In the embryonic yolk sac, ~-globin synthesis was reduced while at the same time "y-globin mRNA synthesis was elevated in all three lines harboring the HS4421 mutant when compared to the pa- rental YAC (Fig. 7; Table 2). During the fetal stage, ex- pression of both -y- and 13-globin was significantly im- paired when HS4421 was compared to the wild-type YAC, while in the adult stage, 13-globin synthesis was attenuated three- to fourfold. Human globin mRNA synthesis in both of the HS de- letion mutants (HS4021 and HS03211 was virtually abol- ished at every stage of erythropoiesis. This effect is con- sistent with, but more extreme than, the stage-specific regulatory characteristics attributed to these HS ele- ments {Fraser et al. 1993). Fetal liver ~3-1ocus transcrip- tion in the HS3 and HS4 deletion mutants also appears to be almost entirely eradicated. The pattern of adult ery- throid expression of these deletion mutants are quite similar to one another: both display significant quanti- tative reductions (three- to ninefold} in adult ~3-globin mRNA synthesis. In summary, three of the four HS mu- tations exhibit phenotypes that differ significantly from the wild-type locus, and (with the exception of the HS3 duplication) each would be predicted to be a lethal mu- tation in humans. Discussion In this study we describe the effects of small, specific LCR HS3 and HS4 mutations, introduced by homolo- gous recombination into a YAC containing the entire human ~3-globin locus, on the expression characteristics of the individual genes in YAC-transformed transgenic mice. No current model fully accounts for all of the ob- served effects of these mutations, but the overriding mechanistic principle in force appears to be that the LCR functions as a coherent macromolecular unit and is com- prised of a specific number ef, but not necessarily 3088 GENES & DEVELOPMENT
`
`SKI Exhibit 2052
`Page 6 of 14
`
`

`

`Human 13-globin switching in YAC transgenic mice Table 1. Oligonucleotides used for amplification of core enhancers and flanking sequences Name Sequence HS4--core US: HS4-core DS: HS3-core US: HS3-core DS: HS4-5'flank: HS4-3'flank: HS3-5'flank: HS3-3'flank: 5'-CGCATCTAGAGGACCCCAGTACACAAGAGG-3' 5'-AGGGCTCGAGTCGGGGAATGGGAGGGAGAG-3' 5'-GCAGTCTAGATGGTGTGCCAGATGTGTCTA-3' 5'-GGCGCTCGAGGCTGCTATGCTGTGCCTCCC-3' 5'-CCGATCTAGAAAGAGCTCTCTAAAAGTGAT-3' 5'-CAGGCTCGAGAGCAGGTTTGCTTATTTATG-3' 5'-GAAGTCTAGAACCGTGAGGTCTTGTGTTTC-3' 5'-TGTACTCGAGATTTTTCATTCTACTACTAC-3' R-armUS: 5'-TCCGTAATCTTGAGATCGGGCGT-3' R-arm DS: 5'-GGTGATGTCGGCGATATAGGCGCCAGCAAC-3' L-arm US: L-arm DS: 5'-GTGATAAATTAAAGTCTTGCGCCTTAAACC-3' 5'-GCTACTTGGAGCCACTATCGACTACGCGAT-3' h~glob. US: 5'-CTTTGGAGATGC TATTAAAAACATG-3' hEglob. DS: 5'-CCAGAATAATCACCATCACG T TAC-3' hTglob. US: 5'-GACCGT TTTGGCAAT C CATTTC-3' h-yglob. DS: 5'-GTATTGCT TGCAGAATAAAGCC-3' h/3glob. US: 5'-ACACAACTGTGTTCACTAGCAACCTCA-3' h[3glob. DS: 5'-GGTTGC C CATAACAGCAT CAGGAGT-3' maglob. US: 5'-GATTC TGACAGACTCAGGAAGAAAC-3' maglob. DS: 5'-CC TTTCCAGGGC TTCAGC T CCATAT-3' Oligonucleotides were designed according to published sequences (GenBank/EMBL). Bold nucleotides represent restriction sites that were generated to facilitate subcloning. Oligonucleotides specific for the right and left vector arms of the YAC were adopted from Peterson et al. (1993). Oligonucleotides used for RT-PCR Chuman E-, ~/-, [3-, and mouse a-globins) were designed according to sequences from the GenBank/EMBL data bank. uniquely specified, constituent elements. An appropri- ately constituted LCR also appears to be capable of dis- criminating between the individual genes by presently obscure competitive regulatory interactions at every stage of erythroid development. Human ~-globin YA C transformation of mice All of the transgenic animals studied here are, to the best of our knowledge, single copy and unrearranged for the microinjected YACs, and thus any regulatory pattern al- terations reported here that differ from the parental YAC transgene expression pattern cannot be due to effects from neighboring, multiply integrated YAC transgene loci. This was initially a cause for concern because it has been shown that multicopy transgenes may behave quite differently than single-copy constructs harboring Table 2. Quantitative analysis of HS3 and HS4 human [3-globin LCR mutations Embryonic yolk sac (9.5 dpc) Fetal liver (14.5 dpc) Ad. spleen Genotype ~ ~/ total ys ~/ [3 total fl [3 HS4321 91/97 96/124 188/221 6.7/8.4 64/78 71/86 46/( N. D. } HS3321 117/91 111/94 228/185 6.5/7.5 70/71 76/78 39/47 HS0321 2.1 / 1.2 2.3/5.8 4.4/7.0 i. 7/2.1 1.3/2.1 3.0/4.2 5.7/9.8 HS4421 8.4/6.9/7.5 168/185/195 176/192/202 1.5/2.7/2.6 0/1.9/3.2 1.5/4.6/5.8 13/14/(N.D.) HS4021 8.7/6.4 0/43 8.7/49 0/3.2 0/3.8 0/7.0 7.3/15 Expression of human [3-type globin transgenes from the wild-type (HS43211 and mutated (HS3321, HS0321, HS4421, HS4021) YACs in transgenic mice, normalized to expression of the mouse ~-globin gene (four copies, arbitrarily set as 100% I. The data were obtained from quantitation of multiple PCR analyses {as represented by the data shown in Fig. 6J on a PhosphorImager as described in Materials and methods. The first number shown in each box is from transgenic line a, the second from line b, and Ifor HS4421) the third from line c, for each YAC. (N.D.) not determined. GENES & DEVELOPMENT 3089
`
`SKI Exhibit 2052
`Page 7 of 14
`
`

`

`Bungert et al. Figure 6. Erythroid expression of the hu- man e-, ~/-, and [3-globin genes in HS3 and HS4 YAC mutant transgenic mice. RNA was prepared from transgenic embryos or adult mouse erythroid tissues, reverse tran- scribed, amplified into the linear range by PCR and then electrophoresed as described (see Materials and methods). Erythroid tis- sues representing embryonic, fetal and adult stages (Strouboulis et al. 1992; Gaensler et al. 1993) were analyzed. (Lanes 1-5) 9.5-day embryonic yolk sac RNA; (lanes 6-10) 14.5- day embryonic liver RNA and (lanes 11-16) adult spleen RNA. The expression of the fol- 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 hu[5- 212bp m / + 4-~3 A4 3-~ A3 + 4-~3 A4 3-~ A3 nt + 4-~3 A4 31,4 A3 L 1 J Embryonic yolk sac I I Adult spleen (e9.5) Fetal liver (e14.5) lowing YACs (Fig. 3) was examined in each erythroid tissue: HS4321 llanes 1