Nucleic Acids Research, Vol. 18, No. 23 7093

Developmental regulation of the human zeta globin genein transgenic mice

E.A.Spangler, K.A.Andrews and E.M.Rubin*Divisions of Cell and Molecular Biology and Research Medicine, University of California, LawrenceBerkeley Laboratory, Berkeley, CA 94720, USA

Received June 22, 1990; Revised and Accepted November 9, 1990

ABSTRACT

We have characterized the expression of the humanzeta (f) gene, which encodes an embryonic a-llkeglobin, in transgenic mice. We find that a 777 base pairfragment spanning erythroid specific hypersensitivesite II (HSII) from the distal 5- region of the human 0globin gene cluster potentiates expression of the fglobin gene. In the absence of the HSII fragment, nof expression is observed. Expression of the human fgene In mice parallels expression of a murineembryonic a-like globin gene (x). Thus, expression ofthe human f gene in mice requires linkage to anerythroid-specific enhancer sequence, but thepresence of the enhancer does not affect thedevelopmental regulation of the transgene. Our resultsindicate that the factors involved In switching fromembryonic to adult a globin gene expression duringdevelopment are evolutionarily conserved, and suggestthat the transgenic mouse is an In vivo system in whichthe requirements for the developmental switch in aglobin gene expression can be analyzed in detail.

INTRODUCTION

The genes of the human a globin and /3 globin loci are expressedin a specific temporal pattern, and in specific hemopoietic tissuesduring development. The embryonic globins (f and e) areexpressed in the yolk sac blood islands until about the fifth weekof gestation. At that time, adult a globin (a) and fetal j3 globin(A7 and G7) genes begin to be expressed in the liver. The liveris gradually replaced as the major site of hemopoiesis by thespleen and bone marrow, which express the adult globin genes(a, 8 and /3) (1). A similar developmental pattern of globin geneexpression exists in mice. Hemopoiesis begins in the yolk sacat about day eight of gestation, and the fetal liver becomes themajor site of hemopoiesis at about day ten. There is no fetal j3globin in mice, thus the mouse fetal liver produces only adulthemoglobins (2,3).

Although the mechanisms that regulate developmental changesin globin gene expression are not completely understood, severallines of evidence indicate that two categories of DNase Ihypersensitive sites (major and minor) are important in human

/S globin gene regulation. Minor hypersensitive sites are locatedclose to each of the genes of the /3 cluster, and their presenceor absence correlates with the expression of those genes (4,5).Major hypersensitive sites map upstream and downstream of the/3 globin cluster, and are present in erythroid tissues throughoutdevelopment (5,6).

When constructs containing only the minor hypersensitive sitestogether with either a human fetal or adult /S globin gene areintroduced into mice, the genes are developmentally regulated,but the level of expression is variable (7,8,9,10,11,12). Theaddition of DNA fragments containing some or all of the majorhypersensitive sites allows -y and /3 globin transgenes to beexpressed consistantly at levels approaching those of endogenousglobin genes (13,14). This suggests that sequences proximal tothe genes are essential for the developmental switch in /? globingene expression, while the role of distal sequences, includingthe major hypersensitive sites, is to potentiate high levelexpression of those genes.

Distal DNA sequences normally involved in the regulation ofa globin gene expression have not been reported, and expressionof human a globin in transgenic mice has been achieved onlywhen the a gene is coupled to DNA fragments spanning someor all of the major hypersensitive sites from the human /S globincluster (15,16,19). We describe here the expression in transgenicmice of the human f gene, which encodes an embryonic a-likeglobin. Moderate levels of adult human a and )3 globin geneexpression can be obtained in mice when a small fragmentencompassing major hypersensitive site II (HSII) alone isassociated with the gene (17,18,19). We demonstrate that thisHSII fragment is also sufficient to potentiate expression of thehuman embryonic a-like globin (f) in mice. In addition, we findthat the human embryonic gene follows the same developmentalpattern of regulation as the murine embryonic a-like globin gene

METHODSTransgenic Mice

Transgenic mice were generated essentially as described byHogan et al. (20). Embryos were obtained by breedingC57BL/6XSJL Fl hybrid parents. DNA was microinjected into

* To whom correspondence should be addressed

7094 Nucleic Acids Research, Vol. 18, No. 23

the pronuclei of one-cell embryos at a concentration of 2.0 ng//tl.Injected embryos were cultured overnight in M16 medium (20),and those that divided to two cells were transferred to the oviductof a pseudopregnant female and allowed to develop to term.Transgenic animals were identified by Southern blot analysis oftail DNA, using a 4.8 kb genomic EcoRI fragment that spansthe entire f gene as a probe. The probe fragment was labelledwith 32P to a specific activity of ~ 1X109 cpm//tg using therandom primer method (21).

Preparation of DNA for MicroinjectionAn 882 bp fragment spanning HSU (17) was provided by PeterCurtin (University of California, San Francisco), and was insertedinto the BamHI site of vector pUC19. A 4.8 kb EcoRI fragmentcontaining the f gene was isolated from cosmid pCL9, whichcontains the entire human a globin cluster (22), and was ligatedinto an EcoRI site located 21 bp from the HSII fragment. The5' to 3' orientation of the f gene relative to the HSII fragmentwas the same as is found in the genome for the genes of the /3globin cluster.

For injection into mouse embryos, a 5.7 kb Pvul-Hindmfragment containing 777 bp of the original HSU insert linked tothe f gene was isolated by preparative agarose gel electrophoresis.The fragment was recovered by electroelution, followed byconcentration on an Elutip column (Schliecher & Schuell, Inc.)and ethanol precipitation. The DNA pellet was resuspended ininjection buffer (10 mM Tris, 0.1 raM EDTA, pH 7.5) andfiltered (0.2 /im) prior to microinjection.

Preparation of Mouse RNA and DNARNA was prepared as described by Chirgwin et al. (23). DNAwas prepared by digestion of tissue with proteinase K, followedby chloroform extraction and ethanol precipitation. To obtain fetaland embryonic tissue samples, transgenic males were bred withnormal females. The day that the copulation plug was observedwas designated as day 0. Embryonic yolk sacs were collectedat day 9 and fetal livers were collected at day 16. In each case,10 or more tissue samples were pooled in the final RNApreparation. In some experiments RNA was prepared at day 16from individual fetal livers, and DNA was prepared from theremainder of each fetus. Adult transgenic mice were treated withphenylhydrazine (25 ng/gm body weight injected intraperitoneallydaily for 4 days) to induce production of nucleated red bloodcells, and RNA was prepared from ~ 200 fi\ of peripheral bloodon the fifth day.

Primer Extension AssayOligonucleotide primers were 5'-end-labelled with 32P using T4Kinase (Bethesda Research Laboratories), and were separatedfrom unincorporated label using Nuctrap columns (Stratagene).The sequences of the primers were as follows: human f,5'-TCTTGGTCAGAGACATGGCG-3'; mouse x, 5'-TGAGGG-TTGGAGGCTGCGCT-3'; mouse a, 5'-CAGGCAGCCTTG-ATGTTGCTT-3'. Approximately 10,000 cpm of labelled primerwas added to 10 /ig RNA. The RNA/primer mixture wasevaporated to dryness and resuspended in 7.5 n\ of 0.2M NaCl.Samples were sealed in glass microcapillaries, heated to 80°Cfor 2 minutes, and then incubated at 30°C for 1 hour. Afterhybridization, each sample was added to a tube containing areverse transcription reaction mixture. The final reaction volumewas 50 /tl, and contained 50 mM Tris, pH 8.3, 30 mM NaCl,75 mM KC1, 10 mM DTT, 3 mM MgCl2, 1 mM each dGTP,dATP, dCTP and TTP, 20 units of RNasin (Promega) and 200

units of M-MLV Reverse Transcriptase (Bethesda ResearchLaboratories). Reactions were incubated at 37°C for 1 hour.Nucleic acids were precipitated with ethanol, resuspended indenaturing sample buffer (95 % formamide) and analyzed on 8 %acrylamide/7M urea gels.

RESULTS

In order to study the regulation of the human zeta gene duringdevelopment, we established lines of transgenic mice andexamined expression in staged embryos. The construct used togenerate those mice is shown in figure 1. It consisted of a4.8 kb DNA fragment containing the human f globin gene, linkedto a 777 bp fragment spanning erythroid-specific hypersensitivesite II (HSU). Three founder transgenic animals were identifiedby Southern blot hybridization, and were designated Zl , Z5 andZ l l .

To characterize the inheritance of the transgene, each of thefounder animals was bred and offspring were screened bySouthern blot hybridization. A typical result obtained using ahuman f probe is shown in figure 2. In DNA from the founder

0 10 20 30 40 50 60 70 80 90 100 kb

r irm n iHuman p Locus

Human a Locus

HSII-Zsto Construct

i I i i I I I0 1 2 3 4 5 6 kb

Figure 1. HSII-f construct. The human a and /3 globin loci are shown, withthe approximate positions of the genes represented by black boxes. The locationsof the major DNasel hypersensitive sites are indicated by arrows, and are identifiedby roman numerals. DNA fragments used to build the HSII-f construct areindicated by dotted lines, and features of the construct are represented as follows:HSII fragment, black line; f exons, black boxes; f introns, open boxes; f 5'and 3' flanking sequences, hatched boxes.

.5 4 5 6 7 8

Figure 2. Southern blot analysis of mice from the Zl 1 transgenic line. DNAfrom the Zl 1 founder (lane 1) and 5 offspring (lanes 2 -6 ) was digested withEcoR], blotted and probed with a 4.8 kb genomic DNA fragment spanning thehuman f globin gene. Transgenic animals are characterized by a band at 4.8kb that hybridizes to the human probe. Additional bands visible in lane 1 arethe result of partial digestion of the DNA. Lanes 7 and 8 contain an amount ofplasmid DNA equivalent to 1 and 2 copies of human f per haplokl genomerespectively.

Nucleic Acids Research, Vol. 18, No. 23 7095

(Zll , lane 1), and in DNA from each of the transgenic offspring(lanes 2,4,5), there is a 4.8 kb fragment that hybridizes to thef probe. By comparing the intensity of the transgenic f signalto the intensity of the signal in lanes 7 and 8 which contain knownamounts of f plasmid DNA, we estimate that the transgene copynumber in the Zl 1 line is ~ 2 - 4 per haploid genome. Similarly,four of the eight offspring of Z5 were found to be transgenic,and the transgene copy number was estimated to be —1—2 perhaploid genome. No transgenic offspring of Zl were identified,and we presume that this animal is a mosaic. To verify that theentire HSII-f construct was present in the Z5 and Zl l lines,duplicate blots were probed for HSII. A band of the expectedsize was detected in each transgenic lane (data not shown). Onthe basis of these results, the Z5 and Zl l transgenic lines wereboth chosen for further analysis.

A primer extension assay was used to determine if the humanf globin gene was expressed in the two transgenic lines, and atwhat stages of development. Total RNA was isolated from theyolk sacs of day 9 embryos and from the peripheral blood ofadult transgenic animals. The embryonic tissues were derivedfrom matings between transgenic males and normal females, andRNA was prepared from pools of 10 or more tissue samples tcinsure an adequate supply of material. Embryonic RNA sample;were primed with a 1:1 mixture of the mouse x and human iprimers. Adult RNA samples were primed independently withthe mouse a primer and the mixed embryonic primers (x ancf), because the a and f extension products migrate at overlappingpositions on the gel. The results of this experiment are showrin figure 3. In both transgenic lines, correctly initiated humarf and mouse x transcripts are present in embryonic erythroiccells (lane 1). In adult erythroid cells, mouse a globin transcript;are present (lane 3), but human £ and mouse x transcripts areundetectable (lane 2).

The relative intensities of the primer extension band;approximate the relative amount of each transcript in the RNAsample. By assuming that 50% of the pooled tissue from whichthe embryonic RNA samples were prepared was transgenic, weestimate that human f transcripts are present at greater than 609?the level of mouse x transcripts in embryonic erythroid tissueof the Z5 line, and at greater than 30% the level of mouse >transcripts in embryonic tissue of the Zl l line. The level 01

Mx[

Z5

1 2 3

I

Z l l

1 2 3

- M o -Ma

Figure 3. Primer extension analysis of RNA from embryonic and adult erythroidtissues. In the panel on the left, RNA samples are from the Z5 line. In the panelon the right, RNA samples are from the Zl 1 line. Lane 1 contains RNA preparedfrom embryonic yolk sacs, primed with a 1:1 mixture of the human f and mousex primers. Lane 2 contains adult RNA primed with a 1:1 mixture of the humanf and mouse x primers, and lane 3 contains adult RNA primed with the mousea primer.

human f and mouse x transcripts in transgenic adult RNApreparations is below the limit of detection of our assay. Fromthis we conclude that transcription of the human f transgene isreduced to less than 1 % the level of the mouse a globin genein adult animals.

The mouse x globin gene is not expressed in the definitiveerythroid cells of the fetal liver. To establish if there is a parallelreduction in the number of human f transcripts per animal atthe same developmental stage, transgenic males were mated withnormal females and total RNA from individual day 16 fetal liverswas analyzed. DNA was isolated from each fetus, and transgenicswere identified by Southern blot hybridization. Representativetransgenic and non-transgenic RNA samples were assayed forthe presence of transcripts from the mouse a, the mouse x andthe human f globin genes by primer extension. The results ofthis experiment are shown in figure 4. In both transgenic andcontrol samples there is a strong mouse a signal, and nodetectable signal for mouse x or human f. Thus, transcriptionof the human f transgene has ceased in parallel with the switchfrom expression of murine embryonic to adult globin genes.

A. Z5

T 1 T C C T T T C C

r - 1 1 2 3 4 5 1 2 3 - 1 ' 5 dQC

V

i Mx

primer Mx • HE

3. Z11

C T C T

d9C 1 2 3 4

- M a

-Hi

HC

Figure 4. Primer extension analysis of RNA prepared from individual fetal livers.In panel A, RNA samples are from the Z5 line. In panel B, RNA samples arefrom the Zl 1 line. The number above each lane identifies the RNA preparationused in that primer extension reaction. Transgenic samples are designated by aT and non-transgenic controls are designated by a C. The primer(s) used in eachreaction are indicated below the lanes. Lane d9C contains RNA prepared fromyolk sacs of transgenic embryos at day 9, primed with a 1:1 mixture of the mousex and human f primers. Lane C contains the products of extension by all threeprimers with no added template RNA. Lane M contains a DNA size marker.

7096 Nucleic Acids Research, Vol. 18, No. 23

DISCUSSION

We have constructed transgenic mice that harbor the humanembryonic f globin gene associated with a 777 bp DNA fragmentthat spans erythroid-specific DNase I hypersensitive site II fromthe distal 5' region of the /3 globin gene cluster, and haveexamined the expression of the human gene during development.In six lines of transgenic mice that contained the same f globinDNA fragment without HSII, no expression of the human genewas detected at any stage of development (E. R., unpublished).This result mirrors what is observed when human a globin isintroduced into mice without an enhancer (15,16). Using theHSII-f construct, we have observed correctly initiatedtranscription of the human gene in the erythroid cells of theembryonic yolk sac. We conclude that the embryonic gene, likethe adult a globin genes, requires the presence of an erythroid-specific enhancer for expression in transgenic mice.

In addition, we find that the timing of expression of the humanf gene in erythroid tissues parallels that of the mouse x gene.From this we infer that like the murine embryonic globins, thehuman embryonic globin transgene is expressed only in theprimitive erythroid cells of the yolk sac. The results of this studysuggest that the 4.8 kb fragment containing the zeta globin genewith 0.55 kb of 5' and 2.75 kb of 3' flanking sequence containsall of the information necessary to specify that the gene beexpressed in primitive erythroid cells and not be expressed indefinitive erythroid cells. The evolutionary conservation of anapproximately 300 bp region located immediately 5' to both thehuman f and the mouse x genes suggests that this noncodingsequence may have an important regulatory function (24).

It is unlikely that expression of the human f gene in murineerythroid cells is limited to a particular time in development asa direct result of its association with the HSII fragment. Whenthe HSII fragment is linked to an adult human globin gene (aor /3) in transgenic mice, the adult gene can be expressed in adulttissues (17,18,19). Moreover, a human /3 globin transgene in anHSII-/3 construct is not developmentally regulated; the adult geneis expressed in both primitive and definitive murine erythroidcells (E.S, K.A. and E.R., unpublished).

Developmental regulation of the human f globin gene whenlinked to one of the major hypersensitive sites does not requirethe presence of other globin genes in cis. A similar observationwas reported by Raich et al. (25), who found that developmentalregulation of the human embryonic /3 globin gene («) is correctwhen that gene is linked to four major hypersensitive sites. Incontrast, the loss of developmental regulation of human y and)3 globin gene expression when the individual genes are introducedinto mice linked to the major hypersensitive sites of the £ globincluster is well documented (26,27,28).

It has been suggested that the switch from fetal to adult globingene expression involves a competition between the individualglobin genes for interaction with the major hypersensitive sites(27,28). The outcome of this competition is thought to beinfluenced by positive and/or negative factors whose presenceis developmentally regulated. According to this model, in theabsence of competition the major hypersensitive sites form astable interaction with a single gene and drive expression of thatgene at all developmental stages (27,28). This model cannotaccount for the observed developmental regulation of embryonicglobin transgenes, thus it seems likely that the mechanism forsilencing expression of embryonic globin genes at later stagesof development is fundamentally different from that whichregulates the transition from fetal to adult gene expression.

Further gene transfer experiments will allow clarification of therole of proximal regulatory sequences and of their interactionswith the major hypersensitive sites in the developmental regulationof globin gene expression.

The mechanism of a globin gene switching during developmenthas been refractory to experimental studies because of the lackof a cell culture or transgenic mouse system in which those geneswere appropriately regulated. The use of constructs includingmajor hypersensitive sites from the human /S globin cluster hasmade it possible to obtain tissue-specific a globin expression(15,16), and our results suggest that this is a system in whichthe developmental regulation of expression of the a-like globingenes can also be studied. Because the murine pattern ofembryonic a globin expression mirrors what occurs in humans,the transgenic mouse should provide an accurate and informativemodel system in which to study human a globin gene switching.

After the submission of this manuscript Higgs et al. (29)reported the identification of sequences located far upstream ofthe human a globin genes that activate high level a globinexpression both in MEL cells and in transgenic mice. The useof constructs containing those sequences, in place of the activatingsequence from the j3 globin cluster, may provide additionalinsights into the requirements for a globin switching.

ACKNOWLEDGEMENTS

We thank Shirley Clift and Deepa Kumar for identifying the zetatransgenic founders, and David Asarnow and Linda Couto forcritical reading of the manuscript. This work was supported byNIH-HL2O985 (EMR) and NRSA-HL07279 (EAS). Thisresearch was conducted at the Lawrence Berkeley Laboratory(Department of Energy Contract DE-AC03-76SF0OO98 to theUniversity of California).

REFERENCES1. Stamatoyannopoulos, G. & Nienhuis, A.W (1987) The Molecular Basis of

Blood Diseases, eds. Stamatoyannopoulos, G., Nienhuis, A.W., Leder, P.& Majerus, P.W. (Saunders, Philadelphia, PA).

2. Craig, M.L. & Russell, E.S. (1964) Developmental Biology 10, 191 -201.3. Fantoni, A , Bank, A. & Marks, P.A. (1967) Science 157, 1327- 1328.4. Groudine, M., Kohwi-Shigematsu, T., Gelinas, R., Stamatoyannopoulos,

G. & Papayannopoulou, T. (1983) Proc. Natl. Acad. Sci. 80, 7551-7555.5. Tuan, D., Solomon, W., Li, Q. & London, I.M. (1985) Proc. Natl. Acad.

Sci. 82, 6384-6388.6. Forrester, W.C., Takegawa, S., Papayannopoulou, T., Stamatoyannopoulos,

G. & Groudine, M. (1987) Nuc. Acids Res. 15, 10159-10177.7. Chada, K., Magram, J., Raphael, K., Radice, G., Lacy, E. & Costantini,

F. (1985) Nature 314, 377-380.8. Magram, J., Chada, K. & Costantini, F. (1985) Nature 315, 338-340.9. Townes, T.M., Lmgrel, J.B., Chen, H Y., Brinster, R.L. & Palrruter, R.D.

(1985) EMBO J. 4, 1715-1723.10. Chada, K., Magram, J. & Costantini.F. (1986) Nature 319, 685-689.11. KoUias, G., Wrighton, N., Hurst, J. & Grosveld, F. (1986) Cell 46, 89-94.12. Rubin, E.M., Ronghua, L., Cooper, S., Mohandas, N. & Kan, Y.W. (1988)

Am. J. Hum. Genet. 42, 585-591.13. Grosveld, F., Blom van Assendelft, G., Greaves, D.R. & Kollias, G. (1987)

Cell 51, 975-985.14. Talbot, D., Collis, P., Antoniou, M., Vidal, M., Grosveld, F. & Greaves,

D.R. (1989) Nature, 338, 352-355.15. Ryan, T.M., Behringer, R.R., Townes, T.M., Palmiter, R.D. & Brinster,

R.L. (1989) Proc. Nad. Acad. Sci. 86, 37-41 .16. Hanscombe, O., Vidal, M., Kaeda, J., Luzzatto, L., Greaves, D.R. &

Grosveld, F. (1989) Genes & Dev. 3, 1572-1581.17. Curtin, P.T., Liu, D., Liu, W., Chang, J.C. & Kan, Y.W. (1989) Proc.

Nad. Acad. Sci. 86, 7082-7086.18. Ryan, T.M., Behringer, R.R., Martin, N.C., Townes, T.M., Palmiter, R.D.

& Brinster, R.L. (1989) Genes & Dev. 3, 314-323.

Nucleic Acids Research, Vol. 18, No. 23 7097

19. Rubin, E.M., Spangler, E.A., Curtin, P., Witowska, E., Clift, S. & Lubin,B. (1989) Blood, 74, 823a.

20. Hogan, B., Costamini, F. & Lacy, E. (1986) Manipulating the Mouse Embryo:A Laboratory Manual (Cold Spring Harbor Laboratory, Cold SpringHarbor.N.Y.)

21. Feinberg, A.P. & Vogelstein, B. (1983) Anal. Biochem. 132, 6 -13 .22. Lau, Y.-F. & Kan, Y.W. (1983) Proc. Natl. Acad. Sci. 80, 5225-5229.23. Chirgwin, J.M., Przybyla, A.E., MacDonald, R.J. & Rutter, W.J. (1979)

Biochem. 18, 5294-5299.24. Leder, A., Weir, L. & Leder, P. (1985) Mol, & Cell. Biol. 5, 1025-1033.25. Raich, N., Ebens, A.J., Papayannopoulou, T., Enver, T. &

Stamatoyannopoulos, G. (1989) Blood 74, suppl. 1, 7a.26. Enver, T., Ebens, A.J., Forrester, W.C. & Stamatoyannopoulos, G. (1989)

Proc. Natl. Acad. Sci. 86, 7033-7037.27. Enver, T., Raich, N., Ebens, A.J., Papayannopoulou, T., Costantini, F.

& Stamatoyannopoulos, G. (1990) Nature 344, 309-313.28. Behringer, R.R., Ryan, T.M., Palmiter, R.D., Brinster, R.L. & Townes,

T.M. (1990) Genes & Dev. 4, 380-389.29. Higgs, D.R., Wood, W.G., Jarman, A.P., Sharpe, J., Lida, J., Pretorius,

I.-M. & Ayyub, H. (1990) Genes & Dev. 4, 1588-1601.