Gene expression pattern of chicken erythrocyte nuclei in heterokaryons

MATHIAS BERGMAN* and NILS RINGERTZ

Department of Medical Cell Genetics, Medical Nobel Institute, Karolinska Instituted Box 600400, S-104 01, Stockholm, Sweden

* Author for correspondence from September 1990 at: Department of Pathology, University of Helsinki, Haartmaninkatu 3, SF-00290Helsinki, Finland

Summary

Expression of erythro-specific chick genes was stud-ied in heterokaryons prepared by fusing chick eryth-rocytes (CE) with rat myoblasts. In this type ofheterokaryon the inactive erythrocyte nucleus takesup nuclear proteins of myoblast origin and under-goes transcriptional reactivation. In order to studythe stability of the genetic programming of the reacti-vated CE nucleus, chick gene expression was exam-ined by analysis of RNA from the heterokaryons.Probes for several erythro- and chick-specific geneswere used. The heterokaryons showed strong ex-pression of the chick histone H5 and adult a-globingenes, while other genes, e.g. the transcription factorEryfl gene, normally expressed during erythroiddifferentiation, were not transcribed. Although theCE used were of the definitive lineage, the hetero-karyons showed activation of the chick embryonic

/J-globin gene, i.e. a gene normally expressed only inCE of the primitive lineage.

We conclude that the reactivation of the mature CEnucleus in a rat cytoplasm resulted in a more imma-ture erythroid gene expression pattern. The acti-vation of the embryonic /J-globin gene indicated aswitch of the lineage-specific gene expression pat-tern. This switch occurred in the absence of DNAreplication. The strong expression of the globin andH5 genes in heterokaryons, in the absence of ex-pression of the regulatory factor Eryfl, suggested theexistence of Eryfl-independent regulatory mechan-isms for erythroid gene expression in these cells.

Key words: heterokaryon, differentiation, erythrocyte, Eryfl.

Introduction

Differentiated cells in the adult organism and in lateembryos are programmed (committed) to express only atissue-specific subset of genes. This state is sometimesreferred to as the determined state. Once established, thisstate is stable in the sense that it may be inherited bydaughter cells, and even propagated in somatic cellhybrids for many generations (for review, see Ringertz andSavage, 1976). The mechanisms that control cell differen-tiation can be studied in multinucleate heterokaryons.Fusion of normal, differentiated cells of different tissueshas resulted in heterokaryons showing co-expression orextinction of differentiation markers (Ringertz andSavage, 1976; Blau et at. 1985). Reprogramming of nucleifor a new type of tissue-specific gene expression in hetero-karyons also occurs. Thus, Blau et al. (1985) have showninduction of nuclei from non-muscle cells to express myo-specific genes in heterokaryons and Baron and Maniatis(1986) have described activation of globin genes in non-erythroid nuclei. In view of these observations it was ofconsiderable interest to examine if and to what extent thedifferentiated chick erythrocyte (CE) nucleus retained itsprogramming for erythroid gene expression in hetero-karyons.

The advantage of the heterokaryon system, compared toproliferating cell hybrids, is that the unfused parentalcells express characteristic phenotypic markers and givehigh fusion yields. Furthermore, problems arising fromselection of cells and chromosome segregation can beJournal of Cell Science 97, 167-175 (1990)Printed in Great Britain © The Company of Biologists Limited 1990

avoided, since the heterokaryons can be maintained in anon-dividing state for at least two weeks. Also, since thetwo types of nuclei are maintained as separate compart-ments, the presence and effects of fra/is-acting factors canbe examined.

Avian erythropoiesis is a complex process during whichthe types of precursor cells and the site of development ofthese cells change. Two main lineages of erythroid cellsexist during embryonic development. These cell types areclassified by the types of globins they produce. Cells of theprimitive lineage arise from precursors in the yolk sac andexpress embryonic globin genes (e.g. the p gene of the/S-type globins) as they mature. After day 6 of embryonicdevelopment, the primitive lineage is replaced by defini-tive erythroid cells. These originate from stem cells in thebone marrow and express adult globin genes (e.g. the aA-gene of the a--globin cluster), and are the erythroid cellsfound in the circulation of the adult animal (Ingram, 1981;Tobin et al 1981; Groudine and Weintraub, 1981). As thecells of both lineages mature, they accumulate theirspecific types of hemoglobin. The nuclei of fully differen-tiated definitive CE have undergone a drastic chromatincondensation. During this process they become transcrip-tionally inactivated and cease to synthesize DNA(Ringertz and Bolund, 1974; Williams, 1972). As hemoglo-bin accumulates in the cytoplasm of the cells, the erythro-specific linker histone H5 accumulates in the nucleus,possibly causing the inactivation of the genome (Appelsand Wells, 1972; Bergman etal. 1988; Ringertz etal. 1985).After replication has stopped, synthesis of the other

167

histones is markedly reduced but adult-type globin and H5continue to be synthesized (Affolter et al. 1987). Transfec-tion assays have suggested that the erythro-specific ex-pression of both the embryonic and adult globin genes iscontrolled by binding of a nuclear protein, Eryfl, to aspecific enhancer element (Evans and Felsenfeld, 1989;Evans et al. 1988), which results in a 60- to 80-fold increasein the transcription of a reporter gene (Hesse et al. 1986;Evans et al. 1988). The enhancer is located in the promoterregion of the /3p-globin gene and in the 3' flanking region ofthe adult o--globin gene (Evans et al. 1988). Interestingly,the erythro-specific expression of the H5 gene is alsomediated by an identical enhancer element located down-stream from the H5 gene (Trainor et al. 1987; Trainor andEngel, 1989).

During differentiation of the definitive CE, expression ofthe genes encoding the cytoskeletal protein band 4.1 (Yewet al. 1987), the erythro-specific isoform of carbonic anhyd-rase (CA, type II; Yoshihara et al. 1987) and the erythro-specific type of (5-aminolevulinate synthase (ALA-S; Rid-dle et al. 1989) is also induced. Expression of the band 3gene, encoding the erythro-specific trans-membraneanion-transport protein also increases during differen-tiation of the definitive CE, although this gene is alsoexpressed in primitive erythroid cells (Kim et al. 1988).

After fusion of the mature, inactive CE with transcrip-tionally active cells, such as rat myoblasts, the erythrocytenucleus undergoes transcriptional and replicational reac-tivation (reviewed by Harris, 1970; Ringertz and Savage,1976).

We have studied the stability and regulation of the celltype-specific gene expression pattern of the erythroidnucleus in inter-specific (chick x rat) heterokaryons byexamining mRNA expression of genes transcribed duringerythroid differentiation (H5 histone, adult and embryonicglobin, Eryfl, CA II, band 3, ALA-S, band 4.1) as well as agene typically active in precursor cells (myb). We showthat the erythrocyte genome largely retained its erythroidexpression pattern, in the absence of Eryfl gene ex-pression. However, the embryonic /S-globin gene, normallyexpressed only in primitive erythroid cells, and the mybgene, were activated.

Materials and methods

CellsMature, definitive chick erythrocytes (CE) were collected fromblood of 18-day-old embryos. In order to obtain RNA fromprimitive cells to detect a clear Eryfl hybridization signal, CEwere also collected from 5-day-old embryos. The cells were washedtwice in phosphate-buffered saline (PBS) and then used for thedifferent experiments.

The rat myoblast cell line, L6J1 (Ringertz et al. 1978), wasgrown in Dulbecco's modified Eagle's medium (DMEM), sup-plemented with 5% fetal calf serum (FCS).

Cell fusions using CE and u.v.-irradiated CEFor the fusion experiments, L6J1 cells were seeded at 2.5xlO6

cells per 9 cm dish and treated with O^mgml"1 mitomycin C for20 h to inhibit DNA replication. This treatment prolongs the life-span of the heterokaryons dramatically, and prevents overgrowthby non-fused cells. The drug was washed away with PBS and thecells were fused for 20h with CE, using u.v.-inactivated Sendaivirus (Bolund et al. 1969). The rat myoblasts were incubated incold Earle's balanced salt solution (EBSS) for 15 min at 4°C. Theywere then incubated for 20 min in the cold with the inactivatedvirus, suspended in cold EBSS. The CE, suspended in cold EBSSwere then added and the cell mixture was kept in the cold for

another 60 min. After that, the mixture was incubated at 37 °C for4-5 h. The moment when cells were transferred to 37 °C wasdefined as the zero time point of the incubation. FCS was added to5 % and the cells were incubated for 20 h. After fusion, unfused CEwere thoroughly washed off and the heterokaryons incubated inDMEM/5 % FCS for different periods of time.

The percentage fusion efficiency was determined as the numberof chick nuclei per myoblast (e.g. 100 chick nuclei per 100myoblasts=100% fusion). In the experiments with u.v.-inacti-vated CE, these cells were u.v.-irradiated with a dose of 2340Jm~2 (Harris, 1967) in 9 cm dishes, with occasional stirring, andthen used for fusion as described above.

AutoradiographyThe transcriptional activity of the CE nuclei in heterokaryonswas determined by pulse labelling of cells grown on coverslipswith [3H]uridine. The heterokaryons were incubated withlOmCiml"1 for lh . DNA synthesis of the heterokaryons wasanalyzed by labelling the cells with [3H]thymidine at 50 mCi ml"1

for 2 h. In both labelling procedures the coverslips were exposed toKodak AR 10 film for 5 days at 4°C. After development of the film,the number of silver grains over the nuclei was counted.

RNA extraction, Northern blotting and hybridizationTotal RNA was extracted with hot phenol (Edmonds and Cara-mela, 1969). Poly(A)+ RNA was prepared by selection on anoligo(dT)-cellulose column (Pharmacia). The RNA was run on1.5 % agarose formaline/formamide gels (Lehrach etal. 1977), andblotted onto Hybond-N membranes (Amersham) overnight, ac-cording to the manufacturer's recommendations. The membraneswere prehybridized, then hybridized with nick-translated probes,or probes labelled by random priming (Feinberg and Vogelstein,1984) according to standard protocols (1.5xl06ctsmin~1ml~1,overnight). The membranes were then washed at high stringency,O.lxSSC (SSC is 0.15 M NaCl, 0.015 M sodium citrate) at 60°C andexposed to Fuji ARG-film at -70°C for 20 h to 6 days. Under theseconditions, none of the chick probes hybridized to rat RNA.

Run-on transcription assayNuclei from mature CE were isolated by disrupting the cells inreticulocyte standard buffer (RSB; 10 mil Tris-HCl, pH7.4,10 mM NaCl, 3mM MgCl2), containing 0.5% NP-40 with 20passages through a 19 guage needle. The nuclei were washed 1-3times in RSB/0.005% NP-40, and then resuspended in 29mlRSB/0.005% N P - 4 0 / I M sucrose and layered on top of 10ml ofRSB/2.3M sucrose and centrifuged at 25 000 revs min"1 (AH 627swing-out rotor) for l h at 4°C. The nuclear pellet was washedonce in RSB and 100 X106 nuclei were resuspended in 100/d2xreaction buffer (Marzluff, 1978): 5mM Mg(CH3COO)2, 50mMTris-HCl (pH 8), 25 % glycerol, 5 mM DTT (dithiothreitol). Then alOxsalt solution (1.2 M KC1, 25 mM Mg(CH3COO)2) and RNasin(1200 Uml"1; Promega) were added. Nucleotides were added tofinal concentrations of 0.4 mM ATP, 0.2 mM CTP, 0.2 mM GTP,7.5 mM cold UTP, and 1.5 mM [32P]UTP (3000 Ci minor1). Thevolume of the mixture was adjusted to 200/d with water, andincubated for 40 min at 25 °C with shaking. After incubation, themixture was treated with DNase I (20^gml~1 DNase I,300/igml"1 yeast tRNA, 2mM MgCy for 15min at room tem-perature. The RNA was extracted with hot phenol, ethanol-precipitated, resuspended in 100;il TEN buffer (10 mM Tris-HCl,pH 8,100 mM NaCl, 1 mM EDTA), and separated from unincorpor-ated nucleotides on a G-50 (Pharmacia) spun column. The radio-activity of the sample was determined, the sample was boiled for1-2 min to destroy possible RNA secondary structures andRNA-RNA hybrids, and then added to the hybridization solution.The radioactive RNA was hybridized overnight at 42 °C(1.5xl06ctsmin~1ml~1) to Hybond-N niters containing dots ofdifferent insert-containing plasmids (0.1//g insert/dot). The dotblots were prepared according to Amersham's protocol. The filterswere then washed under stringent conditions (O.lxSSC, 60°C,1 h), treated with RNase (20 ̂ gml"1 TEN) at 37 °C for 15 min, andthen exposed to film at -70°C for a week. Samples of the labelledRNA were taken and run on 7 % polyacrylamide/urea denaturing

168 M. Bergman and N. Ringertz

gels to determine the size of the RNA. Gels were dried andexposed to film overnight.

PlasmidsThe chicken histone H5 probe, pCH5-03, contained a 913 basepair (bp) cDNA insert cloned into pBR322 (Krieg et al. 1982). Itwas a gift from Dr Julian Wells (University of Adelaide, Austra-lia).

The plasmid cr14, containing a 1.7 kilobase (kb) genomic insertof adult chicken cAglobin gene in pBR322 (Stalder et al. 1980)and the plasmid /32H2, carrying a 4.3 kb genomic insert contain-ing the chicken embryonic /3-globin gene (the p-gene) in pBR322(Dolan et al. 1981) were provided by Dr Harold Weintraub (FredHutchinson Cancer Research Center, Seattle, USA). For the dotblots, an 850 bp Sacl fragment, containing only coding sequence,was excised from f}2H2.

The chicken Eryfl probe p20-l, containing the 1030 bp full-length Eryfl cDNA inserted into the £coRI site of pSK II(Bluescript) (Evans and Felsenfeld, 1989), was kindly provided byDr Gary Felsenfeld (NIH, Bethesda, USA).

The plasmid pCA-1.2, carrying the cDNA encoding the erythro-specific isoform II (CA II) of chicken carbonic anhydrase, con-tained a 1.2 kb fragment cloned in the -BcoRI site of pBR 325(Yoshihara et al. 1987). The chicken band 3 cDNA was a 2.8 kbPsfl-fragment derived from the plasmid pcB3-3.1 (Zenke et al.1988). Both cDNAs were kindly provided by Dr Bjttrn Vennstrom(Karolinska Inst, Stockholm, Sweden).

The plasmid pAE18 (Riddle et al. 1989), containing a 1.7 kbcDNA encoding the erythro-specific form of the <5-aminolevulinatesynthase (ALA-S) gene in pGEM4, and the plasmid p4.1A (Yew etal. 1987), carrying a 1.9 kb insert ofcDNA encoding protein 4.1, inpGEM2, were both gifts from Dr Douglas Engel (NorthwesternUniversity, Evanston, USA).

The chicken-specific myb probe pMl, a pUC19 derivative,contained an 810 bp Xbal-EcoRl fragment of AMV (Avian mye-loblastosis virus) v-myb, corresponding to exons 3-7 of thechicken c-myb gene (Klempnauer et al. 1982). This probe waskindly provided by Dr Thomas Graf (EMBL, Heidelberg, FRG).

The pCRlov2.1, carrying a 2.1kb chicken ovalbumin cDNA(Humphries et al. 1977) was kindly provided by Dr PierreChambon (Institute de Chimie Biologique, Strasbourg, France).

The actin probe, pAM91 (Minty et al. 1981), contained a 1350 bpinsert of mouse o"-actin cDNA, cloned in pBR 322. This sequencehybridizes to both the ubiquitous cytoskeletal )5- and y-types of ratand chicken actin mRNA, and the o"-type mRNA, expressed atmyotube formation (Minty et al. 1981). The mouse skeletal musclemyosin heavy chain (MHC) probe was a 1252 bp insert in pBR322(Weydert et al. 1983). This sequence detects rat and chick myosinmRNA. These probes were kindly provided by Dr Margaret E.Buckingham (Inst. Pasteur, Paris, France).

Results

mRNA expression in chick erythroid cells prior toheterokaryon formationIn order to be able to establish whether or not activation ofgenes occurred in erythrocyte nuclei after fusion with ratmyoblasts, the original transcriptional activity of CE wasdetermined in vitro, using a run-on assay with nuclei fromunfused CE from 18-day-old embryos. Fig. 1 A shows thatthe H5 and the adult o--globin genes as well as the carbonicanhydra9e gene (CA II) were transcribed. The band 3 andband 4.1 genes produced small amounts of RNA, while theEryfl, ALA-S, myb and ovalbumin genes failed to producedetectable hybridization signals. As expected (cf. Groudineet al. 1981), the embryonic /3-globin gene was not tran-scribed in these cells. No in vitro synthesized RNA hybrid-ized to the actin and myosin sequences (not shown).

The results from the run-on assay were largely consist-ent with Northern analyses of RNA from 18-day-old CE(Fig. 1 B). The CE contained large amounts of H5 (Fig. 1

B; lane 1) and adult type a-globin (lane 2) mRNAs, someCA II (lane 4) and band 3 mRNAs (lane 4). Despite the lackof detectable transcription, they also contained embryonic/3-globin mRNA Gane 3) and large amounts of ALA-SmRNA (lane 5). No Eryfl, band 4.1 or myb mRNAs weredetected in the CE (not shown).

Expression of chick mRNAs in rat myoblastxCEheterokaryonsCE were fused with L6 rat myoblasts, the replication ofwhich was inhibited with mitomycin-treatment beforemyotube formation. Heterokaryons resulting from suchfusions fail to enter mitosis and survive for long periods,thereby allowing more chick RNA to accumulate. In thesecells transcription in the CE nuclei is strongly reactivated(Table 1), but DNA synthesis is not (Ringertz and Savage,1976; Table 2).

A

H5 a g Eryl 1

CA B3 ALA-S B 41

myb ov pBR

B CE

H5 a j3

4.5- -B3

-CA

-ALA-S

Fig. 1. (A) Run-on transcription on nuclei of mature chickenerythrocytes from 18-day-old embryos. The embryonic /3-globin(yS), Eryfl, ALA-S, myb and ovalbumin genes are nottranscribed, (a, adult o^globin; CA, carbonic anhydrase; B3,band 3; B4.1, band 4.1; ov, ovalbumin; pBR, pBR 322, control.(B) Presence of transcripts in mature (18-day) CE prior tofusion. The size markers at the left indicate the positions of the18 S and 28 S ribosomal RNAs. The probe used for detection isindicated above or to the right of the lanes; 6 days exposure ofall samples. Lane 1 contains 20 //g poly(A)+ RNA, lanes 2-5contain 20 /Jg total RNA. Lane 1, H5 mRNA; lane 2, adult a-globin mRNA; lane 3, embryonic y3-globin mRNA; lane 4; CA IIand band 3 (B3) mRNAs; lane 5; ALA-S mRNA.

Chick gene expression in heterokaryons 169

Table 1. Uridine incorporation in heterokaryons

Type of experimentIncubation

time (h)% of labelled

rat nuclei% of labelled

CE nuclei

L6xCE

L6xUV-CE

20144

20

100100

100100

100100

0a

100100

100100

Incorporation of [3H]uridine in CE and rat nuclei in heterokaryons during reactivation. Nuclei were scored after labelling of the cells andautoradiography as described in Materials and methods. Nuclei showing >10 silver grains were scored as positive. The data were collected from 1+1-heterokaryons only (1 L6 nucleus + 1 CE nucleus per cell).

Table 2. Thymidine incorporation in heterokaryons

Type of experimentsIncubation

time (h)

% of labelledrat nuclei

O10 grains/nucleus)

% of labelledCE nuclei

(>5 grains)

L6mit+xCE

L6 mit+, mock fusion*

L6mit~xCE

L6 mit~, unfused

2012020

120

2012020

120+

80•81

50.560S

00

- i

m0.6

5050

100100

5050

100100

Incorporation of [3H]thyniidine in CE and rat nuclei in 1 + 1— heterokaryons during reactivation, and in nuclei of unfused rat myoblasts. Nucleishowing a defined number of silver grains in autoradiographic preparations were scored as positive after labelling as described in Materials andmethods. mit+ and mit~ indicate addition or omission of mitomycin C.

•Fusion procedure without addition of CE.t Myotube formation.

The expression pattern of muscle-specific genes was firstcharacterized in the mitomycin-treated, unfused rat myo-blasts. RNA was extracted at day one, three, six and tenafter drug treatment. In these cells, transcription of themyotube-specific »-actin gene (Caravatti et al. 1982) wasactivated (Fig. 2 A), and the myotube-specific myosinheavy chain gene (Weydert et al. 1983) was expressed atincreasing levels during incubation (Fig. 2 B). Thus, therat myoblasts expressed markers of differentiation duringthe incubation, in the absence of DNA synthesis andmyotube formation.

To ensure that the chick probes could be used fordetection of chick-specific gene expression in the hetero-karyons producing both rat and chick transcripts, theprobes were hybridized to the polyadenylated RNA fromthe mitomycin-treated rat myoblasts. No myb mRNA wasdetected with the chick-specific probe, and none of theother chick-specific probes hybridized to these blots (notshown).

In heterokaryons between CE and the mitomycin-treated rat myoblasts, the fate of the pre-existing chickmRNAs was studied, as well as the possible occurrence ofde novo produced chick transcripts. The chick mRNAs,introduced with the CE into the myoblasts, were degradedupon fusion. Fig. 3A, B, C; lanes 1, shows H5, a- and)3-globin mRNAs, respectively, after a 6h incubation ofheterokaryons. For these transcripts, degradation wasdetected as a smear of small RNAs. This was not due togeneral degradation of the RNA sample, since a correctactin mRNA signal was seen (not shown). The othererythro-specific mRNAs were not detected at this timepoint.

After 20 h, the cells expressed large amounts of intactH5 and o--globin mRNAs (Figs 3A, B; lanes 2). CA IImRNA was weakly expressed (Fig. 3D; lane 1), while band3 and band 4.1 mRNAs were not (not shown).

A Actin20 72 144 240 h

4.5-

1.8-\-oc

B MHC

4.5- A . * .

Fig. 2. Expression of myo-specific genes in mitomycin C-treatedrat myoblasts, not fused with erythroid cells. Cells incubated20, 72, 144 and 240 h. The actin probe hybridizes to both the /?-and )«-type8 of cytoskeletal actin mRNAs, as well as to the lowermolecular weight myotube-specific a<-actin mRNA. (A) Actin.Note the activation of o^actin mRNA expression, despite theabsence of myotube formation. 20,ugpoly(A)+ RNA applied persample. (B) Myosin heavy chain (MHC) mRNA expression.Same filter as in A.

Interestingly, large amounts of intact embryonic /3-globinmRNA were found (Fig. 3C; lane 2), as well as lowamounts of ALA-S (Fig. 3 E; lane 1) and myb mRNAs

170 M. Bergman and N. Ringertz

H520

B240 h

Of

20 240h 20 240 h

I 1 2 34

1 2 3

1 2 3

D CA

20 72 240h

1.8-

ALA-S20 72 240 h

1.8-

F myb20 72 144 240 h

4.5-

1 2 3 1 2 3 1 2 3 4

G Eryfl20 72 240h 5d CE

1.8-

, r.^'.-,-,-

Fig. 3. Occurrence of chick transcripts during reactivation of CE nuclei in rat myoblastxCE heterokaryons. Equal exposure timesfor all autoradiograms. (A) H5. Note that the introduced, pre-existing H5 mRNA is degraded at fusion, but then large amounts ofpolyadenylated H5 mRNA accumulates. Lane 1. Heterokaryons incubated 6h after fusion. Fusion frequency: 60%, 20 j/g total RNAapplied; lane 2, 20h incubation, fusion frequency: 160%, 15 fig poly(A)+ RNA; lane 3, 240 h incubation. Fusion frequency: 66%, 25 figpoly(A)+ RNA. (B) Adult c-globin. The same filter as in A was used. (C) Activation of chicken embryonic /3-globin mRNA expressionin the absence of DNA replication in the CE genome. Same filter as in A. (D) Expression of chicken CA II mRNA duringreactivation. The expression is down-regulated. Lane 1, 20h after fusion; 160% fusion, 20/Jg total RNA; lane 2, 72 h after fusion;130% fusion, 30 fig poly(A)+ RNA; lane 3, 240h after fusion; 65% fusion, 25 fig poly(A)+ RNA. (E) Reactivation of ALA-S mRNAexpression. Lanes as in D. (F) Reactivation of myb mRNA expression. Lanes 1, 2 and 4 as in D; lane 3, 144 h after fusion; 120 %fusion, 20 fig total RNA applied. (G) Eryfl mRNA expression. This gene is not expressed at all during the incubation, but the probedetects moderate amounts of the Eryfl mRNA in RNA from CE from 5-day-old embryos (lane 4). Lanes 1-3 as in D; lane 4 contains20 fig total RNA. The size markers indicate the positions of 18 S RNA and H5 mRNA.

(Fig. 3F; lane 1). The occurrence of these mRNAs was aresult of transcriptional activation, since the correspond-ing genes were not expressed in the mature CE prior tofusion (Fig. 1 A) and the cytoplasmic mRNA pools weredegraded at fusion.

At 72 and 144 h after fusion, the heterokaryons con-tained large amounts of intact, polyadenylated H5 mRNA,and even larger amounts of adult and embryonic globinmRNA (not shown). CA II mRNA was also weakly ex-pressed at 72 h (Fig. 3D; lane 2), but not at 144 h (notshown).

Heterokaryons incubated for 240 h still contained largeamounts of H5 and globin mRNAs, but no CA II expressionwas detected (Fig. 3A, B, C, D; lanes 3). ALA-S mRNA wasvery weakly expressed up to 240h (Fig- 3E; lanes 2-3).The myb expression, however, showed different kinetics.After 72 h, the myb signal was almost undetectable (Fig. 3

F; lane 2), after 144 h quite strong (Fig. 3 F; lane 3), andthen again strongly reduced after 240 h (Fig. 3 F; lane 4).The band 3 and band 4.1 genes were not expressed duringthe nuclear reactivation.

In spite of the strong H5 and globin gene expression, theEryfl gene was not expressed at any time during incu-bation of the heterokaryons (Fig. 3 G, lanes 1-3). The lackof signal was not due to inability of the probe to hybridizeto the corresponding mRNA, since moderate levels ofEryfl mRNA were detected in CE from 5-day-old embryos(Fig. 3 G, lane 4).

Activation of the chick ovalbumin gene was not detectedat any time during the reactivation (not shown).

To prove further that the detected chick transcripts wereproduced by transcription of the chick genome, u.v.-irradiated CE were fused to L6 rat myoblasts as above.Such irradiated CE nuclei swell and take up proteins in

Chick gene expression in heterokaryons 171

u.v.-fusion144 h

B4.1-

B 3 -myb —

C A -A L A - S -

Eryf1,H5-

globins—

-4.5

_0-actin-1.8

t

Fig. 4. Gene expression in heterokaryons between ratmyoblasts and u.v.-irradiated CE, incapable of transcription.Only 6 days incubated (144 h) shown. Fusion frequency: 92%,5/(g poly(A)+ RNA applied. Lane 1. Hybridization with probesfor H5, Eryfl, embryonic /S- and adult a<-globin, CAII, band 3,ALA-S, band 4.1 and myb. No signals detected; lane 2,hybridization with the actin probe. No switch has occurred,since only /J-actin mRNA is detected. The same filter as inlane 1.

heterokaryons, as do unirradiated nuclei in hetero-karyons, but transcription is not activated (Bolund et al.1969; Table 1). Heterokaryons using irradiated CE wereincubated for 144 h after fusion, and RNA was extracted.None of the chick probes hybridized with mRNA extractedfrom these cells (Fig. 4; lane 1). In these heterokaryons, nomyotube-specific cr-actin mRNA could be seen; only /3-actinmRNA was expressed (Fig. 4; lane 2). The absence of chickmRNAs in this experiment showed that the mRNAsdetected in heterokaryons 20 h and later after fusion withnon-irradiated CE, were synthesized de novo from therespective chick genes.

To determine if the non-u.v.-irradiated CE nuclei affec-ted the myogenic properties of the rat nuclei duringreactivation, the expression of actin and myosin mRNAsin heterokaryons was studied. At 20 h after fusion, theexpression of /3-actin mRNA was very strong and nocr-actin mRNA was detected (Fig. 5, lane 1). During theincubation, the total expression decreased but at 240 h thea'-actin mRNA expression was activated (Fig. 5, lanes2-3). Compared to the mitomycin-treated, unfused rat

20 144 240 h

Fig. 5. Actin mRNA expression in rat myoblastxCEheterokaryons. Note the strong expression at 20 h after fusionand the decline at the later time points. Equal exposure timesof all lanes; Lane 1. 20 h after fusion; 140% fusion; 16 [igpoly(A)+ RNA; lane 2, 144h incubation; 145% fusion; 15/igpoly(A)+ RNA; lane 3, 240 h incubation; 65 % fusion, 25 /igpoly(A)+ RNA.

myoblasts after 20 h incubation (Fig. 2 A), the /3-actinmRNA signal was very strong. Whether this signal wasproduced by activation of the chick /3-actin gene or byinduction of the rat gene as a consequence of the fusioncould not be further elucidated, since the probe used wasnot species-specific (see Materials and methods). Themyosin mRNA expression, however, was totally sup-pressed (not shown), in contrast to the increasing ex-pression seen during incubation of unfused myoblasts(Fig. 2 B).

Effects of mitomycin C on chick transcription inheterokaryonsIn the above experiments, gene activation occurred in theabsence of DNA synthesis in the heterokaryons. In orderto study how reactivation of DNA synthesis in the CEnuclei influenced chick transcription, the mitomycin treat-ment of the L6 cells was omitted prior to fusion. As shownin Table 2, the CE nuclei in mitomycin-treated heterokar-yons were incapable of activating replication. In the non-treated heterokaryons, however, DNA synthesis was acti-vated in the CE nuclei early after fusion. At 20 h some18 % of the CE nuclei actively incorporated [3H]thymidine.After 5 days, replication ceased both in the L6 nuclei andthe CE nuclei (Table 2). RNA was extracted from bothtreated and untreated heterokaryons and the expression ofthe erythroid genes was compared. The results show thatthe heterokaryons capable of DNA synthesis containedmore H5 and globin transcripts than the ones treated withmitomycin, after both short and long incubations (Fig. 6,compare + and — lanes). CA II expression increased onlyat 20 h (Fig. 6). The increase in mRNA levels peruntreated heterokaryon was even greater than that re-vealed by the hybridization, since the fusion frequency ofthe untreated cells was only half that of the treated ones.However, expression of ALA-S was not affected and theEryfl, band 3, band 4.1 genes were not activated (notshown). Also, the ovalbumin gene was not activated (notshown).

Discussion

In the present experiments the stability of the erythroiddetermined state has been analysed by fusing maturechick erythrocytes (CE) of the definitive lineage with ratmyoblasts. The erythrocytes contain a small condensednucleus, which is inactive with respect to transcriptionand replication. The cells do, however, still contain smallresidual amounts of RNA.

In order to establish a base-line for the fusion exper-iments, the CE were first analysed for the presence oferythro-specific transcripts, and whether or not thesemRNAs were being actively transcribed before fusion. Theresults showed that mature CE contained transcripts forH5 histone, the adult type of o--globin, as well as smalleramounts of CAII and band 3 mRNAs. Run-on transcrip-tion assays demonstrated that at least a portion of thesetranscripts was produced by ongoing transcription. Othergenes, some expressed during erythropoiesis (e.g. theALA-S gene), myogenesis (the actin and myosin genes), orin other differentiation pathways (the ovalbumin gene),were inactive. Thus, at the time of fusion, some erythro-specific genes were transcribed in the CE.

After fusion with transcriptionally active cells, the CEnuclei underwent marked changes in morphology andbiochemical composition in parallel with a general reacti-

172 M. Bergman and N. Ringertz

H5 P CA

120 144 120 144 120 144 20 20 120 144 h

*

Fig. 6. Effect of mitomycin C on chicken gene expression in heterokaryons 20 h, 120 h and 144 h after fusion, respectively.+ indicates mitomycin C-treatment before fusion; - indicates omission of the drug. The H5, adult (<x) and embryonic QS) globinmRNAs and CAII mRNA are more strongly expressed in untreated cells. Note the fusion frequencies: + lanes, 120%; — lanes, 60 %-,20 \i% total RNA applied in all lanes. The same filters used for the different probes.

vation of transcription of the CE genome, as indicated bythe incorporation of [3H]uridine. These changes have beendescribed in detail elsewhere (for a review, see Ringertzand Savage, 1976). Particularly relevant in relation to thepresent study is the previous finding that in chick x mam-malian heterokaryons, mammalian nuclear proteins enterthe chick nuclear compartments at the same time as theH5 histone is lost from the CE nuclei (Appels et al. 1974).As an extension of these observations, we wished to studythe stability of the erythroid programming of the CEnucleus, when reactivated in a rat myoblast cytoplasm.

Expression of chick genes in heterokaryonsNorthern blot analysis of chick transcripts present inheterokaryons examined shortly after fusion (6 h) showeddegradation of pre-existing RNA species from the erythro-cyte fusion partner. De nouo chick mRNA synthesis wasdetected at 20 h post-fusion and increased thereafter.These RNA molecules were not detected if, prior to fusion,the chick erythrocytes were irradiated with u.v.-light,which abolishes transcriptional reactivation. It was con-cluded, therefore, that in heterokaryons with non-irradiated CE, the new RNA species detected 20 h andlater after fusion, were due to the transcriptional reacti-vation of CE nuclei.

The expression of the chick genes in heterokaryons fallsinto five categories. First, the H5, adult a--globin and CAIIgenes, which were actively transcribed before and afterfusion. Second, the band 3 gene that was weakly active inthe CE before fusion but suppressed in the heterokaryons.Third, the myb and the ALA-S genes, which were inactivebefore fusion, but were reactivated in the heterokaryons.Fourth, the embryonic /J-globin gene, which is not nor-mally expressed in the definitive lineage of erythrocytes,was activated in heterokaryons. The fifth category isrepresented by the ovalbumin gene, which is not expressedin erythroid cells or myoblasts and was not activated afterfusion.

Thus, after fusion with rat myoblasts, the CE nucleusexpressed some of the genes that were active duringerythropoiesis. It is noteworthy that the H5 gene wasstrongly expressed during the reactivation, since the H5histone is thought to be a factor involved in the excessivecondensation and transcriptional inactivation of the CEgenome during the final stages of erythropoiesis (for areview see Ringertz and Bolund, 1974; Bergman et al.1988). The expression of the a-globin gene is in agreementwith previous observations by Linder et al. (1981, 1982)and Lanfranchi et al. (1984). On the other hand, the

activation of the embryonic /S-globin gene, a gene normallyexpressed in the primitive erythroid cell lineage, wasunexpected, and has not been seen in previous studies.Together with the expression of the myb gene, this obser-vation suggests that the CE nuclei retained their eryth-roid nature but reverted to a more primitive/embryonictype of erythroid gene expression. This might be caused bythe general activation of the CE nucleus, which alsorenders it morphologically more erythroblast-like. At themolecular level, this might be induced and maintained bythe reactivation of the myb gene, since high expression oftransfected c-myb cDNA has been shown to inhibit DMSO(dimethylsulfoxide)-induced erythroid differentiation ofmouse Friend erythroleukemia cells (Clarke et al. 1988).

It is worth noting that no expression of the gene for theerythroid enhancer-binding protein Eryfl was detected inheterokaryons, while at the same time the enhancer-containing genes were strongly expressed (H5, adult andembryonic globin genes). Previous analyses of Eryfl func-tion by transfection assays or using nuclear extracts havesuggested that transcription of the target genes wasdependent on the presence of Eryfl (Evans and Felsenfeld,1989; Evans et al. 1988; Hesse et al. 1986; Trainor et al.1987; Trainor and Engel, 1989). However, our resultssuggest that H5 and globin gene transcription may besustained and even activated by other factors. In fact,Eryfl may not be absolutely or only required for erythroidgene expression, since Martin et al. (1990) and Romeo et al.(1990) have shown that the Eryfl gene is expressed also inmegakaryocytes and mast cells.

Stability of the determined state of chick erythrocytenucleiThe data suggest that the programming for erythroid geneexpression is quite stable in the present system, since theheterokaryons showed expression of H5 and adult n-globingenes, as well as some other genes typically expressedduring erythropoiesis. The only sign of reprogramming ofthe erythroid nucleus was the reactivation of the myb geneand the activation of the embryonic /3-globin gene. Theactivation of these two genes reflects different phenomena,however. The myb gene is expressed in immature, defini-tive cells, while the embryonic globin gene is normallyexpressed only in cells of the primitive lineage. Thus, theembryonic globin gene activation in heterokaryons rep-resents a reprogramming of erythroid lineage-specificgene expression. Although this is probably the first reportof globin switching in heterokaryons with non-erythroidcells, activation of embryonic globins in adult-type cells

Chick gene expression in heterokaryons 173

has previously been described in heterokaryons betweenmouse erythroleukemia (MEL) cells and the human eryth-roid cell line K562 (Baron and Maniatis, 1986). However,using the same human erythroblast cell line in fusionswith adult CE, Lanfranchi et al. (1984) found expression ofonly adult chick globins.

To examine if the CE nuclei were induced to expressmyogenic properties, expression of the muscle-specific a-subunit of the acetylcholine receptor (ACR) was alsostudied. However, no expression of chick ACR mRNAcould be detected (data not shown). Furthermore, theovalbumin gene remained inactive in all heterokaryons.Thus, no evidence could be found for a genetic reprogram-ming of the CE nucleus. These results agree with aprevious study by Carlsson et al. (1974), showing thatCEx primary rat myoblast hybrids produced only ratmyosin, while primary chick myoblast x primary rat myo-blast hybrids produced both chick and rat myosin.

DNA replication and gene activationOur results differ from those of Blau et al. (1985), whofound activation of human myogenic gene expression infusions of a variety of human non-myogenic cell types witha mouse myoblast line. This system is, however, quitedifferent from ours, since it uses two mammalian cell typesboth of which are transcriptionally active at the time offusion. Experiments by DiBernardino and Hoffner (1983)have shown that frog erythrocyte nuclei, if first con-ditioned in developing frog oocytes and then re-injectedinto enucleated oocytes, became totipotent and capable ofsupporting development of living tadpoles. These resultssuggest that factors regulating DNA synthesis might berequired for reprogramming of gene expression. However,Chiu and Blau (1984) have shown that myo-specific geneswere activated in nuclei from non-myogenic cells in mousemyotube heterokaryons treated with an inhibitor of DNAsynthesis. In our experiments, using mitomycin-treatedmyoblasts for fusion, chick erythroid genes were alsoactivated in the absence of DNA synthesis. This suggeststhat gene activation can occur independently of DNAsynthesis. Interestingly, however, in the experimentswhere the drug treatment of the rat myoblasts wasomitted and replication was initiated in the CE nucleiafter fusion (Table 2), the heterokaryons expressed thechick erythroid genes more strongly. Nevertheless, therewas no evidence of reprogramming. However, under theseconditions, the heterokaryons that eventually proceed tomitosis undergo abnormal chromosome segregation anddie. Therefore, the cells analysed were the remainingintact heterokaryons that had not completely replicatedtheir genomes.

We express our gratitude to Marianne Frostvik-Stolt and EviMellqvist for excellent technical assistance. We also thank Drs D.Forrest, B. Vennstrom and J. Thyberg for critical reading of themanuscript and many helpful suggestions.

This work was supported by the Swedish Medical ResearchCouncil, Karolinska Institutet Funds, The Finnish Academy ofScience and Nordiska Forskarkurser.

References

AFTOLTER, M., COTE, J., RENAUD, J. AND RUIZ-CARRILLO, A. (1987).Regulation of histone and /J*-globin gene expression duringdifferentiation of chicken erythroid cells. Molec. cell. Biol. 7,3663-3672.

APPELS, R., BOLUND, L. AND RINOERTZ, N. R. (1974). Biochemical analysisof reactivated chick erythrocyte nuclei isolated from chick/HeLaheterokaryons. J. molec Biol. 87, 339-365.

APPELS, R. AND WELLS, J. R. E (1972). Synthesis and turnover of DNA-bound histone during maturation of avian red blood cells. J. molec.Biol. 70, 675-679.

BAHON, M. AND MANIATIS, T. (1986). Rapid reprogramming of globin geneexpression in transient heterokaryons. Cell 46, 591-602.

BERGMAN, M. G., WAWRA, E. AND WINGE, M. (1988). Chicken hiatone H5inhibits transcription and replication when introduced intoproliferating cells by microinjection. J. Cell Sci. 91, 201-209.

BLAU, H. M., PAVLATH, G. K., HARDBMAN, E. C, CHIU, C-P.,SlLBERSTEIN, L., WEBSTER, S. G., MlLLER, S. C. AND WEBSTER, C. (1985).Plasticity of the differentiated state. Science 230, 758-766.

BOLUND, L., DAHZYNKIEWICZ, Z. AND RINGERTZ, N. R. (1969). Growth ofhen erythrocyte nuclei undergoing reactivation in heterokaryons. ExplCell Res. 56, 406-410.

CABAVATTI, M., MINTY, A., ROBERT, B., MONTAHRAS, D., WKYDERT, A.,COHEN, A. AND BUCKINGHAM, M. (1982). Regulation of muscle geneexpression. The accumulation of messenger RNAs coding for muscle-specific proteins during myogenesis in a mouse cell line. J. molec. Biol.160, 59-76.

CARLSSON, S.-A., LUGER, O., RINGERTZ, N. R. AND SAVAGE, R. E. (1974).Phenotypic expression in chick erythrocyte x rat myoblast hybrids andin chick myoblastxrat myoblast hybrids. Expl Cell Res. 84, 47-55

CHIU, C.-P. AND BLAU, H. M. (1984). Reprogramming cell differentiationin the absence of DNA synthesis. Cell 37, 879-887.

CLARKE, M. F., KUKOWSKA-LATALLO, J. F., WETSIN, E., SMITH, M. ANDPROCHOWNIK, E. V. (1988). Constitutive expression of a c-myb cDNAblocks Friend murine erythroleukemia cell differentiation. Molec. cell.Biol. 8, 884-892.

DIBERARDINO, M. A. AND HOFFNER, N J. (1983). Gene activation inerythrocytes: nuclear transplantation in oocytes and eggs of Rana.Science 219, 862-864.

DoLAN, M., SuGABMAN, B. J., DODGSON, J. B. AND ENGEL, J D. (1981).Chromosomal arrangement of the chicken fi-type globin genes Cell 24,669-677.

EDMONDS, M. AND CARAMELA, M. G. (1969). The isolation andcharacterization of adenosine monophosphate-nch polynucleotidessynthesized by Ehrlich ascites cells. J biol. Chem. 244, 1314-1324.

EVANS, T. AND FELSENFELD, G (1989). The erythroid-specifictranscription factor Eryfl: A new finger protein. Cell 58, 877-885.

EVANS, T., REITMAN, M. AND FELSENFELD, G. (1988). An erythrocyte-specific DNA binding factor recognizes a regulatory sequence commonto all chicken globin genes. Proc. natn. Acad. Sci. U.S.A. 86,5976-5980.

FEINBERG, A. P. AND VOGELSTEIN, B. (1984). A techique for radiolabellingDNA restriction endonuclease fragments to high specific activity.Analyt. Biochem. (addendum) 137, 266-267.

GROUDINB, M., PERETZ, M. AND WEINTRAUB, H. (1981). Transcriptionalregulation of hemoglobin switching in chicken embryos. Molec. cell.Biol. 1, 281-288.

GROUDLNE, M. AND WEINTRAUB, H. (1981). Activation of globin genesduring chicken development. Cell 24, 393-401.

HARRIS, H. (1967). The reactivation of the red cell nucleus. J. Cell Sci. 2,23-32.

HARRIS, H. (1970). Cell Fusion. Oxford: Oxford University Press.HESSE, J. E., NICKOL, J. M., LJEBER, M. R. AND FBLSBNFEID, G. (1986).

Regulated gene expression in transfected primary chickenerythrocytes. Proc. natn. Acad. Sci. U.Sji.. 83, 4312-4316.

HUMPHRIES, P., COCHET, M., KRUST, A., GERLINOER, P., KOURILSKY, P.AND CHAMBON, P. (1977). Molecular cloning of extensive sequences ofthe in vitro synthesized chicken ovalbumin structural gene. Nucl.Acids Res. 4, 2389-2406.

INGRAM, V. M. (1981). Hemoglobin switching in amphibians and birds. InHemoglobins in Development and Differentiation (ed. G.Stamatoyannopoulos and A. W. Nienhuis), pp. 147-160. New York:Alan R. Liss, Inc.

KIM, H.-R. C, YEW, N. S., ANSORGE, W., VOSS, H., SCHWAGER, C,VENNSTROM, B., ZENKE, M. AND ENGEL, J. D. (1988). Two differentmRNAs are transcribed from a single genomic locus encoding thechicken erythrocyte anion transport proteins (band 3). Molec. Cell.Biol. 8, 4416-4424.

KLEMPNAUBR, K.-H., GONDA, T. J. AND BISHOP, J. M. (1982). Nucleotidesequence of the retroviral leukemia gene v-myb and its cellularprogenitor c-myb: The architecture of a transduced oncogene. Cell 31,453-463.

KRIEG, P. A., ROBINS, A. J., COLMAN, A. AND WELLS, J. R. E. (1982).Chicken histone H5 mRNA: the polyadenylated RNA lacks theconserved histone 3' terminator sequence. Nucl. Acids Res. 10,6777-6785.

LANFRANCHI, G., LLNDER, S. AND RINGERTZ, N. R. (1984). Globinsynthesis in heterokaryons formed between chick erythrocytes andhuman K562 cells or rat L6 myoblasts. J. Cell Sci. 66, 309-319.

LEHRACH, H., DIAMOND, D., WOZNBY, J. M. AND BOEDTKER, H. (1977).RNA molecular weight determinations by gel electrophoresis under

174 M. Bergman and N. Ringertz

denaturing conditions, a critical reexamination. Biochemistry 16,4743-4749.

LJNDER, S., ZUCKBRMAN, S. H. AND RINGKKTZ, N. R. (1981). Reactivationof chicken erythrocyte nuclei in heterokaryons results in expression ofadult chicken globin genes. Proc. natn. Acad. Sci. U.SA. 78,6286-6289.

LINDKR, S., ZUCKERMAN, S. H. AND RINGERTZ, N. R, (1982). Pattern ofchick gene activation in chick erythrocyte heterokaryons. J. Cell Biol.98, 885-892.

MARTIN, D. I. K., ZON, L. I., MUTTER, G. AND ORKIN, S. H. (1990).Expression of an erythroid transcription factor in megakaryocytic andmast cell lineages. Nature 344, 444-447.

MARZLUFT, W. JR (1978). Transcription of RNA in isolated nuclei. InMethods in Cell Biology, vol. 19 (ed. L. Wilson), pp. 317-332. NewYork: Academic Press.

MINTY, A. J., CARAVATTI, M., ROBEBT, B., COHEN, A., DAUBAS, P.,WKYDEET, A., Gsoe, F. AND BUCKINGHAM, M. E. (1981). Mouse actinmessenger RNAs. Construction and characterization of a recombinantplasmid containing a complementary DNA transcript of mouse a^actinmRNA. J. biol. Chem. 286, 1008-1014.

RIDDLE, R. R., YAMAMOTO, M. AND ENGEL, J. D. (1989). Expression of 6-aminolevulinate synthase in avian cells: separate genes encodeerythroid-specific and nonspecific isozymes. Proc. natn. Acad. Sci.U.SA. 86, 792-796.

RINGERTZ, N. R. AND BOLUND, L. (1974). The nucleus during avianerythroid differentiation. In The Cell Nucleus, vol. 3 (ed. H. Busch), pp.417 446. San Francisco: Academic Press.

RINGERTZ, N. R., KRONDAHL, U. AND COLEMAN, J. (1978). Reconstitutionof cells by fusion of cell fragments. I. Myogenic expression after fusionof minicells from rat (L6) myoblasts with mouse fibroblast (A9)cytoplaste. Expl Cell Res. 113, 233-246.

RINGEHTZ, N. R., NYMAN, U. AND BERGMAN, M. (1985). DNA replicationand H5 histone exchange during reactivation of chick erythrocytenuclei in heterokaryons. Chromosojna 91, 391-396.

RINGERTZ, N. R. AND SAVAGE, R. E. (1976). Cell Hybrids. London, NewYork: Academic Press.

ROMBO, P.-H., PRANDINI, M.-H., JOULIN, V., MIONOTTB, V , PRENANT, M.,VAINCHENKER, W., MARGUBRIE, G. AND UZAN, G. (1990).

Megakaryocytic and erythrocytic lineages share specific transcriptionfactors. Nature 344, 447-449.

STALDER, J., LARSEN, A., ENGEL, J. D., DOLAN, M., GROUDINE, M. ANDWEINTRAUB, H. (1980). Tissue-specific DNA cleavages in the globinchromatin domain introduced by DNAase I. Cell 20, 461-460.

TOBIN, A. J., HANSBN, D. A., CHAPMAN, B. S., MCCABE, J. B. AND SBFTOR,E. A. (1981). Globin gene expression during chicken development. InHemoglobins m Development and Differentiation (ed. G.Stamatoyannopoulos and A. W. Nienhuis), pp. 203-213. New York:Alan R. Liss, Inc.

TRAINOR, C. D. AND ENGEL, J. D. (1989). Transcription of the chickenhistone H5 gene is mediated by distinct tissue-specific elements withinthe promoter and the 3' enhancer. Molec. cell. Biol. 9, 2228-2232.

TRAINOR, C. D., STAMLER, S. J. AND ENGBL, J. D. (1987). Erythroid-specific transcription of the chicken histone H5 gene is directed by a 3'enhancer. Nature 328, 827-830.

WKYDBHT, A., DAUBAS, P., CARAVATTI, M., MINTY, A., BUGAISKY, G.,COHEN, A., ROBERT, B. AND BUCKINGHAM, M. (1983). Sequentialaccumulation of mRNAs encoding different myosin heavy chainisoforms during skeletal muscle development in vivo, detected with arecombinant plasmid identified as coding for an adult fast myosinheavy chain from mouse skeletal muscle. J biol. Chem. 258,13867-13874.

WII.I JAMB A. F. (1972). DNA Bynthesis in purified populations of avianerythroid cells. J. Cell Sci. 10, 27-46.

YEW, N. S., CHOI, H.-R., GALLARDA, J. L. AND ENGEL, J. D. (1987).Expression of cytoskeletal protein 4.1 during avian erythroid cellularmaturation. Proc. natn. Acad. Sci. U.S.A. 84, 1035-1039.

YOSHIHARA, C. M., LEE, J.-D. AND DODGSON, J. B. (1987). The chickencarbonic anhydrase II gene: evidence for a recent shift in intronposition. Nucl. Acids Res. 15, 753-770.

ZENKE, M., KAHN, P., DISBLA, C, VENNSTROM, B., LEUTZ, A., KEEGAN,K., HAYMAN, M. J., CHOI, H.-R., YEW, N., ENGEL, J. D. AND BEUG, H.(1988). v-erbA specifically suppresses transcription of the avianerythrocyte anion transporter (band 3) gene. Cell 52, 107-119.

(Received 27 March 1990 - Accepted 2 July 1990)

Chick gene expression in heterokaryons 175