5
Proc. Natl. Acad. Sci. USA Vol. 92, pp. 10713-10717, November 1995 Developmental Biology Intraembryonic hematopoietic cell migration during vertebrate development H. WILLIAM DETRICH TI1*t, MARK W. KIERANt, FUNG YEE CHANt#, LAUREN M. BARONEtt, KAREN YEEt, JON A. RUNDSTADLERt, STEVEN PRAT[t, DAVID RANSOMtt, AND LEONARD I. ZONt#§ *Department of Biology, Northeastern University, Boston, MA 02115; and tHoward Hughes Medical Institute, and tDivision of Hematology/Oncology, Children's Hospital and Dana-Farber Cancer Institute, Department of Pediatrics, Harvard Medical School, Boston, MA 02115 Communicated by Elizabeth S. Russell, The Jackson Laboratory, Bar Harbor, ME, June 30, 1995 ABSTRACT Vertebrate hematopoietic stem cells are de- rived from ventral mesoderm, which is postulated to migrate to both extra- and intraembryonic positions during gastrula and neurula stages. Extraembryonic migration has previously been documented, but the origin and migration of intraem- bryonic hematopoietic cells have not been visualized. The zebrafish and most other teleosts do not form yolk sac blood islands during early embryogenesis, but instead hematopoi- esis occurs solely in a dorsal location known as the interme- diate cell mass (IM) of Oellacher. In this report, we have isolated cDNAs encoding zebrafish homologs of the hemato- poietic transcription factors GATA-1 and GATA-2 and have used these markers to determine that the IM is formed from mesodermal cells in a posterior-lateral position on the yolk syncytial layer of the gastrula yolk sac. Surprisingly, cells of the IM then migrate anteriorly through most of the body length prior to the onset of active circulation and exit onto the yolk sac. These findings support a hypothesis in which the hematopoietic program of vertebrates is established by vari- ations in homologous migration pathways of extra- and in- traembryonic progenitors. In 1872, Joseph Oellacher demonstrated that embryonic he- matopoiesis in most teleosts occurs in a dorsal intraembryonic region of the tail bud called the intermediate cell mass (IM) (1). The intraembryonic site of hematopoiesis in bony fishes has perplexed developmental biologists, because at an analo- gous developmental stage, all other vertebrate species that have been examined form blood on the extraembryonic yolk sac (1-13). In 1885, Wenckebach (12) hypothesized that the IM was formed by the convergent migration of two bilateral populations of hematopoietic progenitors; nevertheless, care- ful studies in the 20th century have failed to define the developmental origin of the IM (2-11). Vertebrate embryonic hematopoiesis involves critical tran- scriptional regulation of coordinately expressed genes such as the globins, heme biosynthetic proteins, and cell-surface re- ceptors. Functionally defined GATA motifs, which are recog- nized by GATA-binding transcription factors, are present in the promoters and enhancers of many hematopoietic-specific genes (14). In mice, disruption of the Gatal or Gata2 gene leads to severe hematopoietic defects, demonstrating a re- quirement for these factors for normal erythroid maturation (15, 16). GATA-1 and GATA-2 have recently been used in Xenopus as markers of hematopoietic mesoderm induction (17). GATA-1 and GATA-2 are initially expressed during gastrulation, and GATA-1 expression is restricted to hemato- poietic cells during embryogenesis. GATA-2 is expressed in early ventral ectoderm and hematopoietic cells and later is expressed in the central nervous system. In each vertebrate species studied thus far, GATA-1 levels increase and GATA-2 levels decrease during erythroid differentiation (17). The existence of distinct ventral (yolk sac) and dorsal (intraembryonic) hematopoietic compartments in vertebrate embryos has been known for many years (18-24), but the origin and migration pathway of the dorsal compartment and its relationship to the ventral compartment remain to be determined. To address this issue, we have studied blood formation in the zebrafish, Danio rerio, which utilizes only a dorsal compartment during embryogenesis. Using GATA-1 and GATA-21 as early markers of hematopoietic tissue (14, 17), we have demonstrated that the dorsal population of cells containing these markers is derived from posterior-lateral mesoderm at the gastrula stage, and we have delineated the migration of the marked cells during development to the dorsal mesentery. These results suggest that similar migration path- ways are used in vertebrates for the formation of both ventral and dorsal hematopoietic populations. METHODS Zebrafish Maintenance. Wild-type zebrafish stocks were obtained from Ekk Will Waterlife Resources (Gibsonton, FL). Zebrafish matings were performed according to standard protocols (25). Wild-type embryos from crosses of spadetail heterozygous mutants (spadetail provided by C. Kimmel, Uni- versity of Oregon, Eugene) were occasionally used. Isolation of cDNA Clones Encoding Zebrafish GATA-1 and GATA-2. ZG1 and ZG2 were isolated from 36-h Agtll and neurula AZapII cDNA libraries, respectively, by screening at reduced stringency (26) with Xenopus GATA-1 (XG1) cDNA. From 5 x 105 plaques per library, one GATA-1 cDNA and 23 GATA-2 cDNAs were isolated. Phage cDNA inserts were subcloned into Bluescript KS1I- (Stratagene) and sequenced by the dideoxynucleotide chain-termination method (26, 27). Whole Embryo Staining for Globin Expression. o-Dianisidine staining was used to study the expression of globin (28). Decho- rionated (nonfixed) embryos were stained for 15 min in the dark in o-dianisidine (0.6 mg/ml), 0.01 M sodium acetate (pH 4.5), 0.65% H202, and 40% (vol/vol) ethanol. Stained embryos were cleared with benzyl benzoate/benzyl alcohol (2:1, vol/vol) and examined by differential interference contrast microscopy. Whole Embryo in Situ Analysis. In situ hybridization was performed by a modification of the method of Harland (29). After digestion with proteinase K, treatment with acetic anhydride, and prehybridization, embryos were incubated with digoxigenin-labeled antisense or sense RNA probes (1 ,ug/ml in hybridization buffer) for 14-16 h at 60°C. The embryos were Abbreviations: IM, intermediate cell mass; DM, dorsal mesentery. §To whom reprint requests should be addressed at: Division of Hematology/Oncology, Children's Hospital, 300 Longwood Avenue, Enders 650, Boston, MA 02115. 1The sequences reported in this paper have been deposited in the GenBank data base [accession nos. U18311 (GATA-1) and U18312 (GATA-2)]. 10713 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Downloaded by guest on August 4, 2020

Intraembryonic hematopoietic cell migrationduring ... › content › pnas › 92 › 23 › 10713.full.pdf10713 Thepublication costs ofthis article weredefrayed in part bypage charge

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Intraembryonic hematopoietic cell migrationduring ... › content › pnas › 92 › 23 › 10713.full.pdf10713 Thepublication costs ofthis article weredefrayed in part bypage charge

Proc. Natl. Acad. Sci. USAVol. 92, pp. 10713-10717, November 1995Developmental Biology

Intraembryonic hematopoietic cell migration duringvertebrate developmentH. WILLIAM DETRICH TI1*t, MARK W. KIERANt, FUNG YEE CHANt#, LAUREN M. BARONEtt, KAREN YEEt,JON A. RUNDSTADLERt, STEVEN PRAT[t, DAVID RANSOMtt, AND LEONARD I. ZONt#§*Department of Biology, Northeastern University, Boston, MA 02115; and tHoward Hughes Medical Institute, and tDivision of Hematology/Oncology, Children'sHospital and Dana-Farber Cancer Institute, Department of Pediatrics, Harvard Medical School, Boston, MA 02115

Communicated by Elizabeth S. Russell, The Jackson Laboratory, Bar Harbor, ME, June 30, 1995

ABSTRACT Vertebrate hematopoietic stem cells are de-rived from ventral mesoderm, which is postulated to migrateto both extra- and intraembryonic positions during gastrulaand neurula stages. Extraembryonic migration has previouslybeen documented, but the origin and migration of intraem-bryonic hematopoietic cells have not been visualized. Thezebrafish and most other teleosts do not form yolk sac bloodislands during early embryogenesis, but instead hematopoi-esis occurs solely in a dorsal location known as the interme-diate cell mass (IM) of Oellacher. In this report, we haveisolated cDNAs encoding zebrafish homologs of the hemato-poietic transcription factors GATA-1 and GATA-2 and haveused these markers to determine that the IM is formed frommesodermal cells in a posterior-lateral position on the yolksyncytial layer of the gastrula yolk sac. Surprisingly, cells ofthe IM then migrate anteriorly through most of the bodylength prior to the onset of active circulation and exit onto theyolk sac. These findings support a hypothesis in which thehematopoietic program of vertebrates is established by vari-ations in homologous migration pathways of extra- and in-traembryonic progenitors.

In 1872, Joseph Oellacher demonstrated that embryonic he-matopoiesis in most teleosts occurs in a dorsal intraembryonicregion of the tail bud called the intermediate cell mass (IM)(1). The intraembryonic site of hematopoiesis in bony fisheshas perplexed developmental biologists, because at an analo-gous developmental stage, all other vertebrate species thathave been examined form blood on the extraembryonic yolksac (1-13). In 1885, Wenckebach (12) hypothesized that the IMwas formed by the convergent migration of two bilateralpopulations of hematopoietic progenitors; nevertheless, care-ful studies in the 20th century have failed to define thedevelopmental origin of the IM (2-11).

Vertebrate embryonic hematopoiesis involves critical tran-scriptional regulation of coordinately expressed genes such asthe globins, heme biosynthetic proteins, and cell-surface re-ceptors. Functionally defined GATA motifs, which are recog-nized by GATA-binding transcription factors, are present inthe promoters and enhancers of many hematopoietic-specificgenes (14). In mice, disruption of the Gatal or Gata2 geneleads to severe hematopoietic defects, demonstrating a re-quirement for these factors for normal erythroid maturation(15, 16). GATA-1 and GATA-2 have recently been used inXenopus as markers of hematopoietic mesoderm induction(17). GATA-1 and GATA-2 are initially expressed duringgastrulation, and GATA-1 expression is restricted to hemato-poietic cells during embryogenesis. GATA-2 is expressed inearly ventral ectoderm and hematopoietic cells and later isexpressed in the central nervous system. In each vertebrate

species studied thus far, GATA-1 levels increase and GATA-2levels decrease during erythroid differentiation (17).The existence of distinct ventral (yolk sac) and dorsal

(intraembryonic) hematopoietic compartments in vertebrateembryos has been known for many years (18-24), but theorigin and migration pathway of the dorsal compartment andits relationship to the ventral compartment remain to bedetermined. To address this issue, we have studied bloodformation in the zebrafish, Danio rerio, which utilizes only adorsal compartment during embryogenesis. Using GATA-1and GATA-21 as early markers of hematopoietic tissue (14,17), we have demonstrated that the dorsal population of cellscontaining these markers is derived from posterior-lateralmesoderm at the gastrula stage, and we have delineated themigration of the marked cells during development to the dorsalmesentery. These results suggest that similar migration path-ways are used in vertebrates for the formation of both ventraland dorsal hematopoietic populations.

METHODSZebrafish Maintenance. Wild-type zebrafish stocks were

obtained from Ekk Will Waterlife Resources (Gibsonton, FL).Zebrafish matings were performed according to standardprotocols (25). Wild-type embryos from crosses of spadetailheterozygous mutants (spadetail provided by C. Kimmel, Uni-versity of Oregon, Eugene) were occasionally used.

Isolation of cDNA Clones Encoding Zebrafish GATA-1 andGATA-2. ZG1 and ZG2 were isolated from 36-h Agtll andneurula AZapII cDNA libraries, respectively, by screening atreduced stringency (26) with Xenopus GATA-1 (XG1) cDNA.From 5 x 105 plaques per library, one GATA-1 cDNA and 23GATA-2 cDNAs were isolated. Phage cDNA inserts weresubcloned into Bluescript KS1I- (Stratagene) and sequencedby the dideoxynucleotide chain-termination method (26, 27).Whole Embryo Staining for Globin Expression. o-Dianisidine

staining was used to study the expression of globin (28). Decho-rionated (nonfixed) embryos were stained for 15 min in the darkin o-dianisidine (0.6 mg/ml), 0.01 M sodium acetate (pH 4.5),0.65% H202, and 40% (vol/vol) ethanol. Stained embryos werecleared with benzyl benzoate/benzyl alcohol (2:1, vol/vol) andexamined by differential interference contrast microscopy.Whole Embryo in Situ Analysis. In situ hybridization was

performed by a modification of the method of Harland (29).After digestion with proteinase K, treatment with aceticanhydride, and prehybridization, embryos were incubated withdigoxigenin-labeled antisense or sense RNA probes (1 ,ug/mlin hybridization buffer) for 14-16 h at 60°C. The embryos were

Abbreviations: IM, intermediate cell mass; DM, dorsal mesentery.§To whom reprint requests should be addressed at: Division ofHematology/Oncology, Children's Hospital, 300 Longwood Avenue,Enders 650, Boston, MA 02115.1The sequences reported in this paper have been deposited in theGenBank data base [accession nos. U18311 (GATA-1) and U18312(GATA-2)].

10713

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Aug

ust 4

, 202

0

Page 2: Intraembryonic hematopoietic cell migrationduring ... › content › pnas › 92 › 23 › 10713.full.pdf10713 Thepublication costs ofthis article weredefrayed in part bypage charge

10714 Developmental Biology: Detrich et al.

treated with RNase and washed to a stringency of 0.2x SSC/0.3% CHAPS (60°C). Transcripts were detected by incubationwith alkaline phosphatase-conjugated anti-digoxigenin Fab frag-ments (Boehringer Mannheim), followed by development withnitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate(Promega). Sense control transcripts produced no signal. Someembryos were dehydrated, infiltrated with JB4 resin (Poly-science), and sectioned (5 ,um).

RESULTS

Cloning of the Zebrafish GATA-1 and GATA-2 cDNAs.cDNA clones encoding GATA-1 and GATA-2 were obtainedby screening zebrafish cDNA libraries with the zinc fingerregion of Xenopus GATA-1 cDNA (Fig. 1) (30). Sequenceanalysis of the encoded polypeptides demonstrated that theprimary structure of zebrafish GATA-1 is very similar to thatof zebrafish GATA-2, particularly in the zinc finger region.The extreme N termini of the two zebrafish transcriptionfactors are significantly more homologous to each other thanare the N termini of Xenopus GATA-1 and GATA-2 (Fig. 1;XG1 and XG2, respectively). This suggests that GATA-1 arose

1 50ZG2 EVAADQSRW MAHHHA QGOHSHHHGL TH EPMAP LLPPDEVDVZGl MENSSEPSRW VS... PG Sp ...V . ..rTDSGL LPPVDVDEPXGI1 . N DY.. .T qT Q P .... 4. ESGL ASTSE.DSOEXG2 MEVATDQPRW MAH.NAV G QH SHHPGL AHN MEP.TQ LLPPDEVDVM

51 100ZG2 LNHLDSQGN .Y SNS... ARVSYGQAHA RLTGSQVCRP HLIHSPGIPWZGI YSSSETDLLP SY STSVQS SSYRHSPV RQVYSSQSIL GNIQWL....XG1 LYGLGGES G GGAVSS GFRSPV FQTFPVGR.LHWP....XG2 FNHLDSQGN .YIANSAHA a RVSYSQAHA RLTGSQMCRP HLLHSPGLPW

101 150ZG2 LDSGKAALSA A.HHNAWAVS HFSKPGLHPA ... SAAY CS SSSSTAPVSSZG1 .......... DNSAGH SLN.SPYNPT STVWSSSFP KTPLHSHTSTXG1 .......... .... ETSAGI PSNLTAYGRS TGTLSF A ASALGPITSPXG2 LESGKTALSA AHHHNPWTVS PFGKAPLHPA ARGGSLY GT GSSACP....

151 200ZG2 LTSATHSSPH PLYNLPPTPP KDVSPDPGPS SPTSTTARMD EKE KYQVSZG1 .......... SIYQNTATPS FTSPKEGFPS PSR......D GK RLQ..XG1 .......... PLY.. SASS FLLG..... SA PPA....E RE KFL..XG2 . .SSSHSSPH .LFGFPPTPP KDVSPDPGPA SPPS.SSRLE DK KYQMS

201 _ 250ZG2 IADG C SPLRGS.... LAMSAQTPST HHPIPTYP.. .TYSLRAPHDZGI . . E L SPMSGSGSSF LSSATGGV YGPSPHMLSP YGSYMSTSQDXG1 .ET A SP ... TSDL PLEPRSPSI LQV......GYIGGGGQEXG2 LSE G SP LRSS.... LAPMGTQCST HHPIPTYP.. SY.VPAAHD

251 300ZG2 YGGGLFHPGA LLSGSASSFT PKCKSKTRSC SEFCTEG EC VNCGATZG1 YSSAALYSTG GPWMSPSSYS PKLRNKMRLS P ... PEA EC VNCGAITArPLlXG1 FS ..... LFQ S.. TED REC NCGAT VrP1IXG2 YSSGLFHPG SLLGGPASSFT PKQRSKSRSC S....EGCVNCGAT

301 350ZG2 HY ACGLY QMRPL CZG1 HY AC LY GQNRPL LVS SXG1 HY ACOLY GQNRPL VS S SXG2 NPLGLYNCR

351 400ZG2 rTTLWRGN VNACG L LT IQ KZG1 rTTLWRRAS VCNACG Y L LTMK IQXG1 rITTLWRRAS NACGL Y L LT IQ RXG2 rTLWRR NACGS L

401 450ZG2 RSFqRSGEGFE ELSKCMQDKT SPFGTASA . ASHMPHM PPFSHSZG1 K NSEV YPDMSHMAPP DEHVGAYSI PGPLLSY ..... PXG1 K.WQLDNPF EPPKAGVEEP SPYPFGPI4 HGQMPPM . INPPXG2 KEJGSECFE ELSRCMQEKS SPFS. AAPAI. ASHMAPM APFSHS

451 480ZG2 PTPTPIHP.. ..TFSHPHHS GRSPAWAEPHZG1 PTST.LHSST TLPYTHHPNFI GMMPTLV ...XG1 QSPR.ISHSA PAVSYRQAA84 GVTPP.....XG2 QTPTPIHPSS SLSFGHPHSW TAMG...

FIG. 1. Primary sequences of zebrafish GATA-I and GATA-2. Forpurposes of comparison, zebrafish GATA-1 (ZG1) and GATA-2(ZG2) have been aligned with Xenopus GATA-1 (XG1) and GATA-2(XG2) (30). Periods denote gaps introduced to establish maximalsequence similarity. Boxed residues are shared by all four GATA-binding proteins, and the dashed lines indicate the two zinc fingers.Based on GAP analysis (Genetics Computer Group format), ZG1 is45.2% identical to ZG2 and 48.3% identical to XG1, while ZG2 andXG2 are 78.2% identical. The longest ZG2 cDNA obtained lacked theinitiator codon and started with the second codon; we infer thatmethionine is the first amino acid based on homology to all othervertebrate GATA-2 polypeptide sequences (30).

Proc. Natl. Acad. Sci. USA 92 (1995)

from an ancestral GATA-2 molecule before or during theevolution of fish and that the structure of GATA-1 hasdiverged from that of GATA-2 in more advanced vertebrates.GATA-1 Expression Delineates the Migration of Hemato-

poietic Cells During Development. The erythroid cells of thezebrafish IM are easily distinguished by their small size, ovalshape, slightly condensed chromatin (3), hemoglobin expres-sion based on o-dianisidine staining (Fig. 2Q and R) (3, 4), andGATA-1 mRNA expression (Fig. 2H and data not shown).Based on whole embryo in situ analysis, GATA-1 expression isnot detected in any other embryonic tissue. Therefore, as in thedevelopment of other vertebrates (17, 32), GATA-1 mRNAexpression is restricted to hematopoietic cells during zebrafishembryogenesis. GATA-1 mRNA is initially detected at the2-somite stage in putative progenitors that reside in two stripesflanking the paraxial mesoderm of the posterior embryo andthat contact the yolk syncytial layer [Fig. 2A and B (arrows)].This embryonic region contains lateral mesoderm that someinvestigators consider extraembryonic (33-35) (Y. Kunz, per-sonal communication) and others regard as embryonic (W.Ballard, personal communication). The two cell masses thatexpress GATA-1 converge medially between the 2- and5-somite stages (Fig. 2 C and D), meet anteriorly at the18-somite stage, and fuse completely by the 24-somite stage(Fig. 2 E-H). Examination of the GATA-1-expressing cellsfrom the 2- to 18-somite stages by differential interferencecontrast microscopy demonstrated that they are large, roundcells with prominent nuclei that differentiate to the smallererythroblast stage as the cells meet at midline. These eryth-roblasts are distinct from larger interspersed cells of the IMthat differentiate into endothelial cells between 18 and 24somites (3). At 23 h [ust before the prim 5 stage (25)],GATA-1-expressing cells of the IM migrate anteriorly (Fig. 2I and J). By 24 h, the cells have reached the midtrunk region(Fig. 2K, arrow) and appear to exit onto the yolk sac (Fig. 2L-N) between the ectoderm and the yolk surface (sections notshown). On the yolk sac, the erythroid cells mature, charac-terized by increased nuclear condensation and a slightly moreelliptical cell shape. The migration of hematopoietic cellsinitiates before an active circulation is established at 24 h (6).Between 25 and 29 h, the common cardinal veins and ducts ofCuvier form (Fig. 20), and the blood cells appear to bechanneled toward the heart (Fig. 2P, arrow). The anterioraccumulation of blood cells on the yolk sac has also beenvisualized by time-lapse video microscopy of zebrafish em-bryos (C. Kimmel, personal communication; M. Thompsonand L.I.Z., unpublished observations).GATA-2 Is Expressed in an Early Hematopoietic Progenitor

Population. GATA-2 expression during development is simi-lar, although not identical, to that of GATA-1. GATA-2mRNA is initially detected in the ventral ectoderm at 75%epiboly (Fig. 3A), and high levels are expressed at the bound-ary of the embryo and in the yolk syncytial layer at 90% epiboly(Fig. 3 B and C). By the 2- to 5-somite stages, GATA-2expression is observed in cells that border the anterior (Fig.3D, closed arrow) and posterior (open arrow) embryo, includ-ing the presumptive hematopoietic progenitors. From the 5- to20-somite stages, these progenitors continue to expressGATA-2 mRNA (data not shown). Large cells in the posteriorIM express abundant GATA-2 mRNA but do not expressGATA-1 (compare Fig. 2G, open arrow with Fig. 3E, openarrow). Because of their blast-like morphology (3) and theabsence of GATA-1 expression, this cell population mayrepresent undifferentiated embryonic or larval/adult stemcells. After the 20-somite stage, the level of GATA-2 mRNAin the anterior cells of the IM decreases substantially (data notshown), consistent with its declining expression in embryonicerythroid cells after blood island formation in higher verte-brates (30). The intense GATA-2 expression throughout theIM (Fig. 3E) may be due to both hematopoietic and vasculo-

Dow

nloa

ded

by g

uest

on

Aug

ust 4

, 202

0

Page 3: Intraembryonic hematopoietic cell migrationduring ... › content › pnas › 92 › 23 › 10713.full.pdf10713 Thepublication costs ofthis article weredefrayed in part bypage charge

E

Proc. Natl. Acad. Sci. USA 92 (1995) 10715

C D

.. .,.... ..

...

..a t. ... . ..

F Ge

* .

-

/y

n....

... ............ ....

...

*.id-f..":...... ...... 9.

K

4%;'

L - N 0M

p

FIG. 2. Expression of GATA-1 and globin during zebrafish embryogenesis. In situ hybridization was performed with fixed zebrafish embryosat the following stages: (A and B) Two somites. (C and D) Five somites. (E and F) Eighteen somites. (G and H) Twenty-three hours (before prim5 stage). (I) Twenty-four hours (prim 5). (J) Twenty-five hours. (K) Twenty-six hours. (L-P) Twenty-six to 29 h. (B, D, F, and H) Cross-sectionsthrough the posterior region or tail bud (IM) of embryos corresponding to (A, C, E, and G). For each section, e denotes the embryo and y indicatesthe yolk. The arrows in B show GATA-1-expressing progenitors that reside in contact with the yolk syncytial layer. The closed arrow in G indicatesthe posterior boundary of GATA-1 expression in the IM. The open arrow in G points to undifferentiated, GATA-2-expressing progenitors in theposterior IM (see Fig. 3E). The arrows in J and K show hematopoietic progenitors migrating anteriorly and onto the yolk sac. Erythroid cells laterenter the heart (P, arrow). [M, 0, and P are normal sibs from a cross of two spadetail (31) heterozygotes.] (Q and R) o-Dianisidine staining of 24-hwild-type zebrafish embryos. The brown color indicates erythroid cells containing hemoglobin in the IM region.

Developmental Biology: Detrich et al.

jj O,.f

Dow

nloa

ded

by g

uest

on

Aug

ust 4

, 202

0

Page 4: Intraembryonic hematopoietic cell migrationduring ... › content › pnas › 92 › 23 › 10713.full.pdf10713 Thepublication costs ofthis article weredefrayed in part bypage charge

10716 Developmental Biology: Detrich et al.

Y A

D

FIG. 3. Expression of GATA-2 mRNA in zebrafish. (A) Seventy-five percent epiboly. Open arrows represent the border of the embryo,and closed black arrows indicate expression in ventral ectoderm. (Band C) Ninety percent epiboly. GATA-2 is expressed within theadvancing yolk syncytial layer, with higher levels at the embryonicboundary (white arrows). (D) Five somites. A "baseball" pattern ofexpression encompasses the anterior (solid arrow) and posterior (openarrow) regions of the embryo. (E) Twenty somites. Note the expressionthroughout the IM, including the posterior undifferentiated progen-itors (open arrow). The solid arrow indicates the posterior boundaryof GATA-1 expression in the IM.

genic transcription. GATA-2 has been implicated as an endo-thelial transcription factor involved in the regulation of pre-pro-endothelin and the cell adhesion molecule VCAM-1 (14).Zebrafish GATA-2 mRNA is also detected in discrete neuralstructures including the hindbrain-midbrain border and indi-vidual spinal neurons, similar to higher species (17).

DISCUSSIONThe IM-forming teleosts have been regarded as exceptions tothe general rule of yolk sac hematopoiesis in vertebrates. Somefish, such as killifish (Fundulus spp.) (10, 11, 36), angelfish(Pterophylum scalare) (2), and chondrichthyans (1, 3, 37) formblood predominantly on the yolk sac. Fundulus blood islands

Zebraf ish Killifish

EARL r

LATE

are derived from cells that occupy the rim of the posteriorembryo proper, similar to those that give rise to the IM inzebrafish (10, 11, 36). Preliminary in situ analyses in angelfishembryos demonstrate GATA-1 expression in two bilateral cellpopulations on the yolk sac (data not shown). Therefore, basedon GATA-1 expression detected during gastrulation, yolk sacblood islands and the IM are likely to derive from similarlylocalized stem cell populations.The spatial distribution of hematopoietic stem cells in fish

embryos may be established by variations in morphogeneticmovements. In embryos, such as Fundulus, that have extendedepiboly and rapid gastrulation, hematopoietic progenitorsremain at the lateral border of the embryo proper andsubsequently migrate onto the yolk sac to form blood islands.In embryos characterized by rapid epiboly and slower gastru-lation, such as the zebrafish, IM formation predominates,apparently because hematopoietic cells are subjected to con-vergent mesodermal migration. Alternatively, specific induc-ers or chemotactic molecules may regulate the localization ofblood cell populations.Our studies on the zebrafish demonstrate a propensity for

fish hematopoietic progenitors, which may constitute theembryonic or postembryonic programs, to migrate to thedorsal mesentery (DM). This process in teleosts, which we havedirectly visualized using antisense RNAs for the DNA-bindingproteins GATA-1 and GATA-2 as cell markers, is remarkablysimilar to proposed hematopoietic cell migration pathways inother vertebrates. Reciprocal transplantations between cyto-genetically marked amphibians (21) have demonstrated ante-rior migration of hematopoietic progenitors from a posteriorregion surrounding the pronephric duct to the dorsal mesen-teric region including the ducts of Cuvier, dorsal aorta, andpronephros. Intraembryonic hematopoiesis (including theDMregion) has also been documented in avian development usingchicken-quail chimeras (24) and in mammalian developmentusing early progenitor assays (22, 23).

Migration may have a prominent role in the induction of thehematopoietic program by establishing the relative distribu-tion of extra- and intraembryonic hematopoietic stem cells.Moore and Metcalf (38) demonstrated that removal of themurine yolk sac prior to appreciable blood island formation

Rana Chicken Mouse

V. X \>YX2

Vt DM

E

FIG. 4. Unifying hypothesis for hematopoietic cell migration during vertebrate embryogenesis. (Upper) Early stage. Hematopoietic stem cellsoccupy a ventral or lateral position of the posterior embryo (E) proper and either migrate to the yolk (Y) sac or to a dorsal, intraembryonic position.In the zebrafish, ventral mesoderm migrates medially to a position adjacent to the pronephric duct, but in the killifish, it migrates laterally ontothe yolk sac. In amphibians (i.e., Rana), the embryonic ventral blood islands are v-shaped, and dorsal progenitors are located adjacent to thepronephros. In birds, the lateral progenitors are initially detected in the area opaca; in mammals, they reside at the boundary of the intra- andextraembryonic tissues during late gastrula stages. (Lower) Late stage. Yolk sac hematopoietic cells enter circulation, whereas dorsal hematopoieticcells migrate into a dorsal mesenteric (DM) region. From the latter position, the dorsal cells either migrate to the yolk sac, invade the adjacentvasculature to enter the circulation to the heart (H), remain in situ, or migrate to the thymus (T) to form T cells. Sites such as the human fetalliver are colonized by circulating stem cells.

Proc. Natl. Acad. Sci. USA 92 (1995)

Dow

nloa

ded

by g

uest

on

Aug

ust 4

, 202

0

Page 5: Intraembryonic hematopoietic cell migrationduring ... › content › pnas › 92 › 23 › 10713.full.pdf10713 Thepublication costs ofthis article weredefrayed in part bypage charge

Proc. Natl. Acad. Sci. USA 92 (1995) 10717

(day 7) led to a marked deficiency of fetal liver hematopoiesis,suggesting an early medial migration of yolk sac hematopoieticcells to the embryo. It appears that the dorsal hematopoieticpopulation mostly contributes to fetal and adult hematopoiesisin higher vertebrates, but in lower vertebrates, embryonichematopoietic progenitors are also found in the DM. Forinstance, in Rana catesbeiana, it is likely that the larval/adultprogenitors of the DM colonize the larval liver, while theembryonic progenitors of the DM colonize the larval prone-phros (39). Such migration, which appears independent ofactive circulatory flow, may be mediated by a selective, direc-tional gradient of cell adhesion with the developing vasculatureand pronephric duct or by long- or short-range chemotaxis. Inaddition, the migration of extraembryonic mesoderm to formhematopoietic cells in the dorsal mesentery is reminiscent ofthe migration of extraembryonic germ cell progenitors to anintraembryonic compartment (40-42).

In 1963, Colle-Vandevelde concluded that "a better knowl-edge of the origin of those blood cells might enable us toestablish a connection between the classical 'dorsal' origin ofblood in the fish and its 'ventral' origin in other vertebrates"(4). Clearly, each vertebrate species has unique features ofembryonic hematopoiesis, and it is difficult to relate theextraembryonic membranes of amniotes to the yolk sac tissuesof fish. Nevertheless, a unifying hypothesis (Fig. 4), based onprevious studies and our studies of the expression of GATA-1and GATA-2 in fish, postulates that homologous migrationpathways play a central role in establishing the vertebratehematopoietic program. Two primordial pathways are exem-plified by migration to the yolk sac in killifish and by intraem-bryonic migration in the zebrafish. Blood is derived fromventral or lateral mesoderm of the posterior embryo. Thismesoderm either travels further laterally and colonizes theyolk sac as blood islands (19, 43) or migrates medially. In-traembryonic cells then move anteriorly into the DM region.Some DM hematopoietic cells populate the yolk sac, someenter the vasculature into circulation, and others move to thethymus and form T cells. From the yolk sac, all hematopoieticcells travel to the ducts of Cuvier to enter the embryoniccirculation. Thus, we postulate that the yolk sac receives twowaves of hematopoietic stem cells during development: cellsfrom lateral extraembryonic mesoderm that form embryonicblood islands and some later larval/adult colonies (44) andcells from dorsal intraembryonic mesoderm that form pre-dominantly larval/adult hematopoietic cells (24). In summary,our studies indicate that the hematopoietic program, includinghomologous transcription factors and migration pathways, isconserved in vertebrate evolution.

The zebrafish cDNA libraries were provided by K. Zinn (CaliforniaInstitute of Technology) and R. Riggleman (University of Utah Schoolof Medicine). We thank Y. Kunz, A. Colle-Vandevelde, W. Ballard, J.Trinkaus, J. Turpen, D. Stainier, and F. Dieterlen-Lievre for theirhelpful discussions. We thank S. Lux, S. Orkin, M. Thompson, L.Schneider, and C. Kimmel for critically reviewing the manuscript.L.I.Z. is an Assistant Investigator of the Howard Hughes MedicalInstitute. H.W.D. was supported by a U.S. Public Health ServiceNational Research Service Award Senior Fellowship (HL09061). Thiswork was supported by grants from the National Institutes of Health(HL48801; L.I.Z.), National Science' Foundation (OPP-9120311;H.W.D.), and Medical Research Council of Canada (M.W.K.).

1. Oellacher, J. (1872) Z. Wiss. Zool. 23, 373-421.2. Al-Adhami, M. A. & Kunz, Y. W. (1976) Wilhelm RouxArch. 179,

393-401.

3. Al-Adhami, M. A. & Kunz, Y. W. (1977) Dev. Growth Differ. 19,171-179.

4. Colle-Vandevelde, A. (1963) Nature (London) 198, 1223.5. Iuchi, I. (1973) J. Exp. Zool. 184, 383-396.6. Reib, J. (1973) Ann. Embryol. Morphog. 6, 43-54.7. Zapata, A. (1983) Bull. Inst. Pasteur (Paris) 81, 165-186.8. John, C. C. (1931) Clupea 110, 112-119.9. Ruckert, C. B. & Mollier, S. (1906) Die erste Entstehung der

Gefasse und de Blutes bei Wirbeltieren, Handbuch der Ver-gleichenden und Experimentellen Entwickelungslehre der Wir-beltiere (Hertwig, Germany), Vol. 1, pp. 1125-1153.

10. Stockard, C. R. (1915) Am. J. Anat. 18, 227-327.11. Stockard, C. (1915) Am. J. Anat. 18, 525-594.12. Wenckebach, K. F. (1885) J. Anat. Physiol. 19, 231-236.13. Ziegler, H. E. (1887) Arch. Mikrosk. Anat. Entwicklungsmech. 30,

596-665.14. Orkin, S. H. (1992) Blood 80, 575-581.15. Tsai, F.-Y., Keller, G., Kuo, F. C., Weiss, M., Chen, J., Rosen-

blatt, M., Alt, F. W. & Orkin, S. H. (1994) Nature (London) 371,221-226.

16. Pevny, L., Simon, M. C., Robertson, E., Klein, W. H., Tsai, S. F.,D'Agati, V., Orkin, S. H. & Costantini, F. (1991) Nature (Lon-don) 349, 257-260.

17. Kelley, C., Yee, K., Harland, R. & Zon, L. (1994) Dev. Biol. 165,193-205.

18. Iuchi, I. (1985) Zool. Sci. 2, 11-23.19. Romanoff, A. L. (1960) in The Avian Embryo, ed. Romanoff,

A. L. (Macmillan, New York), pp. 569-678.20. Ingram, V. M. (1972) Nature (London) 235, 338-339.21. Turpen, J. B. & Knudson, C. M. (1982) Dev. Biol. 89, 138-151.22. Medvinsky, A. L., Samoylina, N. L., Muller, A. M. & Dzierzak,

E. A. (1993) Nature (London) 364, 64-67.23. Godin, I. E., Garcia-Porrere, J. A., Coutinho, A., Dieterlen-

Lievre, F. & Marcos, M. A. R. (1993) Nature (London) 364,67-70.

24. Dieterlen, L. F. & Martin, C. (1981) Dev. Biol. 88, 180-191.25. Westerfield, M. (1989) The Zebrafish Book: A Guide for the

Laboratory Use of Zebrafish (Brachydanio rerio) (Univ. OregonPress, Eugene, OR).

26. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) MolecularCloning: A Laboratory Manual (Cold Spring Harbor Lab. Press,Plainview, NY), 2nd Ed.

27. Sanger, F. & Coulson, A. R. (1975) J. Mol. Biol. 94, 441-448.28. Iuchi, I. & Yamamoto, M. (1983) J. Exp. Zool. 226, 409-417.29. Harland, R. M. (1991) Methods Cell Biol. 36, 675-685.30. Zon, L. I., Mather, C., Burgess, S., Bolce, M., Harland, R. &

Orkin, S. H. (1991) Proc. Natl. Acad. Sci. USA 88, 10642-10646.31. Kimmel, C. B., Kane, D. A., Walker, C., Warga, R. M. & Roth-

man, M. B. (1989) Nature (London) 337, 358-362.32. Orkin, S. H. (1990) Cell 63, 665-672.33. Nelsen, 0. E. (1953) in Comparative Embryology of the Verte-

brates, ed. Nelsen, 0. E. (Blakiston, New York), pp. 436-441.34. Fioroni, P. (1987) in Embryologie, ed. Fioroni, P. (Springer,

Berlin), Vol. 183, pp. 230-231.35. Trinkaus, J. P. (1993) J. Exp. Zool. 265, 258-284.36. Stockard, C. R. (1915) Anat. Rec. 9, 124-127.37. Swaen, A. & Brachet, A. (1901) Arch. Biol. 18, 73-190.38. Moore, M. A. & Metcalf, D. (1970) Br. J. Haematol. 18,279-296.39. Broyles, R. H. (1981) in Changes in the Blood DuringAmphibian

Metamorphosis, eds. Gilbert, L. I. & Frieden, E. (Plenum, NewYork), pp. 461-490.

40. Rosenquist, G. C. (1966) Contrib. Embryol. Carnegie Inst. 38,71-110.

41. Gardner, R. L., Lyon, M. F., Evans, E. P. & Burtenshaw, M. D.(1985) J. -Embryol. Exp. Morphol. 88, 349-363.

42. Bianchi, D. W., Wilkins-Haug, L. E., Enders, A. C. & Hay, E. D.(1993) Am. J. Med. Genet. 46, 542-550.

43. Tavassoli, M. (1991) Blood Cells 1, 269-281.44. Toles, J. F., Chui, D. H. K., Belbeck, L. W., Starr, E. & Barker,

J. E. (1989) Proc. Natl. Acad. Sci. USA 86, 7456-7459.

Developmental Biology: Detrich et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 4

, 202

0