Development 103. 379-390 (1988)Printed in Great Britain © The Company of Biologists Limited 1988

379

The allocation of cells in the presomitic mesoderm during somite

segmentation in the mouse embryo

PATRICK P. L. TAM

Department of Anatomy. Faculty of Medicine. The Chinese University of Hong Kong. Shattn. NT. Hong Kong

Summary

Orthotopic grafts of wheat germ agglutinin-colloidalgold conjugate (WGA-gold) labelled cells were used todemonstrate differences in the segmental fate of cellsin the presomitic mesoderm of the early-somite-stagemouse embryos developing in vitro. Labelled cells inthe anterior region of the presomitic mesoderm col-onized the first three somites formed after grafting,while those grafted to the middle region of this tissuewere found mostly in the 4th-7th newly formedsomites. Labelled cells grafted to the posterior regionwere incorporated into somites whose somitomereswere not yet present in the presomitic mesoderm at thetime of grafting. There was therefore an apparentposterior displacement of the grafted cells in the

presomitic mesoderm. Colonization of somites byWGA-gold labelled cells was usually limited to two tothree consecutive somites in the chimaera. The distri-bution of cells derived from a single graft to twosomites was most likely due to the segregation of thelabelled population when cells were allocated to adjac-ent meristic units during somite formation. Furtherspreading of the labelled cells to several somites insome cases was probably the result of a more extensivemixing of mesodermal cells among the somitomeresprior to somite segmentation.

Key words: presomitic mesoderm, somitomere. cellmovement, microsurgical grafting, mouse embryo.

Introduction

Early differentiation of the paraxial mesoderm in themouse embryo is characterized by the sequentialformation of epithelial somites at the cranial end ofthe presomitic mesoderm (Flint et al. 1978; Tarn,1981). Morphological studies using scanning electronmicroscopy have revealed the presence of a meristicpattern of somitomeres in this overtly unsegmentedpresomitic mesoderm (Tarn et al. 1982). Experimen-tal studies on the development of this tissue havedemonstrated that somites are formed at the expenseof somitomeres and the number of prospectivesomites contained in the presomitic mesoderm corre-lates closely with the number of pre-existing somito-meres (Tarn, 1986). The recruitment of cells into theparaxial mesoderm has been studied using orthotopicgrafts of [3H]thymidine-labelled primitive streak cellsand specific labelling of ectodermal cells with wheatgerm agglutinin conjugated to colloidal gold (WGA-gold). The primitive streak of the gastrulating embryoand the early-somite-stage embryo was shown to beresponsible for the continuous addition of cells to the

paraxial mesoderm in a craniocaudal sequence (Tam& Beddington, 1987). However, it was found in suchstudies that the appearance of primitive-streak-derived cells did not begin with the newly establishedsomitomeres but labelled cells seemed to have col-onized those somitomeres that were already installedin the presomitic mesoderm. Such a pattern ofdeployment of primitive-streak-derived cells to theparaxial mesoderm in the experimental embryosseems to be incompatible with the presence of astable prepattern of selfcontained meristjc units in thepresomitic mesoderm. In certain cases where labelledcells were only confined to the sclerotomal popu-lation, it could arguably be the result of cell mixingoccurring during the dispersion and resegmentationof sclerotome of the developing somite (Flint, 1977;Stern & Keynes, 1986). However, the presence oflabelled cells in the epithelial component of thesomites would indicate that some cell mixing musthave occurred in the somitomeres because it would beunlikely that donor cells could invade the dermamyo-tome after somite segmentation has taken place (Tam& Beddington, 1987). The present study was carried

380 P. P. L. Tarn

out to investigate the extent of cell mixing within thepresomitic mesoderm by following the temporal andspatial pattern of allocation of graft-derived cells tothe newly formed somites. The segmental fate of cellsof specific somitomeres were also examined usingorthotopic graftings of labelled cells to different sitesin the presomitic mesoderm of early-somite-stageembryos.

Materials and methods

Recovery, culture and examination of embryosEarly-somite-stage mouse embryos were obtained from aclosed bred colony of ICR strain mice. At 8-5 days p.c.,embryos were dissected from the uterus in PB1 mediumcontaining 10% fetal calf serum (FCS. GIBCO). Theparietal yolk sac was removed and the embryos werewashed in several changes of fresh PB1 + FCS medium.After manipulation or labelling, embryos were cultured inrotating (30revmin~') 50ml glass bottles (Wheaton) con-taining 3-4 ml of culture medium made up of equal parts ofDulbecco's modified Eagle's medium (DMEM, GIBCO)and rat serum (Tarn & Beddington, 1987). For culturing of[3H]thymidine-labelled embryos, cold thymidine(8x 10~b M) was added. Five or six embryos were cultured ineach bottle and cultures were maintained under an atmos-phere of 5 % COS in air at 37°C. After culture, embryoswere examined under the dissecting microscope for thepresence of heart beat and vitelline circulation beforetransfer to phosphate-buffered saline (PBS). Various devel-opmental features, such as the closure of the cephalicneural tube, the formation of optic and otic primordia, theappearance of forelimb buds, the degree of axial rotationand the number of somites were noted. Grossly abnormaland retarded embryos were discarded. The embryos werethen assayed for the protein content (Lowry el al. 1951) orexamined by light or electron microscopy (Chan & Tarn,1988). To visualize the colloidal gold marker, the sectionswere treated with a silver developer (Tam & Beddington,1987). The presence of [3H]thymidine-labelled cells wasstudied by autoradiography. The preparation of the auto-radiographs and the criteria used for identifying and count-ing colonizing donor cells labelled with the two differentmarkers were the same as those described by Beddington(1981) and Tam & Beddington (1987).

Preparation of labelled grafts(1) Wheat germ agglutinin (WGA)-gold

conjugate labellingFollowing the removal of the parietal yolk sac, the posteriorportion of the 4- to 7-somite-stage embryo was isolated bycutting with a pair of polished alloy needles made fromorthodontic wire. The endodermal layer was then peeledoff from the ventral surface of the posterior fragment toexpose the somites and the presomitic mesoderm. Byscraping with a finely drawn glass needle, clumps ofmesenchymal cells were obtained from the anterior, middleand posterior regions of the presomitic mesoderm. Thesecell clumps were then transferred to a 50u\ drop of PB1

medium to which about 50 nl of a concentrated WGA-goldsuspension was added. The WGA-gold conjugate wasprepared as described by Tam & Beddington (1987).Labelling of the tissue was carried out under room atmos-phere at 35-37°C for 60min. After labelling, the cellclumps were washed in four changes of fresh PB1 medium.The labelled tissue was further subdivided with glassneedles into clumps of a suitable size for injection, eachconsisting of about 30 cells.

(2) [3H]thymidine labellingIntact 4- to 7-somite-stage embryos obtained from thedecidua were transferred directly to a labelling mediummade up of the DMEM-rat serum medium supplementedwith 4xlO~7M-thymidine and 0-74MBqml~' of [3H]thymi-dine (specific activity l-55TBqmM~': Radiochemicals,Amersham). The embryos were labelled for 2h at 37°C inrotating bottles gassed with 5 % CON in air. After labelling,embryos were washed in three changes of PB1 mediumcontaining 8xl0~bM-cold thymidine. Some embryos werefixed immediately to serve as "uptake control'. In eachexperiment, labelled embryos ("labelled control) werecultured for the same duration and in the same culturebottle with the experimental embryos. The remainingembryos were dissected to obtain tissue fragments forgrafting as previously described for WGA-gold labellingstudies.

Grafting experimentsThe general approach of the experiments was to makesynchronous grafts of WGA-gold-labelled mesenchymalcells orthotopically into different regions of the presomiticmesoderm of early-somite-stage embryos (mean somitenumber: 6-1 ± 0-2, n — 65). The initial incorporation of thelabelled cells was studied by examining the embryo atvarious time intervals after grafting. The final distributionof labelled cells in the paraxial mesoderm and othermesodermal tissues of the embryo after 20-21 h of furtherdevelopment in vitro was studied microscopically followingsilver enhancement or autoradiography.

The manipulation of the early-somite-stage embryos andthe grafting of labelled tissues into the presomitic meso-derm was the same as previously described by Tam &Beddington (1987). The presomitic mesoderm was dividedroughly by visual inspection into three equal-size portionsusing the posterior border of the last-formed somite and thetip of the embryonic axis as the reference points. Grafts oflabelled cells were places in the anterior (ANT-PSM),middle (MID-PSM) and posterior (POST-PSM) regions ofthe unsegmented presomitic mesoderm. Some embryosreceiving grafts of WGA-gold-labelled cells were immedi-ately dissected to isolate the posterior embryonic fragmentcontaining the presomitic mesoderm. Camera-lucida draw-ings of the embryonic fragments were made and thepositions of the graft in the presomitic mesoderm weremarked on the drawings. The length of the presomiticmesoderm and the distance of the graft from the posteriorborder of the last-formed somite were measured using acomputer-driven sonic digitizer (Grafbar. Science Access-ories Corporation).

The following grafting experiments were performed.

Somite formation in the mouse 381

(1) WGA-gold-labelled cells were grafted orthotopi-cally into the anterior and middle regions of the presomiticmesoderm of 4- to 7-somite-stage embryos. Graftings werealways made to the left side of the embryo. Embryos wereharvested either immediately after grafting or after furtherdevelopment in vitro for 3, 6 or 9-10h. The embryos werefixed in Carnoy fluid for light microscopy. Specimens usedfor transmission electron microscopy were postfixed with1 % osmium tetroxide and embedded in Spurr plastic.Uranyl acetate and lead citrate stained ultrathin sectionswere examined with a JEOL CXII-100 microscope at 80 kV.Specimens for scanning electron microscopy were furtherdissected to remove the surface ectoderm and the neuralplate in order to expose the underlying paraxial mesoderm.The dissected specimens were then postfixed with 1 %osmium tetroxide and critical-point dried with Freon as theexchange fluid. Gold-palladium-coated specimens wereexamined with a JEOL JSM-35CF microscope operated at15 kV.

(2) WGA-gold-labelled cells were grafted orthotopi-cally to the ANT-PSM, MID-PSM and POST-PSM sitesof the left presomitic mesoderm of 4- to 7-somite-stageembryos and cultured for 20-21 h in vitro. After culture,the embryos were examined and processed for light mi-croscopy. Serial sections were cut at 5^m, enhanced bysilver reagent and examined for the presence of labelledcells in the paraxial mesoderm and adjacent mesodermaltissues. Specimens with incomplete series of sections wereexcluded from this study.

(3) The pattern of distribution of grafted cells betweenthe two sides of the embryo was compared by following thedistribution of WGA-gold-labelled cells which weregrafted orthotopically to the MID-PSM site of both preso-mitic mesoderm of the embryo. After culturing for 20-21 hin vitro, the embryos were processed for light microscopyand the segmental distribution of labelled cells in thesomites was examined.

(4) In order to compare the pattern of tissue colonizationby cells labelled with different markers, orthotopic graft-ings of WGA-gold-labelled and [3H]thymidine-labelledcells were made to the MID-PSM region of the presomiticmesoderm. Tissue fragments were obtained from donor

embryos labelled with either WGA-gold or radioactivethymidine and were disaggregated into clumps of about 20cells. During grafting, two differently labelled fragmentswere injected simultaneously into the presomitic mesodermof the experimental embryo. After culturing for 20-21 h,the embryos were fixed in Carnoy fluid and processed forhistology. Serial sections were cut at 5/.im and mounted onchromic-acid-cleaned glass slides. The sections were firsttreated by silver deposition to reveal the location of theWGA-gold-labelled cells and after a preliminary micro-scopic examination the sections were processed for auto-radiography by coating with Kodak nuclear track emulsion.After 3 weeks of exposure, the autoradiographs weredeveloped with Kodak D19 developer, fixed in Kodak rapidfixer and counterstained with haematoxylin and eosin.

Results

Orthotopic graftings were made in three differentregions of the presomitic mesoderm which corre-sponded to somitomeres I/II, III/IV and V/VI, asindicated by the distance from the posterior border ofthe last-formed somite in the paraxial mesoderm(Table 1). The position of the graft in the presomiticmesoderm could be easily ascertained by virtue of thered coloration of the WGA-gold conjugate labelledcells. Figs 1 & 2 show the location of the grafted cellsin the presomitic mesoderm of the early-somite-stageembryo. Among the 18 experimental embryos fixedimmediately after grafting and examined histologi-cally, the graft was found to have been placed wellwithin the presomitic mesoderm in 16 cases but waslocated in the visceral endoderm and the lateralmesoderm, respectively, in two cases. When theoperated presomitic mesoderm (11 specimens) wasexamined using stereoscopic technique under thescanning electron microscope, five to six somitomereswere observed (Fig. 3). The morphology and sizes ofthe somitomeres were similar to those seen in four

Table 1. The position of grafts in the presomitic mesoderm of early-somite-stage mouse embryos

Group

ANT-PSM (n = 22)MID-PSM (n = 22)POST-PSM (n = 19)

Total PSMlength (fim)

mean ± S.E.M.

767 ±28699 ±20664±25

Position of graft from

Relative distance*mean ± S.E.M.

0-13 ±0-010-37 ±0020-68 ±002

last formed somite

Calculated distance!mean ± S.E.M. (pm)

93 ±7242 ±12484 ±12

Somitomeres:):

I, IIIII, IVV, VI

Note: ANT-PSM, anterior; MID-PSM, middle; POST-PSM, posterior regions of the presomitic mesoderm.* Relative distance = distance of graft from last formed somite divided by total PSM length.t Calculated distance = relative length x mean total PSM length of the group.t Somitomeric units determined according to the map described by Tam (1986).Statistical analyses: Significant between-group differences in the total PSM length (P<001) and the position of the graft (P<0014)

by ANOVA.Total length: [ANT-PSM] > [MID-PSM] and [POST-PSM] by Duncan's multiple range comparison at /><005.Relative distance and calculated distance: [ANT-PSM] < [MID-PSM] < [POST-PSM] by Duncan's multiple range comparison at

P<005.

382 P. P. L. Tarn

^ '

Fig. 1. An early-somite-stage embryo with an orthotopic graft of WGA-gold-labelled cells (arrowhead) in the anteriorregion of the presomitic mesoderm. Silver enhanced and fast green counterstained. Bar, 100\.im.Fig. 2 A clump of WGA-gold-labelled cells in the presomitic mesoderm of an early-somite-stage embryo fixedimmediately after grafting. Silver enhanced and fast green counterstained. ne, neural plate ectoderm; en, visceralendoderm. The arrow points anteriorly. Bar, 100 fim.Fig. 3 An overview of the presomitic mesoderm of an experimental embryo 3 h after grafting. The neural plateectoderm was removed and the five somitomeres were identified with stereoscopic techniques using pairs of scanningelectron micrographs with a tilting of 10°. The intersomitomeric boundaries were marked by (>). 5, somite. Bar. 100/<m.

specimens of intact presomitic mesoderm. Whenexamined with the transmission electron microscope,the graft tissue was found to be heavily labelled withgold particles. Most of the gold particles adhered tothe cell membrane and lined the invaginating pits andgutters, but some gold particles were already in thecytoplasm of the mesenchymal cells within 1 h oflabelling (Fig. 4). Incorporation of gold particles intocytoplasmic vacuoles was more evident by 6h aftergrafting (Fig. 5). However, a considerable amount ofgold particles was still found on the cell surface(Fig. 5). At 9-10 h after grafting, there was only asmall quantity of gold particles on the cell membraneand most of the intracellular gold particles werepacked into cytoplasmic vacuoles (Fig. 6A) or werefound as aggregates in the cytoplasm (Fig. 6B).

Experimental embryos receiving orthotopic graftsdeveloped normally in culture and were morphologi-cally similar to intact embryos developing undersimilar culture conditions. The cultured embryosformed fewer somites (average 18 pairs) than em-bryos at an equivalent stage in vivo (average 20 pairs,Table 2). Microscopic examination revealed thatsomite formation was not adversely affected in theexperimental embryos and similar numbers of somitewere formed in these embryo as that of intactembryos cultured in vitro (Table 2). A comparison ofthe number of somites present on the operated side

(left side) and the unoperated side (right side) of theexperimental embryos also revealed no significantdifferences (paired Mest, f>0-05, n = 50) betweenthe two paraxial mesoderms. The morphology of thesomite in the chimaeric embryos was essentially thesame as that in the control embryos. Among the 76specimens examined, abnormally large somites wereseen in the operated paraxial mesoderm of sevenchimaeras. Despite the presence of the enlargedsomite, equal numbers of somites in correct segmen-tal register were formed in the operated and theopposite unoperated paraxial mesoderm of the exper-imental embryo, suggesting that there is not a disturb-ance of somite number by the enlarged somites.

Histological examination of the presomitic meso-derm of the experimental embryo showed that,shortly after grafting, the donor cells dispersed andwere then incorporated into the segmenting somites(Fig. 7). The allocation of labelled cells was notalways limited to one somite. Donor cells were oftendistributed in apparently equal numbers to adjacentsomites (Fig. 8) but sometimes one somite was moreheavily colonized than its neighbour (Fig. 9).Labelled cells were usually found in both the derma-myotome and sclerotome of the somite (Fig. 10).Colonization of the more lateral mesoderm (inter-mediate and lateral plate mesoderm) was also ob-served (Fig. 11, Table 3) and the labelled cells in the

Somite formation in the mouse 383

lateral mesoderm were usually localized at the samesegmental level as those graft-derived cells in theparaxial mesoderm. In the paraxial mesoderm ofANT-PSM and MID-PSM embryos, all the labelledcells were confined to the segmented somites but in

the POST-PSM, some labelled cells still remained inthe presomitic mesoderm after 20-21 h of develop-ment (Fig. 13, Table 3).

The segmental fate of cells in different regions ofthe presomitic mesoderm was studied by making

Fig. 4. (A) A clump of presomiticmesodermal cells labelled for 1 h byimmersion in the WGA-goldpreparation. A smaller fragment wouldbe obtained from this clump for theactual grafting. Toluidine blue stained.Bar. lOfim. (B) WGA-gold-labelledcells in the presomitic mesoderm of anearly-somite-stage embryo fixedimmediately after grafting. Theundulating cell surface is heavilydecorated with gold particles and somegold particles were locatedintracellularly. Bar, 0-5 fim.Fig. 5. WGA-gold-labelled cells in thepresomitic mesoderm of an early-somite-stage embryo examined 3 h aftergrafting. Many gold particles arecontained in large vacuoles (va) ordispersed as clumps in the cytoplasm(arrow). Some gold particles are stilladhering to the cell surface(arrowheads). Bar, 2jum.Fig. 6. (A) WGA-gold-labelled cell inthe paraxial mesoderm of an early-somite-stage embryo examined 6h aftergrafting. Clumps of gold particles (Au)are found in the cytoplasm of themesodermal cells, which are joinedtogether by intracellular junctions(arrowheads). Bar, 2/im. (B) Amagnified view of the intracellularaggregates of gold particles. Bar,0-1/im.

384 P. P. L. Tarn

Table 2. The development of mouse embryos in vitro compared to equivalent stages in vivo

No. of embryos showing development of

Group

Otic andNo. of optic

embryos primordiaClosed brain

tubeCompletebody turn

Circulationand heart

beatForehmb

budSomite no.

mean ± S.E.M.

Protein contentmean ± S.E.M.

(Pg)

8-5-day embryos cultured for 20-21 h in vitroOperated embryos 34 34 (100 %)Intact embryos 51 50 (98 %)

9-5-day embryos in vivo50 49 (96 %)

23 (68 %)38 (74 %)

48 (96 %)

17(50%)35 (71 %)

45 (90 %)

33 (97%)50 (98%)

25 (68 %)28 (55 %)

49(98%) 43(86%)

18-0 ±0-618-3 ±0-5

20-2 ±0-4

276 ±8321 ± 32

351 ± 24

Statistical analyses:(1) Extent of development. No significant differences among the three groups by Kendall tau test for ranked categones.(2) Somite number. No significant differences between [Operated embryos] and [Intact embryos]. [In vitro embryos] < [In vivo embryos] by ANOVA

and Duncan's multiple range comparison (P<0-05).(3) Protein content. No significant differences among the three groups by ANOVA and Duncan's multiple range comparison of log-transformed

values.(4) Protein content (log values) correlates with somite number of embryos (r = 0-676, P < 0001, n = 135) by 2-tailed correlation analysis.

orthotopic graftings of labelled cells followed byculturing the experimental embryos in vitro. Fig. 13summarizes the distribution of WGA-gold-labelledcells in the first 9-10 h after orthotopic grafting andFig. 14 shows the distribution of labelled cells graftedto three different sites in the presomitic mesoderm tothe somites after 20-21 h of in vitro development.There were slight variations in the pattern of coloniz-ation by labelled cells between embryos but this wasnot unexpected given the small size and the lack ofdistinctive positional markers within the presomiticmesoderm. At 3h after grafting labelled cells toANT-PSM, WGA-gold-positive cells were found inthe most recently segmented somite and the cranialend of the presomitic mesoderm. After 6h, labelledcells were present in the first two somites formed aftergrafting but were no longer found in the presomiticmesoderm. When the embryos were examined after9-10 h and 20-21 h of in vitro development, labelledcells were consistently found in the first three somitesformed after grafting (Figs 13, 14). When labelledcells were grafted to the 3rd and the 4th somitomeres(MID-PSM), WGA-gold-positive cells remained in-itially within the presomitic mesoderm at 3 and 6hafter grafting. After 9-10 h and 20-21 h of develop-ment, labelled cells which were previously found inthe presomitic mesoderm were incorporated into the3rd-8th newly formed somites (Figs 13, 14). Thegrafting of labelled cells to POST-PSM resulted inthe colonization of more posterior somites, i.e.6th-9th somites formed after grafting (Fig. 14). Itmust be noted, however, that Figs 13, 14 are showingthe segmental distribution of labelled cells for groupsof embryos receiving grafts at three specific sites ofthe presomitic mesoderm and the actual spreading oflabelled cells in any one chimaeric embryo was less

extensive than what is depicted in the diagrams. MostANT-PSM embryos had one or two somites whichwere colonized by labelled cells but up to four somiteswere colonized by donor cells in the MID-PSM andPOST-PSM chimaeras (Table 4). There is, there-fore, an apparently wider spreading of the graft-derived cells in the middle and posterior regions ofthe presomitic mesoderm.

Of 25 embryos that received grafts at theMID-PSM site on both sides of the body, bilateralcolonization of the paraxial mesoderm was observedin 14 chimaeras. The pattern of distribution oflabelled cells in the paraxial mesoderm of theseembryos tended to be symmetrical. Statistical analy-sis using the test for concordance (Fleiss, 1981)revealed a significant tendency for somites located atthe same segmental level on both sides of the embry-onic axis to be colonized by the graft-derived cells(2 = 6-14, P<0-001).

When a mixture of WGA-gold-labelled cells and[3H]thymidine-labelled cells were grafted to the pre-somitic mesoderm, the two differently marked popu-lations were distributed to similar sets of somites inthe resultant chimaera. There were only five excep-tions among 34 cases where the somite was colonizedby only one type of labelled cells.

Discussion

Summary of resultsThe most significant finding of this study was thedemonstration that cells in the three regions of thepresomitic mesoderm differed in their segmental fate.Cells labelled with the WGA-gold marker andgrafted to the first two somitomeres (I, II) in the

Somite formation in the mouse 385

Fig. 7. WGA-gold-labelled cells (arrowhead) in the anterior regions of the presomitic mesoderm in an ANT-PSMexperimental embryo 6h after grafting showing a slight dispersion of the labelled population which seems to be splitbetween two potential somites. Silver enhanced and fast green counterstained. Arrow points cranially. Bar, 50^m.Fig. 8. Two adjacent somites of an ANT-PSM chimaeric embryo cultured for 20-21 h after grafting containingapproximately equal number of WGA-gold-labelled cells. The labelled population tends to localize on both sides of theintersomitic fissure (arrows). Silver enhanced and fast green counterstained. Bar, 25f/m.Fig. 9. Two adjacent somites of an ANT-PSM chimaeric embryo cultured for 20-21 h after grafting showing unequalpartitioning of the WGA-gold-labelled cell population, se. surface ectoderm; en, endothehum. Silver enhanced and fastgreen counterstained. Bar, 25nm.Fig. 10. Somites in a MID-PSM chimaeric embryo cultured for 20-21 h after grafting showing the colonization of thedermamyotome (dm) and the sclerotome (sc) by WGA-gold-labelled cells (arrowheads). Silver enhanced and fast greencounterstained. Bar, 25/im.

presomitic mesoderm colonized the first three newlysegmented somites. Cells placed in the middle regionof the presomitic mesoderm which corresponded tothe third and fourth somitomeres (III, IV) weredistributed mostly to the 4th-7th somites formedafter grafting, but labelled cells were not found insomites cranial to the 3rd newly formed somite nor inthe presomitic mesoderm of the chimaeric embryoafter 20-21 h of in vitro development. Labelled cellsgrafted to the posterior region of the presomiticmesoderm in the vicinity of the primitive streak (i.e.somitomeres V and VI) were subsequently found inthe 6th-9th newly formed somites as well as in theunsegmented presomitic mesoderm. Colonization ofthe paraxial mesoderm by labelled cells apparently

began with the somite corresponding to the somito-mere that initially received the graft and, further-more, there was a tendency for the labelled cells tospread to somites located more posteriorly to theinitial site of grafting. These observations seem there-fore to be compatible with the hypothesis of a generalposterior movement of mesenchymal cells in theposterior regions of the presomitic mesoderm. Theanalysis of the degree of spreading of labelled cellswithin individual chimaeric embryo revealed thatsignificantly more somites were colonized by labelledcells placed in the posterior region of the presomiticmesoderm than those grafted to the anterior region,suggesting that there might be more extensive cellmixing in the nascent somitomeres.

386 P. P. L. Tarn

11Fig. 11. A parasagittal section of an ANT-PSM chimaeric embryo cultured for 20-21 h after grafting showing thecolonization of the lateral plate mesoderm by WGA-gold-labelled cells (arrowheads), dm, dermamyotome of thesomite. Silver enhanced and fast green counterstained. Bar, 50;im.Fig. 12. The presence of WGA-gold-labelled cells (arrowhead) in the presomitic mesoderm of a POST-PSM chimaericembryo cultured for 20-21 h after grafting showing that these cells have not yet been incorporated into the somites.nt, neural tube. Silver enhanced and fast green counterstained. Arrow points cranially. Bar, 50^m.

Table 3. The distribution of WGA—gold-labelled cells in embryonic tissues following orthotopic grafting andcultured for 20-21 h

Positionof graft

ANT-PSM

MID-PSM

POST-PSM

No. of embryos

grafted labelled

43 41

No. of embryos with

ia is

No of embryos with

i§ a

No. of embryos with

No ofcases

127

•i

4• 1

3

labelled cells in:

98J

labelled cells in:

44iIIi,

T

labelled cells in:

Somites

**

38(93 %)

+

18(95 %)

-1-

13(81 %)

Distribution of labelled cells in:

Lateral Presomiticmesoderm mesoderm

•*•

H.

23(56%)

+

9(47 %)

:

7(44%)

-

-

0

_

0

I

1(44%)

Neuraltube

li

lt

-

7(17%)

+

1(5%)

1 +

1 1

1 f

1 1

1(6%)

Bloodvessels

I

7(17%)

-

ft

4(25%)

Note: ANT. anterior: MID. middle: POST, posterior regions of the presomitic mesoderm (PSM) +. presence and - . absence oflabelled cells in the tissue.

Lateral mesoderm = intermediate mesoderm and lateral plate mesoderm.

Somite formation in the mouse 387

ANT-PSM MID-PSM

3h

6h

9-10h

11

jjuunnc j

II : : 1

I •y II II : i

..tilXIDDDDDL

1

Fig. 13. The segmental location of the WGA-gold-labelled cells in the paraxial mesoderm of the early-somite-stage embryo examined 3, 6 and 9-10 h aftergrafting to the anterior (ANT-PSM) and the middle(MID-PSM) regions of the presomitic mesoderm. Eachbox represents a somite in the cultured embryo formedafter grafting. The rectangle at the posterior endrepresents the presomitic mesoderm which is divided intothree portions corresponding to the anterior, middle andposterior regions of the tissue. Each dot represents oneembryo having a labelled somite at the specific segmentallevel.

20 -i

15-

B 10-

• ANT-PSMES MID-PSM• POST-PSM

I I1 2 3 4 5 6 7 8 9

Newly formed somites

Fig. 14. The segmental distribution of labelled cells inthe newly formed somites of the chimaeric embryosfollowing orthotopic grafting at the anterior(ANT-PSM). middle (MID-PSM) and posterior(POST-PSM) regions of the presomitic mesoderm at theearly-somite stage and cultured for 20-21 h. Significantdifferences in the segmental distribution of labelled cellsbetween groups by Kendall tau B test at P< 0-001.

Cell labelling by WGA-goldThe binding of WGA to newly formed mesenchymeof primitive-streak-stage mouse embryos and theembryonic ectoderm of the neurulating rat and mouseembryos has recently been demonstrated (Kimber,1986; Dealtry & Sellen, 1987; Smits-van Prooije etal.1986; Tan & Morriss-Kay, 1986; Chan & Tarn, 1988).

Table 4. The number of somites colonized by theWGA-gold-labelled cells following grafting todifferent regions of the presomitic mesoderm

No. of somiteswith labelledcells

No. of embrvos receiving grafts at:

ANT-PSM MID-PSM POST-PSM

OneTwoThreeFour

Mean no oflabelled somites

61920

1-85 ±0-10(" = 27)

6

25

II2

2-22 ±0-11

0791

2-64±0-15(n = 17)

Note: ANT-PSM. anterior: MID-PSM. middle andPOST-PSM. posterior regions of the presomitic mesoderm(PSM)

Statistical analyses.Significant between Experimental group differences by

ANOVA (P< 0-004).[POST-PSM] = [MID-PSM] and both groups > [ANT-PSM]

by Duncan's multiple range comparison ill / ' < 0 (15.

Using WGA molecules conjugated to either horse-radish peroxidase or fluorescein isothiocyanate. cellsurface receptors for this lectin could be detected inthe posterior paraxial mesoderm of the early-somite-stage to forelimb-bud-stage mouse embryos (i.e.8-5-9-5 days). WGA-binding activity is limited to thepresomitic mesoderm and becomes greatly dimin-ished in the segmented somite (P.P.L.T., unpub-lished observation). The present study has made useof this WGA-binding property of the presomiticmesoderm cells and WGA-gold conjugate wasemployed as the principal cell marker. The WGA-gold marker has the advantage that after appropriatesilver enhancement it could readily be detected evenby low-power light microscopic scanning of serialhistological sections. The present study also demon-strated that the distribution pattern of the WGA-gold-labelled cells was similar to that of cells labelledwith a standard marker - [3H]thymidine - suggestingthat the observation obtained with WGA-gold label-ling is representative of the normal behaviour of thepresomitic mesoderm cells. Results of a graftingexperiment using cells marked with both WGA-goldand [3H]thymidine further suggest that about 90 % ofthe graft-derived population could be detected by theWGA-gold marker (P.P.L.T., unpublished obser-vation).

Cell mixing in the presomitic mesodermIt is clear from this study that graft-derived cellstended to colonize more than one somite and thelabelled cells were found in both the sclerotome andthe dermamyotome of the chimaeric somites. Theincorporation of graft-derived cells into the epithelialcomponent of the differentiating somite strongly

388 P. P. L. Tarn

suggests that colonization must have occurred prior tosegmentation. The colonization of a craniocaudalseries of two to four somites further implicited thatthe graft has contributed cells to several somitomeresin the presomitic mesoderm and this is likely due toconsiderable cell mixing among the maturing somito-meres. Previous studies on the morphology and thedevelopmental potential of the somitomeres has ledto the proposition that the tandem series of somito-meres in the presomitic mesoderm is a prepattern ofprospective somites (Meier & Tarn, 1982; Tarn et al.1982; Tarn & Meier, 1982; Tarn, 1986). The somito-mere is also regarded as a selfcontained meristic unitto which definitive allocation of cells has alreadyoccurred during its inception at the caudal end of thepresomitic mesoderm (Tarn & Beddington, 1986).Therefore, any degrees of cell mixing among thesomitomeres would pose questions on the mechanismof maintaining the early meristic organization ofpresomitic cells and the exact timing of definitiveallocation of cells to the prospective somites.Although there is mounting evidence that once cellsare incorporated into the presomitic mesoderm theyare committed to somite formation (Bellairs, 1985,1986; Tarn & Beddington, 1986; Snow & Gregg, 1986;Stern & Keynes, 1986), the exact stage at whichcomplete somite specification with respect to thesomitic boundary and cell membership will occur isnot fully known. Continuous mixing of cells betweensomitomeres would suggest that the developmentalrestriction of presomitic mesoderm cells is not as tightas previous studies might suggest. The somitomericpattern could simply be the manifestation of someforms of pre-existing meristic framework in the par-axial mesoderm to which allocation of prospectivesomitic cells is entirely opportunistic. The segmentalfate of the mesodermal cells is only broadly delin-eated as revealed by the distribution of graft-derivedcells from different regions of the presomitic meso-derm to a certain set of somites formed after grafting.The final segmental address of the cell is probablyacquired at the last stage before somite segmentation.

The extent of cell mixing is probably greater in theposterior region of the presomitic mesoderm so thatgraft-derived cells were distributed to more somitesthan those in the anterior region. Unrestricted cellmovement in the presomitic mesoderm is to beexpected from the loose arrangement of mesenchy-mal cells (Flint & Ede, 1978; Bellairs. 1979; Meier,1979; Tarn et al. 1982) but cell mobility becomeslimited by the increase in the cell-cell adhesivenessand intercellular junctional binding as cells are be-coming more committed to somite segmentation(Bellairs, 1979; Lash et al. 1984; Chernoff & Lash,1981; Lash & Yamada, 1986; Cheney & Lash, 1984;Bellairs et al. 1978; Hatta & Takeichi. 1986; Hatta et

al. 1987). The present observation that graft-derivedcells in the anterior presomitic mesoderm were dis-tributed to fewer somites than those in the posteriorregions may be indirect evidence for the reduced cellmobility. Alternatively, it may be that graft-derivedcells in the anterior region were given shorter time forintersomitomeric mixing than in other regions of thepresomitic mesoderm.

It is possible that spreading of graft-derived cells toseveral somitomeres might be unrelated to cell mixingbut was a consequence of the disruption of the normalintersomitomeric relationship caused by the graftingof too many cells in the presomitic mesoderm. Duringeach grafting, about 30 cells were added to thesomitomere and this might result in approximately10 % increase in cell number for somitomeres in theanterior region of the presomitic mesoderm andnearly 30 % increase for those in the posterior region(Tarn & Beddington, 1986). However, such an in-crease in tissue mass seemed to be well toleratedduring the development of the paraxial mesoderm.Ultrastructural examination of the operated pre-somitic mesoderm revealed no significant alterationin the number and the morphology of the somitomereand the size of somites was similar between theoperated side and the unoperated side in most em-bryos. Although the formation of abnormally largesomites observed in some experimental embryoswould indicate that such regulatory mechanism mayhave failed occasionally, a careful comparison of thesomite numbers of the operated and unoperatedparaxial mesoderm revealed that a regulation ofsomite number through the adjustment of the somitesize (Flint et al. 1978; Menkes & Sandor, 1977; Tarn,1981) might have been achieved by the embryo duringpostoperative development.

Cell movement in the presomitic mesodermThe present study has demonstrated that the coloniz-ation of paraxial mesoderm began with the somitethat was derived from the somitomere occupying thesite of grafting and extended caudally to the next oneto three somites in the series. There is therefore anapparent posterior movement of the graft-derivedcells in the paraxial mesoderm during somite forma-tion under the present experimental circumstances.However, it should be noted that a similar posteriordisplacement could be achieved by either a passiverelocation of the grafted cells or a retarded anteriormovement of the grafted cells. Both of these possi-bilities have not been strictly excluded by the presentstudy and this could only be resolved by a more directstudy on the movement of cells in the presomiticmesoderm similar to the cinematographic analysis ofthe migration of mesodermal cells in the gastrulatingmouse embryo (Nakatsuji et al. 1986). Nevertheless,

Somite formation in the mouse 389

a posterior displacement of cells of the segmentalplate has been observed in the chick embryo. It isfurther suggested that such posteriorly moving cellsare those that normally reside in the segmental plateand the interaction between them and those recentlyrecruited from the primitive streak results in theformation of a complete somite (Bellairs & Veini,1984). This cell movement is part of the overallmorphogenetic movement associated with gastru-lation and streak regression (Stern & Bellairs, 1984;Ooi et al. 1986). A previous study on the allocation ofcells from the primitive streak to the paraxial meso-derm has shown that these cells could colonize thosesomitomeres that are already established in the preso-mitic mesoderm (Tarn & Beddington, 1987). In orderto achieve this, some cells that are recently recruitedfrom the primitive streak have to move anteriorly inthe presomitic mesoderm and would, therefore, con-front the mesodermal cells that are moving pos-teriorly. Whether this is the same phenomenon asthat described for the avian embryo (Bellairs, 1985)where there is an interaction of two sources of somiticcells resulting in the specification of somitomeres isvet unclear.

References

BEDDINGTON, R. S. P. (1981). An autoradiographicanalysis of the potency of embryonic ectoderm in the8th day postimplantation mouse embryo. J. Embryol.exp. Morph. 64, 87-104.

BELLAIRS, R. (1979). The mechanism of somitesegmentation in the chick embryos. J. Embryol. exp.Morph. 51, 227-243.

BELLAIRS, R. (1985). A new theory about somiteformation in the chick. In Developmental Mechanisms:Normal and Abnormal, pp. 25-44. New York: Alan R.Liss Inc.

BELLAIRS, R. (1986). The tail bud and cessation ofsegmentation in the chick embryo. In Somites inDeveloping Embryos (ed. R. Bellairs, D. A. Ede & J.W. Lash), pp. 161-178. New York: Plenum.

BELLAIRS, R., CURTIS, A. S. G. & SANDERS. E. J. (1978).Cell adhesivenes and embryonic differentiation.J. Embryol. exp. Morph. 46, 207-213.

BELLAIRS, R. & VEINI, M. (1984). Experimental analysisof control mechanisms in somite segmentation in avianembryos. II. Reduction of materials in the gastrulastage of the chick. J. Embryol. exp. Morph. 79,183-200.

CHAN, W. Y. & TAM, P. P. L. (1988). A morphologicaland experimental study of the mesencephalic neuralcrest cells in the mouse embryo using wheat germagglutinin-gold conjugate as the cell marker.Development (in press).

CHENEY, C. M. & LASH, J. W. (1984). An increase incell-cell adhesion in the chick segmental plate results

in a meristic pattern. J. Embryol. exp. Morph. 79.1-10.

CHERNOFF, E. A. G. & LASH. J. W. (1981). Cellmovement in somite formation and development in thechick: inhibition of segmentation. Devi Biol. 87.212-219.

DEALTRY. G. B. & SELLENS, M. H. (1987). Lectinreceptors on the peri-implantation and early post-implantation mouse embryo. Wilhelm Roux Arch, devlBiol. 196, 59-65.

FLEISS, J. L. (1981). Statistical Methods for Rates andProportions. Chapter 12. The Measurement and Controlof Misclassification Error. New York: John Wiley.

FLINT, O. P. (1977). Cell interactions in the developingaxial skeleton in normal and mutant mouse embryos.In Vertebrate Limb and Somite Morphogenesis (ed. D.A. Ede, J. R. Hinchliffe & M. Balls), pp. 465-484.Cambridge: Cambridge University Press.

FLINT, O. P. & EDE, D. A. (1978). Cell interactions inthe developing somite, in vivo comparisons betweenamputated (am/am) and normal mouse embryos.J. Cell Sci. 31.275-291.

FLINT, O. P., EDE, D. A., WILBY, O. K. & PROCTOR, J.(1978). Control of somite number in normal andamputated mutant mouse embryos: an experimentaland a theoretical analysis. J. Embrvol. exp. Morph. 45.189-202.

HATTA, K., TAKAGI, S., FUJISAWA, H. & TAKEICHI, M.(1987). Spatial and temporal expression pattern of N-adherin cell adhesion molecules correlated withmorphogenetic processes of chick embryos. Devi Biol.120, 215-227.

HATTA, K. & TAKEICHI, M. (1986). Expression of N-adherin adhesion molecules associated with earlymorphogenetic events in chick development. Nature,Lond. 320, 447-449.

KIMBER, S. J. (1986). Distribution of lectin receptors inpostimplantation mouse embryos at 6-8 days gestation.Am. J. Anat. 177, 203-219.

LASH, J. W., SEITZ, A. W., CHENEY, C. M. &OSTROVSKY, D. (1984). On the role of fibronectinduring the compaction stage of somitogenesis in thechick embryo. J. exp. Zool. 232, 197-206.

LASH, J. W. & YAMADA, K. M. (1986). The adhesionrecognition signal of fibronectin: a possible triggermechanism for compaction during somitogenesis. InSomites in Developing Embryos (ed. R. Bellairs. D. A.Ede & J. W. Lash), pp. 201-208. New York: Plenum.

LOWRY, O. H., ROSEBROUGH, N. J., FARR, A. L. &RANDALL, R. J. (1951). Protein measurement with theFolin reagent. J. biol. Chem. 193, 265-275.

MEIER, S. (1979). Development of the duck mesoblast:formation of the embryonic axis and establishment ofthe metameric pattern. Devi Biol. 73. 25-45.

MEIER, S. & TAM, P. P. L. (1982). Metameric patterndevelopment in the embryonic axis of the mouse. I.Differentiation of the cranial segments. Differentiation21, 95-108.

MENKES, B. & SANDOR, S. (1977). Somitogenesis.regulation potencies, sequence determination andprimordial interactions. In Vertebrate Limb and Somite

390 P. P. L. Tam

Morphogenesis (ed. D. A. Ede, J. R. Hinchliffe & M.Balls), pp. 405-419. Cambridge: Cambridge UniversityPress.

NAKATSUJI, N., SNOW, M. H. L. & WYLIE, C. C. (1986).

Cinematographic study of the cell movement in theprimitive-streak-stage mouse embryo. J. Embryol. exp.Morph. 96, 99-109.

Ooi, E. C. V., SANDERS, E. J. & BELLAIRS, R. (1986).The contribution of the primitive streak to the somitesin the avian embryo. J. Embryol. exp. Morph. 92,193-206.

SMITS-VAN PROOIJE, A. E., POELMANN, R. E., GESNICK,

A. F., VAN GROENINGEN, M. J. & VERMEJI-KEERS, CHR.(1986). The cell surface coat in neurulating mouse andrat embryos, studied with lectins. Anat. Embryol. 175.111-117.

SNOW, M. H. L. & GREGG, B. C. (1986). The program ofvertebral development. In Somites in DevelopingEmbryos (ed. R. Bellairs. D. A. Ede & J. W. Lash),pp. 301-312. New York: Plenum.

STERN, C. D. & BELLAIRS, R. (1984). The role of noderegression and elongation of the area pellucida in theformation of somites in avian embryos. J. Embryol.exp. Morph. 81,75-92.

STERN, C. D. & KEYNES. R. J. (1986). Cell lineage andthe formation and maintenance of half somites. InSomites in Developing Embryos (ed. R. Bellairs, D. A.Ede & J. W. Lash), pp. 147-161. New York: Plenum.

TAM. P. P. L. (1981). The control of somitogenesis inmouse embryos. J. Embryol. exp. Morph. 65Supplement, 103-128.

TAM. P. P. L. (1986). A study on the pattern ofprospective somites in the presomitic mesoderm ofmouse embryos. J. Embryol. exp. Morph. 81, 75-92.

TAM. P. P. L. & BEDDINGTON. R. S. P. (1986). Themetameric organization of the presomitic mesodermand somite specification in the mouse embryo. InSomites in Developing Embryos (ed. R. Bellairs, D. A.Ede & J. W. Lash), pp. 17-36. New York: Plenum.

TAM, P. P. L. & BEDDINGTON, R. S. P. (1987). Theformation of mesodermal tissues in the mouse embryoduring gastrulation and early organogenesis.Development 99, 109-126.

TAM, P. P. L. & MEIER, S. (1982). The establishment of asomitomeric pattern in the mesoderm of thegastrulating mouse embryo. Am. J. Anat. 164, 209-225.

TAM, P. P. L., MEIER, S. & JACOBSON, A. G. (1982).Differentiation of the metameric pattern in theembryonic axis of the mouse. II. Somitomericorganization of the presomitic mesoderm.Differentiation 21, 109-122.

TAN, S. S. & MORRISS-KAY, G. (1986). Analysis of cranialneural crest cell migration and early fates inpostimplantation rat chimaeras. J. Embryol. exp.Morph. 98. 21-58.

{Accepted 17 March 1988)