RESEARCH ARTICLE

Gastrulation in the Spider Zygiella x-notataInvolves Three Distinct Phases of CellInternalizationR. Crystal Chaw,†‡ Emily Vance,† and Steven D. Black*

The cell movements of gastrulation were analyzed in embryos of the spider Zygiella x-notata, using time-lapse video, cell tracing, and improved histology. Cells are internalized near the center of the germ disc inthree distinct phases. First, cumulus mesenchyme cells ingress and migrate as a group beneath thesuperficial layer. Second, mass internalization through a blastopore yields a diffusely organized deep layer.Third, superficial cells accumulate at the center of the germ disc to form the caudal bud. The floor isinternalized, and the caudal bud moves over the nascent dorsal field to form the caudal lobe. This patternof gastrulation differs from the canonical pattern described in the historical literature: (1) the cumulus ofZ. x-notata is completely formed before any other cells internalize; and (2) the caudal lobe is formed bymeans of the caudal bud, which is a locus of cell internalization. Developmental Dynamics 236:3484–3495, 2007.© 2007 Wiley-Liss, Inc.

Key words: gastrulation; morphogenesis; arthropod; spider; arachnid; chelicerate

Accepted 29 September 2007

INTRODUCTION

Gastrulation is the morphogeneticprocess that establishes the germlayers and converts the simple polar-ities of the egg into the more complexorganization of the later embryo. Aswith most other animal embryos,gastrulation in spiders internalizesboth prospective endoderm and me-soderm. Classic studies of spidergastrulation showed that the embryointernalizes two distinct populationsof cells. First, internalization of aportion of the blastoderm creates asmall region comprising multiplecell layers, which is called the “prim-itive plate” (synonymous with “ante-

rior cumulus” and sometimes with“primary thickening”). Associatedwith the primative plate is a smallsecond population of cells that formsthe “cumulus” (synonymous with“posterior cumulus”). The internal-ized cumulus mesenchyme cells mi-grate as a group underneath theblastoderm and travel to the edge ofthe germ disc. Continued internal-ization of prospective mesoderm andendoderm cells at the center of thegerm disc enlarges the primitiveplate during and after cumulus mi-gration. This general pattern is saidto be followed by most araneomorphspiders and is considered the canon-

ical model of spider gastrulation (seereviews by Anderson, 1973; Foelix,1996).

Both historical and modern litera-ture on spider development highlightthe importance of the second popula-tion of internalizing cells, the cumulusmesenchyme. Holm (1952) used trans-plantation experiments in Agelenalabyrinthica to show that a secondaryaxis can be generated by moving thecumulus to another area of the germdisc. Akiyama-Oda and Oda (2003)found that the cumulus of Achaeara-nea tepidariorum (Achaearanea hasbeen transferred to the genus Paraste-atoda by Saaristo [2006]) is a site of

The Supplementary Material referred to in this article can be found at http://www.interscience.wiley.com/jpages/1058-8388/suppmatKleinholtz Biological Laboratories, Department of Biology, Reed College, Portland, Oregon†Drs. Chaw and Vance contributed equally to this work.‡Dr. Chaw’s present address is Department of Integrative Biology, University of California, Berkeley, CA 94720.*Correspondence to: Steven D. Black, Kleinholtz Biological Laboratories, Department of Biology, Reed College, 3203 S.E.Woodstock Blvd., Portland, OR 97202. E-mail: sblack@reed.edu

DOI 10.1002/dvdy.21371Published online 9 November 2007 in Wiley InterScience (www.interscience.wiley.com).

DEVELOPMENTAL DYNAMICS 236:3484–3495, 2007

© 2007 Wiley-Liss, Inc.

decapentaplegic (dpp) signaling. Theepithelial cells that overlie the path ofcumulus migration extend cytonemesto the cumulus and respond to the dppsignaling by expressing phosphory-lated mothers against dpp (pMAD). Inresponse to cumulus migration anddpp signaling, the embryonic germdisc breaks radial symmetry. Thepoint on the germ disc periphery op-posite the endpoint of cumulus migra-tion becomes the anterior pole of thegerm band, and the epithelial cellsthat overlie the path of cumulus mi-gration become the dorsal extraem-bryonic area. Dpp-depleted embryosdo not break radial symmetry, andshow a phenotype of ventralized seg-mented rings (Akiyama-Oda and Oda,2006). Past and present studies estab-lish the cumulus as an important siteof cell–cell communication that estab-lishes embryonic axis polarity.

A recent study by Oda et al. (2007)further characterized the gastrula ofA. tepidariorum by examining the ex-pression patterns of forkhead, twist,and dpp. Prospective cell fates wereinferred from expression patterns andfrom positions of cells, primarily fromfixed embryos. Oda et al. were able topresent a detailed picture of the orga-nization of the gastrula. First, puta-tive endoderm (including cumulusmesenchyme) ingresses at the centerof the germ disc; then, putativeendoderm and mesoderm internalizeat the periphery of the germ disc; fi-nally, additional putative mesoderminternalizes at the center by means ofa pit (Akiyama-Oda and Oda, 2003).Although the actual fates of the vari-ous cell populations are not known atthis time, the studies by Oda et al.clearly demonstrate that specific re-gions of the embryo can be definedvery early and that the cells can beidentified by in situ hybridization dur-ing subsequent development. Theearly internalization of central, non-cumulus cells seen by Oda et al. can beinterpreted as the initial formation ofthe primitive plate, and is thus consis-tent with the canonical model. What isunexpected is that most of theendoderm and mesoderm in A. tepi-dariorum seems to originate from theperiphery. Classic accounts of spidergastrulation do not recognize the pe-ripheral region as an area of signifi-cant cell internalization, although

Montgomery’s (1909) histologicalstudy of A. tepidariorum (then calledTheridion or Theridium tepidari-orum) noted some vitellophages inter-nalized at the periphery. The later in-ternalization of central cells to formthe interior of the caudal lobe is prob-ably consistent with the canonicalmodel, because Holm (1952) recog-nized that internalization of cells atthe center of the germ disc continuedthroughout gastrulation (see also theAnalpol of Holm, 1940). To summa-rize, gastrulation of A. tepidariorumembryos departs from the canonicalmodel by having large-scale internal-ization at the periphery of the germdisc.

The difference in gastrulationmovements between A. tepidariorumand the canonical model suggest to usthat the model is in need of reevalua-tion. Indeed, the model is based pri-marily on analyses of fixed materialand Holm’s (1952) cell-marking exper-iments of A. labyrinthica. Thus, wehave used modern methods to studygastrulation in a spider embryo. Weanalyze the cell movements of gastru-lation at the cellular level using time-lapse video, cell tracing, and improvedhistological methods. The speciesstudied, Zygiella x-notata (family Ara-neidae), is an araneoid spider as is A.tepidariorum (family Theridiidae). Z.x-notata’s pattern of gastrulation dif-fers from the canonical pattern by ex-hibiting a heterochronic shift in cumu-lus formation relative to formation ofthe primitive plate, and by using anunusual mechanism to internalize pu-tative mesoderm of the caudal lobe.The pattern of cell movements alsodiffers fundamentally from that de-scribed for A. tepidariorum. Our anal-ysis of the morphogenetic movementsof the Z. x-notata gastrula suggeststhat there is more variation in gastru-lation in spider embryos than hasbeen previously appreciated.

RESULTS AND DISCUSSION

Early embryos of Zygiella x-notata un-dergo three phases of cell internaliza-tion, which we designate gastrulationI, II, and III. Signal stages are de-picted in Figure 1, and the timing ofthese stages is given in Figure 2. Sup-plementary Movie S1, which can beviewed at http://www.interscience.

wiley.com/jpages/1058-8388/suppmat,shows time-lapse of Z. x-notata devel-opment beginning with stage C. Be-fore gastrulation, the embryos followthe general pattern described for en-telegyne spiders (Holm, 1952); the700-�m embryos undergo rounds ofnuclear division in what appears to bea common cytoplasm (Fig. 1A). Then,most if not all energids migrate to theperiphery (Fig. 1B). A blastodermforms and the embryo contracts toform the perivitelline space. At the be-ginning of contraction, the blastodermis uniform; by the end of contraction, asmall white spot has appeared at oneside of the blastoderm. This spot is acluster of cells that marks the point ofblastopore formation (Fig. 1C). Asmore cells accumulate locally at theblastopore, most of the other blasto-derm cells migrate to the same hemi-sphere to form the germ disc. Theblastopore is at the center of the germdisc.

Gastrulation I

This phase begins with the formationof the blastopore (equivalent to the“primitive pit” and the “primitivegroove” of Holm [1940, 1952]) and in-volves ingression of fewer than 20blastoderm cells. The primary func-tion of gastrulation I appears to be theinternalization of cumulus mesen-chyme cells. By tracing cells in time-lapse, it is possible to identify the in-gressing cells and their points ofingression. A typical embryo is shownin Figure 3, with the cells that willingress during gastrulation I false-col-ored red. Figure 3 shows that the cen-tral blastoderm region internalizesfirst. These cells include the cumulusmesenchyme cells, which will migrateas a group (white arrowhead) under-neath the surface blastoderm. Cellsfurther away from the blastopore arefalse-colored a variety of colors in Fig-ure 3, and they can be seen to con-verge on the central region withoutinternalizing (see SupplementaryMovie S2, representative of N � 11embryos analyzed by time-lapse). Al-though the majority of ingressing cellsform the cumulus (comprising 8–12cells, N � 6 embryos examined in se-rial sections), a small number of otherblastoderm cells ingress individuallyvery near or with the cumulus, andthese can be seen migrating beneath

SPIDER GASTRULATION 3485

the superficial layer in time-lapse.From our time-lapse observations andsections, we were unable to discern aclear pattern of migration for theseother ingressing cells. They are prob-ably newly formed vitellophages, sim-ilar to those described for A. labyrin-thica (Holm, 1952), Latrodectusmactans (Rempel, 1957), and A. tepi-dariorum (Montgomery, 1909). Vitel-lophages may ingress at multiplepoints in these spiders, but appear toingress near the blastopore in Z. x-notata. At this stage, we typically findcells in the yolk only if they are closeto the blastopore, although in someembryos we have found two or threecells deep within the yolk. In section,the ingressing cumulus mesenchymecells are apically constricted (Fig.4A,B), similar to the bottle cells of theamphibian blastopore. The apicallyconstricted cumulus cells can also beseen externally (they are the unla-beled central cells in Fig. 5A), whereasthe surrounding cells retain an uncon-stricted shape. The bottle morphologysuggests an active role on the part ofthe internalizing cumulus cells. Theblue arrowheads in Figure 4B identifyprobable vitellophages. Internaliza-tion of cells during gastrulation I oc-curs at or near the blastopore in Z.x-notata; no internalization occurs atother regions.

Once internalized, the cumulusmesenchyme cells migrate posteriadas a group underneath the superficiallayer, as described for other species(Holm, 1940; Akiyama-Oda and Oda,2003). The migration can be easilyseen in time-lapse (see white arrow-head in Fig. 3 and SupplementaryMovie S2). Cumulus cell internaliza-tion and cumulus migration take ap-proximately 26 hr (Fig. 2). The mi-grating cumulus mesenchyme cells inZ. x-notata appear vacuolated (likely aresult of our fixation protocol) and areshown in Figure 4C,D. Scanning elec-tron micrographs of A. tepidariorumembryos by Akiyama-Oda and Oda(2003) show that the superficial cellsoverlying the cumulus interact with itby extending cytonemes, and a similarmorphology is revealed at the histo-logical level in Z. x-notata—superficialcells contact the cumulus by means oflong, narrow protrusions (green ar-rowhead, Fig. 4D). These protrusionsare likely to effect cell–cell communi-

cation between the cumulus mesen-chyme cells and overlying superficialcells of the germ disc, as suggested byAkiyama-Oda and Oda (2003) for A.tepidariorum. Up to this point, gastru-lation in Z. x-notata differs from thecanonical model by internalizing onlycumulus mesenchyme and a few vitel-lophages; no primitive plate is formed.In A. tepidariorum, all the early in-gressing cells express forkhead andinclude cumulus mesenchyme and ad-ditional cells designated centralendoderm (Oda et al., 2007; see alsoMontgomery, 1909, whose sectionsshow a considerable number of cellsinternalized before formation of the“cumulus posterior”). Other araneo-morph spiders show internalization of

a large number of noncumulus cellsduring this early period, for example,Latrodectus mactans (Theridiidae;Rempel, 1957), Torania variata (Spar-assidae; Ehn, 1963), Cupiennius salei(Ctenidae; Seitz, 1966), and A. laby-rinthica (Agelenidae; Holm, 1952).

Gastrulation II

A second phase of cell internalizationbegins as the cumulus mesenchymecells are migrating. Two things setgastrulation II apart from cumulusformation in Z. x-notata. First, gastru-lation II involves many more cells andmust be the major source of cells forthe mesoderm and endoderm of theprosoma. Second, the two phases be-

Fig. 1. Normal development. Embryos oriented posterior down in all panels. A–H: View of germdisc side. I–O: Ventral view. A: Prenuclear migration. Embryo appears as a mass of yolk spherules.B: Nuclear migration. C: Gastrulation I begins with cumulus formation; contraction of blastoderm.D: Cumulus migration begins (white arrowhead); germ disc apparent. E: Gastrulation II (massinternalization) begins. F: Cumulus migration ends; continued mass internalization has produced adispersed deep layer of cells. G: Gastrulation III begins with caudal bud formation (red arrowhead);dorsal field begins to form. H: Caudal bud complete; dorsal field expansion. I: Caudal budmovement, dorsal field expansion continues. J: Germ band formation, segmentation apparent.K: Appendage bud formation. L: Inversion begins: ventral sulcus appears along ventral midline;appendage buds elongate. M: Mid-inversion. N: Inversion complete, ventral closure begins.O: Ventral closure complete. Original magnification, �54. Supplementary Movie S1 shows devel-opment of a single embryo from germ disc contraction to hatching.

3486 CHAW ET AL.

gin separately; there is a definitepause (approximately 5 hr) betweencumulus formation and the beginningof gastrulation II (Fig. 2). This pausecan be seen as a period of minimal cellinternalization while the cumulus ismigrating (for example, in the 4 hrspanned by Figure 3D,E, fewer thanfive cells in this embryo are internal-ized in the central region). The pauseis then followed by mass ingressionthrough a central blastopore (Fig. 5and Supplementary Movie S3). Gas-trulation II is the source for much ofthe presumptive mesendoderm anddoes not begin until after the fullyformed cumulus begins to migrate.

Cells internalize during gastrula-tion II via a blastoporal region thatforms in the center of the germ disc.The blastopore cannot be seen easilyin living embryos, as has been re-marked for some other species (e.g.,L. mactans, Rempel, 1957; A. tepidari-orum, Akiyama-Oda and Oda, 2003);however, our time-lapse videos and his-tological sections clearly demonstrateblastoporal function and morphology.Time-lapse analysis shows a pattern ofmovement of superficial cells to thecentral region. This is the same areaof the germ disc at which the presump-

tive cumulus mesenchyme cells in-gressed. This finding is shown in Fig-ure 5 by the central position of boththe presumptive cumulus cells and,hours later, the area in which othersuperficial cells ingress during gastru-lation II. In Figure 5A, the presump-tive cumulus mesenchyme cells are al-ready apically constricted (unlabeledcells at center). Some cells that willinternalize during gastrulation II arefalse-colored to reveal their progressto the blastoporal region and subse-quent internalization. Not all cells in-ternalize at exactly the same point.Internalization of these cells beginsapproximately at Figure 5G and ex-tends beyond Figure 5L; eventuallyeven the turquoise-colored cells willingress (see Supplementary MovieS3). The region of the superficial layerthat will internalize during gastrula-tion II can be approximated by a circlethat includes the turquoise-coloredcells in Figure 5. This circle corre-sponds spatially with the region ofpresumptive mesendoderm in Holm’s(1952) fate map of an earlier stage A.labyrinthica. The period of gastrula-tion II in Z. x-notata is approximately17 hr, ending when the caudal budforms (see below), although there is

some variation between embryos (N �11 embryos analyzed by time-lapse).

Histological sections of embryosduring gastrulation II show a smallblastopore at the center of the germdisc, complete with apically con-stricted cells (Fig. 6A,B). These cellsappear less elongated than the moreclassic bottle cells seen in gastrulationI. Early in gastrulation II, the cumu-lus mesenchyme cells are midwaythrough migration, and only a fewcells have internalized at the blas-topore. Figure 6A,B shows sectionsthrough the early blastopore some dis-tance from the cumulus (Stage E,same embryo as pictured in Fig.4C,D). Together, both figures showthat early in gastrulation II, the onlycells below the superficial layer arenext to the blastopore (with the excep-tion of cumulus mesenchyme cells andvitellophages). As gastrulation II con-tinues, time-lapse and histology showincreased cell internalization. Themovies show movement of superficiallayer cells toward the central region,and migration of internalized cellscentrifugally away from the blas-topore (see Supplementary Movies S4and S5). The direction of movement ofinternalized cells toward the periph-ery of the germ disc is opposite thatdescribed for the same stage of A. tepi-dariorum (Oda et al., 2007). In A. tepi-dariorum, the peripheral rim of thegerm disc is said to be a second site ofcell internalization—both peripheralforkhead-positive cells (designatedendoderm, “pEND”) and forkhead-and twist-positive cells (designatedmesoderm, “pMES”) are found in adeep layer at the rim. Moreover, mi-gration of these internalized cells isthought to be in a centripetal direc-tion, toward the center of the germdisc. A smaller number of twist-posi-tive cells also internalize at the centerof the germ disc (“cMES”), at a laterstage. Careful analysis of our moviesand histological sections show that,unlike A. tepidariorum, all the cells ofthe deep layer in Z. x-notata originatenear the center of the germ disc. Sec-tions near the end of gastrulation IIshow the continued presence of a blas-topore (Fig. 6C,D). The blastopore canspan several sections. At this stage ofgastrulation, there is now a diffusedeep layer of cells underlying the

Fig. 2. Timing of normal development at room temperature. Horizontal bars show duration of eachstage; numbers in bars indicate duration in hours. Each bar represents the average of at least twoembryos from different egg sacs (range � 2–11 embryos). CM, cumulus; BP, blastopore ofgastrulation II; DF, dorsal field; CB, caudal bud.

SPIDER GASTRULATION 3487

Fig. 3.

Fig. 4.

3488 CHAW ET AL.

germ disc. These cells can be seen insections adjacent to the blastopore andat some distance from it (green arrow-heads in Fig. 6C,E). Note that the onlydeep cells at the periphery are cumulusmesenchyme cells (adjacent to blue ar-row, Fig. 6C). Analysis of serial sectionsreveals no accumulation of other deepcells at the periphery, which is consis-tent with an absence of cell internaliza-tion at the rim. The deep layer can beseen in the external view as well; a com-parison of Figure 1E (start of gastrula-tion II) and Figure 1F (near end of gas-trulation II) shows that the germ disc ofthe later stage has become moreopaque, which correlates with the in-creased number of internal cells shownby histology.

In many other spiders, cumulus for-mation is preceded by cellular ingres-sion through a blastopore to form the“primitive plate” or “primary thicken-ing,” which is an area of the earlygerm disc with multiple cell layers(e.g., A. labyrinthica [Holm, 1952],Torania variata [Ehn, 1963], Cupien-nius salei [Seitz, 1966]). The cumulusforms from this area and is describedas a further accumulation of cells, abulge on the primitive plate, or as alarger thickening that appears on one

side of the primitive plate. In Z. x-notata, the internalization of cumulusmesenchyme cells occurs both firstand separately from the gastrulationof the rest of the mesendoderm. Thereis no primitive plate that forms beforethe cumulus. Figure 4 shows that theonly deep layer cells of the early gas-trula are the cumulus mesenchymecells. The primitive plate as describedfor other spiders thus represents rela-tively early internalization of some ofthe presumptive mesendoderm. Theminimal internalization of mesend-oderm before cumulus formation in Z.x-notata may represent a hetero-chrony of gastrulation phases com-pared with the pattern historically de-scribed for many other spiders.

Gastrulation III

The third round of gastrulation in Z.x-notata involves a distinctly differentmechanism that forms a structure wedesignate the caudal bud. Near the endof gastrulation II, the pattern of cell be-havior changes. Superficial cells con-tinue to migrate toward the center ofthe germ disc, but instead of internaliz-ing, many cells remain on the surfaceand pile up around the blastoporal re-gion. As cells continue to layer, themass forms a distinct volcano-likeshape (Fig. 7). The central floor of thevolcano is internalized as superficialcells of the rim move toward the center,over the floor (three such superficialcells are false-colored red in Figure7A–C to show convergence to the cen-tral region). Cell-shape changes ob-served in time-lapse (D. Rasmussen,unpublished results) suggest that clo-sure of the “mouth” may occur througha purse-string–like constriction of thecells of the rim as in Drosophila dorsalclosure (Kiehart et al., 2000). It is alsopossible that the lateral edges of thevolcano are pushed together over thecentral floor by cell crowding, or thatsome combination of constriction andpushing results in internalization of thefloor. In section, the mouth of the mid-stage caudal bud is distinct (Fig. 7E).Cells of the floor also leave the superfi-cial layer by ingression (D. Rasmussen,unpublished results). There is variationin the size of the caudal bud—for exam-ple, the embryo depicted in Figure 7A–D has a larger caudal bud than theembryo in Figure 8—but in all cases,

the caudal bud undergoes the same cel-lular movements to internalize the cen-tral floor (N � 12 embryos analyzed bytime-lapse. Supplementary Movie S5shows a relatively large caudal bud, andMovie S6 shows a relatively small cau-dal bud). Gastrulation III ends with theclosure of the “mouth,” at which pointthe caudal bud begins to move posteriadalong the midline. This typically occursas the dorsal field (see below) expandsbeyond approximately 100 degrees ofarc. Sections of the caudal bud at thisstage show cuboidal cells along theleading edge that appear tightly orga-nized (Fig. 7F), and the impression fromtime-lapse is that the leading edge ismore sharply defined than the trailingedge (see Supplementary Movie S6).Behind the caudal bud is an additionallayer of cells. Time-lapse analysis andour preliminary labeling experimentswith 1,1�, di-octadecyl-3,3,3�,3�,-tetra-methylindo-carbocyanine perchlorate(DiI) suggest that the caudal bud be-comes part of the posterior-most part ofthe embryo, with labeling observed inboth superficial and deep layers.

The caudal bud is probably the pre-cursor in Z. x-notata to the posteriorgrowth zone. The posterior growthzone is the region of the germ bandfrom which the posterior segmentsarise and is a feature typical of short-and intermediate-germ band insects,many other arthropods including spi-ders, and some annelids (Anderson,1973; Davis and Patel, 2002;Stollewerk et al., 2003; Schoppmeierand Damen, 2005; Oda et al., 2007; deRosa et al., 2005). The early caudallobe of A. tepidariorum has beenshown by Oda et al. (2007) to form in asimilar place and time as the caudalbud. Putative mesoderm cells in theearly caudal lobe of A. tepidariorumexpress twist and internalize in smallgroups. However, the caudal bud of Z.x-notata is distinguished from the cau-dal lobe of A. tepidariorum because itis more tightly organized and formsthe distinct “mouth” while internaliz-ing the central floor. The caudal bud isseen to move in a posterior directionbefore the germ band is formed, pass-ing rapidly across the dorsal field (seeSupplementary Movies S5 and S6).This posteriad movement takes onlysome 9 hr, whereas the formation ofthe posterior segments continues forlong afterward, presumably by a

Fig. 3. External view of gastrulation I. The cu-mulus forms by ingression of superficial cells atthe center of the germ disc. The cells that in-gress during gastrulation I are colored red. Stills4 hr apart. Embryo oriented posterior down. A:Gastrulation begins with congregation of cellsat center. B,C: Internalization of cells. D,E: In-ternalized cumulus mesenchyme cells (whitearrowhead) migrate posteriad beneath the su-perficial layer. F: Cumulus reaches margin ofgerm disc. Gastrulation II begins as cells areinternalized at the center of the germ disc (e.g.,pink, purple). Original magnification, �54. Sup-plementary Movie S2 shows gastrulation I (andbeginning of gastrulation II) with false-coloredcells.Fig. 4. Histology of gastrulation I: cumulusformation and migration. A,B: Ingression ofcells at the center of the germ disc to formnascent cumulus (stage C); blue arrowheadspoint to putative vitellophages. C,D: Cumulus inmid-migration (stage E). Parasagittal section,slightly oblique. Curved arrow shows directionof migration of cumulus; black arrowheads in-dicate margin of germ disc. Inset shows thatcumulus mesenchyme cells are vacuolated andhave contact with superficial cells; green arrow-head points to one of the superficial cells withan elongated process in contact with cumulusmesenchyme cells. Others are visible in the fig-ure. Scale bar � 100 �m.

SPIDER GASTRULATION 3489

Fig. 5. External view of gastrulation II. The blastopore will form at the site of the blastopore for gastrulation I. Mass internalization begins only afterthe cumulus has formed. Some superficial cells that will internalize during gastrulation II are false-colored. Cumulus cells are not colored. Stills 2 hrsapart. Embryo oriented posterior down. A–D: Cumulus formation (stage C). E–K: Cumulus migration (stages D–F). G–I: Early mass internalization (stageE). J–L: Continuing mass internalization (stage F). Original magnification, �57. Supplementary Movie S3 shows gastrulation II with false-colored cells;Supplementary Movie S4 shows movement of superficial cells to blastopore; and Supplementary Movie S5 shows centrifugal movement of internalizedcells.

3490 CHAW ET AL.

mechanism that does not involve denovo internalization of cells. Duringthe posterior segmentation process, itis likely that the cells derived from thecaudal bud function as a more typicalposterior growth zone, which is knownfor spiders generally (Anderson, 1973;Stollewerk et al., 2003; Schoppmeierand Damen, 2005). We speculate thatthe early caudal bud of Z. x-notatacontains a significant number of pre-sumptive mesoderm cells and thatthese cells express mesodermal mark-ers such as twist at high levels. A cau-dal bud-like structure may have beenobserved in other species of spiders.Anderson (1973) posits that severalolder studies of liphistiid (Heptathelakimurai, Yoshikura, 1955), mygalo-morph (Atypus karschi, Yoshikura,1958; Conothele, Crome, 1963; Ischno-colus, Schimkewitsch and Schimke-witsch, 1911), and haplogynid em-

bryos (Segestria bavarica and S.sonoculata, Holm, 1940) show a simi-lar structure. Consulting the originalsources reveals, however, that withthe possible exception of H. kimurai(see Yoshikura’s Fig. 16C [Yoshikura,1955]), the structures in the other em-bryos resemble the denser regions ofadjacent segments of the germ bandrather than the highly organized cau-dal bud of Z. x-notata. Neither is thecaudal structure in the other speciessaid to move across the germ disc; pre-sumably extension of the germ bandand creation of the posterior segmentsdrive the separation of anterior andposterior poles.

Dorsal field formation begins as thecaudal bud forms, at the end of gas-trulation II. The dorsal field is an ex-panding area of superficial cells in thepresumptive dorsal region of the germdisc, through which yolk is visible. As

with the caudal bud, the dorsal fieldbegins to form after the cumulus hasreached the edge of the germ disc.From histology and time-lapse, thedorsal field in Z. x-notata seems toarise from thinning and expanding ofsuperficial germ disc cells of the dorsalregion (Fig. 8, Supplementary MovieS7). This expansion begins before thecaudal bud begins to migrate. Dorsalfield cells come to surround the poste-rior germ band on three sides, as canbe seen in Figure 8. Holm (1952) dyedthe surface of an A. labyrinthica gas-trula in similar positions, and ob-served the same spreading of markslaterally, to end up on the sides of theposterior germ band. He suggestedthat the dorsal field arose as a resultof the outer cell layer being pushedtransversely and a deep layer emerg-ing, presumably by movements ofmesendoderm cells, or perhaps from

Fig. 6. Histology of gastrulation II: mass internalization through blastopore. Parasagittal sections, slightly oblique. A,B: Beginning of massinternalization (stage E). Central blastopore is indicated by black arrow. Cumulus is in mid-migration but is not visible in this section. This is the sameembryo as in Figure 4C,D. Not much internalization has occurred by this stage—the only cells in the deep layer are beneath the blastopore or areassociated with the cumulus. C,D: Cumulus end-migration stage (stage F). Cumulus indicated by blue arrow and its direction of migration by heavyblack arrow. Black arrowheads indicate margin of germ disc. Note increased number of cells in diffusely organized deep cell layer (green arrowheads).Blastopore indicated by black arrow in D. E: Section from same embryo as C, D showing large number of internalized cells 24 �m away from theblastopore. Scale bar � 100 �m.

SPIDER GASTRULATION 3491

the cumulus mesenchyme cells. Holmthought that much of the dorsal fieldoriginated from the cumulus. In Z. x-notata, when cumulus mesenchyme

cells reach the margin of the germdisc, they dissipate and appear towander beneath the dorsal field cells.We were unable to determine a consis-

tent pattern of migration or the ulti-mate fate of the cumulus cells. Theexpansion of the superficial layer ofthe dorsal field is probably due to thespreading and movement of the super-ficial cells themselves, as suggested bythe cell tracings in Figure 8. Dorsalfield expansion takes approximately50 hr, at which point the germ band iselongated.

In summary, our study finds that Z.x-notata embryos internalize cells at acentral blastoporal region in threephases. Internalization of cumulusmesenchyme occurs first and is fol-lowed after a pause by internalizationof a large number of cells. The lattercells must be prospective mesodermand endoderm based on their numberand the fact that they constitute a denovo layer of cells underneath the su-perficial layer of the germ disc. Thethird phase of gastrulation internal-izes putative mesoderm to form thecaudal bud, which is the likely precur-sor to the posterior growth zone.

In Z. x-notata a “primitive plate,”that is, a central area of the germ discwith multiple cell layers, does notform before the cumulus nor does thecumulus separate from such an area.The sequence of cell internalization inZ. x-notata is thus reversed from thepattern thought to be typical for spi-der development (Anderson, 1973;Foelix, 1996). However, because manyof the older studies relied on interpre-tation of sectioned embryos ratherthan combining histological and time-lapse analyses, it is possible that thedevelopment of the “typical spider” isnot so different than the pattern wedescribe for Z. x-notata. For example,perhaps in other species cumulusmesenchyme cells also internalizefirst, but delay their migration untilother mesendodermal cells begin to in-gress (and the primitive plate becomesvisible). This would be consistent withHolm’s (1952) fate map of the superfi-cial layer of A. labyrinthica, in whichthe presumptive cumulus is at thecenter of the blastoderm surroundedby presumptive mesendoderm. Thecumulus would form from some of thefirst cells to internalize.

During gastrulation II in Z. x-no-tata, the direction of movement of cellsafter internalization is centrifugal;cells move from the central blastopo-ral region toward the peripheral rim

Fig. 7. Gastrulation III. Caudal bud formation. Close up of center of germ disc, prospective dorsalfield at lower right. Three cells on the rim of the caudal bud are false-colored red. Stills 2 hr apart.A: Accumulation of cells in the center of the germ disc. B: Convergence of central cells makes thecaudal bud visible as a volcano-like structure. C: Continued convergence narrows the mouth of thecaudal bud, which is now several cells thick. D: Mouth has closed; caudal bud beginning itsposteriad movement (cells cannot be traced at this point). E,F: Sagittal sections; direction ofmovement of caudal bud is indicated by black arrow. C,E: stage G. F: Caudal bud duringmovement, stage I. Original magnification, �64. Supplementary Movies S5 and S6 show gastru-lation III and movement of caudal bud from different perspectives.

Fig. 8. Dorsal field expansion. Some prospective dorsal field cells in the superficial layer werefalse-colored and traced. Stills 5 hr apart. Embryo oriented posterior down. White arrowheadindicates cumulus. Black arrowhead indicates forming caudal bud. A: Cumulus reaches posteriormargin of germ disc (stage F). B: Dorsal field emerges, and caudal bud begins to form (stage G).C–F: As the cumulus dissipates and the caudal bud forms and migrates, dorsal field cells expandin area, become thin so that yolk is visible underneath, and move away from each other (stages H,I). Supplementary Movie S7 shows dorsal field expansion with false-colored cells. Original mag-nification, �54.

3492 CHAW ET AL.

of the germ disc. This is the oppositedirection than that seen in A. tepidari-orum embryos, which internalize thebulk of the putative endoderm andmesoderm at the periphery (Akiyama-Oda and Oda, 2003, 2006; Yamazakiet al., 2005; Oda et al., 2007). Thisdifference suggests that the two spe-cies have markedly dissimilar fatemaps, despite being representatives ofthe same superfamily, Araneoidea,with common ancestry estimated atapproximately 150 MYA (Ayoub et al.,2007).

Caudal bud-like structures mayhave been observed in other spiders(Anderson, 1973), and the presence ofa pit in the prospective caudal lobe ofA. tepidariorum may indicate cell in-ternalization (Akiyama-Oda and Oda,2003). However, the caudal bud of Z.x-notata is distinct because of its modeof formation and demonstrated inter-nalization of cells by overgrowth of thesuperficial layer; its rapid movementacross the dorsal field; and the cuboi-dal cells that appear highly organizedalong its leading edge.

The basic pattern of gastrulationseen in Z. x-notata, despite occurringin three phases, resembles that histor-ically described for other species ofaraneomorph spiders. In general, ara-neomorphs internalize two distinctpopulations of cells: the cumulus andthe general mesendoderm. With theexception of the pattern found in A.tepidariorum, gastrulation occurs at acentral, blastoporal region on the ven-tral side of the embryo. Additionally,the araneomorphs internalize most oftheir mesendoderm in one continuousphase as opposed to, for example, theinsects, where internalization of me-soderm and endoderm has been un-coupled in both space and time (Roth,2004). The stomodeum and procto-deum of spiders typically originate asseparate invaginations later in devel-opment, but the bulk of the endodermis thought to internalize with the me-soderm during gastrulation (Holm,1952; Anderson, 1973). The caudalbud of Z. x-notata, which we assume tobe a derived feature, is unique in thatit adds a third phase to the process ofgastrulation, but remains in line withthe spatial pattern of most araneo-morphs because cell internalizationstill occurs in a region central to the

germ disc, on the ventral side of theembryo.

If a cumulus–mass internalizationpattern of gastrulation can be thoughtto represent that of araneomorph spi-ders generally, how does this patterncompare with that of other cheliceratetaxa? Few pertinent embryologicaldata exist for many chelicerate taxaand comparisons are further con-strained by differences in methodol-ogy. Still, to our knowledge, Opilionesis the only other arachnid order thathas a posterior cumulus (function un-known, Holm, 1947). A cumulus hasbeen tentatively identified in embryosof an ovoviviparous scorpion, but itsposition vis-a-vis the anterior–poste-rior axis needs confirmation (Abd-elWahhab, 1954). A cumulus is absentin pseudoscorpions and the Acarina;there are no reliable data for the otherchelicerate orders (Anderson, 1973).Gastrulation in other arachnid groupsseems to follow the pattern seen in thebasal liphistiid spider H. kimurai (Yo-shikura, 1955). Specifically, a ventralblastoderm is formed and gastrulationis thought to proceed through a linearblastoporal groove; no posterior cumu-lus is seen. Scorpions show a variationon this pattern; gastrulation is said toproceed by proliferation from a ridgeon the blastoderm without formationof a blastoporal groove (Abd-el Wah-hab, 1954). In the most basal livingchelicerates, the Xiphosura or horse-shoe crabs, gastrulation also proceedsthrough a ventral blastoporal groove.Interestingly, a posterior cumulus(function unknown) is thought to existin three of the four species (Sekiguchiet al., 1988). Thus, to generalize forthe sake of discussion, the chelicerategastrulation pattern appears to be in-gression of mesoderm and endodermthrough a ventral blastoporal groove.The two-step gastrulation pattern in-volving a cumulus and mass internal-ization of mesendoderm that is seen inaraneomorph spiders is not necessar-ily typical of other chelicerates butcould be similar to the pattern in thehorseshoe crabs.

Gastrulation in other arthropodstypically shows more separation ofpresumptive mesoderm and endodermin both space and time. The Drosoph-ila pattern of separate internalizationof mesoderm at the ventral furrow andendoderm by means of anterior and

posterior midgut invaginations holdsin general for other insects (Roth,2004). This finding is clearly differentfrom the chelicerate pattern, wheremuch of the endoderm is internalizedmore or less continuously with the ad-jacent mesoderm. In crustaceans, thepattern of gastrulation is more vari-able. Of the species reviewed by Ger-berding and Patel (2004), those thathave a pause during gastrulation ap-pear to follow two general patterns,although exceptions exist in eachgroup. The non-malocostracans gener-ally follow a pattern more similar toinsects, wherein mesoderm andendoderm move to the interior sepa-rated by space, time, or both. Malocos-tracans have more examples showinginternalization of a small number ofcells (mesendoderm precursors, vitel-locytes, or germ cells) followed by in-ternalization of adjacent mesendoderm.There remains much controversy re-garding which group and what modeof gastrulation is ancestral. Gastrula-tion in Myriapoda is not well charac-terized and appears to occur by ingres-sion and proliferative mitosis from ablastodisc (Heymons, 1901, as cited inAnderson, 1973).

Finally, a comparison with somespecies of the phylum Onychophora isinteresting because of its phylogeneticposition. Onychophorans are gener-ally considered a closely related out-group to the arthropods (Giribet et al.,2001; Brusca and Brusca, 2002). Wewill focus on the development of ony-chophoran eggs with significantamounts of yolk, as these are thoughtto be the least derived (Anderson,1973). Yolky embryos gastrulate in apattern that appears somewhat simi-lar to the chelicerates; they form ablastoderm and undergo gastrulationby means of a linear blastoporal re-gion on the ventral side (Anderson,1966, 1973). An important differenceis that the majority of the gastrulatingcells (thought to be prospective mid-gut) are organized into two bilaterallysymmetric groups that appear to in-ternalize by means of proliferative mi-tosis or involution or both. Also, theposition of the prospective mesodermis thought to be posterior to the pro-spective midgut. From these compari-sons, it seems that the chelicerategastrulation pattern, namely inter-nalization of spatially continuous me-

SPIDER GASTRULATION 3493

soderm and endoderm through a ven-tral groove, is not ancestral to thearthropods, because such spatial con-tinuity of mesoderm and endoderm isnot typical of yolky onychophorans ormost arthropods. It should be noted,however, that the precise position ofthe germ layers relative to the ventralgroove is uncertain in the Onych-ophora, and in virtually all arthro-pods, because most maps are interpre-tations of fixed and sectionedspecimens. Even Holm’s (1952) mapfor the spider Agelena labyrinthicadoes not positively distinguish be-tween mesoderm and endoderm.

It has become increasingly clearthat early embryonic development isflexible, with considerable variationobserved even within a single genus inyolk content, mechanism of embryonicaxis determination, and mode of gas-trulation. For example, direct develop-ment has evolved multiple times inthe echinoderms and the amphibians(Wilkins, 2002). In some cases, the cellmovements of gastrulation have beenradically altered to accommodate thederived developmental pattern. In thesea urchin Heliocidaris tuberculata,which has a planktotrophic larva, gas-trulation follows the pattern typicalfor indirect-developing sea urchins:the archenteron forms initially by in-vagination and limited involution,and then elongates by way of conver-gent–extension cell rearrangement(Wray and Raff, 1991). Heliocidariserythrogramma has a much larger eggthat develops into a lecithotrophiclarva. In this species, archenteronelongation results primarily from in-volution of sheets of cells (Wray andRaff, 1991). Another example comesfrom eggs of the frog Gastrotheca rio-bambae, which develop into feedingtadpoles, but do so in an unusual man-ner. The embryos form the blastoporenear the vegetal pole, and involutionof mesoderm and endoderm proceedslocally and symmetrically. This pro-cess results in formation of an embry-onic disc that will form the body andthat is attached to a yolk sac that de-velops from the remainder of the em-bryo (del Pino and Elinson, 1983).Other frogs with similar-sized eggstypically do not form an embryonicdisc (Keller and Shook, 2004). Earlyspider development also appears to berather flexible; a variety of patterns is

described in the classic literature, andgastrulation in A. tepidariorum(Yamazaki et al., 2005; Oda et al.,2007) and Zygiella x-notata appears todiffer in fundamental ways. Exploringthe variations of spider gastrulationat both the cellular and molecular lev-els will contribute to an understand-ing of the evolution of these differentstrategies.

EXPERIMENTALPROCEDURES

Collection of Embryos

Adults of Zygiella x-notata Clerk(Aranaeidae) and their egg sacs werecollected locally. Egg sacs contained20–80 embryos that developed syn-chronously. Embryos varied in colorfrom golden-brown (collected in thelaboratory) to lavender to dark purple(collected in the field). After collection,silk was removed and embryos werevisualized by immersing a few in min-eral oil (Sigma). Embryos were thenkept in Petri dishes at room tempera-ture (�23°C). The rate of developmentcould be slowed by keeping embryos atcooler temperatures, although earlystages did not tolerate long periods at�14°C.

Time-Lapse Imaging

Embryos were placed in mineral oil ordechorionated and embedded in 3%gelatin (300 Bloom, Sigma) in 2 mMHEPES with 50 �g/ml Kanamycin, pH7.2. Embryos were imaged withOptronics 750-line video cameras anddigitized by means of Canopus cap-ture boards. Typically, one frame wascaptured every 5 min. Movies weremade of MPEG-4–compressed files byBTV Pro software running on Macin-tosh computers. Cells were traced byfollowing their position frame-by-frame in videos, then exporting imagefiles to Photoshop and false-coloringtheir position at standard time inter-vals. Movies were made of the false-colored files with BTV Pro software.

Fixation

A fixative was developed that providesexcellent results for histology (theheptane-formaldehyde technique gaveinferior results in Z. x-notata). Em-bryos were dechorionated in 50%

household bleach for 15 min. Afterrinsing in DI water, embryos werefixed in Ilsa (58% methanol, 17% chlo-roform, 17% DMSO, 8% acetic acidmade fresh each time) for 20–30 minor until embryos were a very pale or-ange or white. Embryos were thenstepped into 100% methanol for stor-age at �20°C or immediately post-fixed. Fixed embryos were manuallydevitellined with sharpened Dumont#5 forceps and rehydrated into phos-phate buffered saline (PBS) through astepwise methanol series (10 min/step). Embryos were post-fixed 10 minin 2% paraformaldehyde, 5% DMSO,and then washed 3 � 10 min in PBSwith 5% DMSO. Embryos were thenstepped into methanol for histology orstorage at �20°C.

Embedding and Sectioning

Post-fixed, dehydrated embryos wereprestained to facilitate embedding.They were stained with dilute Eos-in–Phloxine (3 drops 0.1% stock in10 ml of methanol) until they werehot pink with distinct cell bound-aries (45–120 min, depending onstage). They were then cleared in tol-uene and passed through threechanges of molten paraffin. Afterembedding, blocks were sectioned at10 �m until tissue was encountered,and then soaked overnight in 5%glycerol, tissue side down. Aftersoaking, blocks were wiped dry andsectioned at 6 –7 �m. Sections weremounted on a degassed solution ofMayer’s albumen (Carolina), stainedwith Delafield’s Hematoxylin andEosin B–Phloxine B, and cover-slipped with Permount.

ACKNOWLEDGMENTSWe thank David Rasmussen forsharing unpublished results and forinteresting discussions. We alsothank Drs. Greta Binford, PaulaCushing, and Susan Masta not onlyfor identifying spiders and clarifyingphylogenetic relationships, but alsofor their enthusiasm, comments, andsupport. This research was sup-ported by a Mellon grant to ReedCollege, by a Betty C. Liu MemorialBiology Research Fellowship (toE.V.), and HHMI student researchgrants.

3494 CHAW ET AL.

REFERENCES

Abd-el Wahhab A. 1954. The formation ofthe germ layers and their embryonic cov-erings in the scorpion, Buthus quinques-triatus (H.E.). Proc Egypt Acad Sci 10:110–118.

Akiyama-Oda Y, Oda H. 2003. Early pat-terning of the spider embryo: a clusterof mesenchymal cells at the cumulusproduces dpp signals by germ disc epi-thelial cells. Development 104:391–402.

Akiyama-Oda Y, Oda H. 2006. Axis spec-ification in the spider embryo: dpp isrequired for radial-to-axial symmetrytransformation and sog for ventral pat-terning. Development 133:2347–2352.

Anderson D. 1966. The comparative earlyembryology of the Oligochaeta, Hirudi-nea and Onychophora. Proc Linn SocN S W 91:10–43.

Anderson D. 1973. Embryology and phy-logeny in annelids and arthropods. NewYork: Pergamon. 495 p.

Ayoub N, Garb J, Hedin M, Hayashi C.2007. Utility of the nuclear protein-cod-ing gene, elongation factor-1 gamma(EF-1�), for spider systematics, empha-sizing family level relationships of taran-tulas and their kin (Araneae: Mygalo-morphae). Mol Phylogen Evol 42:394–409.

Brusca R, Brusca G. 2002. Invertebrates.2nd ed. Sunderland, MA: Sinauer. 936 p.

Crome W. 1963. Embryonalentwicklungohne “Umrollung” (�Reversion) bei Vo-gelspinnen. Dtsch Entomol Zeitschrift 10:83–95.

Davis G, Patel N. 2002. Short, long, andbeyond: molecular and embryological ap-proaches to insect segmentation. AnnuRev Entomol 47:669–699.

de Rosa R, Prudhomme B, Balavoine G.2005. caudal and even-skipped in the an-nelid Platynereis dumerili and the ances-try of posterior growth. Evol Dev 7:574–587.

del Pino E, Elinson R. 1983. A novel devel-opment pattern for frogs: gastrulationproduces an embryonic disk. Nature 306:589–591.

Ehn A. 1963. The embryonic developmentof the spider Torania variata Poc.(Sparassidae). Zool Bidr Uppsala 36:37–48.

Foelix R. 1996. Biology of spiders. 2nd ed.New York: Oxford Thieme. 330 p.

Gerberding M, Patel N. 2004. Gastrulationin crustaceans. In: Stern C, editor. Gas-trulation. New York: Cold Spring HarborLaboratory Press. p 79–89.

Giribet G, Edgecombe G, Wheeler W. 2001.Arthropod phylogeny based on eight mo-lecular loci and morphology. Nature 413:157–161.

Holm A. 1940. Studien uber die Entwick-lung und Entwicklungsbiologie der Spin-nen. Zool Bidr Uppsala 19:1–215.

Holm A. 1947. On the development ofOpilio parietinus Deg. Zool Bidr Uppsala29:409–422.

Holm A. 1952. Experimentelle Untersu-chungen uber die Entwicklung und En-twicklungsphysiologie des Spinnenem-bryos. Zool Bidrag Uppsala 25:293–424.

Holm A. 1954. Notes on the development ofan orthognath spider, Ischnothele kars-chi Bos. & Lenz. Zool Bidr Uppsala 30:199–221.

Keller R, Shook D. 2004. Gastrulation inamphibians. In: Stern C, editor. Gastru-lation. New York: Cold Spring HarborLaboratory Press. p 171–203.

Kiehart D, Edwards K, Rickoll W, Mon-tague R. 2000. Multiple forces contributeto cell sheet morphogenesis for dorsalclosure in Drosophila. J Cell Biol 149:471–490.

Montgomery T. 1909. The development ofTheridium, an araneid, up to the stage ofreversion. J Morphol 20:297–352.

Oda H, Nishimura O, Hirao Y, Tarui H,Agata K, Akiyama-Oda Y. 2007. Progres-sive activation of Delta-Notch signalingfrom around the blastopore is required toset up a functional caudal lobe in thespider Achaearanea tepidariorum. De-velopment 134:2195–2205.

Rempel J. 1957. The embryology of theblack widow spider, Latrodectus mac-tans. Can J Zool 35:35–74.

Roth S. 2004. Gastrulation in other insects.In: Stern C, editor. Gastrulation. New

York: Cold Spring Harbor LaboratoryPress. p 105–121.

Saaristo M. 2006. Theridiid or cobweb spi-ders of the granitic Seychelles islands(Araneae, Theridiidae). Phelsuma 14:49–89.

Schimkewitsch L, Schimkewitsch W. 1911.Ein Betrag zur Entwicklungsgeschichteder Tetrapneumones. Bull Acad Imp SciSt. Petersbourg 8:637–654.

Schoppmeier M, Damen W. 2005. Expres-sion of Pax group III genes suggests asingle-segmental periodicity for opis-thosomal segment patterning in thespider Cupiennius salei. Evol Dev 7:160 –169.

Seitz K-A. 1966. Normale Entwicklungdes Arachniden-Embryos Cupienniussalei Keyserling und seine Regula-tionsbefahigung nach Rontgenbe-strahlungen. Zool Jahrb Anat Ontog-enie 83:327–447.

Sekiguchi K, Yamamichi Y, Seshimo H,Sugita H. 1988. Normal development.In: Sekiguchi K, editor. Biology ofhorseshoe crabs. Tokyo: Science House.p 133–224.

Stollewerk A, Schoppmeier M, Damen W.2003. Involvement of Notch and Deltagenesinspidersegmentation.Nature423:863–865.

Wilkins A. 2002. The evolution of develop-mental pathways. Sunderland, MA: Si-nauer. 603 p.

Wray G, Raff R. 1991. Rapid evolution ofgastrulation mechanisms in a sea urchinwith lecithotrophic larvae. Evolution 45:1741–1750.

Yamakazi K, Akiyama-Oda Y, Oda H.2005. Expression Patterns of a twist-re-lated gene in embryos of the spiderAchaearanea tepidariorum reveal diver-gent aspects of mesoderm developmentin the fly and spider. Zool Sci 22:177–185.

Yoshikura M. 1955. Embryological studieson the liphistiid spider Heptathela kimu-rai. Kumamoto J Sci Ser B 2:1–86.

Yoshikura M. 1958. On the developmentof the purse-web spider, Atypus kars-chi Donitz. Kumamoto J Sci Ser B 3:73–85.

SPIDER GASTRULATION 3495