Development 103, 317-324 (1988)Printed in Great Britain © The Company of Biologists Limited 1988

317

The role of secondary mesenchyme cells during sea urchin gastrulation

studied by laser ablation

JEFF HARDIN*

Biophysics Group and Department of Zoology, University of California, Berkeley, CA 94720, USA

* Current address: Department of Zoology, Duke University, Durham, NC 27706, USA

Summary

It has long been thought that traction exerted byfilopodia of secondary mesenchyme cells (SMCs) is asufficient mechanism to account for elongation of thearchenteron during sea urchin gastrulation. The filo-podial traction hypothesis has been directly tested hereby laser ablation of SMCs in gastrulae of the seaurchin, Lytechinus pictus. When SMCs are ablated atthe onset of secondary in vagina t ion, the archenterondoubles in length at the normal rate of elongation, butadvance of the tip of the archenteron stops at the §gastrula stage. In contrast, when all SMCs are ablatedat or following the § gastrula stage, further elongationdoes not occur. However, if a few SMCs are allowed toremain in §-f gastrulae, elongation continues,

although more slowly than in controls. The final lengthof archenterons in embryos ablated at the i-i gastrulastage is virtually identical to the final length of evertedarchenterons in LiCl-induced exogastrulae; since filo-podia] traction is not exerted in either case, analternate, common mechanism of elongation probablyoperates in both cases. These results suggest thatarchenteron elongation involves two processes: (1)active, filopodia-independent elongation, whichdepends on active cell rearrangement and (2) filo-podia-dependent elongation, which depends on mech-anical tension exerted by the filopodia.

Key words: filopodia, sea urchin, gastrulation, secondarymesenchyme, laser ablation, Lytechinus pictus.

Introduction

The migration of mesenchymal cells and the invagi-nation of epithelial sheets are ubiquitous phenomenaduring gastrulation (Gustafson & Wolpert, 1963,1967; Holtfreter, 1943, 1944). For 30 years, the seaurchin embryo has been a useful system for studyingthe interaction between these two fundamental cellbehaviours. Gastrulation in the sea urchin embryooccurs in two phases (Dan & Okazaki, 1956; Gustaf-son & Kinnander, 1956). Primary invagination in-volves the inward bending of the vegetal plate to formthe archenteron, which is roughly cylindrical in shapeand extends one third of the way across the blasto-coel. During secondary invagination the archenteronelongates and this phase is marked by the extensionof long filopodial protrusions towards the animalregion of the embryo by secondary mesenchyme cells(SMCs) at the tip of the gut rudiment.

In 1956, two landmark studies independentlysuggested that traction exerted by filopodia could

account for secondary invagination. Dan & Okazaki(1956) found a correlation between the presence andproper extension of filopodia and the completion ofgastrulation under a variety of circumstances. Treat-ments or agents that disrupt filopodia or their contactwith the animal ectoderm, including pancreatin, os-motic swelling and low calcium, are correlated withfailure of the archenteron to elongate fully. In ad-dition, the lateral margin of the tip of the archenteronappears to be under tension, reminiscent of a sus-pended fabric (Dan & Okazaki, 1956). Gustafson &Kinnander (1956) noticed a reproducible pause in theprogress of invagination prior to secondary invagi-nation, followed by a resumption of invagination at adistinctly more-rapid pace after the onset of filopodialactivity. The lateral wall of the archenteron, incontrast, exhibits little, if any, obvious protrusiveactivity (Kinnander & Gustafson, 1960). This lack ofprotrusive activity led Gustafson & Kinnander toconclude that filopodial traction is the most-import-ant mechanism for elongating the gut rudiment.

318 J. Hardin

These ideas have been discussed in two reviews byGustafson & Wolpert (1963, 1967).

The simplicity and economy of the filopodial trac-tion hypothesis, coupled with the strong circumstan-tial evidence supporting it, have made it the acceptedmodel of secondary invagination. However, severalquestions can be raised regarding its adequacy. First,the evidence supporting it is largely based on obser-vation and it has proven difficult to devise an exper-imental treatment that can be demonstrated to affectSMCs specifically (reviewed in Hardin & Cheng,1986). Second, the mechanical consequences of filo-podial traction are unclear; mechanical simulationssuggest that filopodial pulling alone cannot accountfor the shape of the embryo at the end of secondaryinvagination (Hardin & Cheng, 1986). Third, cellrearrangement has been demonstrated to occur in thewall of the gut rudiment as it elongates (Ettensohn,1985; Hardin & Cheng, 1986); whether this cellrearrangement is normally due to filopodial traction,to active forces within the archenteron or to acombination of the two is not known. Finally, thefindings that autonomous thinning of the archenteronwall (Dan & Okazaki, 1956; Dan & Inaba, 1968) andepithelial cell rearrangement (Hardin & Cheng, 1986)occur in exogastrulae strongly suggest that in this casearchenteron elongation involves an active componentwhich is independent of filopodial pulling, since thefilopodia in exogastrulae cannot exert traction. Inview of these facts, the relationship between filo-podial traction and cell rearrangement should beclarified.

Laser microsurgery (reviewed by Berns et al. 1981)permits selective alteration or destruction of regionsof a cell or embryo without disruption of neighbour-ing tissues. These advantages are afforded by acombination of high energy intensity at the site of thefocused laser spot, combined with rapid attentuationof energy intensity away from the focal plane of thespot (Berns et al. 1981). The transparency of the seaurchin embryo makes it a good candidate for theapplication of the laser microbeam and this techniquenow allows the role of filopodial traction in archen-teron elongation to be examined directly for the firsttime through laser ablation of SMCs and their filo-podia. The results indicate that significant, auton-omous extension of the archenteron can occur with-out filopodial traction, although the filopodia have animportant role during the final one third of gastru-lation.

Materials and methods

Procurement and measurement of embryosGametes of L. pictus (Marinus, Westchester, CA) were

obtained, fertilized and cultured as described by Hardin &Cheng (1986). Exogastrulae were produced using LiCI asdescribed by Hardin & Cheng (1986). 'Archenteron length'was defined as the distance from the vegetal edge of theblastopore lip to the middle of the archenteron roof, asmeasured from differential interference contrast (DIC)micrographs or video images. Some micrographs wereobtained by photographing the video screen using a Polar-oid CU-5 Land camera and Polaroid 611 positive film.

Laser ablation of secondary mesench'yme cellsLaser ablations were performed at the Beckman LaserInstitute of the University of California, Irvine, CA. Thebasic apparatus and techniques involved are described inBerns et al. (1981). The chambers used for irradiatingembryos were constructed using poly-L-lysine-coated cover-slips mounted on microscope slides as described in Hardin& Cheng (1986). Irradiations were performed using a ZeissAxiomat inverted microscope equipped with DIC opticsand infrared and orange glass thermal filters connected to aDage/MTI Newicon videocamera and a Sony three-quarterinch time-lapse video recorder. The second harmonicwavelength (532 nm) of a Quantel neodymium-YAG 481laser was used to irradiate the specimens through a x25DIC objective with individual pulses of 10-15 nanoseconds,with an average energy per pulse of 8-4 mj and an attenua-tion of 15-17 dB. Embryos were irradiated by directing oneto several pulses at either the filopodia or cell bodies ofvisible secondary mesenchyme cells. Irradiated embryoswere then followed individually, using time-lapse videomic-roscopy or sequential 35 mm still micrography and visuallyinspected at 15 min intervals. Embryos showing evidence ofadditional incipient filopodia were reirradiated to ensurethat all secondary' mesenchyme cells and their filopodiawere ablated.

Results

Microsurgery using the laser microbeamSecondary mesenchyme cells or their filopodia can beablated with the laser microbeam by focusing atvarious optical planes of section with a x25 DICobjective. Unfortunately, for several reasons it wasnot technically possible to ablate all filopodia (asopposed to the cell bodies of SMCs) in a singleembryo simultaneously. The diameter of the embryo(~120jim) requires the use of an objective thatprovides substantial depth of field, resulting in acompromise in magnification and optical resolutionwhich in turn limits the spatial resolution of the laser.Moreover, the sheer number of filopodia and aninability to visualize all of them with total confidencemakes their complete ablation difficult. Finally, be-cause filopodia are continually extended and retrac-ted by SMCs throughout secondary invagination, it isnecessary to ablate not only individual filopodia, butthe cell bodies of SMCs as well, if the concertedaction of many filopodia over time is to be prevented.

Filopodial traction and gastrulation 319

Indeed, SMCs in which individual filopodia wereablated invariably extended new filopodia within1-3 min after the ablation (data not shown). There-fore most experiments involved the irradiation of thecell bodies of SMCs. The archenterons of controlembryos in which other cell populations were ir-radiated (e.g. primary mesenchyme cells, lateralectoderm) displayed no detectable alterations inelongation compared to unirradiated controls.

Cells often lyse upon irradiation; ablated SMCssometimes liberate microplasts, which move throughthe blastocoel. If the cell does not lyse, it retracts orceases extension of its protrusions. In both cases, it isapparent when an irradiated SMC has suffered per-manent damage; inactivated or lysed cells are ident-ified in real time by their damaged cell membrane andcellular debris, or by their subsequent quiescence andcessation of motility within 1-2 min. The results oflaser ablation are even more dramatic when viewed intime lapse; the rapid 'boiling' protrusive activity ofthe basal surfaces of SMCs prior to irradiation ab-ruptly ceases following irradiation with the micro-beam. Because the SMCs are clustered together soclosely at the tip of the archenteron, it is not alwaysimmediately clear that all SMCs have been ablatedfollowing the first series of laser pulses. Therefore itwas impossible to irradiate large numbers of embryosand to score them later for their ability to gastrulate ina batch-wise fashion. Instead, periodic inspections ofindividual irradiated embryos were performed at10-15 min intervals; if new filopodia appeared, theprotrusive cells were ablated. The experimentsreported below represent time-lapse and real-timeobservations of five to ten individual embryos overthe period of interest for each experimental series.

Mechanical consequences of ablation of filopodiaIf an individual filopodium is severed, the base of thefilopodium may either be resorbed or remain ex-tended but inactive. In either case, little retraction ofthe archenteron tip occurs (Fig. 1A-C). When indi-vidual SMCs are ablated at any stage of elongation, alocal retraction (~ 1 /zm) of a small region of the tip ofthe archenteron usually occurs (Fig. 3A,B), but littleor no retraction of the archenteron as a whole takesplace. When a group of SMCs are ablated simul-taneously, a small, immediate retraction of the tip ofthe archenteron occurs. However, even when theentire tip of the gut rudiment is rapidly irradiated, theamount of retraction is ^4/im. Archenterons ablatedat the J-i gastrula stage retract l-7±0-3/zm (mean±S.E.M.; n = 6), while those ablated at the §-$gastrula stage retract 3-0 ± 0-6/xm {n = 3), which is atbest a slight difference ( O K P<0-05). After com-plete destruction of SMCs at the §-i gastrula stage,the archenterons of 20 % of the embryos show a small

degree of lateral buckling near their bases (Fig. ID),suggesting that some axial tension is present in the gutrudiment at this time. However, such buckling wasnot observed when the ablation was performed priorto the § gastrula stage. If the filopodia on theuppermost portion of the archenteron tip are ablatedat any time during secondary invagination, thearchenteron is deflected towards the remaining lat-erally located filopodia, with the direction dependingon which filopodia are left intact (Fig. IE). This resultprovides experimental evidence that lateral move-ment of the tip of the gut rudiment can be guided byan asymmetric distribution of filopodia.

Ablation of filopodia at the onset of archenteronelongationWhen individual embryos are followed using time-lapse videomicroscopy after laser ablation of SMCs, aconsistent pattern emerges. If the tip of the archen-teron is irradiated at the 4 gastrula stage, the archen-teron elongates until it reaches two thirds of the wayacross the blastocoel (diamonds, Fig. 2). Within thefirst hour after ablation, the rate of advance ofirradiated archenterons is identical to controls(~0-4^mmin~'), but slows asymptotically, finallyapproaching zero after 2-2^h (Fig. 2). In contrast,control embryos continue elongating at a constantrate of ~0-4^mmin~1 (dotted line, Fig. 2). Thus theelongation of archenterons ablated at the % gastrulastage occurs despite the absence of all motile SMCsand represents up to a 100 % increase in the length ofthe archenteron compared to its length at the end ofprimary invagination. In time-lapse studies, archen-teron elongation occurs without any obvious pulsatileor protrusive activity on the part of endoderm cellsafter ablation of SMCs, suggesting that pulsatileactivity may be a misleading guide as to which cells inthe archenteron are actively involved in morphogen-etic movements and suggesting that a thorough reexa-mination of motile activity in the archenteron duringits elongation is in order.

In addition to changes in the length of ablatedarchenterons, their blastopores close by 10-20%,depending on the initial length of the gut rudiment, toa final diameter of 33-0 ± 0-4 j/m (mean ± S.E.M.,n = 4). Similarly, the walls of the gut rudiment thin by15-40 %, to a final thickness of 4-4 ± 0-4/zm. In bothcases, the final dimensions agree closely with pre-viously determined values for normal archenterons ata comparable stage of elongation (Hardin & Cheng,1986; Hardin, 1986). The contours of the archenteronoften change as well, from slightly dome-shaped ornearly cylindrical with straight walls to a shape inwhich the central region of the gut rudiment is slightlynarrower than the base or tip regions (Fig. 3). In 35 %of the cases, the tip of the gut rudiment bends to one

320 J. Hardin

r \

if

Fig. 1. Immediate effects of laser ablation of filopodia. (A) Amidgastrula following ablation of a single filopodium (small arrow).An intact filopodium is also visible (large arrow). (B) The sameembryo following ablation of the intact filopodium. Littleretraction of the archenteron occurs. (C) The same embryofollowing ablation of all SMCs. (D) \ gastrula following rapidablation of all SMCs. The original outline of the archenteron tip isshown as a dotted line and the outline after ablation is indicated bya solid line. A slight retraction of the archenteron occurs,accompanied by buckling in the thinnest region of the gut rudiment(arrows). (E) A i gastrula in which all secondary mesenchyme cellswere ablated except for a single prominent cell (arrow). Note thedeflection of the archenteron tip in the direction of the active cell.Bar, 20fim.

side slightly and the archenteron tip, which appears tobe somewhat sticky, makes contact with the ectoder-mal epithelium somewhere in the animal-most onethird of the blastocoel wall. In other cases, thearchenteron simply stops without significant deflec-

tion. Other developmental processes appear unaffec-ted by this treatment; primary mesenchyme cellscontinue aggregation, syncytium formation and theearly phases of spicule formation in irradiated em-bryos in synchrony with unirradiated controls in the

100

90

80

70

60

50

40

Control

Late,

three remaining

Late, all ablated

iEarly, all ablated

0 50 100 150Time after onset of elongation (min)

200

Fig. 2. Advance of the tip of the archenteron over timein ablated and normal embryos. Archenteron length (urn)was measured from time-lapse video images and plottedagainst time after onset of secondary invagination (min).Corresponding stages of elongation are shownschematically for reference. Time points for archenteronsablated at the J - | gastrula stage (mean ± S.E.M. bothaxes; n = 4 except for the final time point, where n = 2)are shown (diamonds). A representative control embryofollowed from the 4 gastrula stage onward is provided forcomparison (dotted line). Two records of archenteronsablated at the f gastrula stage are also shown (squares).One embryo (filled squares) retained three active SMCs,while the other had no intact filopodia (open squares).

same culture chamber (Fig. 3). Unirradiated controlsin the same chamber complete gastrulation normally(data not shown).

Ablation of filopodia in late gastrulaeWhen SMCs are ablated after the tip of the archen-teron has advanced two thirds of the way across theblastocoel, no further extension of the gut rudimentoccurs (open square, Fig. 2). In all cases, advanceceases immediately upon complete ablation of allSMCs. If a few (=s3) SMCs are left intact, advance ofthe archenteron continues, albeit more slowly thanusual (Figs 2, 4A,B). The rate of advance appears tobe proportional to the number of remaining SMCs,and is significantly slower than in unirradiated con-trols. For example, in an embryo in which three intactSMCs remained, the rate of advance of the archen-teron tip was ~0-15|immin~1 (filled square, Fig. 2),as compared to the control rate of —0-4nm min"1.This result again indicates that general cytotoxiceffects of the ablation procedure are not significant,since SMCs immediately adjacent to ablated cellsshow completely normal protrusive activity. More-over, it suggests that the traction of at least a fewfilopodia is necessary for continued elongation of thegut rudiment at or beyond the § gastrula stage.

Filopodial traction and gastrulation 321

V

Fig. 3. Results of ablation of secondary mesenchymecells at the onset of archenteron elongation (A). Videoimage of i gastrula just prior to ablation. Note theprominent secondary mesenchyme cell (arrow). (B)Immediately after ablation. The remnants of the SMC arestill visible (arrow). (C) 2h 17min after ablation.Significant elongation has occurred. Note the normalaggregation of primary mesenchyme cells intoventrolateral clusters (large arrow). Remnants of severedfilopodia ablated after the first series of laser pulses arestill visible (small arrows). Bar, 20//m.

322 J. Hardin

treatment

Fig. 4. Results of ablation of secondary mesenchyme cells at the § gastrula stage. (A) Video image of a § gastrula justprior to ablation. Many secondary mesenchyme cells are visible (arrows). (B) The same embryo following ablation of allbut three secondary mesenchyme cells. Little retraction of the archenteron has occurred. (C) The same embryo 45 minafter ablation; the archenteron has advanced noticeably. Protrusive cells are marked by arrows. Bar, 20fim.

average final length is 67-8 ± 1-5/im (mean±s.D.);for ablated archenterons the average length is68-5±2-8/xm, which is not a significant difference(P> 0-5). The final average length of the archenteronin both cases is 73-74 % of the final length of normalgastrulae (Fig. 5).

Discussion

Archenteron elongation involves two processesThe results presented here are the first direct test ofthe filopodial traction hypothesis of gastrulation inthe sea urchin embryo. The results of laser ablation ofsecondary mesenchyme cells (SMCs) at the tip of thegut rudiment suggest that secondary invaginationitself actually involves two processes: active extensionand filopodia-dependent elongation. The first processis capable of extending the archenteron to two thirdsof its final length and is independent of filopodia, asevidenced by the ability of archenterons to double inlength when SMCs are ablated at the i gastrula stage.This lengthening presumably involves autonomouscell rearrangement in the archenteron wall, sincesecondary invagination is accompanied by cell re-arrangement at all stages (Ettensohn, 1985; Hardin &Cheng, 1986).

The extension accomplished by embryos ablated atthe J-i gastrula stage corresponds very closely to theextent of elongation achieved by LiCl-induced exo-gastrulae. In the case of exogastrulae, the walls of thearchenteron thin, the 'blastopore' region decreases incircumference, and the archenteron lengthens to twothirds of the length of normal gastrulae by active cellrearrangement (Hardin & Cheng, 1986). The striking

Fig. 5. Comparison of final lengths (mean ± S.D.) ofexogastrulated archenterons (n = 6), archenterons ablatedat the i - i gastrula stage (n = 4), and normal archenterons

Comparison of ablated embryos and exogastrulaeIt has been noted by several workers that the archen-terons of exogastrulated embryos are intermediate inlength between those of normal embryos at the onsetand end of secondary invagination (Dan & Okazaki,1956; Horstadius, 1973; Ettensohn, 1985). In ad-dition, the gut rudiments of exogastrulae undergoautonomous cell rearrangement as they elongate,though to a lesser extent than normal embryos(Hardin & Cheng, 1986). Therefore it is of interest tocompare the final lengths of exogastrulated archen-terons and gut rudiments ablated at the onset ofsecondary invagination, since in both types of em-bryos elongation must result from processes indepen-dent of filopodial traction. As shown in Fig. 5, thefinal lengths of archenterons in the two types ofembryos are virtually identical. For exogastrulae, the

Filopodial traction and gastrulation 323

similarities between these two cases suggest a com-mon mechanism: a filopodia-independent extensionresulting in sufficient cell rearrangement and cellshape changes to achieve a doubling of archenteronlength. In bothcases, filopodia cannot extend prop-erly following this partial extension and furtherlengthening of the gut rudiment is prevented. Ad-ditional evidence for autonomous extension duringsecondary invagination comes from observations offilopodial activity in Lytechinus variegatus gastrulae.In this species, filopodia are directed either laterallyor even vegetally until the § gastrula stage, butextension and thinning of the archenteron nonethe-less occur (Trinkaus, 1984, pp. 438-439; Morrill &Santos, 1985). Since the upward (animalward) forcecomponent due to filopodial traction would appear tobe negligible in these embryos until the § gastrulastage, autonomous extension could account for theinitial elongation.

Mechanisms of action of secondary mesenchymefilopodiaSuccessful completion of gastrulation requires intactfilopodia, since ablation of all SMCs after the §gastrula stage results in cessation of further exten-sion, whereas the presence of a small number ofactive SMCs allows elongation to continue. Theexperiments presented here do not address the pre-cise mechanism of action of the filopodia. The most-likely possibility is that filopodial traction is necessaryto achieve the final elongation of the archenteron bystretching the gut rudiment. Several lines of evidencesuggest that filopodial traction may be more substan-tial late in gastrulation. The lateral buckling observedin some §-J gastrulae following ablation may reflectincreased axial stress within the gut rudiment late ingastrulation. In L. pictus, there is a transient rise inthe length/width ratio along the animal-vegetal axisof cells in the archenteron at the f gastrula stage(Hardin, 1986), again suggesting that considerabletension is being exerted on the gut rudiment at thistime. The reversion to a more isodiametric shape bythe end of gastrulation (Hardin & Cheng, 1986;Hardin, 1986) suggests that the archenteron respondsto this tension by cell rearrangement, thereby reliev-ing stress generated at the I gastrula stage. Alterna-tively, SMCs may transmit some sort of signal to thecells of the archenteron wall which they require tocontinue active extension beyond the § gastrula stage.This signal could presumably be mediated by mech-anical, chemical, cell contact or other cues. A finalpossibility that cannot be ruled out is that by ablatingcells at the archenteron tip, enough cells weredamaged so that the animalmost region of the archen-teron wall was unable to participate in cell rearrange-ment late in gastrulation. Further experiments are

needed to distinguish between these possibilities,although prima facie a direct role for mechanicaltension seems to be most plausible.

Whatever the mechanism is by which SMCs facili-tate elongation of the gut rudiment, individual filo-podia do not contribute significantly to the steady-state extension of the archenteron. The absence ofappreciable retraction of the archenteron tip follow-ing local ablation of SMCs or their filopodia, or evenfollowing ablation of all active SMCs at once, stronglysuggests that the deformation of the archenteron ispermanent at intermediate stages of elongation. Thisis perhaps not surprising in light of the fact that cellrearrangement occurs at all stages of elongation(Hardin & Cheng, 1986).

The role of filopodial traction in guiding andstabilizing the archenteronIn addition to their role in elongating the archen-teron, the filopodia are probably important for sev-eral other reasons. First, it is likely that filopodialtraction lends lateral stability to the elongating gutrudiment. In ablated embryos, the tip of the archen-teron sometimes leans to one side, or buckles slightly,suggesting that filopodial tension normally helps tokeep it erect. Second, filopodia appear to guide thearchenteron to the site of the oral primordium. Theobservations of Gustafson & Kinnander (1960) pro-vide strong evidence for such a guidance mechanism.In Psammechinus miliaris, once the archenteron hasspanned the blastocoel, filopodia are sent out lat-erally to the site of the stomodeal invagination,apparently pulling the tip of the gut rudiment to thecorrect site for fusion with the oral ectoderm. Basedon the ablation studies presented here, filopodia canguide the archenteron from side to side if filopodialtension is asymmetrically distributed around thearchenteron; in the normal embryo, such asymmetrycould be provided by the preferential attachment offilopodia to specific adhesion molecules expressed inthe vicinity of the oral ectoderm (Gustafson &Wolpert, 1963; McClay & Ettensohn, 1987). How-ever, further experiments designed to determine thespecific molecular basis of filopodial attachment andguidance are needed if the role of filopodia in seaurchin gastrulation is to be more thoroughly under-stood.

I thank Dr Saul Zackson for suggesting the use of highenergy lasers to study sea urchin gastrulation and Dr BruceNicklas for the use of his Polaroid camera and three-quarterinch video recorder. The tremendous helpfulness of thestaff of the Beckman Laser Institute was greatly appreci-ated, especially that of Jeff Andrews, Glenn Profeta andTao Wen. I am especially grateful to Drs Ray Keller andFred Wilt for their constant encouragement during mygraduate studies. Discussions with Drs Steve Black and

324 J. Hardin

Dave McClay concerning the manuscript were extremelyhelpful. This work was supported by a NSF predoctoralfellowship and by NIH grants HD14503 to F. Wilt andHD18979 to R. Keller.

References

BERNS, M. W., AIST, J., EDWARDS, J., STRAHS, K.,GIRTON, J., MCNEILL, P., RATTNER, J. B., KITZES, M.,HAMMER-WILSON, M., LIAW, L.-H., SIEMENS, A.,KOONCE, M., PETERSON, S., BRENNER, S., BURT, J.,WALTER, R., BRYANT, P. J., VAN DYK, D., COULUMBE,J., CAHILL, T. & BERNS, G. S. (1981). Lasermicrosurgery in cell and developmental biology.Science 213, 505-513.

DAN, K. & INABA, D. (1968). Echinoderma. InInvertebrate Embryology (ed. M. Kume & K. Dan),pp. 280-332. Belgrade: NOLIT Publishing House.

DAN, K. & OKAZAKI, K. (1956). Cyto-embryologicalstudies of sea urchins. III. Role of secondarymesenchyme cells in the formation of the primitive gutin sea urchin larvae. Biol. Bull. mar. biol. Lab., WoodsHole 110, 29-42.

ETTENSOHN, C. A. (1985). Gastrulation in the sea urchinembryo is accompanied by the rearrangement ofinvaginating epithelial cells. Devi Biol. 112, 383-390.

GUSTAFSON, T. & KINNANDER, H. (1956). Microaquariafor time-lapse cinematographic studies ofmorphogenesis of swimming larvae and observations ongastrulation. Expl Cell Res. 11, 36-57.

GUSTAFSON, T. & WOLPERT, L. (1963). The cellular basisof morphogenesis and sea urchin development. Int.Rev. Cyt. 15, 139-214.

GUSTAFSON, T. & WOLPERT, L. (1967). Cellular

movement and cell contact in sea urchinmorphogenesis. Biol. Rev. Camb. Phil. Soc. 42,442-498.

HARDIN, J. D. (1986). Active epithelial cellrearrangement in the archenteron: a new mechanismfor driving sea urchin gastrulation. J. Cell Biol. 103, 3a.

HARDIN, J. D. & CHENG, L. Y. (1986). The mechanismsand mechanics of archenteron elongation during seaurchin gastrulation. Devi Biol. 115, 490-501.

HOLTFRETER, J. (1943). A study of the mechanics ofgastrulation. Part I. J. exp. Zool. 94, 261-318.

HOLTFRETER, J. (1944). A study of the mechanics ofgastrulation. Part II. /. exp. Zool. 95, 171-212.

HORSTADIUS, S. (1973). Experimental Embryology ofEchinoderms. Oxford: Clarendon Press.

KINNANDER, H. & GUSTAFSON, T. (1960). Further studieson the cellular basis of gastrulation in the sea urchinlarva. Expl Cell Res. 19, 276-290.

MCCLAY, D. R. & ETTENSOHN, C. A. (1987). Cellrecognition during sea urchin gastrulation. In GeneticRegulation of Development (ed. W. Loomis),pp. 111-128. New York: Alan R. Liss.

MORRILL, J. B. & SANTOS, L. L. (1985). A scanningelectron microscopical overview of cellular andextracellular patterns during blastulation andgastrulation in the sea urchin, Lytechinus variegatus. InThe Cellular and Molecular Biology of InvertebrateDevelopment (ed. R. H. Sawyer & R. M. Showman),pp. 3-33. Columbia, SC: Univ. South Carolina Press.

TRINKAUS, J. P. (1984). Cell Into Organs: The Forces ThatShape the Embryo, 2nd ed. Englewood Cliffs: Prentice-Hall.

(Accepted 11 March 1988)