Cellular Interactions in the Development of the Shell Gland of the Gastropod, Ilyc~Nc~Ssa

Cellular Interactions in the Development of the Shell Gland of the Gastropod, Ilyc~nc~ssa

JAMES N. CATHER Associate Professor of Zoology, The University of Michigan, Ann Arbor, Michigan

ABSTRACT To determine the cell lineage of the shell gland in Ilyunassa, the 2d and 2c micromeres were marked with carbon granules. The 2d micromere derivatives form the shell gland, but 2c derivatives are later incorporated into the mantle. Dr?- letion of 2d causes shell abnormalities but, in about 10% of the cases, the shell ap- pears normal although somewhat small. Deletion of 2c also causes characteristic shell abnormalities. After deletion of both 2d and 2c, no shell forms. When 2c is marked and 2d deleted, the shell gland is marked. Any one-quarter embryo can form internal shell masses. In 1013 isolated ectoblasts (-3A, -3B, -3C, -3D), no shell material formed. Ectoblast with any single macromere formed shell with good morphogenesis, as did ectoblast plus 4d. Posterior ectoblast plus 4d, equivalent in size to isolated whole ectoblast, formed external shell. Properly oriented ectoblast, grown in contact with a polar lobe, formed shell in up to 70% of the cases; in one-half of these, the shell was external. Deletion of the macromere from an ectoblast plus 3C series of embryos at intervals through the first day resulted in embryos which did not form shell. Ectoblast in contact with polar lobe for six days, then separated, developed shell on the seventh or eighth day. (1) Neither ectoderm nor endoderm in isolation will form shell but any ectoderm-endoderm combination has the histogenetic ability to form shell na- terial; (2) the endoderm exerts its influence between the second and sixth days of development; (3) only those combinations including or in contact with polar lobe cytoplasm through the third quartet stage undergo normal morphogenesis; (4) trans- planted polar lobes can induce ectoderm to form shell; (5) the histogenetic ability to form shell is suppressed or unactivated when polar lobe cytoplasm is present, except in the D quadrant; (6) specialization of the D quadrant is one of the major evo- lutionary trends in spiralians.

Evidence for correlative differentiation self-differentiate, so that each cell will - the developmental modifications due form the same structure in isolation as it to the interaction between cells and tis- would in the intact embryo. Therefore the sues -in embryos from spirally cleaving prospective potency of the isolated or trans- mosaic eggs has generally eluded investi- planted blastomere is said to be the same gators since the origin of the theory of as the fate of the blastomere in the em- mosaic development three-quarters of a bryo. The potency may or may not equal century ago. Classical studies of embryos the fate but will not surpass it, according from eggs from which the polar lobe was to the theory. Other evidence indicates removed (lobeless embryos) by Crampton that the differentiative capacities of the (1896) and Wilson ('04a) and of isolated blastomeres are fixed through a precocious blastomeres by Wilson ('04b) and Costello segregation of qualitatively different cyto- ('45) indicate a progressive segregation of plasmic areas, especially the cortex, by capacities during development, so that at a highly determined cleavage pattern to each cleavage there is a part of the em- form a mosaic of self-differentiating cells bryo formed with the potentiality to form (Raven, '58). a particular structure and a remainder Although self-differentiation is dominant which does not have such a potential. This, during the early development of spiralians, in itself, does not necessarily imply that great correlative capacities are shown in the whole embryo had such a capacity the regenerative ability of these forms after prior to the segregative cleavage, as is metamorphosis. At present, little is known often stated; and the evidence for such an of the transition of the embryo from a idea is, at best, equivocal. Isolated blasto- group of independent blastomeres, pre- meres or groups of blastomeres tend to cisely arranged and tightly bound to one

J. EXP. ZOOL., 166: 205-224. 205 206 JAMES N. CATHER

another, into a highly integrated organism the polar lobe or D quadrant cytoplasm can capable of delicate responses to the devel- be considered the primary morphogenetic opmental environment. Whether the capac- organizer of molluscan development. ity for correlation develops gradually or Other evidence for regulative ability instantaneously is unknown, as is the tem- comes from the studies of Penners ('37), poral assumption of correlative abilities in who showed that the mesentoblast cell the different structures or systems in the (4d) of Tubifex, under suitable conditions, individual. can form ectodermal structures normally Although early development was demon- formed from the primary somatoblast (2d), strated to be primarily due to self-differen- which derives pole plasm from the same tiatlon, a large number of studies have area of the egg as the mesentoblast. Un- indicated a special developmental signifi- fortunately, this work has never been cance for the D quadrant, in addition to confirmed or extended to other species. the origin of the mesoderm. Clement's in- Novikoff ('38a) showed that twins could vestigations ('52, '56, '62) are the most develop from a Sabellaria egg when the complete in this regard. He has shown that formation of the polar lobe was suppressed D quadrant material, in this case derived so that the first cleavage was equal. Rather from the polar lobe, is necessary for the than a true correlative interaction between establishment of normal symmetry, as a cells, this can be interpreted as an equal regulator of the rate and pattern of cleav- distribution of lobe substance providing age in the D quadrant, and for the normal each of the two cells with its organizer. morphogenesis of the foot, shell, eyes, oto- Carefully designed experiments (Novikoff, cysts, and heart. The potency of the foot '38b) failed to demonstrate cellular or lobe and shell does not exceed the fate but interaction in Subellaria. Costello ('45) D quarter embryos can produce eyes which showed that gastrulation would occur in are normally derived from la and lc Nereis only in those blastomere combina- (Clement, personal communication) so tions which include both endoderm and that the D quadrant retains a significant ectoderm. Using combined animal and veg- ability to regulate. Thus its potency can etal tiers or vertical half-combinations, exceed its fate. Furthermore, Clement has Horstadius ('37) failed to demonstrate cel- clearly demonstrated that the morpho- lular interactions in embryos of the nemer- genetic influence of the polar lobe region tine, Cerebratulus. No comparable experi- is exerted during the formation of the four ments have been done with mollusks. quartets of micromeres but that not all Raven's series of studies of lithium- lobe-dependent larval features are deter- treated Limnaea embryos provide two val- mined at the same time. External shell uable cases in the analysis of regulative appears only after the formation of the ability in mollusks (Raven, '58). The first second quartet; however, normal morpho- involves the differentjation of head struc- genesis of the shell, as well as the foot, tures, particularly the eyes and tentacles. velum and eyes, occurs only after the for- A gradient-field concept was developed in mation of the third quartet. The heart and which there is a peak of activity at the ani- intestine are determined with the forma- mal pole, the region which later becomes tion of 4d. The 4D macromere is evidently the apical plate. The lateral cephalic plates, nutritive and of no morphogenetic value. where the eyes and tentacles form, are Although cytoplasmic segregation appears areas of intermediate dominance. Raven adequate to account for the determination explained the abnormalities produced by of the intestine and heart, it is necessary lithium - synopthalmia, cyclopia, aceph- to look to a mechanism of inductive nature aly - on the basis of gradient suppression. to explain the relation of the polar lobe to Although this concept has provoked valu- the morphogenesis of shell, foot, velum, able experimental analysis the work of and eyes. Clement's ('62) experiments in- Verdonk ('65) carried out in Raven's lab- dicate that the fate of all the cells in mo- oratory makes the concept questionable. saic eggs is not irrevocably determined at Verdonk showed that lithium treatment some early stage but is progressively de- modifies the direction of cleavage in spe- termined through development. Certainly cific cells. These cells, which cease division ILYANASSA SHELL GLAND DEVELOPMENT 207 in the normal embryo, then continue to gastrulates by epibole. This is true of a divide to form a cluster of small cells. The large part of molluscan species where the treatment', in this case, appears to release shell gland forms prior to the formation of certain cells from a division inhibitor. the endodenn in the proximity of the shell However, other cells which normally con- gland. tinue division are blocked by such treat- The present investigation was initiated ment so that their progeny are not formed. to determine the role of induction in the The end result is that the anlagen of the formation of the shell gland in mollusks. eyes and tentacles are displaced so that they give rise to the structure equal to their METHODS fate but in the wrong position. Modifica- tion of cleavage pattern alone can account The experiments were performed on the for the abnormalities and regulation is not prosobranch gastropod Nassarius obsoletus demonstrated. Although the cortex may Say, commody known to embryologists by represent a prepatterned series of anlagen its subgeneric name Ilyanassa, which will sites, no gradient hypothesis need be in- be used hereafter. Animals were obtained voked. from the Marine Biological Laboratory, Raven's ('52) analysis of the develop- Woods Hole or were collected on the coast ment of the molluscan shell gland showed of the Olympic Peninsula in Washington. that its differentiation is dependent on tis- Operation and culture methods are given sue interaction. Raven found that lithium in detail by Clement ('62) and will only treatment at certain stages leads to a dis- be summarized here. Eggs or embryos placement of the archenteron. Normally, removed from a capsule were carefully the shell gland forms from the ectoderm washed in pasteurized sea water and at the point where it contacts the arch- divided into control and experimental enteron. When the archenteron is dis- groups. When eggs had reached the stage placed, the shell gland forms wherever con- of development required, the cell which tact with the ectoderm is made, whether was to be deleted from the experimental pretrochally or post-trochally. In exogas- embryo was punctured with a fine glass trulae, where the endoderm fails to contact needle. The injured cell cytolysed within the ectoderm, no shell gland forms. a few minutes and fell away from the op- Hess ('56a, '57), while studying Bithynia erated embryo, which was then checked for and later Limnaea, found that a secondary other damage and transferred into fresh invagination of the digestive gland endo- pasteurized sea water. The number of em- derm can induce shell gland but that never bryos from a capsule which could be sub- more than one shell gland forms in any jected to the same operation depended on embryo. A number of investigators (Hess, the complexity of the operation and the '56a, '57; Morrill and Gottesman, '62; length of time that the embryo remained Clement, '62) have shown that shell can be at a suitable stage. The experimental em- formed from AB half-embryos, and Clem- bryos were observed several times during ent has shown that any quarter-embryo the first day to make sure that no cells can form shell. Hess ('56b) also deter- killed unintentionally were present. mined that more than one part of the The embryos were allowed to develop endoderm has the capacity to induce the until a well-formed shell was present in ectoderm to form shell and that this capac- the controls or until larval differentiation ity is retained for some time, as judged was completed and all of the visible yolk from the response of the ectoderm to mi- was utilized. Embryos were observed in gration of endodermal fragments into the living condition, then relaxed in chlore- embryo in late exogastrulae. Ranzi ('31) tone, fixed in 70% ethanol, and preserved found that even isolated fragments of the as whole mounts in diaphane (Clement shell gland formed shell in the cephalopod, and Cather, '57). All specimens were ob- Loligo. served in polarized light to determine Collier ('64) has pointed out that the the extent of the highly birefringent ara- interaction between the archenteron and gonite of the shell. Embryos for cell lineage ectoderm cannot occur in Ilyanassa, which studies were prepared in a similar manner 208 JAMES N. CATHER

or according to the method of Clement plastic Petri dish. To extract toxic ma- ('52) or Cather ('58). terials from the hardened wax, the dish Marking experiments were done by two was filled with glass distilled water over- methods which are standard for the study night. The dishes were drained and filled of vertebrate embryos but which, to my with pasteurized sea water, after which knowledge, have not been previously ap- depressions were made in the wax, which plied to the much smaller invertebrate em- were larger in diameter and somewhat bryos. In the first method, a piece of agar, deeper than an isolated polar lobe. An iso- containing nile blue sulfate was embedded lated ectoblast was first placed in the de- in the wax-side of a depression previously pression, the polar lobe was placed over it formed in an operating dish. An embryo and the two pushed into close contact. in a drop of pasteurized sea water was then appropriately positioned in the de- RESULTS pression so that the proper cell was in I. Origin and development of the contact with the dyed agar. The cell re- shell gland mained in contact with the agar for 30 minutes to one hour after which the em- A. Nmal Histogenesis and Morphogen- bryo was placed in pasteurized sea water esis and checked to determine that only one The development of the shell gland in blastomere was stained. The development Ilyanassa is typical of that described in of the stained and unstained areas was other prosobranch gastropods (Raven, '58; observed for as long as they could be dis- Manigault, '60; Fretter and Graham, '62; tinguished, at best for four or five days. By Wilbur and Yonge, '64). this itme the shell gland was functional. In Ilyanassa, five stages of shell gland In the second method, carbon marking was development can be recognized. The time done wtih a 12 I-I glass ball formed on the given for the initiation of each stage was end of a thin glass rod. The ball was im- measured from anaphase of Meiosis I, the mersed in sea water, touched to carbon earliest stage readily available (Cather, particles spread on a paper towel and then '63), for embryos reared at 20 * 1°C. Vari- brought in contact with the appropriate ation in the initiation of stages by embryos blastomere. Most of the carbon was usu- within a single capsule increased with ally incorporated into the cell but often time. There was as much as five hours some would remain on the surface. In both difference between the first embryo to go cases, however, the carbon appeared to be into stage three and the last one to do so, limited only to the daughter blastomeres and as much as ten hours difference by of the marked cell. In some cases, carbon stage five. could not be detected after fixation. It was The ectodermal cells which eventually probably either so dispersed as to be in- form the shell gland show no visible signs visible or it became separated from the of differentiation until about 66 hours, cell after the period of initial observation when the embryo elongates anteriorly as instead of being incorporated. Carbon- the stomodeum is formed. However, the marked embryos were prepared as whole ectodermal area which forms the shell mounts or fixed in 2% buffered osmium gland can be followed, using the posterior tetroxide, embedded in Epon, and sec- turret cell and the 4D macromere as nat- tioned at two I-I with a Porter-Blum micro- ural markers. tome. The specimens were then stained Stage 1 -Plate fornation. The ecto- with 0.1 % azur B bromide in Clark-Lubbs derm cells which form the shell gland near buffer at pH 8.5 or in 0.1% basic fuchsin the mid-dorsal area of the posterior end in 50% acetone. Thin sections were ob- of the embryo become cuboidal while the served with RCA electron microscopes. rest of the posterior ectoderm remains Other histochemical techniques used were more or less squanous. The nuclei enlarge those from Pearse ('61). and nucleoli persist throughout interphase. Operating dishes for polar lobe trans- There is a progressive development of plantation were made by melting dental basophilia during this stage which persists wax and pouring a layer into a disposable until stage five. This is evidently due to ILYANASSA SHELL GLAND DEVELOPMENT 209

RNA synthesis, since it is lost after ribo- ten or more hours the shell gland evagi- nuclease digestion (Pearse, '61). By 76 nates, first to form a cup-shaped structure, hours, the cells have become columnar and then flattening to become saucer-shaped, form a flat plate which extends along and and finally becoming convex (fig. 4). Other to the left of the dorsal midline of the em- than the development of a very strong alka- bryo's posterior end (fig. 1). line phosphatase reaction near the end of Stage 2-Invagination. At about 80 the stage, no obvious changes in cellular hours, the center of the flattened plate be- components are apparent. gins to invaginate to form a small pocket Stage 4 - Shell formation. Deposition which, for a time, is almost closed from of calcium carbonate starts at about 120 the exterior. The surrounding cells form hours. This occurs immediately before or the typical rosette which makes the shell during the first stage of torsion (fig. 5). gland particularly obvious at this stage. A thin membrane is formed at the apical The operculum also becomes birefringent ends of the cells, where it remains attached at about the same time but there is no con- for a time by microvilli before lifting from sistency in the order of appearance be- the cells to close the opening of the invag- tween the shell and operculum. The cal- ination. The membrane persists through cium carbonate in the larval shell consists the early development of the shell gland entirely of aragonite (the crystalline struc- and shell is continuously formed and al- ture was determined through x-ray diffrac- ways attached at the edge of the shell tion by Dr. Donald Peacor, University of gland or mantle. This becomes the peri- Michigan) and is highly birefringent. The ostracum of the larval shell. The invagi- thin membranous periostracum, formed at nation is completed by 90 hours (fig. 2) the time of invagination, becomes rather when a somewhat viscous material - rel- closely associated with the cells of the ative to sea water-is present in the in- shell gland and there appear to be some vagination. fusions which persist as minute holes in The highly organized cytoplasm devel- the larval shell. The underlying viscous oped at this stage is shown semidiagram- material evidently forms the matrix for matically in figure 3. The mitochondria crystal formation, which occurs beneath become polarized so that their long axes the periostracum. Only one morphological parallel that of the cell. Mitochondria, in cell type can be distinguished in the shell this stage, are most abundant in the apical gland and it' evidently forms the mem- half of the cell whereas they were more brane, secretes the viscous material and or less evenly distributed earlier. A large the calcium carbonate of the shell. vesicle appears basally to the nucleus in Stage 5 -Mantle formation. After the each cell and almost completely fills the secretion of the shell, the basal vesicles area between the nucleus and cell mem- are no longer present in the central cells brane. The matrix of this vesicle is in- of the shell gland. As these cells lose their tensely stained by basic fuchsin and azure basophilia, they become very flat and show B bromide. Minute birefringent granules no morphological signs of activity. The appear within the vesicular matrix near marginal cells retain the characterisitcs of the end of the stage. These are only visible the stage 2-3 shell gland cells. From this in living material but are evidently repre- time on, the term mantle is preferable sented by small clear circles in fixed and since the initial phase of development is sectioned material. The endoplasmic retic- completed with the secretion of shell ma- ulum is well developed as cisternae and terial. tubules in the medial part of the cell sur- The mantle edge appears to grow in the rounding the nucleus. Rough vesicular ele- early stages through the incorporation of ments persist throughout the cytoplasm. surrounding ectodermal cells, as well as The Golgi apparatus is most often seen, through proliferation. Division figures seen in the early part of this stage, near the nu- in the shell gland never occur in cells cleus on the apical side. There is a loss of which have major morphological criteria yolk granules and lipid during this period. for differentiation present. The source of Stage 3 -Evagination. Over the next such dividing cells is unknown. Fig, 1 Control embryo, 72 hours old. Shell gland stage 1, plate formation. Shell gland (sg). Fig. 2 Control embryo, 90 hours old. Shell gland stage 2, invagination, Fig. 3 Shell gland cell €rom 96-hour-old control embryo near the end of stage 2. Endoplasmic re '- ticulum (er), Golgi complex (g>, Mitochondria (m), Veside (v). Fig. 4 Control embryo, 108 hours old. Shell gland stage 3, evagination. Fig. 5 Control embryo, 120 hours old, during early torsion. Shell gland stage 4, sheU. formation '* Shell (s). ILYANASSA SHELL GLAND DEVELOPMENT 21 1

The veliger shell is almost bilaterally of the mantle cavity somewhat later. This symmetrical. Its early growth is almost was observed in four embryos which were exclusively on the posterior and postero- fixed, after torsion had moved the intestine ventral side but the first phase of torsion dorsally. In seven embryos which were rapidly shifts this small plate of shell to marked, no carbon could be detected on the right, so that the right side and pos- the fourth or fifth day. terior end are covered first by shell. Later More than 100 embryos (87 of 2d and the mantle edge grows forward dorsally 23 of 2c) were stained with nile blue sul- and from the posterior along the left side fate. The marked areas on these embryos to cover the visceral hump. showed essentially the same pattern as the B. Origin of the Shell Gland carbon-marked embryos of the same stages Lillie (1895) originally described the but the dye became progressively mare shell gland to be a derivative of the left diffuse, rendering these experiments less posterior micromere of the second quartet valuable than the carbon-particle experi- (2d) in Unio, and Conwin (1897) agreed, ments. although with some reservation, in his Second quartet micromere deletion ex- study of Crepidula. Later studies have all periments. The 2d micromere was de- supported this lineage. However, my early leted from 100 embryos from 15 different operations in this series and the work of capsules. Of these, 11 embryos formed Clement ('56) and Hess ('56a) raised con- shells which appeared to be perfectly nor- siderable doubt about the completeness of mal, except that they were 7585% the the early observations since shell was be- size of the shells of controls, on the ninth ing formed from half and quarter embryos day of development (fig. 17). Thirty-two which lacked, not only the 2d micromere, embryos formed cup-shaped shells, one- but in some cases the whole D quadrant. To half to one-third the size of the controls. determine the origin of the shell gland and Variable dumbbell-shaped shell plates were mantle, the following experiments were formed in 28 embryos (fig. 9). Internal done. shell masses were formed in 21 cases, all Cell marking experiments. Cells were in the posterior half of the embryos. No marked with particles of carbon or stained shell formed in five cases. with nile blue sulfate. In 21 embryos with Thirty-two embryos from six capsules the 2d micromere carbon marked, the par- in which the 2c micromere was deleted ticles were found in the invaginating shell formed cup-shaped shells of one-half 00 gland at 84 hours, as well as on the dorsal one-third normal size. These were always and ventral left side posterior to the larval severely flattened on the right side (fig. kidneys, and along the dorsal and right 19). The umbilicus was usually present. posterior surface. The marked areas were In contrast to the 2d-deletion syndrome not identical in every embryo but always which is highly variable the 2c-deletion included the area shown in figure 6. Many syndrome is extremely uniform. Deletion marked embryos were discarded because of 2C has no effect on shell formation of questionable marking procedure or be- (fig. 18). cause of abnormalities. In six embryos In 27 embryos from six capsules in which appeared to be normal, no carbon which the 2c and 2d micromeres were both could be detected on the fourth day after deleted, no shell was visible either inter- marking. nally or externally (fig. 8). The embryos In 12 embryos which had the right pos- appeared normal except for some shorten- terior micromere of the second quartet ing of the visceral hump. In three of (2c) marked with carbon particles, the in- the embryos, only one statocyst was pres- vaginated shell gland was unmarked but ent but, in all others, the foot was well carbon was observed in the right lateral formed and the statocysts and operculum and ventral ectoderm. By 120 hours, a part were typically birefringent. Deletion of 2a, of the marked area was being incorporated which is in a roughly comparable position into the mantle edge (fig. 7), while the to 2c on the left side, has no effect on shell rest of this area appeared to form a part formation. 212 JAMES N. CATHER

Fig. 6 Composite illustration where the hatched area represents the minimum distribution of car- bon granules at 84 hours (stage 2) when the 2d micromere was marked. Note the complete cover- age of the shell gland at the upper left. Fig. 7 Composite illustration where the hatched area represents the distribution of carbon gran- ules at 120 hours (stage 4) when the 2c micromere was marked. Viewed from the upper right side. Shell (s). Fig. 8 Larva, 168 hours old, from an egg lacking the 2c and 2d micromeres. No shell is present. Fig. 9 Larva, 168 hours old, from an egg lacking the 2d micromere. The shell (s) is represented by the hatched area.

Combination of cell marking and cell de- delay was apparent after the formation of letion experiments. In six embryos from the shell gland, the rate of development six capsules, the 2d micromere was deleted thereafter was normal. One embryo was and the 2c micromere was marked with grossly monstrous and was not analyzed. carbon particles. In five cases, the shell 11. Cellular interaction in shell gland gland was considerably displaced to the development right and appeared to be lagging behind the controls so that the marked area was A. Th-e Development of Isolated Ectoblasts comparable to an 84-hour shell gland in Since all of the ectoderm of gastropods the controls, when the experimental em- comes from the first, second, and third bryo was 96 hours old. Since no additional quartets of micromeres, an embryo with- ILYANASSA SHEU GLAND DEVELOPMENT 213 out mesoderm or endoderm - an ectoblast Several hundred macromeres, unidenti- embryo - can be prepared by deletion of fied as to quadrant, were isolated in cal- the 3A, 3B, 3C, and 3D macromeres after cium-free artificial sea water but these the third quartet is formed. One thousand developed so poorly and were so grossly and thirteen operations of this type have abnormal that they were considered to be been done and the resulting embryos have of no value. been studied for as long as 14 days of de- No obvious larval structures developed velopment. in any case but the development of such The ectoblast embryo forms a hollow cells in isolation is poor at best. The mes- ball-like structure within 24 hours of the entoblast derivatives formed no structures operation. Over the next 2-4 days, it be- but were never observed as a compact mass comes ciliated along a circumferential ring, so their real capacity in isolation is un- near the more or less flattened base of the known. embryo. On the more pointed end, an api- cal tuft of cilia develops, associated with C. Shell Development in Ectoblast-ento- large cells which are probably apical plate blast Combinatisrts cells. In some embryos, distinct areas Half- and quarter-embryos. Since calci- lateral to the apical plate appear to be um carbonate was dissolved by the method cephalic plates (figs. 10,20). However, of fixatSon used by Clement ('56), he de- these areas cannot always be distinguished, scribed shell only in CD and D isolates. possibly due to the difficulty in orienting However, he has more recently reported the embryos. ('62) that A, B, C, or AB can form internal Although these embryos live and are shell masses. motile for at least ten days, no further dif- In 40-50 embryos of each type (A, B, C, ferentiation of larval structures has been and AB) more than 50% developed inter- observed. In no case was any pigment, bi- nal shell masses. In about 10% of the refringent material, or other structural en- cases, more than one mass was present, tity observed. with several containing three or more. His- tological examination indicates that these B. The Development of Isolated Macrcr- masses are formed by typical shell gland meres cells arranged in closed sacs of various A series of experiments were done in forms (fig. 15). The cells undergo the which quarter embryos were isolated. same sequence of changes as those in the When the micromeres formed, they were normal shell gland so that, when shell is deleted, so that isolated 3A, 3B, 3C, and present, the surrounding cells are flattened 3D macromeres were obtained. The first with little basophilia. three macromeres are entirely endodermal, In 40 cases of AD half-embryos, 33 de- while the 3D macromere forms the mesen- veloped external shells, most of which toblast cell (4d) as well as the 4D macro- were small and cap-shaped; seven of these mere. Twelve partial embryos of each type had better morphogenesis characterized by were isolated. the presence of an umbilicus. Four em- The endodermal blastomeres (3A, 3B, bryos formed internal shell masses, while 3C) all developed similarly, undergoing a three had no apparent shell. In 53 CD half- few divisions, then becoming inactive and embryos, 37 showed external shells which degenerating after a few days. The 3D were about a third larger than those of the macromere formed the mesentoblast cell AD half-embryos. Morphogenesis appeared which then divided a number of times. essentially normal in 11 of these, although The 4D macromere was never observed to the shells were relatively smaller than the develop further. The mesentoblast deriva- controls. Four embryos developed internal tSves often would adhere to the bottom of shell masses; the remainder developed no the dish and formed spindle shaped cells shell and were often quite abnormal in which persisted for from 10-15 days. En- general development. dodermal cells also frequently sent out In 12 D quarter-embryos, one developed pseudopodia and stuck to the substrate be- an external shell tag, eight had internal fore their degeneration. shell masses, and three had no shell ma- 214 JAMES N. CATHER

terial visible. Development of D quarter- Clement (personal communication) noted embryos was poor with only 12 out of 60 that, in a case of ectoblast plus 4d, a shell, experimental embryos growing to a stage operculum, and eyes developed. The em- roughly comparable to the veliger. bryo had fair velar development. When Ectoblast plus one macromere embryos. this experiment was repeated, by deleting In one of the early attempts to remove the the 4D macromere about 40 minutes after macromeres to make an ectoblast embryo, mesentoblast formation and then waiting an anucleate yolk tag (12 by 16 u) re- for an hour before deleting 3A,3B, and mained attached to the ectoblast. The re- 3C, I was able to confirm and extend his sulting embryo formed a small cap of shell result. In 32 cases from five capsules, four (fig. 21). embryos lacked shell, nine had an internal This result led to a systematic series of shell mass, and 19 an external shell (figs. experiments in which three of the four 11,23) and foot, an intestine, heart, and macromeres were removed. In 12 ectoblast some large muscles. Velar development plus 3A embryos, all formed shell which, was variable and either one or two eyes in four cases, was external, with an um- were present. In the best cases, these em- bilicus and characteristic form. Of 12 ecto- bryos were small but at least superficially blast plus 3B embryos, one had no shell, normal, although the endoderm was too six had internal shells, and five developed compact to be adequately analyzed. external shells, three of which showed Posterior one-half ectoblast plus mesen- good morphogenesis. In 12 ectoblast plus toblast embryos. In 12 embryos from 3C embryos, nine developed external well seven capsules, the AB cell was killed at formed shells, each with some evidence of the 2-cell stage; the CD half was allowed an umbilicus (fig. 14); three had internal to develop until an hour or more after the shell masses. In 26 ectoblast plus 3D em- mesentoblast cell formed, at which time bryos, two formed internal shell masses, 3C and 4D were deleted. The embryo from while 24 formed external shells with good such an operation therefore included the morphogenesis. These latter embryos had posterior half of the ectoblast and the mes- consistently well developed endodermal entoblast cell, and was equal in size or structures, as well as the heart and intes- smaller than an isolated whole ectoblast. tine which were absent or poorly developed All such embryos formed shell; in three in the other types. All ectoblast plus mac- cases it was internal but, in nine, an ex- romere embryos showed good morphogen- ternal shell and foot developed. The ve1v.m esis of foot, velum, and eyes. was variable and reduced and only one em- One series of experiments was done in bryo had one eye. Pigment of the intestine which the 3A, 3B, and 3D macromere were was visible but the rest of the endoderm deleted at the third quartet stage and then, Fig. 10 Larva, 168 hours old, from an egg at intervals of 4,8, 12, and 16 hours after lacking the 3A, 3B, 3C and 3D macromeres (an the stage at which the mesentoblast forms ectoblast embryo). Note the pattern of cilia and in the controls, the 3C macromere was de- the absence of differentiated structures. leted (four cases at each time). These ex- Fig. 11 Larva, 168 hours old, from an egg lacking 3A, 3B, 3C and 4D (an ectoblast plus mes- periments were done to determine the time entoblast embryo). The area of the shell (s) is of interaction between the ectoderm and hatched. Note the eyes and velar cilia. endoderm. All embryos in the series de- Fig. 12 Larva, 168 hours old, lacking AB, 3C veloped as typical ectoblast embryos and and 4D (a posterior ectoblast plus mesentoblast embryo). Note the large cells underlying the shell lacked shell, foot, and pigmentation. (s) in this case. Ectoblast plus mesentoblast embryos. Fig. 13 Larva from an ectoblast and polar In 42 embryos from 11 capsules, all of the lobe combination where the ectoblast cells have macromeres were removed about 40 min- completely overgrown the yolky polar lobe. Note the internal shell (s) masses. utes after the formation of the mesento- Fig. 14 Larva from an ectoblast plus 3C mac- blast cell. These embryos formed no shell romere combination. Note the form of the shell, but did show a light pigmentation. Clusters eyes, and velum as well as the absence of the in- of small yolk granules which were presum- testine. Fig. 15 Section of an AB half-embryo (-CD) ably from the mesentoblast were present showing two internal shell (s) masses and the in the internal cells of the embryo. typical shell gland cells surrounding them. ILYANASSA SHELL GLAND DEVELOPMENT 215

- Figure 10-15 Fig. 16 Control veliger larva, 192 hours old. Fig. 20 Ectoblast larva (-3A, 3B, 3C, 3D), Ventral view. Note the pattern of pores in the 144 hours old. Note the absence of differentiated shell in the lower part of the picture. structures. Fig. 17 Larva, 184 hours old, from an egg Fig. 21 Embryo formed from an ectoblast plus lacking the 2d micromere. The animal is partly a tag of yolky cytoplasm from an undetermined retracted into the shell. macromere. Photographed in polarized light to Fig. 18 Larva, 168 hours old, from an egg show the highly birefringent shell seen in the lacking the 2C macromere. Note the form of the lower part of the picture. shell. Fig. 22 Eight-day-old embryo, separated from Fig. 19 Larva, 168 hours old, from an egg the polar lobe after six days of contact. Photo- lacking the 2c micromere. Note the characteristic graphed in polarized light to show the shell. flattening on the right side of the reduced shell. Fig. 23 Ectoblast plus mesentoblast larva showing the shell, operculum, velar pigment and eye. Photographed in polarized light. ILYANASSA SHELL GLAND DEVELOPMENT 217 and mesoderm was crowded and difficult cilia were present but these did not appear to analyze (fig. 12). as large as velar cilia. One embryo had an Ectoblast plus polar lobe embryos. To internal cavity surrounded by columnar combine ectoblasts with polar lobes the cells. method was modified in several ways which affected the results. In all, however, DISCUSSION ectoblasts were obtained as previously de- In order to evaluate critically the results scribed and then eggs in the trefoil stage of deletion and transplantation experi- from another capsule were delobed with ments done on the embryos of animals a fine glass hair. The ectoblast and the with mosaic development, the normal fate anucleate polar lobe were then placed in of the cells under consideration must be a suitably sized depression in a wax op- known. Otherwise it is not possible to dis- erating dish. In the first series of 100 tinguish between the self-differentiation of embryos from 18 capsules left in the blastomeres and correlated interactions be- depression for seven days, six formed tween them. All of the classic cell lineage shell material of which four had internal studies derive the shell gland from the 2d masses (fig. 13) and two had small ex- micromere (Raven, ’58). It is not known ternal pieces. Long cilia similar to velar which of the derivatives of 2d actually cilia developed on the end opposite the forms the shell gland but 2dl1 forms part shell in three of the elongate embryos. No of the head vesicle in animals as widely other differentiated structures could be de- separated phylogenetically as Limnaea tected. Only in those cases where the ecto- (Verdonk, ’65) and Crepidula (Conklin, blast fused to the polar lobe so that the 1897) and probably does also in Ilyanassa, ectoblasts grew down over it, was shell although there is no specific information formed. Only those ectoblasts placed so to support this point. Other structures that the surface next to the original macro- ranging from the stomodeum to anal cells meres was in contact with the polar lobe and foot have also been attributed to 2d, fused with it and only those placed to- but these will require confirmation before gether within about two hours after the they can be generally accepted. From macromere deletion fused. C1ement”s (’58) study of the development In a second series of 15 embryos from of the foot, we now know that it originates eight capsules, which were oriented so that from the third quartet micromeres, 3c and the polar lobe was placed in contact with 3d, in 12yanassa. Usually, in gastropods, the ventral surface of the ectoblast and in the primary somatoblast (2d) is not as which the duration of the operation was large as in lamellibranchs and annelids kept as short as possible, five developed and it may not have as extensive a role in shell. One had a cluster of internal shell the formation of the posterior ectoderm. granules, one formed an internal mass, The experimental evidence derived from and three developed small posterior shell the marking and deletion experiments pre- plates. All were ciliated, but velar-like cilia sented here strongly supports the cell line- were not observed. age studies on the origin of the shell gland. In a third series, the polar lobes were The participation of 2c in the formation removed in calcium-free artificial sea wa- of the shell and mantle cavity has not been ter and then placed with the ectoblasts. previously described. In neither case does They were left in contact for six days and 2c form all of the structure. In the shell, then removed from the depression. No 2d is the major participant while in the fusion had occurred, but six of the 13 had case of the mantle cavity the left side re- shell which was internal in three and ex- mains unmarked after 2c has been marked. ternal in three. Three of the ectoblasts Although this could be artifactual, it may which lacked shell on the sixth day, when also indicate the participat’ion of another they were separated from the lobe, devel- blastomere, possibly 2d, but this could not oped shell by the eighth day (fig. 22). One be determined in the preparations made embryo had internal granules and two had so far. Torsional movements make the de- small shell plates which curved around the velopment of the mantle cavity difficult to posterior end of the embryo. Some long follow at the cellular level. 218 JAMES N. CATHER

f 9 24 Fig. 24 A diagrammatic summary of the interactions preceding shell formation. Isolated ectoblast, (a). Larva from an isolated ectoblast, (b). Larva developed from an ectoblast plus mesentoblast (4d) combination in which the shell is hatched, (c). Larva developed from an ectoblast and macro- mere (M), (a). Larva developed from an ectoblast plus polar lobe (pl) combination. Dotted lines show cells which have been deleted, (e). Isolated posterior ectoblast, (f). Larva from a posterior ectoblast and mesentoblast cell combination, (g).

The later development of the shell gland shell gland of Ilyanassa, as it does in the is typical of that previously described in mantle of many adult mollusks, nor are other species, but it should be noted that separate prismatic and nacreous layers the development of more than one type of formed in the larval shell. The actual for- differentiated cell does not occur in the mation of shell material is extracellular ILYANASSA SHELL GLAND DEVELOPMENT 219

and occurs in the viscous material between account for shell gland formation in Lim- the periostracum and the shell gland. naea as it cannot in Ilyanassa. Clement ('62) and Collier ('64) have Specificity and histogenesis already noted that it is highly unlikely that Combinations of ectoderm and endo- the small-celled endoderm plays any role derm from any half- or quarter-embryo in the development of the shell gland in are capable of shell histogenesis. The pres- those animals with epibolic gastrulation ent studies, however, indicate that neither such as Ilyanussa, because of the lack of ectoblasts nor endoblasts give rise to cells proximity of the layers prior to the forma- which will form shell material when iso- tion of the shell gland. In Ilyanassa, there lated from the other type. Raven ('52) is never any small-celled endoderm in the originally proposed that only small-celled vicinity of the shell gland. The 4D macro- endoderm could serve to induce the shell mere, which is in continuous contact with gland. Since the small-celled endoderm the derivatives of 2d, probably does not originates in each quarter in most mol- divide, at least not in the region of the lusks, the apparent diversity of origin of shell gland, and the isolated polar lobe shell material does not disprove his con- does not undergo divisions, yet both are tention. Hess ('62) believes there is no capable of inductive interaction. great specificity of endoderm required, A possible interaction between the mes- since even the digestive gland diverticu- entoblast (4d) and the somatoblast (2d), lum formed in exogastrulae can serve as during late cleavage when these cells are the inducer. Unfortunately, the origin of in close contact, must also be considered. the digestive gland is not well enough Since embryos resulting from properly ex- known for one to critically evaluate this ecuted operations to form combinations of evidence, although it appears to indicate ectoblast and mesentoblast almost always that the large-celled endoderm is also capa- form shell, this is a distinct possibility. In ble of inductive interaction. An attempt those cases where shell was not formed, has been made to clarify this problem and there was probably a displacement of the to determine the inductive capacity of the mesentoblast which caused too great a sep endoderm and the competence of the ecto- aration for interaction to occur. The se- derm. Since any quarter-embryo can form quential macromere deletion experiment, shell material, both anterior and posterior however, indicates that the endoderm must and right and left ectodermal areas are be present for more than 24 hours of de- competent to undergo the histogenetic dif- velopment for the interaction to occur. ferentiation necessary for shell secretion. During this period in normal development, External shell, however, has never been the mesentoblast derivatives are shifted observed in Ilyanassa embryos which lack away from the site of the shell gland so posterior ectoderm, but this appears to be the mesentoblast is probably not the nor- related to the factors controlling morpho- mal inducer. genesis rather than histogenesis. There As Collier ('64) has pointed out, the does not appear to be great specificity of shell gland develops in ectoderm which has the posterior ectoderm derived from 2d contact only with the 4D macromere. Con- and 2c, since either can form the shell trary to his conclusion that induction is not gland as well as other structures. The evi- probable, it appears that the 4D macro- dence indicates that specificity is estab- mere is the inductive cell. lished through correlated interaction with Costello ('45) has pointed out that the the endoderm, and possibly with the meso- failure of differentiation of certain struc- derm, in the formation of mantle cavity tures in small isolates and production of structures. Raven ('52) showed that there these structures by larger combinations of is a considerable inherent cytodifferentia- blastomeres is not, in itself, proof of in- tive ability in cells in exogastrulae even duction. There are many cases known when morphogenesis does not occur, and where a minimal cellular mass is required when there is no close apposition of layers for differentiation. which might interact. He pointed out, how- When the posterior ectoblast is isolated ever, that inherent histogenesis cannot with the mesentoblast, the mass is less 220 JAMES N. CATHER

than the whole isolated ectoblast yet the shell is formed. Since 3d does not con- resulting embryo almost always secretes a tribute to the shell and since its removal small external shell. The cytoplasmic par- has no effect on shell formation or mor- ticles of the mesentoblast, such as fine phogenesis, then there must be an organiz- yolk granules and lipid droplets distinguish ing influence exerted by the 3D macromere it from the ectodermal cells. It appears on the overlying ectoderm. The presence that qualitative differences, whatever their of the 3D macromere during this period bases may be, rather than quantitative is then critical for morpohgenesis even ones determine the capacity to interact. though removal of its progeny 4d and 4D, Thus it appears that any endoderm can two hours later, has no effect on morpho- serve to induce the ectoderm to carry out genesis of the shell. The ectoblast plus the necessary histogenetic changes for macromere combinations give added sup- shell secretion within wide limits of size port to Clement's proposal for a temporally and relative proportions. specific morphogenetic influence exerted at the third quartet stage. Removal of the M orphogenesis 3D macromere and either two of the other If the endoderm has such broad induc- three macromeres after the third quartet tive capacity and the ectoderm such wide has formed results in an embryo with a competence, how is it that shell does not small but essentially normal shell. The form in each quadrant of the intact em- morphogenetic influence therefore has bryo? In those cases where most of the been exerted prior to removal of 3D. The posterior ectoderm is removed (-2c2d), other macromeres retain the capability to no shell forms in any part of the embryo. induce the histogenetic differentiation nec- Yet in isolated half-, quarter-, or lobeless essary for shell secretion by the ectoderm embryos, shell is secreted. Is there an in- at some time between the first and sixth tegrative phenomenon which allows only days of development. The mesentoblast one shell to form in an embryo? If there and ectoblast combinations also give simi- is, it can only occur when the polar lobe lar support. In all of these cases, the mor- is present, for in its absence several inter- phogenesis of the foot, velum, and eyes nal shell masses may be formed. Only are also regulated just as Clement de- when the D quadrant (the polar lobe cyto- scribed from his D macromere deletion plasm) is present do the other quadrants studies. fail to form shell, even in those cases Clement ('66) has further shown that where in the absence of ectoderm the D the morphogenetic influence is not simply quadrant will not. It then appears that the due to the nutritive value of the yolk since polar lobe exerts an inhibitory influence on nucleated vegetal halves containing almost the other quadrants which prevents differ- no yolk will develop into small but morpho- entiation leading to shell secretion. genetically complete veligers. The D quadrant has a complex role in The fact that an isolated yolk tag or shell formation. It gives rise to the 2d polar lobe can induce shell material points micromere as the source of cells, it inhibits to the uniqueness of the vegetal half and histogenesis leading to shell secretion in particularly that of the D quadrant. Cos- the A, B, and C quadrants, and it has a tello ('45) has made clear the necessity for morphogenetic role, which not only affects critical evaluation of the quality of the con- the shell but also the form of the foot, tact between the inducing and responding velum, and eyes. Clement ('62) has shown, tissues in studying induction in mosaic through his D quadrant macromere dele- development, especially in experimental tion studies, that external shell develops systems such as those utilizing polar lobes. only after 2d is formed. Even though all It appears that an actual fusion of the ec- of the ectoderm used in the formation of toblast cells to the polar lobe is required the shell is present at this time, the shell for shell to be formed. Whether or not the is never morphologically normal. It is only polar lobe membrane persists after this after the formation of the third quartet, fusion has not been determined satisfac- which contributes the 3d micromere to torily but the observations so far made form the left side of the foot, that a normal suggest that it does persist. The 4D macro- ILYANASSA SHELL GLAND DEVELOPMENT 22 1 mere does retain its membrane intact, at entiation which precedes the secretion of least in the region of the shell gland. The shell. same kind of contact appears to be re- The paper by Davidson et al. ('65) pre- quired for the various ectoderm-endoderm sents a very attractive hypothesis for the combinations for shell secretion. On the role of the polar lobe as a differential acti- other hand, when the lobe is removed in vator of specific genes but the evidence calcium-free sea water, the lobe can induce presented to support the concept is yet in- the ectoblast to form shell after close con- adequate. It is hoped that more evidence tact. If there is any fusion in these cases, will be forthcoming and that it can bridge it is not strong enough to keep the ectoblast the gap between the time of known polar and lobe together during their removal lobe interaction at the third quartet stage from the depression in the operating dish. and the synthesis of RNA, some 24 hours Since short treatment with calcium-free later. artificial sea water at this stage has no major effects on later development if the Evolution of D quadrant specialization lobe is not separated, it appears that the Costello ('55) pointed out that spiral cell membrane over the polar lobe serves cleavage is characteristic of many more as a significant barrier to the inductive process. animal groups than was originally thought. Since shell can be secreted for some time Many such animals have undergone exten- after the separation of the ectodem from sive modifications in their cleavage pat- the endodenn, it appears that the induc- terns but enough of the basic characteris- tive process is not simply a passage of tics of spiral cleavage remain so that there molecules through the ectoderm to form is little doubt about the origin. The acoel shell in the overlying matrix as has been flatworms were the first group to develop proposed by some investigators (Fretter spiral cleavage. It appears, on the basis of and Graham, '62). This could not be the meagre evidence, that the three duets of case in the early shell gland anyway be- micromeres give rise to the ectoderm and cause an elaborate series of differentiative some to the ectomesenchyme while the events must precede the secretion of shell pair of macromeres and the fourth duet material. form the endoderm and inner mesodem. This of course raises the question as to (For a review, see Hyman, '51.) the nature of the interaction which can Along the main evolutionary line of ani- occur between cells and a nonnucleate re- mals which have the basic spiralian plan gion of cytoplasm where the plasma mem- are the polyclad flatworms which have de- brane of the latter may play a significant veloped the typical four quadrants, with role. At present, one attractive possibility the first three quartets forming the ecto- involves that portton of the yolk platelets derm. The macromeres and all of the which shows birefringence as first de- fourth quartet cells except 4d are nutritive. scribed by Clement ('62). These are dis- Some micromeres form ectomesenchyme tributed initially throughout the yolk but but the mesoderm and all of the endoderm later appear to be predominantly localized come from the left posterior micromere of in the D macromere. They disappear near the fourth quartet, the mesentoblast cell the shell gland during shell synthesis and (4d). This is the first and most constant are completely lost from the endoderm by specialization of the D quadrant. the time the larval shell is completed. The The nemertines have evidently experi- presence of alkaline phosphatase associ- mented extensively with mesoderm forma- ated with some yolk granules gives support tion, and mesoderm has been attributed to to this idea but one cannot be certain that the micromeres, the 3D macromere, and these are the same granules as those which the mesentoblast. are birefringent. If the birefringent yolk In annelids, the D quadrant also has granules did serve as a calcium reserve additional significance in that the primary for shell secretion, the level of the sub- somatoblast (2d) and the mesentoblast are strate itself could be a major factor in the necessary for the formation of most of the endoderm, causing the histogenetic differ- adult worm and are held in reserve in the 222 JAMES N. CATHER

trochophore. The segmentation character- ACKNOWLEDGMENTS istic of the group requires the D quadrant I wish to thank Professor A. C. Clement material present in the mesoderm. The for his encouragement during the course highly determined cleavage pattern ap- of this investigation. pears to be of adaptive value because it This work was supported by grants from allows the larva to form rapidly, with the the University of Michigan Cancer Re- least possible chance for error from cellu- search Committee and the National Sci- lar displacement due to the stress of the ence Foundation (GB-1035). marine environment. Even here though an interaction is required between the ecte LITERATURE CITED Cather, J. N. 1958 Fixing and staining the derm and endoderm for gastrulation to eggs of invertebrates. Stain Tech., 33: 146-147. occur (Costello, ’45). 1963 A time schedule of the meiotic In mollusks, the D quadrant becomes and early mitotic stages of Ilyanassa. Caryo- even more important and begins its influ- logia, 16: 663-670. 1966 Induction of the shell gland by ence early in development (Clement, ‘52, transplanted polar lobes in Ilyanussa. Biol. ’56, ’62), thereby controlling the symmetry Bull., 131: 306. and pattern of cleavage as well as the gen- Clement, A. C. 1952 Experimental studies on germinal localization in Ilyanassa. I. The role eral form of the embryo by acting as of the polar lobe in determination of the cleav- the “morphogenetic organizer.” The major age pattern and its influence in later develop modifications which transform the troche ment. J. Exp. Zool., 121: 593-626. 1956 Experimental studies on germinal phore into the veliger, the foot, shell, velar localization in Ilyanussa. 11. The development lobe and paired eyes, are dependent on this of isolated blastomeres. J. Exp, Zool., 132: organizer. The ontogeny of the veliger is 427-446. 1958 Discussion of paper by F. E. Leh- then intertwined with that of the trocho- mann in The Chemical Basis of Development. phore, in the early interactions of the polar W. D. McElroy and B. Glass, eds. Johns Hopkins lobe which result in apical tuft formation Press, Baltimore, pp. 92-93, 1960 Development of the Ilyanassa em- and also the third quartet stage morpho- bryo after removal of the mesentoblast cell. genetic interactions. Superimposed on the Biol. Bull., 119: 310. trochophore are such inductive processes 1962 Development of Ilyanassa follow- as that between the ectoderm and endo- ing removal of the D macromere at successive cleavage stages. J. Exp. Zool., 149: 193-216. derm which is required for histogenesis of 1966 Cleavage and differentiation of shell, and there will undoubtedly be many the vegetal half of the Ilyanassa egg after re- others as our information becomes more moval of most of the yolk by centrifugal force. Biol. Bull., 131: 387. complete. Clement, A. C., and J. N. Cather 1957 A tech- The trend in D quadrant specialization nique for preparing whole mounts of veliger is known in at least two other groups, the larvae. Biol. Bull., 113: 340. Collier, J. R., and M. McCann-Collier 1964 cephalopods and the barnacles. In the for- Shell gland formation in the Ilyanassa embryo. mer, the A, B, and C quadrants have been Exp. Cell Res., 34: 512-514. lost or reduced beyond recognition. The Conkkin, E. G. 1897 The embryology of Cre- pidula. J. Morph., 13: 1-226. bilateral divisions of the micromeres of Costello, D. P. 1945 Experimental studies of the remaining D quadrant, which occur germinal localization in Nereis. I. The develop- only in the later stages in the yolky eggs ment of isolated blastomeres. J. Exp. Zool.. 100: 19-66. of some gastropods, are present from the 1955 Cleavage, blastulation and gas- first in the cephalopods. In the barnacles, trulation. In: Analysis of Development. B. H. the A, B, and C quadrants have been com- Willier, P. A. Weiss and V. Hamburger, eds. pletely lost but the cleavage of the re- W. B. Saunders, Philadelphia, chap. 2, pp. 213- mainder is typical spiral cleavage of the 229. Crampton, H. E. 1896 Experimental studies on D quadrant. gasteropod development. Roux’ Arch. Entw.- The modifications and specializations in mech., 3: 1-19. the D quadrant of spiralians evidently rep- Davidson, E. H., G. W. Haslett, R. J. Finney, V. G. Alfrey and A. E. Mirsky 1965 Evidence for resent a very old and plastlc system which prelocalization of cytoplasmic factors affecting has played a major role in the evolution of gene activation in early embryogenesis. Proc. many of the invertebrate phyla. Nat. Acad. Sci., 54: 69ti-703. ILYANASSA SHELL GLAND DEVELOPMENT 223

Fretter, V., and A. Graham 1962 British Proso- of polar lobes and blastomeres as a test of their branch Molluscs. The Ray Society, London. inducing capacities. Biol. Bull., 74: 211-234. Hess, 0. 1956a Die Entwicklung von Halb- Pearse, A. G. E. 1961 Histochemistry, Theoreti- keimen bei dem Siisswasser-Prosobranchier cal and Applied. Second Edition. Little, Brown Bithynia tentaculata. Roux’ Arch. Entw.-mech., and Co., Boston. 148: 336-361. Penners, A. 1937 Regulation am Keim von 1956b Die Entwicklung von Exogas- Tubifex rivulorum Lam. nach Ausschaltung des trula-Keimen bei dem Siisswasser-Prosobranch- ektodermalen Keimstreifs. Zeit. wiss. Zool., ier Bithynia tentaculata. Roux’ Arch. Entw.- f. Bd. 149, S. 86-130. mech., 148: 474488. 1957 Die Entwicklung von Halbkeimen Ranzi, S. 1931 Sviluppo di parti isolate di em- bei dem Siisswasser-Pulmonaten Limnaea stas- brioni di Cefalopodi (Analisi sperimentale dell’ nalis. Roux’ Arch. Entw.-mech., 150: 124-145. embriogenesi). Pub. Staz. Zool. Napoli, 12: 1962 Entwickslungsphysiologie der Mol- 104-146. lusken. Fort. d. Zool., 14: 130-163. Raven. C. P. 1952 Morphogenesis in Limnaea H6rstadius, S. 1937 Experiments on determina- stagnalis and its disturb&& by lithium. J. Exp. tion in the early development of CeTebTatUlUS. ZOO^,, 121: 1-77. Biol. Bull., 73: 317-342. - 1958 Morphogenesis: The Analysis of Hyman, L. H. 1951 The Invertebrates. vol. 11. Molluscan Development. Pergamon Press, New McGraw-Ha, New York. York. Lillie, F. R. 1895 The embryology of the Uni- Verdonk, N. H. 1965 Morphogenesis of the onidae. J. Morph., 8: 569-578. head region in Limnaea stagnalis L. Proeschrift. Manigault, P. 1960 Coquille des mollusques: Rijkuniversiteit Te Utrecht, 1-133. structure et formation. Trait6 de Zoologie, Tome V, fas. 11, pp. 1823-1844. Masson et Cie, Paris. Wilbur, K. M. 1964 Shell formation and regen- Morrill, J. B., and D. M. Gottesman 1960 De- eration. In: Physiology of Mollusca. K. M. velopment of isolated blastomeres of Limnaea Wilbur and C. M. Yonge, eds. Academic Press, palustris. Anat. Rec., 137: 383. New York. Chap. 8, pp. 243-282. Novikoff, A. B. 1938a Embryonic determina- Wilson, E. B. 1904a Experimental studies on tion in the annelid, Sabellaria vulgark. I. The germinal localization. I. The germ regions in differentiation of ectoderm and endoderm when the egg of Dentaliurn. J. Exp. Zool., 1: 1-72. smarated through induced exonastrulation. - 1904b Experimental studies on germi- Biol. Bull., 74: 19&210. nal localization. 11. Experiments on the cleav- - 1938b Embryonic determination in the age-mosaic in Patella and Dentalium. J. Exp. annelid, Subellaria vulgaris. 11. Transplantation ZOO^., I: 197-268.