ELSEVIER Mechanisms of Development 51 (1995) 115-126

Spatially and temporally regulated expression of the LIM class homeobox gene Hrlim suggests multiple distinct functions in

development of the ascidian, Halocynthia roretzi

Shuichi Wada, You Katsuyama, Sadao Yasugi, Hidetoshi Saiga*

Department of Biology, Faculty of Science, Tokyo Metropolitan University, I-I Minamiohsawa. Hachiohji, Tokyo 192-03, Japan

Received 8 December 1994; accepted 24 January 1995

Abstract

Hrlim is a LIM class homeobox gene that was first isolated from the ascidian Halocynthia roretzi. To assess its roles in early development of the ascidian, spatial and temporal expression of Hrlim was examined by whole mount in situ hybridization. This revealed that transcription of Hrlim is activated at the 32-tell stage specifically in the endoderm lineage. Hrlim is also transiently expressed in all notochord precursor cells. Expression in the endoderm lineage continues through to the middle of gastrulation. After gastrulation, Hrlim is expressed in certain lineages that give rise to subsets of cells in the brain and spinal cord. Based on these observations, it is suggested that Hrlim plays multiple distinct roles in ascidian embryogenesis.

Keywords: Ascidian; Gene expression; Homeobox; In situ hybridization; LIM

1. Introduction

Ascidians, members of the chordates, are sessile marine invertebrates which form motile tadpole-shaped larvae. The embryogenesis of ascidians has been noted for several remarkable biological features. Ascidian embryos exhibit similar morphogenesis to that of vertebrates in their early development. The gastrulation and neurulation of ascidians are essentially the same as in vertebrates (Conklin, 1905; Nicol and Meinertzhagen, 1988). It has been suggested that inductive events are required for the formation of the notochord (Nishida, 1992a) and neural tissues (Reverberi et al., 1960; Okado and Takahashi, 1988, 1990; Nishida, 1991) in the asci- dian embryos, which is reminiscent of mesoderm and neural induction in vertebrate embryogenesis, although the mode of induction in the ascidians has not yet been elucidated. Furthermore, ascidian embryos develop into the tadpole larvae with the notochord underlying the neural tube (Kowalevsky, 1866; Conklin, 1905; Katz,

* Corresponding author, Fax: +81 426 77 2559.

1983). These similarities suggest that there may be a common mechanism involved in early development between ascidians and vertebrates. However, little is known about the mechanism for embryonic morphogen- esis in the ascidian development.

On the other hand, ascidian embryos show a unique mode of early development known as mosaic develop- ment. The fate of each blastomere is determined accord- ing to which part of the fertilized egg it comes from. By the llO-cell stage, just before the gastrulation of this organism, the fate of most of the blastomeres is com- pletely restricted (Nishida, 1987). This phenomenon has been interpreted as follows: maternal cell-fate-deter- mining factors called determinants are localized in the fertilized egg cytoplasm and are segregated into appro- priate blastomeres to lead them to cells of a single type specific for a certain tissue (for review, Uzman and Jeffery, 1986; Whittaker, 1987; Satoh, 1987; Nishida, 1992b). Consistent with this hypothesis, it has been shown by cytoplasmic transplantation experiments (Nishida, 1992c, 1993, 1994) that such a specific region is present in the fertilized egg cytoplasm which leads to specific gene expression for epidermis, muscle or

0925-4773/95/$09.50 0 1995 Elsevier Science Ireland Ltd. All rights reserved SSDI 0925-4773(95)00359-Z

116 S. Wada et al. /Mechanisms of Development 51 (1995) 11S-126

endoderm. Similarly, presence of a determinant for initiating gastrulation has been demonstrated by cyto- plasmic deletion and ultraviolet irradiation experiments (for review, Jeffery and Swalla, 1990; Jeffery, 1992). However, the molecular nature of these determinants is yet unknown.

We have decided to undertake molecular analysis of ascidian development. We have chosen to study homeobox genes since homeobox genes have been shown to play key roles during embryogenesis and on- togeny in various animal species. They are subdivided into different classes based on their sequence homology in the homeodomain and other additional properties (Scott et al., 1989). It has been suggested that a number of the HOM/Hox genes are involved in spatial pattern- ing along the body axis (reviewed by McGinnis and Krumlauf, 1992), bicoid class genes are involved in for- mation of head region (e.g. Cho et al., 1992; Driever and Nusslein-Volhard, 1988; Finkelstein et al., 1990; Simeone et al., 1993) and LIM class genes in establishing various cell lineages (e.g. Way and Chalfie, 1988; Freyed et al., 1990; Bourgoin et al., 1992; Cohen et al., 1992). We expect that studies of the homeobox genes in asci- dians should provide clues to understand the nature of ascidian embryogenesis and evolutionary aspects of the mechanisms involved in embryonic morphogenesis in chordates.

Previously, we reported the isolation of the ascidian homeobox gene AHoxl by screening a genomic library of the ascidian, Halocynthia roretzi, with an Anten- napedia homeobox probe (Saiga et al., 1991). AHoxl, which possesses a homeobox similar to that of Droso- phila H2.0, is not expressed in early development but is expressed in the tissues of endodermal origin in juveniles and additionally in coelomic cells in adults (Saiga et al., 1991). In the present study we focused on homeobox genes expressed in early development. By screening a fertilized egg cDNA library with a synthetic oligonu- cleotide probe, we isolated a cDNA clone of a homeobox gene, designated Hrlim. Structural analy- sis revealed that Hrlim belongs to the LIM class homeobox genes that encode proteins containing two cysteine/histidine-rich motifs designated the LIM do- main in addition to a homeodomain (Way and Chalfie, 1988; Karlsson et al., 1990; Freyd et al., 1990). To assess functions of Hrlim. we investigated the spatial and tem- poral expression pattern of Hrlim by whole mount in situ hybridization. We found that transcription of Hrlim is first activated specifically in the endoderm lineage at the 32-cell stage and the expression continues through to gastrulation. Hrlim is also transiently expressed in all the notochord precursor blastomeres before gastrula- tion. After the onset of gastrulation the expression pattern is completely changed. Expression in the mesodermal and endodermal cells disappears and instead Hrlim expression restarts in a pair of cells in the

ectoderm and then in subsets of the cells in the nervous system. These expression patterns suggest that Hrlim plays several distinct roles in the ascidian development. Hrlim may be involved in the determination of cell lineages of endoderm and notochord cells before gastrulation. After gastrulation it may be involved in specification of a distinct group of cells in neural tissues. These possible functions are discussed.

2. Results

2.1. Sequence of the Hrlim cDNA As shown in Fig. 1, nucleotide sequencing and con-

ceptual translation revealed that Hrlim encodes a pro- tein of 514 amino acids, showing the characteristic structure of LIM class homeobox genes with two cystein-rich LIM domains and a homeodomain. Upon sequence comparison of the homeodomain, Hrlim shows high similarity to Xlim-1 and Xlim-3 of Xenopus (Taira et al., 1992), Lim-I (Barnes et al., 1994) and Gsh-4 (Li et al., 1994) of mouse, BK64 and BK87 of Drosophila (Kalionis and O’Farrell, 1993) and linll of nematode (Freyed et al., 1990) (Fig. 2A). Amongst these, Hrlim seems to be most closely related to Xlim-3 and Gsh-4, since their homeodomains are almost identical and they have additionally a similar sequence flanking down- stream of the homeodomain as shown in Fig. 2B.

2.2. Expression of Hrlim: northern analysis Hrlim expression during embryogenesis was examined

by northern blot hybridization. As shown in Fig. 3, a major transcript of approximately 2.6 kb is found throughout embryogenesis, gradually decreasing as development goes on. A smaller transcript of approx- imately 2.4 kb appears transiently through the 64-cell to the gastrula stages. Southern genomic hybridization using the same probe as was used in northern analysis revealed that Hrlim is unique in the Halocynthia genome and no other signal was detected under the same washing conditions (data not shown). This in turn in- dicates that the smaller transcript is also derived from the Hrlim gene.

2.3. Spatial and temporal expression of Hrlim: analyses by whole mount in situ hybridization

In ascidians, the nomenclature for the blastomeres and the cell lineage have been well established (Conklin, 1905; Nishida, 1987). By the llOcel1 stage almost all blastomeres become tissue-restricted, that is, each blastomere has a fate of a single type of tissue (Nishida, 1987). We examined the expression of Hrlim by whole mount in situ hybridization referring to the cell lineage. As described below, we found that the spatial and tem- poral expression of Hrlim in early development can be divided into three phases: maternal expression and zy- gotic expression before and after gastrulation.

S. Wadn et al. /Mechanisms of Development 51 (1995) 115-126 117

1 YmL"rkl*AX~mI 12 1 TT TGACTG AGA cx 5m TMACG AGT CM ATACAT cmrm MTCTC ATO TTC CATCAQOCOCQTAAC AAC MT 74

13 S X W V A A T S I Q Q E 8 I( T A L R V L X D X A S 31 75 ME TTC AT0 OTAOCO @cc ACT TCA MTCAG CMCAC c&c AM ACT QcACrn CAC OTACYC QAAGATAMWC MC 149

I a OMQQA

s v Tc? am3

11 L Aa&-

x ” AM ATQ

P L cc0 OM

T I ACG AK

I( I A-m AK

I r 675 MC A& CCA GiA GiA ACCA MO AAC ATT CCT ACC TtJC TAT MT GOA MC AALT CAC GTT CCT TCA OM TCT ATT TTC 238 D X P L 8 S B AIR X I( T 8 P 0 Id X I Q P 8 T C 8 ?,O QAT AAT CCA cm TCCT TCT CC0 GGT CAT AhA ATA CM CCT TCO ACG TQT TCG 263 D I( L c T 82, QAC ATG CTG TTT ACC 2SS'R T I DXY ILXVQDX PWRIQC LXC X DC 900 COT ATT TK _T CQC TYT ATA CTa AM GTA CM GAC MC cc0 TOG cm Tco CAR TeT CTQ AAA Tee AAC am Tm 313 S A Q L S X X C,S X 0.L V T C T X X R X 975 MT OCT CAO CTA TCOOMAM mT TK TCO COAOOAMT ~n2 am TW MC T T-l-c Tn: MO cm -rK 338 0 T X C T A C 0 I, 0 I P P T X V I R R h Q D 111 V Y

1050 Mu Act - TaC ACC aCA TGr WT CAC WT ATT CCA ccc Act au3 OTC &TA AM cat act CA0 GAC MC OTT TAC 363-•LX CICCXLC t,XXH'=ToDQ9YLLX D

1125 CAC CM OM TGT TT'l TCT mc TTT CTC MC CAC aAhluA~TooMNx3 w OAC CM TTC TAT Trn CM QAa (IAC 388 S X L V C X X D Y X Q A X S II D A D 1 X.0 V X X

1200 MC AQA CM OTT MC MO MO MT TAT GA& CM aCCT AAA TCT CGA aAT oco OAT ATT (1M MT OOF OF0 AhA CGA 413 P R T T T T A X Q L L 'I L X B h Y I Q S x P * R

1275 CCA CGA AC0 MA ATC &CO GCC Ah-4 CM CTT BM ACA TTS AM KA OCG T&C MT CM AOT Ch AM CC0 GCA COC ,38 " V X L Q L 8 8 X T (1 L D II X V V Q V I X Q I I R

1356 CAC ~TACGQ QM CM TTAMT TCC GA&ACAGGC CTC QAC ATQMAQTC QTACM QTT TM TTTCMAATAQAMA 463 A X X X R 2 X R D T Q X Q X I G B I X S X.Q L P

112s - - QM &Q &GA ATC AAl3 MC am ACC aa3 CM CM CQC Too am CAC T?T WC I-CA CGh MC CM cm cc* ,BS 8 a P T S P I8 A P V T T D Q X X X V Q T X.A 0

1500 TCC GQc3 CCC ACA TEA CCC A’lt AGT QCA Ccl’ OTT A’S ACCh WC CM AM AM AkC QTT CM ACA CQA AAT 003 WA 513 T D l

62 224 (1,

299 112 374 13, 449 162 524 187 599 212 674 237 749 262 824 28, 099 312 97. 337

1049 362

1124 38,

1199 412

1274 437

1349 462

1.2. 48,

1499 512

157. 515

1649 1724 1799 1074 1949 2024 2099 217, 2249 232. 2399 2474 2549

Fig. 1. The structure of Hriim of H. roretzi. The nucleotide sequence and deduced amino acid sequence of Hrlim are shown. The amino acids are numbered starting with the putative initiating methionine residue. The homeobox region and two LIM domains are indicated by underlining and boxes, respectively. The nucleotide sequence has been deposited in the DDBJ data base under the accession number D38572.

2.4. Maternal expression of Hrlim before the 32-cell stage In embryos before the 32-cell stage, detection of

Hrlim mRNA was obscured for unknown reasons. After two days of coloring reaction the antisense RNA probe gave a weak signal. There was little if any localization of the signals in unfertilized eggs or embryos of up to the 2-cell stage. Weak localization of Hrlim to anterior ani- mal region was observed at the 4-cell through 16-cell stages (data not shown). Other than this, localization of Hrlim transcript was unclear.

2.5. Zygotic expression of Hrlim before gastrulation 2.5.1. At the 32-cell stage. Hrlim transcripts were first

clearly detected at the 32-cell stage. Signals were observed in three pairs of blastomeres of the vegetal hemisphere. We found that the signals were around nu- clei by comparing with specimens stained with DAPI (data not shown). This indicates that the transcription of Hrlim is activated at this stage (Fig. 4A and B). The three pairs of blastomeres, the A6.1, A6.3 and B6.1 pairs, possess in common a developmental fate that

gives rise to the endoderm, although the A6.3 and B6.1 pairs give rise to trunk lateral cells and endodermal strand as well, respectively (refer Fig. 5 for the cell lineage and the timing of cleavage in early development).

2.5.2. Up to the 64-cell stage. At the end of the 32-cell stage, the blastomeres in vegetal hemisphere except the B6.3 and B6.4 pairs divide to develop the embryos of 44- cell stage (Fig. 5).

At the 44-cell stage, Hrlim expression becomes evi- dent in eight blastomere pairs (Fig. 4C and D). Five pairs out of these are destined for endoderm. These are descendants of the blastomeres that expressed Hrlim at the 32-tell stage. By contrast, Hrlim expression is scarce- ly detectable in A7.6 that gives rise to trunk lateral cells, unlike its counterpart, A7.5. Thus Hrlim expression cor- relates with the endoderm fate.

The other three pairs, A7.3, A7.7 and B7.3, that ini- tiate the expression of Hrlim share a common fate to develop into notochord (Figs. 4C, D and 5), though the B7.3 pair also leads to mesenchyme. Other than these

118 S. Wada et al. /Mechanisms of Development 51 (1995) 115-126

A

Hrlim Xlim-3 Gsh-4 BK64 Xlim- 1 Lim-1 lin-11 BK87 met-3 lmx-1

VKRPRTTITAKQLHTLKSAYNQSPKPARHVRHQLSSETGLDMRWQVNFQNRRAKHKRIK

apterous T--M--SFKHH--R-M--YFAINHN-DAKDLK---QK---QK---PK--L------A---~-~ 47% LB-2 T--M--SFKHH--R-M--YFAINHN-D~D~--AQK---~--L------A---~-NL 45% 191-l TT-V--VLNH ---HT-RTC-AAN-R-DALMK---VEM---SP--IR-----K-C-D-KRS 43%

B Hrlim Xlim-3 Gsh-4 [ 1

RDTGRQRWGHFFS home?- K-A__-___QY_R domain K-A--H-_-Q-~

Fig. 2. Comparison of the homeodomain and its downstream flanking region of Hrlim with those of the LIM class homeobox genes. (A) Comparison of the homeodomain of Hrlim with those of other LIM class homeobox genes. Percent identity to Hrlim homeodomain is indicated in the right column, (B) Comparison of the flanking region downstream of the homeodomain of Hrlim and two vertebrate LIM class homeobox genes. In (A) and (B), the identical residues are shown by a dash (-).

three pairs, no blastomeres give rise to notochord cells. Therefore, Hrlim is expressed in all the notochord precursors at this stage.

The expression pattern of Hrlim in the earlier half of the 64-cell stage is the same as that at the 44-tell stage (Fig. 4E and F) except that the expression level increases in all the Hriim expressing blastomeres. In the latter half

2.0-

1.6-

1.4-

f.O-

Fig. 3. Expression of Hrlim examined by northern blot hybridization. Each lane contains 10 pg of poly(A) RNA from embryos of various stages as indicated at the top of the blot. The blot was hybridized with Hrfim probe (nucleotides 957- 1430 in Fig. I) and washed in 0. I x SSC at 60°C. The size marker used was XDNA fragments digested with EcoRI and HindIII.

of the 64-&l stage, signals become evident in the B7.5 pair that gives rise to the muscle and endoderm (arrows in Fig. 4G). Thus, by this time, all the endoderm- producing blastomeres have started transcription of Hrlim (Fig. 4H). In contrast, the expression in the notochord precursors is decreased at this stage: especial- ly in the B7.3 pair where the Hrlim expression becomes hardly detectable (Fig. 4G and H).

2.5.3. Up to the IIO-cell stage. By the 76-tell stage, three pairs of the notochord progenitors divide to give rise to five pairs of the notochord precursors and a pair of the mesenchyme precursors, while endoderm pro- genitors do not divide in this period (Fig. 5).

The endoderm precursors, except the B7.5 pair, con- tinue to express Hrlim at the 76-tell stage. In com- parison to the 64-cell stage, the signal becomes stronger and extends to the cytoplasm (Fig. 41). In the B7.5 pair, Hrlim expression decreases gradually unlike the other endoderm progenitors. In the six pairs of descendants of the notochord precursors, Hrlim expression becomes scarcely detectable (Fig. 41 and I). Thus the expression of Hrlim in the notochord lineage is restricted to the seventh generation of the cleavages in development (Fig.

5). Embryos of the 1 IO-cell stage exhibit intense expres-

sion of Hrlim in the endoderm precursors except the B7.5 pair (Fig. 4K and L). In the ascidian embryos, gastrulation starts shortly after the 1 IO-cell stage, begin- ning with invagination of the A7.1 and B7.1 pairs. Soon after the beginning of invagination, expression of Hrlim in the endoderm progenitors decreases rapidly and

S. Wada et al. /Mechanisms of Development 51 (1995) 115-126 II9

Fig. 4. Localization of Hrlim transcripts by whole mount in situ hybridization in H. roretzi embryos of the 32-tell through the early gastrula stages. From left to right, first and third colunms show vegetal views of cleared specimens. The hybridization signals are detected as dark purple coloring. Second and fourth columns show schematic representation of vegetal views of the embryos and H&n expression. In the right half of diagrams, the signals are shown as black spots and in the left half the designations of blastomeres are indicated. The blastomeres that give rise to endodetm and notochord cells are colored by orange and pink, respectively. Note that the colored blastomeres are not necessarily destined exclusively for a single fate. With all specimens and diagrams, anterior is to the top. (A and B) A 32-tell stage embryo and its’ diagram. (C and D) A 44-tell embryo and its’ diagram. (E and F) An embryo in the earlier half of the 64-tell stage and its’ diagram. By the 64-tell stage, the B6.3 pair, the B6.4 pair and the animal blastomeres pass through the sixth division. At this stage, expression pattern of Hrlim is identical to that of the 44-ce.ll stage embryos except that the level of.expression has increased. (G) and (H) show an embryo in the later half of the 64-tell stage and its’ diagram. At this stage, Hrlim expression is also detected in the B7.5 pair (arrows) that leads to endoderm. (I and J) A 76-tell stage embryo and its’ diagram. By the 76-tell stage, certain vegetal blastomeres including the notochord precursors undergo the seventh cleavage, while the endodenn precursors remain uncleav- ed. (K and L) An embryo at the early I IO-cell stage and its’ diagram. By the I IO-cell stage, one blastomere pair in the vegetal hemisphere and all animal blastomeres undergo the seventh cleavage. Gastrulation starts just after this stage. (M) An embryo at the early gastrula stage. (N) An embryo at the middle gastrula stage. Hrlim expression in the endoderm precursors decreases rapidly after the onset of gastrulation.

becomes undetectable by the middle of gastrulation (Fig. 4M and N). Hrlim expression between 32-tell and 1 lo-cell stages is summarized in Fig. 5.

2.6. Hrlim expression after gastrulation Whole mount in situ hybridization with tadpole lar-

vae revealed that Hrlim was also expressed in restricted regions of the nervous system. The nervous system in ascidian larvae consists of the brain at the anterior and the spinal cord at the posterior. The brain contains two sensory organs, the otolith, a gravity sensor, and ocellus, a photo sensor, in the anterior and posterior parts,

120 S. Wada et al. /Mechanisms of Development 51 (1995) 115-126

Fig. 5. Zygotic Hrlim expression as summarized referring to the cell lineage up to the I IO-cell stage. The cell lineage is according to Nishida ( The lineage and the names of blastomeres in which Hrlim mRNA was detected are indicated by bold lining and tilled boxes.

%J

A -I b5.4 epidermis

Ab8.18 epidermis

b8.20 epidermis

1987).

respectively (Dilly, 1961). Under the ocellus there are several lens cells (Dilly, 1961,1964). In the larvae, hy- bridization signals were detected in the ventral region of the brain and a part of the spinal cord (Fig. 6A and B). The origin of the expression of Hrlim in these cells could be traced back to the earlier stages as described below.

2.7. Expression in the spinal cord Expression in the spinal cord starts in late gastrula

stage embryos. At the late gastrula stage, when Hrlim expression has disappeared in the vegetal hemisphere, Hrlim expression becomes detectable again, not in the vegetal but in the animal hemisphere, in a pair of single cells located symmetrically on the anterior lip of gastrocoel (arrowheads in Fig. 6C). The Hrlim express- ing cell is one of the presumptive nervous system as judged by the position. This expression continues throughout the neural plate stage (Fig. 6D). At the early neurula stage, a pair of cells on the neural fold begins

to express Hrlim. These cells are located posterior to the Hrlim-expressing cells described above. Thus the stain- ing is observed as two pairs of symmetrically locating spots on the neural folds (Fig. 6E). The two pairs of cells expressing Hrlim move toward the midline as neurula- tion goes on (Fig. 6F). At the early tailbud stage, the two pairs of signals are localized in the dorsal region connec- ting trunk and tail (Fig. 6G and H). The posterior ex- pression decreases gradually and disappears at the end of this stage. In the middle and late tailbud stage em- bryos, the anterior Hrlim expression remains in the anterior-most region of the spinal cord (Fig. 7A-C). The expression is continuously observed until the swim- ming larvae stage (Figs. 7D, E and 6A).

2.8. Expression in the brain In the larva, expression of Hrlim is observed in the

cells underlying ocellus. The expression in the brain lineage starts at the tailbud stage. In early tailbud stage

Fig. 6. Hrlim expression visualized by whole mount in situ hybridization in nervous system of the larva and embryos. With specimens (C-H), dorsal side is shown and anterior to the top. (A) A larva hybridized with anitisense probe and (B) a larva with sense probe. (C) A late gastrula embryo. Arrowheads on the anterior margin of gastrocoel indicate the signals of spinal cord lineage. (D) A neural plate stage embryo. (E) An early neurula embryo. (F) A late neurula embryo. The neural tube is almost closed. A pair of fuzzy signals locating posteriorly on the neural tube is newly observed at this stage. (G) An early tailbud stage embryo. Arrows in the center of trunk indicate a pair of weak staining newly detected. (H) An early tailbud stage embryo. The developmental stage of this embryo is later than that of(G). The posterior pair of signals on the neural tube has almost disappeared at this stage. Arrows indicate the signals of brain lineage. Hrlim expression is detectable on both sides but shortly after this stage the expression on the left side becomes hardly detectable (see Fig. 7).

embryos, expression is first detectable as a symmetric pair of faint signals on the dorsal middle of the trunk (arrows in Fig. 6G and H). By the middle tailbud stage, however, expression becomes asymmetric. The signal on the left side becomes hardly detectable while, by con- trast, in the cells of the right side, Hrlim expression increases (Fig. 7A). This asymmetric expression of Hrlim in the trunk is most clearly observed at the late tailbud stage (Fig. 7B). As development goes on, the signal on the right side becomes much stronger and the Hrlim expression of the left side shows up again (Fig. 7C and D).

In normal development, pigmentation occurs first in otolith and then in ocellus (Nishida and Satoh, 1989). In

the larvae of the otolith pigmentation stage, the two staining spots of Hrlim expression on both sides come together and the expression continues in the posterior region of the brain, the region underlying the presump- tive ocellus (Fig. 7E). Swimming larvae also showed Hrlim expression in the ventral region of the brain at the periphery of the ocellus (Fig. 6A). The ocellus cell itself and lens cells do not express Hrlim as examined with Nomarsky optics (data not shown).

In addition, Hrlim expression is observed transiently in embryos of the latter half of the tailbud stage. The ex- pression is detected as a symmetrical pair of signals in the region anterior to the brain lineage expressing Hrlim (arrowheads in Fig. 7A and B).

122 S. Waab et al. /Mechanisms of Development 51 (1995) iIS-126

Fig. 7. Hrlim expression in the nervous system visualized by whole mount in situ hybridization of the middle tailbud stage embryo through larva. (A) A middle tailbud stage embryo. (B) a late tailbud stage embryo. Arrowheads in (A) or (B) indicate signals observed in the middle or late tail bud stage embryos transiently. (C) a late tailbud stage embryo. The developmental stage of this embryo is later than that of (B). (D) A larva of early stage. (E and F) A larva at the otolith-pigmentation stage hybridized with an antisense (E) or sense (F) probe. In panels (E) and (F), black spots in the head are the pigmented cells in the otolith.

3. Dlscusslon

3.1. Ascidian LIM class borneobox gene

In this paper, we have reported the isolation and ex- pression of an ascidian LIM class homeobox gene, Hrlim. More than a dozen LIM class homeobox genes have been isolated from nematode (Freyd et al., 1990; Way and Chalfie, 1988), Drosophila (Cohen et al., 1992; Bourgouin et al., 1992; Kalionis and O’Farrell, 1993) and vertebrates (Karlsson et al., 1990; Ericson et al., 1992; German et al., 1992; Taira et al., 1992, 1993; Xu et al., 1993; Barnes et al., 1994; Li et al., 1994). Studies

on their expression have shown that they are expressed in a cell-type specific manner and it has been suggested that the homeobox genes of this class might be involved in the specification and differentiation of specific lineages. It is possible that LIM class homeobox genes of ascidians also play such a role as has been suggested in other animal species.

Hrlim is expressed both maternally and zygotically. Whole mount in situ hybridization revealed that zygotic expression of ascidian Hrlim is further divided into two distinct phases. From the 32-tell stage to the early gastrula stage, Hrlim is expressed in the endoderm and notochord precursors. After gastrulation, Hrlim is ex-

S. Wada et al. /Mechanisms of Development 51 (1995) 115-126 123

pressed in specific regions of nervous system in larvae, These lineage-specific expression patterns of Hrlim are compatible with former observations on other LIM class homeobox genes and suggest that Hrlim is involved in the determination of the presumptive endoderm and notochord blastomeres and specific cell types of the nervous system.

3.2. Maternal transcripts of Hrlim do not constitute deter- minants

Maternal transcripts of Hrlim gave weak signals in the anterior animal hemisphere before 32-&l stage. In asci- dian embryos, it has been demonstrated that a certain cytoplasmic region in the egg has activities which lead the cells therewith to an epidermis, muscle or endoderm fate (Whittaker, 1990; Nishida, 1992c, 1993, 1994). For example, by cytoplasmic transfer experiments using alkaline phosphatase activity as an endoderm marker, Nishida showed that cytoplasmic factors which lead to an endoderm fate are present in the unfertilized egg. After fertilization, they become localized to the vegetal pole in the first phase of ooplasmic segregation, spread over the entire vegetal hemisphere during the second phase of segregation, and they are inherited by the four vegetal blastomeres in the 8-cell stage embryos (Nishida 1993). These properties allow this cytoplasmic activity to be regarded as determinants. Similar analyses have been carried out on cytoplasmic activities that lead cells to muscle and epidermis (Nishida 1992c, 1994). How- ever, the localization pattern of the maternal Hrlim transcripts does not correspond to that of any of these cytoplasmic factors. It appears more likely that the zy- gotic expression of Hrlim is under the control of the determinants for endoderm, since transcription of Hrlim starts at the 32&l stage specifically in the endoderm lineage. Considering the timing of the onset of zygotic expression of Hrlim, the expression may be under a direct or otherwise very close control by the deter- minants. Studies on the regulatory mechanisms for Hrlim expression in the endoderm precursors will give us some information about the molecular nature of the endoderm determinants.

3.3. Zygotic expression of Hrlim in the endoderm lineage Transcription of Hrlim is first activated at the 32-cell

stage specifically in the blastomeres that give rise to the endoderm lineage and expression continues until onset of gastrulation. This suggests that zygotic expression is involved in the determination of the endoderm lineage. Intense expression of Hrlim is observed in these en- doderm precursors just before gastrulation but the ex- pression disappears shortly after the beginning of gastrulation. In ascidian embryogenesis, gastrulation begins as the invagination of endoderm precursors that locate in the center of the vegetal hemisphere, followed by the surrounding vegetal blastomeres. The in-

vaginated endoderm precursors develop into the ventral head region of the tadpole-shaped larvae (Bates and Jef- fery, 1987; Nishida, 1987). Thus it is possible that the Hrlim expression in the endoderm precursors of this stage might be involved in the initial steps of invagina- tion or in the formation of the future head region. This is reminiscent of the expression pattern of Xlim-Z around the period of gastrulation in Xenopus (Taira et al., 1992). Interestingly, Hroth, which we have recently isolated and identified as an ascidian homologue of the Drosophila orthodenticle gene (Finkelstein et al., 1990) or the mouse Otx genes (Simeone et al., 1993), is also ex- pressed in the endoderm precursors after the 64-cell stage (unpublished observation). Co-expression of Hrlim and Hroth in the endoderm precursors before gastrulation may suggest that these cells play roles in establishment of body axis of the ascidian embryo.

3.4. Specijication of the notochord lineage and Hrlim ex- pression

Hrlim is also expressed transiently in the notochord lineage. In this lineage, As-T is also expressed. This is an ascidian homologue of the vertebrate Brachyury(T) gene that is expressed exclusively in the notochord lineage up to the early tailbud stage and has been pro- posed to specify the notochord lineage (Yasuo and Satoh, 1994). As-T mRNA is first detected at the 64-tell stage in the anterior-vegetal (A-line) notochord precur- sor cells and at the 1 IO-cell stage in the posterior-vegetal (B-line) precursor cells when the fate of these cells has been destined exclusively for the notochord (Yasuo and Satoh, 1994). Strictly speaking, these notochord precur- sors in the A- or B-line appear at the 44-tell or 76-&l stage, remaining without cleavage up to 64-tell or 1 IO- cell stage, respectively (Nishida, 1987). The As-T expres- sion at these stages, however, has not been reported (Yasuo and Satoh, 1994). Since Hrlim transcription begins at the 44-cell stage both in A- and B-line precur- sors, Hrlim expression precedes or starts at least not later than the As-T expression and disappears at the 76- cell stage. These results suggest that the transient expres- sion of Hrlim may play a role in the initial stage of deter- mination of the notochord lineage. It is a further interesting question to address whether As-T is under the control of Hrlim.

3.5. Hrlim expression is restricted to certain subsets of cells in the nervous system

In the second phase of the zygotic expression, Hrlim transcripts are observed in several specific regions of the nervous system of larvae. In the ascidian larvae, the structure of the central nervous system and cells con- stituting the central nervous system are not well describ- ed, so we could not determine which cell types are positive for Hrlim expression.

In vertebrates, it has been reported that many

124 S. Wadn et al. /Mechanisms of Development 51 (1995) 115-126

members of the LIM class homeobox genes are express- ed in specific regions in the central nervous system. Zsl-Z is expressed in motor neurons in the ventral region of the spinal cord (Ericson et al., 1992), LH-2 in diencephalon and telencephalon (Xu et al., 1993), Lim-I in the lateral diencephalon, hindbrain and dorsal spinal cord (Barnes et al., 1994), Xlim-3 in postmitotic cells of the retina, hindbrain and spinal cord (Taira et al., 1993) and so on. Likewise, Hrlim may also play a role in the commitment and specification of subsets of cells in the nervous system.

3.6. Asymmetric expression of Hrlim in the brain lineage A remarkable feature of Hrlim expression in the brain

lineage is its asymmetry. Hrlim expression in brain lineage is first detectable as two symmetric spots in the middle of the trunk at the early tailbud stage and soon after the expression on the left side disappears while the expression of the right side counterpart becomes stron- ger. In the ascidian brain, it has been reported that a pair of pigment cell precursors develops into otolith and ocellus in a complementary manner (Nishida and Satoh, 1989). During neurulation, the two pigment cell precur- sors that have been positioned bilaterally on the dorsal side of the embryos move to meet and align on the antero-posterior midline. The two cells appeared to be equivalent in developmental potential, that is, either one of the two cells that positions posteriorly becomes ocellus and the other becomes otolith. Thus this proce- dure is independent of their initial right-left position (Nishida and Satoh, 1989). By contrast, asymmetry of Hrlim expression appears to be invariably established. We examined a number of specimens for the expression and found that the expression pattern was always the same. At present it is unclear what the asymmetric ex- pression pattern of Hrlim means. Judging from the posi- tion of the cells expressing Hrlim in the larva in reference to the cell lineage (Nishida, 1987), they are likely to be derived from the a8.17 pair. The pigment cells are derived from the a8.25 pair juxtaposed to the a8.17 cells. The timing when the Hrlim expression is first detected appears to overlap the period that two pigment cell precursors move to align on the midline. These observations suggest that there may be some close rela- tionship between the development of the two pigment cells and the asymmetrical expression of Hrlim.

4. Materials and methods

4.1. Ascidians Adult ascidians, Halocynthia roretzi were purchased

from fishermen near Asamushi Marine Biological Sta- tion, Tohoku University, Aomori, Japan and Otsuchi Marine Research Center, Ocean Research Institute, University of Tokyo, Iwate, Japan. Naturally spawned eggs were fertilized with a suspension of sperm from other individuals and the fertilized eggs were raised in

filtered sea water at l l-13°C. Embryogenesis went on synchronously in all batches of eggs. Samples at appro- priate stages were collected by low speed centrifugation and were fixed for whole mount in situ hybridization or were frozen for RNA extraction.

4.2. cDNA library construction and molecular cloning of Hrlim

cDNA was synthesized from poly(A) RNA isolated from unfertilized egg using a cDNA synthesis kit (Phar- macia). Synthesized cDNA was ligated to the hgtl 1 vec- tor. The ligated materials were packaged in vitro as described previously (Saiga et al., 1991). The cDNA li- brary was screened using degenerate oligonucleotides complementary to the region encoding the third helix of the Antennapedia class homeodomain. The nucleotide sequence was determined by the cycle sequencing meth- od using fluorescent dye primers (ABI) and the ABI 392A sequencing apparatus.

4.3. Isolation and analysis of nucleic acia5 Preparation of genomic DNA and poly(A) RNAs

from various stage embryos was carried out as described previously (Saiga et al., 1991). Northern and Southern blot analyses were also carried out as described previously (Saiga et al., 1991).

4.4. Whole mount in situ hybridization Whole mount in situ hybridization was carried out

according to Holland et al. (1992) with some modifications.

Embryos and larvae were fixed with 4% parafor- maldehyde in 0.5 M sodium chloride, 0.1 M MOPS (pH 7.5) at 4°C overnight, dehydrated in 30%, 50%, 70% ethanol (10 min each) and stored in 70% ethanol at -20°C. The embryos were manually dechorionated under a stereomicroscope just before further treatment. All further steps were performed at room temperature unless otherwise stated. After rehydration by successive incubation in 50%, 30% ethanol and in PBST (PBS con- taining 0.1% Tween20) three times (10 min each), the specimens were treated with 20 &ml (larvae) or 5 &ml (eggs and embryos) proteinase K in PBST (15 min, 37°C) and digestion was stopped by washing with 2 mgml glycine in PBS for 5 min and with PBS twice (5 min each). The specimens were post-fixed with 4% paraformaldehyde in PBS for 1 h, followed by washing with PBST twice (5 min each) and distilled water for 5 min. Acetylation was carried out in 0.1 M triethanol- amine (pH 8.0) supplemented with 0.27% acetic anhyd- ride for 10 min and the specimens were washed in PBST twice (5 min each). The specimens were incubated in prehybridization buffer (50% formamide, 5 x SSC, 100 &ml tRNA, 50 &ml heparin, 1% SDS) for 1 h at 50°C. The prehybridization buffer was replaced with hybrid- ization buffer containing 0.5 &ml digoxygenin (DIG)- labeled anti-sense or sense transcript. Hybridization

S. Wada et al. /Mechanisms of Development 51 (1995) 115-126 125

reaction was carried out at 50°C overnight. At the end of the hybridization, the specimens were washed in 50% formamide, 5 x SSC, 1% SDS (2 x 20 min, SOOC), then 50% formamide, 2 x SSC, 1% SDS (2 x 20 rnin, 37”C), and 2 x SSC, 0.1% Tween20 (2 x 5 min); treated with 20 &ml RNase A in 2 x SSC (20 min, 37°C); and wash- ed in 2 x SSC, 0.1% Tween20 (3 x 5 min, 37°C; 2 x 20 min, 50”(Z), 0.2 x SSC, 0.1% Tween20 (20 min, SO’C) and in PBST for 5 min.

RNA hybrids were detected immunohistochemically. After blocking in 0.1 M Tris-HCl (pH 7.5), 0.15 M sodi- um chloride supplemented with 0.5% blocking reagent in the kit (blocking buffer), the specimens were in- cubated with 1: 2 000 alkaline phosphatase-conjugated anti-DIG antibody (Boehringer Mannheim) in the above blocking buffer (overnight, 4°C). The specimens were washed with PBST four times (20 min each) and alkaline phosphatase buffer (TMN buffer plus 0.1% Tween20) three times (10 min each). Signal detection was performed in TMN buffer (100 mM sodium chlo- ride, 50 mM magnesium chloride, 100 mM Tris-HCl, pH 8.0) with 4.5 ~1 NBT/ml and 3.5 ~1 BCIP/ml added, following the supplier’s instruction (Boehringer Man- nheim DIG RNA Detection Kit), except that 2 mM levamisole was included. When satisfactory signals over the background were obtained, the solution was replac- ed with PBST. In this procedure, the sense probe gave no significant background after coloring reaction for 2 days. For clearing, the specimens were dehydrated by in- cubations in 30%, 50%, 70%, 90%, 95%, 2 x 100% ethanol (10 min each) and mounted in 2:l benzyl ben- zoate: benzyl alcohol.

4.5. Preparation of probes A cDNA fragment (nucleotide 1-2155 in Fig. 1A)

was cloned in the EcoRl site of the plasmid Bluescript KS+ (Stratagene). The plasmid DNA was digested with Hind111 or BarnHI and used as template for in vitro transcription. The RNA probe was synthesized using Boehringer Mannheim DIG RNA labeling Kit accor- ding to the supplier’s instructions. After incubation, the reaction mixture was treated with RQ DNase (Promega) and the transcripts were hydrolyzed by incubation in 40 mM sodium hydrogen carbonate, 60 mM sodium car- bonate at 60°C to an average length of 300 nucleotides.

For northern or Southern hybridization, a cDNA fragment (nucleotide 957-1430 in Fig. 1A) was labeled with [a-32P]dCTP using a random primer labeling kit (Takara).

Oligonucleotide probe was 5’-G(GA)T(TC)TG(GA)- AACCA(AGCT)A(CT)(CT)T(GT)G-3 ’ labeled with [y-32P]ATP using polynucleotide kinase.

Acknowledgements

The authors wish to thank Drs. N. Satoh and H. Yasuo, Kyoto University and Dr. II. Ueda, Hokkaido

University for their help in early stages of in situ hybrid- ization work. Thanks are also due to Dr. Numakunai, Asamushi Marine Biological Station, Tohoku Univer- sity and staffs in Otuchi Marine Research Center, Uni- versity of Tokyo for providing us with research facilities. The authors thank Dr. Paul J. Scatting, University of Nottingham for his critical reading of the manuscript. This work was supported by Grants-in-Aid for Scientific Research on Priority Areas from the Ministry of Educa- tion, Science and Culture of Japan and by the Fund for Special Research Project at Tokyo Metropolitan University to H.S.

References

Barnes, J.D., Crosby, J.L, Jones, CM., Wright, C.V.E. and Hogan, B.L M. (1994) Dev. Biol. 161, 168-178.

Bates, W.R. and Jeffery, W.R. (1987) Dev. Biol. 124, 65-76. Bourgoin, C., Lundgren, SE. and Thomas, J.B. (1992) Neuron 9,

549-561. Cho, K.W.Y., Blumberg, B., Steinbeisser, H. and De Robertis, E.M.

(1991) Cell 67, 111 l-l 120. Cohen, B., McGuffin, M.E., Pfrifle, C., Segal, D. and Cohen, SM.

(1992) Genes Dev. 6, 715-729. Conklin, E.G. (1905) J. Acad. Nat. Sci. (Phila.) 13, I-119. Dilly, N. (1961) Nature 191, 786-787. Dilly, P.N. (1964) Q. J. Microsc. Sci. 105, 13-20. Driever, W. and Nusslein-Volhard, C. (1988) Cell 54, 95-104. Ericson, J., Thor, S., Edlund, T., Jesse], T.M. and Yamada, T. (1992)

Science 256, 1555- 1560. Finkelstein, R., Smouse, D., Capati, T.M., Spradling, AC. and Per-

rimon, N. (1990) Genes Dev. 4, 1516-1527. Freyed, G., Kim, S.K. and Horvitz, H.R. (1990) Nature 344.876-879. German, MS., Wang, J., Chadwick, R.B. and Rutter, W.J. (1992)

Genes Dev. 6, 2165-2176. Holland, P.W.H., Holland, L.Z., Williams, N.A. and Holland, N.D.

(1992) Development 116, 653-661. Jeffery, W.R. (1992) Development 1992 Supplement, 53-63. Jeffery, W.R. and Swalla, B.J. (1990) Sem. Cell Biol. 1, 373-381. Kalionis, B. and O’Farrell, P.H. (1993) Mech. Dev. 43, 57-70. Karlsson, O., Thor, S., Norberg, T., Ohlsson, H. and Edlund, T.

(1990) Nature 344, 879-882. Katz, M.J. (1983) Biol. Bull. 164, l-27. Kowalevsky, A. (1866) Memoires L’Acad. Imperiale Sci. St.

Petersbourg 10, 1 - 19. Li, H., Witte, D.P., Branford, W.W., Aronow, B.J., Weinstein, M.,

Kaur, S., Wert, S., Singh, G., Schreiner, CM., Whitsett, J.A., Scott, Jr, W.J. and Potter, S.S. (1994) EMBO J. 13, 2876-2885.

McGinnis, W. and Krumlauf, R. (1992) Cell 68, 283-302. Nicol, D. and Meinertzhagen, LA. (1988) Dev. Biol. 130, 721-736. Nishida, H. (1987) Dev. Biol. 121, 526-541. Nishida, H. (1991) Development 112, 389-395. Nishida, H. (1992a) Roux’s Arch. Dev. Biol. 201, 81-87. Nishida, H. (1992b) Dev. Growth Differ. 34, 253-262. Nishida, H. (1992~) Development 116, 521-529. Nishida, H. (1993) Development 118, l-7. Nishida, H. (1994) Development 120, 235-243. Nishida. H. and Satoh, N. (1989) Dev. Biol. 132, 355-367. Okado, H. and Takahashi, K. (1988) Proc. Nat]. Acad. Sci. USA 85,

6197-6201. Okado, H. and Takahashi, K. (1990) J. Physiology 427, 603-623. Reverberi, G., Ortolani, G. and Farinella-Ferruzza, N. (1960) Acta

Embryo]. Morph. Exper. 3, 296-336. Saiga, H., Mizokami, A., Makabe, K.W., Satoh, N. and Mita, T.

(1991) Development 111, 821-828.

126 S. Wada et al. /Mechanisms of Development 51 (1995) 115-126

Satoh, N. (1987) BioEssays 7, 51-56. Scott, M.P., Tamkun, J.W. and Hartsell, III, G.W. (1989) Biochim.

Biophys. Acta 989, 25-48. Simeone, A., Acampora, D., Mallamaci, A., Stomaiuolo, A., D’Apice,

M.R., Nigro, V. and Boncinelli, E. (1993) EMBO J. 12,2735-2747. Taira, M., Jemrich, M., Good, P.J. and Dawid, I.B. (1992) Genes Dev.

6, 356-366. Taira, M., Hayes, W.P., Otani, H. and Dawid, I.B. (1993) Dev. Biol.

159, 245-256.

Uzman, J.A. and Jeffery, W.R. (1986) Cell Differ. 18, 215-224. Way, J.C. and ChalIie, M. (1988) Cell 54, 5-16. Whittaker, J.R. (1987) Amer. Zool. 27, 607-622. Whittaker, J.R. (1990) Biol. Bull. 178, 222-230. Xu, Y., Baldassare, M., Fisher, P., Rathbun, G., Oltz, E.M., Yan-

copoulos, G.D., Jessell, T.M. and Alt, F.W. (1993) Proc. Natl. Acad. Sci. USA 90, 227-231.

Yasuo, H. and Satoh, N. (1994) Dev. Growth Differ. 36, 9-18.