The Journal of Neuroscience, August 1995, 15(8): 5551-5559

The Leech Gene Lox4 May Determine Segmental Differentiation of Identified Neurons

Victoria Y. Wong, Gabriel 0. Aisemberg, Wen-Biao Gan, and Eduardo R. Macagno Department of Biological Sciences, Columbia University, New York, New York 10027

We cloned and characterized a new leech homeobox gene, pression and function of regulatory proteins involved in con- Lox4, a homolog of the Drosophila genes Ultrabithorax and trolling early developmentin the nervous system. abdominal-A. Lox4 has a complex and dynamic pattern of The leech Hirudo medicinalis provides a uniquely advanta- expression within a series of segmentally homologous geous system to study these problems at the level of single, neurons. These include a pair of specialized motor neurons identified neurons. Leecheshave a simple CNS comprising a of one segmental ganglion, the rostra1 penile evertors ventral nerve cord with 32 segmentalneuromeres and a supra- (RPEs), and their segmental homologs in other midbody esophagealganglion of nonsegmentalorigin. Four neuromeres ganglia. During gangliogenesis, Lox4 was expressed within fuse to form the head ganglion and seven to form the tail gan- this series of neurons in three different temporal patterns: glion. The remaining 21 neuromeresform the midbody ganglia, (1) it was never expressed in the RPE homologs of ganglia which have about 400 neuronseach (Macagno, 1980). Many of l-3; (2) it was expressed in the RPEs during gangliogene- these neuronscan be reliably identified in the embryo and are sis, but was turned off when these neurons started to dif- amenableto experimental manipulation at early developmental ferentiate after gangliogenesis; and (3) it was expressed in stages.They differentiate in very similar early embryonic en- the RPE homologs of segments 4-5 and 7-21 during gan- vironments, yet many segmentallyhomologous neurons exhibit gliogenesis and the subsequent period of axonogenesis. distinct, segment-specificpatterns of differentiation (Glover and We found that these neurons have three distinct peripheral Mason, 1986; Gao and Macagno, 1987a,b; Jellies et al., 1987; axonal morphologies that correlate with the three temporal Stewart et al., 1991; Jellieset al., 1992; reviewed in French and patterns of expression. Our results suggest that the hom- Kristan, 1994). Although a role for cell-cell interactions during eobox gene Lox4 may determine neuronal identities within developmenthas been demonstratedfor someof theseneurons, this series of segmental homologs. it is likely that certain fates and pathways of differentiation are [Key words: neuron, axon, homeodomain, Hirudo medi- initially determinedby specific, cell-autonomousfactors that in cinalis, Hex genes, CNS, morphology] turn bias the responsesto environmentalsignals. The Hox/HOM genesand other related homeobox-containing There is more cellular diversity within the nervous systemthan genescode for transcription factors which are highly conserved in any other system in an animal. Although essentialfor under- among a wide variety of species(reviewed in Kappen and Rud- standing nervous systemdevelopment and pathology, the mech- dle, 1993; Krumlauf, 1994). In addition to their early function anismsthat generatethis diversity are extremely complex and, in determining anteroposterior(A-P) positional information, the as a consequence,poorly understood. A useful model for ad- Hox/HOM genesare later expressedat high levels in the ner- dressingsome basic aspectsof this issueis the regional differ- vous system (reviewed in Doe and Scott, 1988; Kessel and Gruss, 1990; Boncinelli et al., 1993), where they are likely to entiation in invertebratesof segmentallyrepeated neurons. These neuronsderive from precursorslocated in similar cellular envi- be involved in the determination of neuronalidentities. ronments,but neverthelessacquire unique identities. Although Previouswork in this laboratory hasdetermined that leechHex/ both cell-cell interactions and autonomousfactors appear to HOM genesare expressedin segmentallyrepeated subsets of neu- contribute to their different developmentalfates, the specific in- rons within their particular A-P domains(Wysocka-Diller et al., fluence of the environment on neuronalidentity dependslargely 1989; Aisemberg et al., 1993, 1995; Aisemberg and Macagno, on the intrinsic properties of an individual cell. To gain a better 1994).We presenthere the characterizationof Lo,&, a leechHex/ understandingof the function of autonomousfactors in neuronal HOM gene that displays a complex and dynamic pattern of ex- differentiation, a fruitful approach has been to examine the ex- pressionin the CNS and offers a very interestingmodel for the study of differentiation within a seriesof segmentallyhomologous neurons.The resultswe present here suggestthat Lox4 plays a Received Feb. 16, 1995; revised Apr. 14, 1995; accepted Apr. 20, 1995. key role in the specificationof three different cell fates amongat We thank Laura Wolszon for her critical reading of the manuscript, Tim least one group of segmentallyhomologous neurons. Gershon for removing the male organ from experimental embryos, Mike Ni- tabach for immunizing the rabbits and for helpful discussions, and Nik Necles Materials and Methods for photographic work. We also thank Marty-Shankland for his help with the initial in sine hybridization experiments and Anne Duggan for useful comments Animals. Embryosobtained from our colony of Hirudo medicinalis on an earlier version of the manuscript. Lastly, we thank J.-L. Picard and K. were maintainedat 23°Cwithin their cocoonsuntil embryonicday 8 Janeway. This work was supported by NIH Grant HD-20954. (E8), and in sterileartificial springwater (0.5 gm/literInstant Ocean, Correspondence should be addressed to Gabriel 0. Aisemberg at the above AquariumSystems, Sarrebourg, France) afterwards. Under these con- address. ditions, embryonic development is completed in about 30 d. Copyright 0 1995 Society for Neuroscience 0270-6474/95/155551-09$05.00/O Cloning and sequencing. The Lox4 homeobox was isolated by PCR 5552 Wong et al. - Lox4 May Determine Neuronal Differentiation amplification from genomic DNA. A reaction mixture of 100 p,l con- 260. A probe derived from this cDNA clone, which was as- tained 10 mu Tris-HCl, pH 8.3, 50 mu KCl, 3 mu MgCl,, 200 p,M sumed to contain the entire coding region of Lox4, detected a dGTP, 200 JLM dCTP, 200 PM dATP 200 PM dTTP, 0.8 pg of each oligonucleotide primer, 5 pg of leech DNA, and 2.5 U of Taq DNA transcript of about 3.4 kb in both ElO-El3 embryos and ju-, polymerase (Perkin-Elmer Cetus, Norwalk, CT). After 1 min at 94”C, venile leeches(Fig. 1B). the sample was subjected to 10 cycles of 94°C 35°C and 72°C (1 min The Lox4 protein contained an - (Antp-) class at each temperature), followed by 25 cycles in which the annealing homeodomain(for review, seeBiirglin, 1993). The recognition temperature was increased from 35°C to 50°C and a final 5 min exten- sion at 72°C. The sequences of the primers were GAA/GC/TTGGAA/ helix (helix 3) of Lox4 was identical to the one of Antp-class GAAA/GGAA/GTT for the 5’ primer and CT/GCCT/GG/ATTT/CTGG/ homeodomains(Fig. 1C). In addition, Lox4 containeda YPWM AAACCA for the 3’ primer. The amplified homeobox fragments of 120 motif that is found upstreamof the homeodomainin most Antp- bp were cloned into the vector pIBI31 (International Biotechnologies, classgenes (McGinnis and Krumlauf, 1992). The highestdegree Inc., New Haven, CT). Individual clones were sequenced with the Se- of homeodomainhomology to Lox4 occurred in Hox genesof quenase kit (U.S. Biochemicals, Cleveland, OH). The Lox4 fragment was used as a probe to screen, at high stringency, groups6 and 7 (Scott, 1992), with a 52-53/60 amino acid iden- a XZap (Stratagene, La Jolla, CA) cDNA library of late Hirudo embryos tity. (a generous gift of Dr. James Boulter, The Salk Institute). Two positive The Lox4 homeoboxalso showedhigh homology to the hom- clones were isolated from about 1 X lo6 plaques. They were plaque- eoboxes of leech Lox2 (Wysocka-Diller et al., 1989), and Dro- purified and excised in vivo according to the manufacturer’s instructions, and found to contain inserts of 2.4 and 1.7 kb. A series of nested de- sophila Ultrabithorax (Ubx; Kornfeld et al., 1988) and abdomi- letions of the longest Lox4 cDNA clone were generated using the Erase- nal-A (abd-A; Karch et al., 1990), with 52/60, 51/60, and 52/60 a-Base system (Promega, Madison, WI) and both strands of this Lox4 amino acid identities, respectively. The homology of Lox4 to cDNA clone were sequenced. Lox2, Ubx, and abd-A was found to extend to a sequencead- RNA analysis. PolyA+ RNA was purified by oligo(dT)-cellulose chro- matography from juvenile (lo-30 d after hatching) and embryonic jacent to the C-end of the homeodomain(Fig. 1C). This previ- (ElO-E13) leeches, separated in an agarose gel containing formalde- ously describedconserved motif (Wysocka-Diller et al., 1989) hyde (about 5 pg of polyA+ RNA per lane), transferred onto a nylon has not been found in any other arthropod or leech homeobox membrane, and hybridized at high stringency to a 32P-labeled antisense genes. Within this 8-amino acid-long region, Lox4 had seven riboprobe synthesized from the 2.4 kb Lox4 cDNA clone, as described amino acid identities to Ubx, and four to abd-A. This high de- in Aisemberg and Macagno (1994). For in situ hybridization, a digox- igenin-labeled Lox4 riboprobe was used to probe whole Hirudo embry- gree of similarity in and downstream from the homeodomain os under essentially the same conditions as described in Nardelli-Hae- suggeststhat Lox2 and Lox4 are leech homologsof the arthropod fliger and Shankland (1992). genesUbx and abd-A. Generation qf a Lox4 antibody and antibody stuininR of embryos. A Lox2 and Lox4 shared additional homology within the up- fragment containing the entire &x4 open reading frame was amplified streamregion adjacentto the homeodomain(about 50% homol- using PCR and subcloned into the BamHI site located upstream of the glutathione S-transferase gene (GST) in the pGEX-2T vector (Smith ogy within a 20 amino acid stretch).Lox4 homology to Ubx and and Johnson, 1988). The recombinant plasmid was transformed into E. abd-A was mostly limited to the YPWM sequencein this region. coli. IPTG induction of the Lox4-GST fusion orotein and glutathione- Lox4 was also more homologousto Lox2 within the downstream agarose purification were performed as desciibed in A&be1 et al. region adjacentto the homeodomain(over 50% homology with- (1987). The purified Lox4-GST fusion protein was used to immunize three adult male rabbits and the antisera generated by the rabbits were in a 20 amino acid stretch). A computer-generatedcomparison affinity-purified with chromatography units containing covalently bound of the putative evolutionary distancesbetween thesesequences Lox4-GST as described in Aisemberg and Macagno (1994). showed that leech and arthropod genesform two completely In order to eliminate potentially undesirable binding activities, the separatedgroups (Fig. 10). These results suggestthat the gene purified antibodies were sequentially adsorbed with GST, two other Hi- duplications that produced Ubx/abd-A and Lox2lLox4 occurred rudo homeodomain proteins, Lox1 and Lox3 (Aisemberg and Macagno, 1994; J. W. Wysocka-Diller, G. 0. Aisemberg, and E. R. Macagno, independently in arthropods and annelids. unpublished observations), and total protein from leech embryos. Two of the antisera produced identical staining patterns and one of them was Lox4 is expressedin a subsetof iterated central neurons used for the results presented here. Embryos were stained with this The expressionpattern of Lox4 was determined in whole em- antibody as previously described (Aisemberg and Macagno, 1994). For each preparation, a series of optical sections were recorded and super- bryos by in situ hybridization and antibody staining. Virtually imposed in a single image using a Zeiss Axioplan microscope coupled identical results were obtained with these two assaysof gene to a Bio-Rad MRC-600 laser-scanning confocal system. expression,though antibody staining was more sensitive. Intracellular i?zjection of dyes. For double-labeling, identified neurons The earliest detection of Lox4 transcription was at embryonic of live, dissected embryos were injected with Lucifer yellow using a day 6 (E6), in discrete bilateral patchesstarting at about mid- Drocedure described in Gao and Macagno (1987a). After 20-30 min. ihe embryos were fixed and stained wzh the Lox2 antibody. In other body segment 10 (MlO) (Fig. 2A). Although the ganglia were experiments, the embryos were fixed and the neurons were labeled with not yet formed in thesesegments, the Lox4-positive regions are DiI, as described in Gan and Macagno (1995). known to contain precursorsof central neurons(Fernaudez and Stent, 1992). Results At E9 (Fig. 2&C), after the end of gangliogenesis,the Lox4 Lox4 is an UltrabithoraxZabdominal-Ahomolog mRNA and protein were primarily detected in the CNS, with A 120 bp Lox4 fragment was initially amplified and cloned the majority of expressingneurons being in midbody ganglion from genomic DNA using degeneratehomeobox-specific prim- 10 (MGlO) to MG21. Most of theseneurons were found in the ers. Subsequently, two Lox4 cDNA clones were obtained from posterior half of each ganglion. The anterior boundary of Lox4 a late embryonic cDNA library screenedwith this PCR frag- expressionin the CNS was found to be in MG3, where only ment. The longest cDNA clone was chosen for further char- two neurons, located at the center of the ganglion, expressed acterization (Fig. 1). This cDNA contained a long open reading Lox4 (see high-magnification image of Fig. 3ZJ). There was frame (nucleotides 260-943) that coded for a 228 amino acid also expressionin peripheral tissuesin segmentsMl3 to M21. homeodomainprotein (Fig. 1A). In-frame stop codons at po- At El3 (Fig. 2D), when the level of Lox4 expressionappeared sitions 101 and 140 preceded a methionine codon at position to be the highest, we could count about 100 Lox4-positive neu- The Journal of Neuroscience, August 1995, 75(8) 5553 A 1 GCACOAGCATCTCCACTtGCAOATTATCOACORROOCTACGMGGCTACGMGTGTCCCACCGGGC~GGGCGGCTGCGG~CGMCCCAGCCCCCAGCMCGGA~GA~~TCG~CACACC~GC 121 TCCACCCCGTCTM~GATGACCTC~TT~ATCCA~A~ATCCCCAGCATCA~~ACCACCATCMCAGCMCMCMCMCAG~GCAGCAGCACTGCCGCAGTCAGGMTCT MAGSVSHRSHHPTLTPICQQLUSSPPP 241 GGMCCMTCTCACA~GOATGGCCGG~GQTCAGCCACCGGTCGCACCATCCGACCCTGAC~CCATC~CCAGCAGC~MCTCC~GCCCAGAC~~CGC~GC~~C~CTCC LLAQQQQQQQQQHQQQQSQQQQQQQQQQSQQLQQQKASSK 361 TCTCTTGCATCAGCMCMCM~CMCMC~CAGCATCMCAGCMCMTCTCMCMCMCAGCMC~CAGCMCAGCMCATC~CMCTACMCMC~GGCCTCCTCTM SDDSRQVN SSPSSNGGSSVPPYPWHSVVGPNSSQRRRGR.Q 481 MGCGACGACAGTCGTCAGTACMCAGCTCCCCCTCG~GAGCMCGGGGG~G~CTGTACCC~CTATCCGTGGATGAGCGTGGTCGGACCGMCTCGTCCCA~~A~~~~A 3? $7 S:'R V'Q 'S.L B'L 3.3 3.F Q ,F N R Y L T R R R' R X B X A 3 C t G L T E R Q E R 601 ~~~~;ac;T~C~~CC~G~~G~~~~e~G~~A~~T~~~C~~~AC~A~~~~Q~~~~C~~~~~CG~e~C~A~~ I,WPQPNNMNV,tPN,,NNQQIKKL~KVGKGKBGSVNSNTPGNNN 721 GA~~22TcAGoMCAGGAGGA~~G~MG~GAGM~MCAGATC~G~CTCMCGAGG~GGCMGGGCMGGAGGGCAGCGTCMCAGCMCACCCC~GCMCMCM QNDNDNNNNNNSNLS SDTSDABWBNSVSBKEMBN 841 TCAGMCGACMCGACMCMCMCMCMCMCTCC~TCTGAGCTCGGACACATCCGACGCC~GTGGGAG~TTCCGT~CAGAG~GGAGATGGAGMT~~CTGCMCTGGA 961 1081 1201 1321 1441 1561 1681 CCTTMCATTCATTATGATCACGTCATM~TC~-TACA~TGATGTCCGAGMCAG~CGTCMGTGTCCACGTGAGCAG~C~GTGATG~CATC~ACA~AT~TM 1801 mGTMTTCMTTTACMCATG~G~GTTATTATT 1921 TATTGTTGTTGTTGTTATTC~G~A~MTGACG~~CG~TA~CM~GTGGCGATA~MT~M~CT~CM~TA~TCTC~TACACAG~CA~C~CA 2041 TGTTCTTGCATMCGACGCmGAAATTOOTAACTACTOCCO 2161 ATTATCMTTATATTGTCMTTATATTATAATTGTTRRTTTG AAAAATCTCTTCTTTTE TTTCCCATCTTMT 2281 TTTACAAAGMTGTCGATCRTTGTAAAGTGAATRAAA TTATGACMGTG(A),, B C Hm Lox4 PNSSQ~UGRQTYSRYQTL~EREF~~YL~R~R~~I~,~~~~ER~~~~~QN~~~~~IK~~ kb 12 Nm LOX2 ---N------T------R-----R----LS-T-y------E---"-A-R---- lir Lox2 ---N------T------K------R----LS-T-y------N---"-A-R---- / I Af Ubx ANGL------T------H*-H----R----M--S------L---I-A------“$’ Dm Ubx GTNCL------T------HT-H----R----M--A------L---I-A------Af AM-J, --GCP------T------H--H----R------A------L---L~“--I-- 9.5 - :‘: Sg &d-A --GCP------T--lr------H--H--H------A------L---L~“--I-- m &&A --GCP------T-p------H--H----R----R------A------L--I-- 15 - Dm Ant&, GRCQE-K------T------H------R------KG-pGs

44- . 24 - I-( , D $J?: HmLard

0.24 -

HIOLd H.r Lox2 Al J&A Figure 1. Structure of tix4. A, Nucleotide and deduced amino acid sequence of the Lox4 cDNA. The YPWM motif, the stop codon at the end of the coding region and the polyadenylation site are underlined. In A and C, the homeodomain is shaded and the region immediately downstream that is conserved between Z,ox2Lox4, and lJbx/abd-A is indicated with bold italics. B, Detection of the Lox4 mRNA on a Northern blot of embryonic (lane I) and juvenile (lane 2) leeches hybridized to a cDNA probe. The positions of molecular weight markers (in kb) are indicated. C, Sequence homology between the homeodomain regions of Lox4 and other Ubx/abd-A-like genes from leeches and arthropods. From leeches, three genes are compared: Helobdellu robusta (Hr) Lox2 (Nardelli-Haefliger and Shankland, 1992), and Hirudo medicinalis (Hm) Lox2 (Wysocka-Diller et al., 1989) and Lox4 (this article). From arthropods, six genes available from three major groups are included: the brine shrimp Artemiu franciscana (Af) Ubx and ubd-A (Averof and Akam, 1993), the locust Schisfocerca greguriu (Sg) ubd-A (Tear et al., 1990), and the Drosophila melunogaster (Dm) Antp (Schneuwly et al., 1986), Ubx (Kornfeld et al., 1989) and ubd-A (Karch et al., 1990). The helix 3 is underlined. The dashes represent amino acid identities and the bold letters conservative changes. D, An evolutionary tree deduced from the Ubx/abd-A-likesequences compared in C wasobtained using the MSA program (Lipman et al., 1989). The fact that leech and arthropod genes appear completely segregated in this diagram suggests that the gene duplications that producedUbx and ubd-A in arthropods, and Lox2 and Lox4 in leeches, happened independently in the two lineages. rons per ganglion in MG13 to MG21. After E15, Lox4 expres- Lox4 is expressed transiently in the RPE neurons of MG6 sion was detectable in fewer neurons (data not shown). This We found that a segmentally repeated seriesof paired neurons could be due to downregulationof Lox4, which showed a weak- located at the anterior margins of all midbody ganglia dis- er signal in juvenile leechesthan in embryos on Northern blots played differential expression of Lox4 during early embryo- (Fig. 1B). genesis.We identified theseneurons as the RPEs (rostra1penile E . . . .

c .. .

Figure 2. Lox4 expression in embryos. A and B are Nomarski images of Lox4 in situ hybridiza .tions .c sections of Lox4 antibody stainings, superimposed with transmitted light images of the same fields. The nun A, The ear liest transcription was detected at E6, beginning at segment M 10. The strongest signal w ‘as in MI The Journal of Neuroscience, August 1995, 75(8) 5555 evertor neurons) in MG6 and as their putative homologs in all Table 1. Number of neurons of the RPE-RMV series of other midbody ganglia, which we named RMVs (rostralmost segmental bomologs whose axonal morphology was examined ventral neurons). The RPEs are a single pair of motor neurons found only in Neurons filled with LY Neurons filled with DiI MG6. They are located on the ventral aspect, close to the mid- Stage RPES nRMVsa pRMVsb RPEs nRMVsa pRMVsb line, and are the most anterior neurons in the ganglion. These E8-E9 17 10 27 6 12 13 neurons are known to control penile eversion in adult hirudinid E12-El3 2 1 11 9 15 26 leeches (Zipser, 1979), and they innervate the male organ selec- tively during embryogenesis (Baptista and Macagno, 1988). In u The nRMVs of segments M2 and M3 were examined. contrast, back-fills obtained by placing dye in the male organ b The pRMVs of segments M4, M5, M7, and M8 were examined. indicate that the RMVs in adjacent ganglia do not innervate the male organ (Becker et al., 1995). There is only one pair of RMVs per segment, and they occupy the same position in the RPEs and RMVs have divergent axonal morphologiesthat ganglion as the RPEs. correlate with differencesin Lox4 expression The Lox4 mRNA and protein were initially detected at E7, in Seeking evidence that Lox4 might be involved in the determi- the RPEs of MG6 and in the RMVs of MG4-MG5 and MG7- nation of neuronal identities, we examined the morphology of MG21 (Fig. 3AJ). These RMV cells are henceforth designated the RPEs and RMVs at times when their expressionof Lox4 as Lox4-positive RMVs (pRMVs). At E8, Lox4 expression be- differed. We first determined whether these cells could be ac- gan to decrease in the RPEs, but not in the pRMVs (Fig. 3C). curately identified and filled with dye. PresumedRPEs and By E9, the Lox4 mRNA and protein were not detectable in the RMVs were first injected with the fluorescentdye Lucifer yellow RPEs whereas the pRMVs continued to express Lox4 at about (LY), and then stainedwith the Lox4 antibody to confirm their the same levels as earlier (Fig. 3D,E). The expression pattern of identities. We thereby filled successfullya total of 68 RPEs and Lox4 in RPEs and pRMVs remained unchanged at later stages RMVs in MG2-MG13 (Table 1). The double stainingsconsis- (at least between E9 and E13; see Fig. 3F-H). In contrast, the tently showedthat theseneurons have very similararbors within RMVs of MGI-MG3 (henceforth referred to as the Lox4-neg- the ganglionic neuropil, before their projections exit contralater- ative RMVs or nRMVs) did not express detectable levels of ally through the anterior nerve root. Other neuronswith somata Lox4 mRNA (not shown) or protein (Fig. 3ZJ at any age ex- close to thoseof the RPEs and the RMVs that were intentionally amined (E7-E13). Thus, there were three distinct time courses injected with dye revealed very different projection patterns of Lox4 expression within this series of segmentally repeated (data not shown). neurons: (1) the nRMVs of MGl-MG3 never expressed Lox4, To examine the peripheral axonal morphology in greaterdetail (2) the RPEs of MG6 expressed Lox4 transiently (only from E7 than is possiblewith LY, we injected the lipophilic dye DiI into to E8), and (3) the pRMVs of MG4-MG5 and MG7-MG21 the RPEs and RMVs in MG2-MG8 (Fig. 4). A total of 81 cells expressed Lox4 continuously between E7 and E13, and perhaps were filled (Table 1). At E9, a branchof the peripheralprojection later (illustrated in Fig. 5). of the RPEs begins to grow along the sex nerve toward the Since the phenotype of at least one other neuron, the Retzius primordium of the male organ (Baptista and Macagno, 1988). cell, is known to be affected by contacting the male organ This is the time when Lox4 expressionbecame undetectable in (reviewed in French and Kristan, 1994), we considered the pos- the RPEs. By E12, this branch reachesand innervates the male sibility that the change in Lox4 expression in the RPEs might organ (as in Fig. 4A), whereas all the remaining branchesare also be caused by an inductive effect of the developing male retracting (Baptista and Macagno, 1988). organ. To test this hypothesis, the male organ primordium was In contrast, the Loxl-positive pRMVs displayed a periph- removed from six early E8 embryos, before the RPE projec- eral growth pattern strikingly different from that of the RPEs tions reach this tissue, and then examined the expression of (Fig. 4B,C). At E9, the primary projection of these pRMVs Lox4 at E9. However, all operated embryos were unaffected was longer than that of the RPEs, extending laterally into and showed the same temporal pattern of expression as normal peripheral tissuesof the body wall. In addition, this projection embryos: their RPEs had no detectable Lox4 protein at E9 (data started to extend two pairs of major longitudinal branches(not not shown). shown). The differences in the peripheral growth pattern be- t mouth in A and B (stars) was nonspecific. B, At E9, the strongest Lox4 expression was in the CNS, from MGlO to MG21, as well as in the periphery, from Ml3 to M21. C, At ElO, the Lox4 protein was detected in an iterated subset of neurons of the posterior CNS. D, At E13, about 100 neurons per ganglion were L&-positive from MG13 to MG21. This number decreased by E15. Scale bar: A and B, 250 pm; C and D, 80 pm. Anterior is up in all the panels.

Figure 3. Lox4 expression in RPEs and RMVs. A and D are Nomarski images of Lox4 in situ hybridizations. B, C, and E-J are projections of serial optical sections of Lox4 antibody stainings, superimposed with transmitted light images of the same fields. A-E show the temporal pattern of Lox4 expression in MG5-MG7 during gangliogenesis. In all the panels of this figure, the small arrowheads indicate Lox4-positive pRMVs. In A-E and G the arrows indicate the RPEs of MG6. A and B, Early in gangliogenesis (E7), the Lox4 mRNA and protein were detected in RPEs and pRMVs of MG5-MG7. The numbers in A indicate the segmental positions of these ganglia. C, A day later (E8), there was a lower level of expression of Lox4 in the RPEs than in the pRMVs of MG5 and MG7. D and E, At the end of gangliogenesis (E9), Lox4 was undetectable in the RPEs but maintained approximately constant levels in the pRMVs of MG5 and MG7. F-H, At E13, when axonal growth into the periphery is still in progress, the Lox4 expression pattern established at the end of gangliogenesis was unchanged in MG5 (F), MG6 (G), and MG7 (H). I and J, The nRMVs, anterior to MG4 (large arrowhead), never expressed Lox4. The anterior boundary of Lox4 was found to be in MG3, in a central pair of neurons (arrows), first detected at E7 (1) and were still present at El3 (4. The numbers in I identify MG2-MG5. Scale bar: A, 70 pm; B, 35 pm; C, 45 pm; D, 90 pm; E, 30 pm; F and G, 50 pm. Anterior is up in all the panels. 5556 Wong et al. * Lox4 May Determine Neuronal Differentiation

Figure 4. Axonal morphologies of RPE and RMV neurons. All panels show projections of serial optical sec- tions of RPEs, nRMVs, and pRMVs that were filled with DiI in dissected, fixed El3 embryos. The approximate positions of the cell bodies are indicat- ed with arrowheads. In A, D, and E, some dye diffused into other regions of the ganglion as well, but did not label the axons of any other neurons. A, An RPE of MG6. The branches that arbo- rize over the male organ are indicated by an arrow. B, A pRMV of MG5. C, A pRMV of MG4 (the cell body and ganglion are out of the field of view). D, An nRMV of MG2. A branch that reaches the midline is indicated with an arrow. E, An nRMV of MG3. Note that the nRMVs have shorter, irregular longitudinal branches that are associ- ated with a larger number of fine pro- cesses than the pRMV branches. An asterisk indicates a part of the longi- tudinal connective nerve. Scale bar, 100 pm. Anterior is up in all the pan- els.

came even more evident by El2 (Fig. 4B,C), when the lon- Discussion gitudinal branches of the pRMVs grew further along the A-P Possible roles of Lox4 in neuronal d@erentiation axis of the body wall, extending for approximately 2 body During embryogenesis,the RPE’s and the RPE homologsmay segments. be grouped according to their expressionof L.0~4.Lox4 is tran- The Lox4-negative nRMVs of MG2 and MG3 showeda third, siently expressedin the RPEs of MG6, it is continuously ex- distinct axonal morphology (Fig. 4D,E). Although their main pressedin the pRMVs of MG4-MG5 and MG7-MG21, and it projections exited the anterior nerve root in a similar manner is never expressedin the nRMVs of MGl-MG3. These three and also gave rise to longitudinal branches,these were shorter different temporal patterns of expression within this seriesof and less regular than the ones of Lox4-positive pRMVs (Fig. segmentally homologousneurons correlate with the three dis- 4D,E). Moreover, the branchesof the nRMVs produced an ex- tinct axonal morphologiesof nRMVs, pRMVs, and RPEs. How- tensive network of fine processesthat was not found in the ever, it should be noted that the nRMVs of MG2 and the RPEs pRMVs. Further, one of the branchesof the MG2 nRMVs was of MG6, both of which do not express Lox4 when they are found to arborize over an unknown structure near the midline sendingtheir processesto the periphery, sharea morphological (Fig. 4D), resemblingsomewhat the major branch of the RPEs feature of their peripheral arbors: the presenceof a processthat that innervates the male organ. The differencesbetween the two reachesthe midline and branchesprofusely. The pRMVs, which types of RMVs becameapparent at E9, but were most obvious expressLox4 during the period of processoutgrowth, never pro- at E12. These resultsare summarizedin Figure 5. ject branchesto the midline. The distinct arborization patterns We conclude that the RPEs of MG6, the nRMVs of MG2- of nRMVs and pRMVs suggestthat their peripheral targetsare MG3, and the pRMVs of MG4,5, 7, and 8 have three divergent different not only from those of the RPEs but also from one axonal morphologiesthat correlate with differences in the ex- another.The close correlation of Lox4 expressionand different pressionof LQX~. axonal morphologies(and perhaps targets) suggestthat Lox4 The Journal of Neuroscience, August 1995, 75(8) 5557

necessaryfor the daughter cells of a certain ganglion mother cell to assumedifferent cell fates (Bhat and Schedl, 1994). In the embryonic CNS, the normal phenotype of the lineage of a certain specific neuroblast is controlled by Ubx in the first ab- dominal segment(Al), and by abd-A in A2-A7 (Prokop and Technau, 1994). Ubx and abd-A also specify somelarval sensory organs (Heuer and Kaufman, 1992; Castelli-Gair et al., 1994). When the expressionof grasshopperengrailed is inhibited in the MG3 unpairedmedian neuroblastlineage, extra neuronsare generated insteadof the midline glia (Condron et al., 1994). In C. elegans, lin-39 and mab5 act combinatorially to determine neuronal identities (Clark et al., 1993; Salser et al., 1993), and uric-30 controls the differentiation of GABAergic neurons (Jin et al., MG4 1994). We proposethat the presenceand timing of Lox4 expression play an important role in establishingthe identity of the mem- bers of the RPE-RMV series,allqwing them to adopt three dif- MG5 ferent fates. BecauseLox4 codes for a , this could be accomplishedby differentially regulating the expres- sion of the structural genesor extracellular receptorsthat influ- ence the shapeand branching of the axons of these neurons. According to this model, if any of theseneurons did not express Lox4 by the end of gangliogenesis,they would assumethe fate MG6 of the nRMVs, which would be the default phenotype for this particular cell type. However, in the continuouspresence of Lox4 protein, the developmentalpathway of thesecells would be al- tered and they would adopt the pRMV fate instead.Finally, the downregulationof the Lox4 protein at the end of gangliogenesis would allow the RPEsto follow still another pathway of devel- opment.

Origin of multiple Ubxfabd-A homologs in leeches Figure 5. Lox4 expression and morphology of RPEs, nRMVs, and The homeodomainof Lox4 can be classifiedas belonging to the pRMVs. This schematic drawing summarizes the results presented in Antp-class on the basisof its overall homology to Drosophila Figures 2-4. The Lox4-negative nRMVs are represented by open cir- Antp, conservation of the recognition helix and presenceof a cles, the Lox4-positive pRMVs by solid circles, the transiently express- YPWM motif. This type of Hox/HOM genehas beenpreviously ing RPEs by shaded circles, and the male organ by a shaded ellipse. found in leeches(Wysocka-Diller et al., 1989; Shanklandet al., 1991; Nardelli-Haefliger and Shankland, 1992; Aisemberg and may determinethe emergenceof three distinct phenotypeswith- Macagno, 1993), and in other organismsas diverse as cnidarians in an initially homogeneousset of segmentalhomologs. and vertebrates(reviewed in Kappen and Ruddle, 1993). Among It is not known what mechanismgenerates the differential the Hox/HOM genes,only the homologsof the fly geneslabial expressionof Lox4 in the RPEs. The fact that the Lox4 mRNA (lb), proboscipedia Gb), Deformed (Dfd), and abdominal-B and protein disappearat about the sametime suggeststhat Lox4 (abd-B) can be unequivocally identified in many different phyla expression is regulated in these neurons at the transcriptional and are likely to have been representedin a primitive Hox/HOM level. One possibility would be that other regulatory proteins complex of a common ancestorof all metazoans.The remaining present in the RPEs turn off L0.~4in a cell-autonomousfashion. Hox/HOM genes[i.e., fly Sex combs reduced (Scr), fushi tarazu An alternative possibility is that cell-cell interactions are in- (fiz), Antp, Ubx, and abd-A, and the vertebrate Hox genesof volved. A potential influence is the male organ, whose presence groups6-81 are so similar within any given speciesthat it is not is known to be required for the mature morphology of the RPEs possibleto unambiguouslyidentify their orthologs in distantly (Baptista and Macagno, 1988) and the Retzius cells (reviewed related organisms(reviewed in Krumlauf, 1994). in French and Kristan, 1994). However, removal of the male Contrary to this general finding, leech Lox2 (Wysocka-Diller organ at early E8 did not change the time course of Lox4 ex- et al., 1989; Nardelli-Haefliger and Shankland, 1992) and Lox4 pressionin the RPEs and RMVs. This suggeststhat the male can be identified as homologsof the insect Ubx and abd-A genes organ does not play a role in the downregulation of Lox4. It (Komfeld et al., 1989, Karch et al., 1990), on the basisof se- should be noted that its removal at E8 may not be early enough quencesimilarity and expressionpattern. Thesefour genesshare if induction occurs via long-range diffusible factors before this a unique conserved motif, adjacent to the C-end of the homeo- age. domain, that has not been found thus far in any phylum other The specification of neuronal fate by a homeobox gene has than Arthropoda and Annelida. Within this motif, Lox;! and Lox4 been observed in other systems.For example, the Drosophila are more similar to Ubx than any of thesethree genesto abd- segmentationgenes ftz and eve are required to determine the A, suggestingthat insect Ubx is closer to the original sequence morphology of identified motor neurons(Doe et al., 1988a,b). of a putative Ubx/abd-A-like gene in a common ancestorof both The downregulationof the Drosophila POU gene miti-mere is annelids and arthropods. In addition to these homologies,the 5558 Wong et al. * Lox4 May Determine Neuronal Differentiation primary embryonic domain of expression of these genes is sim- Biirglin T (1993) A comprehensive classification of homeobox genes. ilarly restricted to posterior segments. However, we cannot dis- In: Guide to the homeobox genes (Duboule D, ed), pp 25-71. Oxford: Oxford UP tinguish between Lox2 and Lox4 as the homolog of Ubx or Castelli-Gair J, Greig S, Micklem G, Akam M (1994) Dissecting the abd-A on the basisof sequencehomology and expressionpat- temporal requirements for function. Development 120: tern. Lox2 and Lox4 are more similar to each other than either 1983-1995. of these leech genesto Ubx or abd-A, suggestingthat the du- Clark S, Chisholm A, Horvitz HR (1993) Control of cell fates in the central body region of C. elegans by the homeobox gene lin-39. Cell plications giving rise to Lox2/Lox4 and Ubx/abd-A occurred in- 74:43-55. dependentlywithin the annelid and arthropod lineages. Condron BG, Pate1 NH, Zinn K (1994) engruiled controls glialhreu- It has been postulated that the duplication events leading to ronal cell fate decisions at the midline of the central nervous system. the array of trunk HOM genes(Scr, Antp, Ubx, abd-A) present Neuron 13:541-554. in modern insects took place before the divergence of insects Doe CQ, Scott MP (1988) Segmentation and homeotic gene function in the developing nervous system of Drosophila. Trends Neurosci and crustaceansfrom their last common ancestor(Averof and ll:lOl-106. Akam, 1993). When these results and our data are considered Doe CQ, Hiromi Y, Gehring WJ, Goodman CS (1988a) Expression together,they also imply that theseduplications occurred before and function of the segmentation genefushi turuzu during DrosophiZu the divergence of annelidsand arthropods. Surprisingly, we re- neurogenesis. Science 239:170-175. cently identified a new leech homeobox gene, LoxIS, which Doe CQ, Smouse D, Goodman CS (1988b) Control of neuronal fate by the Drosophila segmentation gene even-skipped. Nature 333:376 showsextensive homology to Lox2 and Lox4 (our unpublished 378. observations).The previously characterizedgenes Lox1 (Aisem- Fernandez J, Stent G (1982) Embryonic development of the hirudinid berg and Macagno, 1994) and Lox5 (Shankland et al., 1991) are leech Hirudo medicinalis: structure, development and segmentation also homologs of trunk HOM genes. of the germinal plat. J Embryo1 Exp Morph01 72:71-96. Such a large number of trunk HOM genesis somewhatun- French KA, Kristan WB (1994) Cell-cell interactions that modulate neuronal development in the leech. J Neurobiol 25:640-651. expected, given the relative morphological simplicity of the Gan W-B, Macagno ER (1995) Interactions between segmental ho- leech. It is possible, however, that only in certain groups of mologues and between isoneuronal branches guide the formation of animals these Hox/HOM genesacquired morphogenetic func- sensory terminal fields. J Neurosci 15:3243-3253. tions such as the specification of segmentalidentities. In this Gao W-Q, Macagno ER (1987a) Extension and retraction of axonal projections by some developing neurons in the leech depends upon context, it is not surprisingthat in the leech Helobdella the ex- the existence of neighboring homoloaues. I. The HA cells. J Neu- pressionof Lox2 doesnot seemto be regulatedby A-P positional robiol 18:43-59. - - - information, but only by the birth order of the neuronalprecur- Gao W-Q, Macagno ER (1987b) Extension and retraction of axonal sors(Nardelli-Haefliger et al., 1994). Since leechesdo not sub- projections by some developing neurons in the leech depends upon divide their midbody regionsto form specializedfunctional units the existence of neighboring homologues. II. The AP and AE neu- rons. J Neurobiol 18:295-313. like a thorax or an abdomen,the expressionof numerousHOM Glover JC, Mason A (1986) Morphogenesis of an identified leech neu- genesmay be primarily required in these animals to specify ron: segmental specification of axonal outgrowth. Dev Biol 115:256- individual neuronalfates, rather than segmentidentities. The fact 260. that Loxl, Lox2, Lox4 and Lox15 have their primary sites of Heuer J, Kaufman T (1992) Homeotic genes have specific functional expressionin specificsubsets of central neuronsthat are repeated roles in the establishment of the Drosophila embryonic peripheral nervous system. Development 115:35-47. within different but partially overlapping long A-P domainsalso Jellies J, Loer CM, Kristan WB Jr (1987) Morphological changes in supportsthis idea. leech Retzius neurons after target contact during embryogenesis. J Neurosci 7:2618-2629. References Jellies J, Kopp DM, Bledsoe JW (1992) Development of segment- and Aisemberg GO, Macagno ER (1994) L,oxl, an Antennupedia-class target-related neuronal identity in the medicinal leech. J Exp Biol homeobox gene, is expressed during leech gangliogenesis in both 170:71-92. transient and stable central neurons. Dev Bioi 161:455-465. Jin Y, Hoskins R, Horvitz HR (1994) Control of type-D GABAergic Aisemberg GO, Wysocka-Diller J, Wong VY, Macagno ER (1993) An- neuron differentiation by C. eleguns Uric-30 homeodomain protein. tennupedia-class homeobox genes define diverse neuronal sets in the Nature 372:780-783. embryonic CNS of the leech. J Neurobiol 24:1423-1432. Kappen C, Ruddle F (1993) Evolution of a regulatory gene family: Aisemberg GO, Wong VY, Macagno ER (1995) Genesis of segmental HOM/HOX genes. Curr Opin Genet Dev 3:931-938. identity in the leech nervous system. In: The nervous systems of Karch E Bender W, Weiffenbach B (1990) abdA expression in Dro- invertebrates: an evolutionary and comparative approach (Breidbach sophila embryos. Genes Dev 4:1573-1587. 0, Kutsch W, eds), pp 77-87. Basel: Birkhatiser. Kessel M, Gruss P (1990) Murine developmental control genes. Sci- Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, Smith ence 249:374-379. JA, Struhl K, eds (1987) Current protocols in molecular biology. Kornfeld K, Saint RB, Beachy PA, Harte PJ, Peattie DA, Hogness DS New York: Wiley. (1989) Structure and expression of a family of Ultrubithorux mRNAs Averof M, Akam M (1993) HOM/Hox genes of Artemia: implications generated by alternative splicing and polyadenylation in Drosophila. for the origin of insect and crustacean body plans. Curr Biol 3:73- Genes Dev 3:243-258. 78. Krumlauf R (1994) Hox genes in vertebrate development. Cell 78: 191- Baptista CA, Macagno ER (1988) Modulation of the pattern of axonal 201. projections of a leech motor neuron by ablation or transplantation of Lipman DJ, Altschul SE Kececioglu JD (1989) A tool for multiple its target. Neuron 1:949-962. sequence alignment. Proc Natl Acad Sci USA 86:4412-4415. Becker T, Berliner AJ, Nitabach MN, Gan W-B, Macagno ER (1995) Macagno ER (1980) Number and distribution of neurons in leech seg- Target-induced neurogenesis in the leech CNS involves efferent pro- mental ganglia. J Neurosci 7:1911-1918. jections to the target. Development 121:359-369. McGinnis W, Krumlauf R (1992) Homeobox and axial patterning. Cell Bhat KM, Schedl P (1994) The Drosophila miti-mere gene, a member 68:283-302. of the POU family, is required for the specification of the RP2/sibling Nardelli-Haefliger D, Shankland M (1992) Lox2, a putative leech seg- lineage during neurogenesis. Development 120: 1483-1501. ment identity gene, is expressed in the same segmental domain in Boncinelli E, Simeone A, Acampora D, Gulisano M (1993) Homeobox different stem cell lineages. Development 116:697-710. genes in the developing central nervous system. Ann Genet 36:30- Nardelli-Haefliger D, Bruce AEE, Shankland M (1994) An axial do- n-s main of HOM/Hox gene expression is formed by morphogenetic The Journal of Neuroscience, August 1995, 15(8) 5559

alignment of independently specified cell lineages in the leech He- Smith DB, Johnson KS (1988) Single-step purification of polypeptides lobdella. Develonment 120:1839-1849. expressed in Escherichia coli as fusion with glutathione S-transferase. Prokop A, Technai GM (1994) Early tagma-specific commitment of Gene 67:37-40. Drosophila CNS progenitor NBl-1. Development 120:2567-2578. Stewart RR, Gao W-Q, Macagno ER (1991) Segmental differentiation Salser SJ, Loer CM, Kenyon C (1993) Multiple HOM-C gene inter- in the leech central nervous system: proposed segmental homologs actions suecifv cell fates in the nematode central nervous svstem., of the heart accessory neurons. J Comp Neurol 313:431-440. Genes D& 7:i714-1724. Schneuwlv. S. Kuroiwa A. Baumgrtner F! Gehring W (1986) Structural Tear G, Akam M, Martinez-Arias A (1990) Isolation of an organization and sequence of-the homeotic iene kterkapedia of abdominal-A gene from the locust Schistocerca gregaria and its ex- Drosophila. EMBO J 5:733-739. pression during early embryogenesis. Development 110:915-925. Scott MP (1992) Vertebrate homeobox gene nomenclature. Cell 71: Wysocka-Diller Jw, Aisemberg GO, Baumgarten M, Levine M, Ma- 551-553. cagno ER (1989) Characterization of a homologue of bithorax-com- Shankland M, Martindale MQ, Nardelli-Haefliger D, Baxter E, Price DJ plex genes in the leech Hirudo medicinalis. Nature 341:760-763. (1991) Origin of segmental identity in the development of the leech Zipser B (1979) Identifiable neurons controlling penile eversion in the nervous system. Development [Suppl] 2:29-38. leech. J Neurophysiol 42:455-464.