Proc. Natl. Acad. Sci. USA Vol. 95, pp. 10671–10675, September 1998 Evolution

Expression of homeobox genes shows chelicerate retain their deutocerebral segment

MAXIMILIAN J. TELFORD AND RICHARD H. THOMAS*

Department of Zoology, The Natural History Museum, Cromwell Road, London, SW7 5BD, United Kingdom

Communicated by Edward B. Lewis, California Institute of Technology, San Marino, CA, July 7, 1998 (received for review April 14, 1998)

ABSTRACT Expression patterns of six homeobox con- posterior to the ‘‘missing’’ deutocerebral region and therefore taining genes in a model chelicerate, the oribatid mite Arche- for the understanding of morphological evolution in the ar- gozetes longisetosus, were examined to establish homology of thropods. chelicerate and insect head segments and to investigate claims In the insects, a secondarily reduced intercalary segment, that the chelicerate deutocerebral segment has been reduced corresponding to the crustacean second antennal (tritocere- or lost. engrailed (en) expression, which has been used to bral) segment, was demonstrated unequivocally by a stripe of demonstrate the presence of segments in insects, fails to expression of the gene engrailed (en) (14, 15). Region-specific demonstrate a reduced deutocerebral segment. Expression expression of other genes has been used as a means of inferring patterns of the chelicerate homologs of the Drosophila genes homology of body regions (16), segments (17), and Antennapedia (Antp), Sex combs reduced (Scr), Deformed (Dfd), even parts of arthropod limbs (18). We have looked at the proboscipedia (pb), and orthodenticle (otd) confirm direct cor- expression of several homeobox-containing genes in the em- respondence of head segments. The chelicerate deutocerebral bryos of a model chelicerate, the oribatid mite Archegozetes segment has not been reduced or lost. We make further longisetosus, by using both in situ hybridization to see whether inferences concerning the evolution of heads and Hox genes in the segment corresponding to the insect͞crustacean deutoce- arthropods. rebrum has been reduced or lost secondarily in chelicerates and to deduce homologies between head segments of cheli- Head segments and their appendages have long been consid- cerates and other arthropods. This gives us a powerful tech- ered of great importance in understanding the relationships of nique for predicting the ancestral morphology of the crusta- the main extant arthropod classes—insects, crustaceans, myri- cean͞insect clade by using the chelicerates as an outgroup. apods, and chelicerates (arachnids and horse-shoe crabs). The posterior regions of their bodies vary greatly, but several MATERIALS AND METHODS authors have found evidence for homologies between specific segments of the head in support of different phylogenetic DNA Sources and Extraction and cDNA Preparation. DNA schemes relating the arthropods (1–3). There is recently a extraction followed described methods (19). Reverse transcrip- general acceptance of the close relationship of insects and tion–PCR used the primer CDNAI (GGATTTAGGTGAC- crustaceans and of direct homologies of their head segments ACTATAGCGGCCGCTTAAGA(T15)NN) attached to and associated appendages (antennae, mandibles, etc.), some Dynabeads (Dynal, Great Neck, NY) to capture mRNA and to of the evidence for which comes from studies of gene expres- prime the first strand cDNA production by using Boehringer sion (1, 4–6). Myriapod head segments also are widely pre- Mannheim products and protocols. Subsequent PCR amplifi- sumed to be directly homologisable with those of insects and cation used a gene-specific 5Ј primer (below) and CDNAB crustaceans, grouping the three classes into the mandibulates (TATAGCGGCCGCTTAAGA) at the 3Ј end. (3, 7). The chelicerates, on the other hand, are seen as PCR Amplification, Genomic Clone Isolation, and in Situ fundamentally different in most schemes of arthropod evolu- Probe Template Cloning. Archegozetes homologs of Antenna- tion, and suggested homologies of their head segments with pedia (Antp), Sex combs reduced (Scr), and probiscipedia (pb) those of other arthropods seem, by their very variety, uncon- (AlAntp, AlScr, and Alpb) were amplified by using a degenerate vincing (8–10). primer screening for all Hox genes. Primers were TLELEKEF One of the most widely held views concerning the cheli- [a 1:1 mixture of ACT TTG GAR TTR GAR AAR GAR TTY cerate head is that they have secondarily lost or reduced the and ACT TTG GAR CTI GAR AAR GAR TTY (I ϭ homolog of the front-most appendage-bearing segment, which inosine)] at 5Ј and WFQNRRXK (TTT NRY TCT TCT ATT in the other three classes carries the antennae (3, 7, 8, 11) (Fig. YTG RAA CCT) at 3Ј. Clones isolated from this PCR screen 1). This idea stems from two observations. Firstly, unlike all were used to probe a genomic ␭ library for full length other groups (including the extinct trilobites and the onych- sequences. The Archegozetes homolog of orthodenticle (otd) ophorans), the chelicerates have no antennae, the front-most (Alotd) was PCR amplified by using degenerate primers OTD5 appendages being the chelicerae. Secondly, the homolog of the (TTC ACA CGT GCN CAR YTN GAY GT) at 5Ј and OTD3 mandibulate deutocerebrum, the brain region associated with (TGC AGY TGY TGN CKA CAY TTN GC) at 3Ј. The the antennal segment found anterior to the mouth, is thought amplified and cloned product was used to screen the ␭ genomic to have been lost or greatly reduced in chelicerates (8). Support library for full length sequences. ␭ clones were sequenced by for this view comes from studies claiming evidence of coelom using outward facing primers within the homeobox. Inward remnants corresponding to the deutocerebral segment in a facing primers then were designed to amplify a coding region spider (10, 12, 13). This idea has direct consequences for suitable in length for in situ hybridization. The Archegozetes attempts to assign homology to the segments and appendages homolog of deformed (Dfd)(AlDfd) was amplified by using DFD5B (STC GAY CCN AAR TTY CCN CC) at 5Ј and The publication costs of this article were defrayed in part by page charge WFQNRRXK at 3Ј and was used directly for in situ probe payment. This article must therefore be hereby marked ‘‘advertisement’’ in accordance with 18 U.S.C. §1734 solely to indicate this fact. Abbreviation: PTW, PBS with 0.2% Tween 20. © 1998 by The National Academy of Sciences 0027-8424͞98͞9510671-5$2.00͞0 *To whom reprint requests should be addressed. e-mail: R.Thomas@ PNAS is available online at www.pnas.org. nhm.ac.uk.

10671 Downloaded by guest on September 24, 2021 10672 Evolution: Telford and Thomas Proc. Natl. Acad. Sci. USA 95 (1998)

FIG. 2. Conceptual translations of homeodomains of Archegozetes (Al) genes compared with homologs from other arthropods and an FIG. 1. Two hypotheses of segmental homologies between cheli- onychophoran. (A) Engrailed (en) and invected (inv). (B) proboscipedia cerate and crustacean͞insect head regions. (A) Deutocerebral seg- (pb). (C) Deformed (Dfd). (D) Sex combs reduced (Scr). (E) Anten- ment missing. The cross indicates the antennal segment thought to napedia (Antp)(F) orthodenticle (otd). Insect sequences obtained from have been reduced or lost in the chelicerates relative to the insects and the GenBank database: Dm, Drosophila melanogaster; Af, Artemia crustaceans. This theory supposes that the chelicerae (Ch) are ho- franciscana; Sg, Schistocerca gregaria; Tc, Tribolium castaneum. Non- mologous to the second antennal segment of crustaceans (An2) and insect sequences from ref. 36: Cen, (Ethmostigmus rubripes); the intercalary segment of insects (Ic). (B) Deutocerebral segment Ony, onychophoran (Acanthokara kaputensis). Matches are indicated present. Diagonal lines between A and B emphasize the shift in with a period and missing data by a dashed line. segmental register. This theory supposes that the chelicerae are homologous to the first antennal segment of insects and crustaceans. in1mg͞ml BSA, 5 mg͞ml Boehringer Mannheim block, and Lr, labrum; Oc, ocular; Md, mandible; Pp, pedipalp; Mx1 and 2, 50 ␮l͞ml normal goat serum in PTW for 1 hr followed by maxilla 1 and 2; Lb, labium; L1–4, legs 1–4; T1–3, thorax 1–3; Op1–2, incubation overnight at 4°C with a preabsorbed Boehringer opisthosomal 1–2. More posterior segments are not shown. Mannheim alkaline phosphatase conjugated antidigoxygenin antibody diluted 1:2000 in PTW followed by washes in PTW production. The Archegozetes homolog of en (Alen) was am- and detection with nitroblue tetrazolium͞5-bromo-4-chloro- plified by using primers JM10 (GAI AAG CGI CGC ACI Ј Ј Ј 3-indoyl phosphate p-toluidine salt. Detailed protocols are GCC TTC AC) at 5 and WFQNRRXK at 3 . The 3 terminus available from the authors. of the Alen gene was reverse transcription–PCR amplified by using a 5Ј primer designed according to the amplified sequence (CCT GAA ATT AAA TGA ATC A). All sections of DNA RESULTS for in situ hybridizations were cloned into pGEM T or pGEM Expression of Alen. Alen is expressed at the rear of segments Ј TEASY (Promega) with the 3 end closest to the T7 promoter. in a fashion identical to that seen in other arthropod groups (6, Alignment of these Archegozetes genes with homologs from 20, 21) (Fig. 3A). Five stripes appear initially, corresponding to other arthropods are shown in Fig. 2. DNA sequences used for the five visible limbs (the sixth limbs—fourth walking legs—are in situ hybridizations are available from the GenBank database suppressed until after hatching in oribatid embryos), and more (accession nos. AF071402 to AF071407). posterior stripes, including that corresponding to the sixth limb Embryo Preparation. Embryos were dechorionated in 5% bearing segment, are added progressively (results not shown). bleach for 5 min, were fixed in n-heptane over 4% formalde- If the chelicerate deutocerebral segment has been suppressed hyde in PBS for 1 hr, and were devitellinized by placing the but not entirely lost, we would expect, by analogy with the embryos in n-heptane prechilled on dry ice, adding room insect intercalary segment, to see two regions of Alen expres- temperature methanol and shaking vigorously for 2–3 min. sion anterior to the cheliceral stripe corresponding to the Embryos were digested partially in 7.5 ␮g͞ml proteinase K for reduced deutocerebral segment and to the ocular segment. In 8 min, were washed, were refixed in 4% formaldehyde in PBS fact, no Alen expression is seen anterior to the cheliceral stripe. with 0.2% Tween 20 (PTW) for 20 min, and were rewashed in We find no support from Alen expression patterns for a PTW. secondarily reduced deutocerebral segment in chelicerates. In Situ Hybridizations and Color Detection. Digoxygenin The lack of expression corresponding to the ocular expression labeled riboprobe production used Boehringer Mannheim T7 of insects is puzzling but probably is explained by the tendency polymerase and buffers and Boehringer Mannheim digoxyge- for this region of expression to be small (6, 20, 21). Further- nin-labeled ribonucleotides according to the manufacturer’s more, the size of the ocular en expression, at least in insects, instructions. Prehybridization was in hybe (50% form- seems to correspond with the size of the eyes (22); Archegozetes amide͞5ϫ standard saline citrate͞150 ␮g/ml yeast RNA͞0.1% has no eyes. Tween 20͞50 ␮g/ml heparin) at 65°C for 24 hr. Hybridizations Expression of Alpb, AlDfd, AlScr, and AlAntp. A lost or were carried out by using 500 ␮l of 0.4–1 ␮g͞ml of antisense reduced segment has consequences for the assignment of riboprobe in hybe at 65°C overnight. Washes were 1 ϫ 30 min homology of more posterior segments and appendages. If the in hybe at 65°C, 1 ϫ 30 min in 1:1 hybe͞PTW at 65°C, and 5 ϫ deutocerebral segment has been reduced as generally sup- 20 min in PTW at room temperature. Embryos were blocked posed, then the first chelicerate appendages, the chelicerae, Downloaded by guest on September 24, 2021 Evolution: Telford and Thomas Proc. Natl. Acad. Sci. USA 95 (1998) 10673

FIG. 3. Expression patterns of Homeobox-containing genes in Archegozetes demonstrating homology of anterior segments. The embryo is Ϸ170 ␮m long in each case. Abbreviations as in Fig. 1. (A) Expression of the Archegozetes engrailed (Alen) homolog in an embryo after the formation of the labrum. Alen is expressed in the posterior portion of each segment with visible limbs. Despite strong staining in these segments, no engrailed expression is evident anterior to the chelicerae. (B) Alpb is expressed in all of the visible walking legs with an anterior boundary at the front of the pedipalp. (C) AlDfd is expressed throughout the posterior of the embryo, excepting the terminal segments, with an anterior boundary at the front of the first walking leg. (D) AlScr is expressed strongly in the third leg and weakly in the second. It appears to be expressed only in the more distal regions of the second leg though the staining is diffuse. (E) Close up of opisthosoma and fourth leg bud. AlAntp expression is strong in the opisthosoma and continues into the rear portion of the fourth leg bud. Because of the small size of the fourth limb bud and its position, tucked in next to the opisthosoma, it has proved very hard to photograph; this expression therefore is schematized in a camera lucida drawing in F.(F) Camera lucida drawing highlighting expression of AlAntp in the opisthosoma and rear of fourth leg bud. (G) Optical mid-sagittal expression showing AlAntp expressed strongly in the opisthosoma with no expression in the chelicerae, pedipalps, or legs 1–3. Leg 4, a small bud, is not in focus in this picture. (H) Alotd is expressed in the ocular lobes and in the ventral midline. The photograph is a sagittal optical section focusing on the expression in the ventral midline. No segmental expression is seen posterior to the ocular segment. (I) Alotd expression is seen in the ocular lobes but not in the chelicerae. View is anterior-ventral with the anterior to the left.

correspond to the second appendages of insects and crusta- meobox amino acid sequence (Fig. 2), only a single nucleotide ceans (crustacean second antennae, insect intercalary). The sequence was found. This suggests strongly that only a single second chelicerate appendages (pedipalps) correspond to the Hox cluster exists in Archegozetes. We looked at expression third appendages of insects and crustaceans (mandible) (8), patterns of four of these genes that specify anterior regions of etc. (Fig. 1A). If, however, the deutocerebrum has not been metazoan embryos. The remaining genes we have identified lost or reduced, as indeed is suggested by lack of Alen are homologs of the Drosophila genes labial, Ultrabithorax, expression, then the appendage bearing segments would align Abdominal-B (data not shown), and zen (26). directly: chelicerae with first antennae, pedipalps with second We first looked at expression of the Archegozetes probosci- antennae, and first legs with mandibles (9), etc. (Fig. 1B). By pedia homolog (Alpb). One of the most striking aspects of Hox comparing the anterior boundaries of expression of segment- gene expression is the co-linearity between the order of the specifying homeobox genes, we sought to infer homology of genes on the chromosome and their anterior boundaries of segments directly (23). Anterior boundaries are stable evolu- expression. This co-linearity is maintained in both the verte- tionarily and hence are good positional markers whereas brates and insects. Drosophila proboscipedia (pb) is an excep- posterior boundaries vary across different taxa under the tion to this rule because, according to its chromosomal posi- influence of a variety of transrepressors. We make the as- tion, it would be expected to have an anterior expression sumption that the arthropods are monophyletic (24) and come boundary in front of that of Dfd in the intercalary segment from a segmented ancestor with appendages whose head whereas it actually is expressed principally in the labial seg- segments were specified by Hox genes. All comparisons are ment. There is, however, a limited amount of ectodermal made only with insects because expression data are not yet expression in the maxillary, mandibular, and intercalary seg- available outside the insects for the more anteriorly expressed ments in crickets and milkweed bugs (22), placing its anterior arthropod Hox genes. Results from insects come from Rogers boundary of ectodermal expression—as predicted from its and Kaufman (22), who provide comparative expression data chromosomal position and by comparison with its vertebrate from distantly related insect orders. Results for Antp come homolog (Hox2)—anterior to that of Dfd in the second from ref. 25. limb-bearing segment. In Archegozetes, Alpb is expressed The arthropod Hox cluster contains nine genes, and Arche- equally in all three visible walking leg segments and in the gozetes homologs of eight of these were discovered in a pedipalpal segment (Fig. 3B) so its anterior expression bound- degenerate PCR screening (results not shown). For each ary is in the second limb-bearing segment. From this compar- separate gene unambiguously identified based on full ho- ison, we infer that the ancestral anterior boundary of pb Downloaded by guest on September 24, 2021 10674 Evolution: Telford and Thomas Proc. Natl. Acad. Sci. USA 95 (1998)

expression was in the second limb-bearing segment as seen in a reduced fashion in crickets and milkweed bugs. We predict that full pb expression will be found in this segment in crustaceans and that the differences seen within the insects are secondary modifications. The Archegozetes Deformed homolog (AlDfd) is expressed throughout the rear of the embryo with a discrete anterior boundary at the anterior of the third limb segment (first walking leg) (Fig. 3C). Insect Dfd is expressed in the fourth limb (maxillary) segment and has its anterior boundary at the front of the third limb (mandibular) segment. In Archegozetes embryos, Sex combs reduced (AlScr)is expressed in two segments: strongly in the fifth limb (third leg) segment and less strongly in the fourth limb (second leg) segment (Fig. 3D). In the fourth limb segment, expression is in the limb rather than the ventral blastoderm. Insect Scr also is expressed strongly in the fifth limb (labial) segment and more weakly and restricted to the posterior portion of the fourth limb (maxillary) segment. Later expression of insect Scr in the fourth limb is in lateral epidermis (22). The final Hox gene we studied, AlAntp, is expressed strongly in the more posterior segments of the opisthosoma. There is weaker expression in the first opisthosomal segment and the anterior boundary of expression is the rear of the fourth legs (sixth limb segment), which are small buds (Fig. 3 E–G). This corresponds closely with the early expression of Antp in FIG. 4. Homologies between anterior segments of chelicerates and Drosophila in which expression is in a broad band in paraseg- insects as determined by overlapping patterns of expression of ho- ment 4 (anterior of the second thoracic and rear of the first mologous genes. Posterior boundary of otd and anterior boundaries of thoracic) (25). In other words, initial expression in insects has Dfd, Scr, and Antp are identical in insects and chelicerates. The its anterior boundary, like Archegozetes, in the rear of the sixth anterior boundary of Archegozetes pb is as seen in some but not all insect groups. Appendages with Dll expression are colored black limb segment. Later, modulated expression of Antp in Dro- (results not shown). The inferred chelicerate homolog of the insect sophila also includes expression of a new transcript in a small mandible (leg 1) does express Dll. Abbreviations as in Fig. 1. subset of more anterior cells. AlAntp expression corroborates the inference drawn from Alpb, AlDfd, and AlScr expression the trivial observation that the anteriormost chelicerate ap- that chelicerate and insect limb-bearing segments are directly pendages are not antennae, the most important idea suggesting alignable. that the deutocerebral segment is lost is the conviction that the Expression of Alotd. In Drosophila, otd is expressed in the deutocerebrum is missing from the chelicerate brain. This idea ocular and first antennal segments as well as along the length stems from the observation that insects have two segmental of the ventral midline (27). In the beetle Tribolium, there are ganglia anterior to the mouth whereas, ancestrally at least, two otd genes expressed in similar but not identical patterns; chelicerates only have one (8). As a result, the postoral both of these genes differ from the pattern seen in Drosophila cheliceral ganglion was homologized to the postoral tritoce- in that, at the later stage, when expression includes the ventral rebrum of the insects. In fact, having the deutocerebrum midline, there is no expression in the antenna (28). We anterior to the mouth is probably a derived condition in interpret the persistent expression in the Drosophila antennal crustaceans (and also, therefore, in insects) because certain segment as a Drosophila specific secondary adaptation. The crustacean orders have their first antennal ganglion lateral to expression seen in Archegozetes is strikingly similar to that seen the esophagus (29). The same tendency is seen in chelicerates in Tribolium (Fig. H and I), being restricted to expression in the in which the cheliceral ganglion is primitively on either side of ocular segment and the ventral midline. the esophagus, but, in many chelicerate orders, the cheliceral ganglion is actually preoral and continuous with the protoce- DISCUSSION rebrum (8, 30), though it only moves to this position during Assignment of Homology Between Chelicerate and Insect embryogenesis as the chelicerae move anterior. This presum- Anterior Segments. Each of the comparisons of genes expres- ably indicates a convergent tendency for ganglionic fusion and sion patterns described support the conclusion proposed based cephalization as well as suggesting that all of the appendages on the lack of Alen expression anterior to the cheliceral we see in extant arthropods were primitively postoral. The segment, namely, that chelicerate and insect head segments initial postoral position of the chelicerae also has been used to line up directly: (Fig. 4) the antennal with the cheliceral, the imply that they could not be homologs of the first antennae, second antennal (intercalary) with the pedipalpal (expressing which are thought of as preoral (31). In fact, studies of pb), the mandibular with the chelicerate first leg (expressing embryogenesis of insects, crustaceans, and myriapods show Dfd), the first maxillary with chelicerate second leg (partially that, in all cases, the first antennae are initially postoral and the expressing Scr), the labial with the chelicerate third leg (fully subsequent anteriorwards morphogenetic movements are expressing Scr), and the first thoracic with the chelicerate strikingly similar in each case (32). This can be seen as further fourth leg (expressing Antp in the rear). We conclude that evidence supporting our view that cheliceral and first antennal there has not been a loss or reduction of a segment in segments are homologous. chelicerates relative to insects and crustaceans. Chelicerates as an Outgroup of Insects͞Crustaceans. Mo- Reappraisal of Evolution of Arthropod Anterior Segments. lecular phylogenetic studies show that the chelicerates are an In the light of our conclusion that the chelicerate deutocere- outgroup to the insect͞crustacean clade (33–36). This idea is bral segment is not missing and indeed corresponds to the confirmed by our demonstration that Dll is expressed in all cheliceral segment, we looked at the evidence that had given chelicerate limbs (results not shown) whereas it is secondarily rise to this idea and the ramifications of its rejection. Ignoring missing in crustacean and insect mandibles (4). As such, the Downloaded by guest on September 24, 2021 Evolution: Telford and Thomas Proc. Natl. Acad. Sci. USA 95 (1998) 10675

chelicerates make an ideal outgroup to allow inference, 11. Wegerhoff, R. & Breidbach, O. (1995) in Comparative Aspects of through outgroup comparison, of the ancestral condition and, the Chelicerate Nervous Systems, eds. Breidbach, O. & Kutsch, W. hence, the subsequent evolution of mandibulate anterior seg- (Birkhau¨ser, Basel), pp. 159–180. ¨ ments, their appendages, and the expression of the genes that 12. Pross, A. (1966) Z. Morphol. Okol. Tiere 58, 38–108. 13. Pross, A. (1977) Zoomorphologie 86, 183–196. specify them. 14. Schmidt-Ott, U., Gonza´lez-Gaita´n,M., Ja¨ckle,H. & Technau, We already have inferred, for example, that pb was ex- G. M. (1994) Proc. Natl. Acad. Sci. USA 91, 8363–8367. pressed as far forward as the second appendage in the common 15. Schmidt-Ott, U. & Technau, G. M. (1992) Development (Cam- ancestor of insect͞crustaceans and chelicerates. Furthermore, bridge, U.K.) 116, 111–125. expression of AlDfd in the third limbs is contrary to the 16. Holland, P. W. H., Holland, L. Z., Williams, N. A. & Holland, hypothesis (22) that the anterior boundary of Dfd was ances- N. D. (1992) Development (Cambridge, U.K.) 116, 653–661. trally in the fourth limb-bearing segment and that there was a 17. Averof, M. & Akam, M. (1993) Curr. Biol. 3, 73–78. 18. Averof, M. & Cohen, S. M. (1997) Nature (London) 385, 627–630. migration forwards of this boundary in the insects and a 19. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) Molecular concomitant loss of Dll expression and mandible tip. Further Cloning: A Laboratory Manual (Cold Spring Harbor Lab. Press. studies on chelicerates should be invaluable for the under- Plainview, NY). standing of the evolution of the arthropods, and expression 20. Schmidt-Ott, U., Sander, K. & Technau, G. M. (1994) Roux’s data from the anteriorly expressed genes described here from Arch. Dev. Biol. 203, 298–303. crustaceans and myriapods would be particularly welcome. 21. Scholtz, G. (1995) Zoology 98, 104–114. 22. Rogers, B. T. & Kaufman, T. C. (1997) Int. Rev. Cytol. 174, 1–84. Note Added in Proof. Comparable results have been obtained from 23. Abouheif, E., Akam, M., Dickinson, W. J., Holland, P. W. H., spiders, which are chelicerates phylogenetically distant from the mites Meyer, A., Patel, N. H., Raff, R. A., Roth, V. L. & Wray, G. A. (37). (1997) Trends Genet. 13, 432–433. 24. Fortey, R. A. & Thomas, R. H. (1997) Arthropod Relationships (Chapman & Hall, London). We are grateful to J. A. Mackenzie-Dodds, T. Snell, and D. Goode 25. Bermingham, J. R., Martinez-Arias, A., Petitt, M. G. & Scott, for technical assistance, R. A. Norton for providing our original M. P. (1990) Development (Cambridge, U.K.) 109, 553–566. Archegozetes stock, P. W. H. Holland for donating PCR primer JM10, 26. Telford, M. J. & Thomas, R. H. (1998) Dev. Genes Evol., in press. and N. H. Patel and N. A. Williams for sharing protocols. This work 27. Finkelstein, R., Smouse, D., Capaci, T. M., Spradling, A. C. & was funded by the Biotechnology and Biological Sciences Research Perrimon, N. (1990) Genes Dev. 4, 1516–1527. Council (Grant G01972). 28. Li, Y., Brown, S. J., Hausdorf, B., Tautz, D., Denell, R. E. & Finkelstein, R. (1996) Dev. Genes Evol. 206, 35–45. 1. Averof, M. & Akam, M. (1995) Philos. Trans. Roy. Soc. Lond. B 29. Wallosek, D. & Mu¨ller, K. J. (1997) in Arthropod Relationships, 347, 293–303. eds. Fortey, R. A. & Thomas, R. H. (Chapman & Hall, London), 2. Briggs, D. E. G. & Fortey, R. A. (1989) Science 246, 241–243. pp. 139–153. 3. Nielsen, C. (1995) Evolution: Interrelationships of the 30. Babu, K. S. (1985) in Neurobiology of Arachnids, ed. Barth, F. G. Living Phyla (Oxford Univ. Press, Oxford). (Springer, Heidelberg), pp. 3–19. 4. Popadic, A., Rusch, D., Peterson, M., Rogers, B. T. & Kaufman, 31. Willmer, P. (1990) Invertebrate Relationships: Patterns in Animal T. C. (1996) Nature (London) 380, 395. Evolution (Cambridge Univ. Press, Cambridge). 5. Telford, M. J. & Thomas, R. H. (1995) Nature (London) 376, 32. Anderson, D. T. (1973) Embryology and Phylogeny in Annelids 123–124. and Arthropods (Pergamon, Oxford). 6. Scholtz, G. (1997) in Arthropod Relationships, eds. Fortey, R. A. 33. Regier, J. C. & Shultz, J. W. (1997) Mol. Biol. Evol. 14, 902–913. & Thomas, R. H. (Chapman & Hall, London), pp. 317–332. 34. Wheeler, W. C., Cartwright, P. & Hayashi, C. Y. (1993) Cladistics 7. Meglitsch, P. A. & Schram, F. R. (1991) Invertebrate Zoology 9, 1–39. (Oxford Univ. Press, Oxford). 35. Turbeville, J. M., Pfeifer, D. M., Field, K. G. & Raff, R. A. (1991) 8. Weygoldt, P. (1985) in Ontogeny of the Arachnid Central Nervous Mol. Biol. Evol. 8, 669–686. System, ed. Barth, F. G. (Springer, Heidelberg), pp. 20–37. 36. Grenier, J. K., Garber, T. L., Warren, R., Whitington, P. M. & 9. Raff, R. A. (1996) The Shape of Life (University of Chicago Press, Carroll, S. (1997) Curr. Biol. 7, 547–553. Chicago). 37. Damen, W. G. M., Hausdorf, M., Seyfarth, E.-A. & Tautz, D. 10. Legendre, R. (1979) Bull. Soc. Zool. Fr. 104, 277–287. (1998) Proc. Natl. Acad. Sci. USA 95, 10665–10670. Downloaded by guest on September 24, 2021