The Origin of Multiple Sex Chromosomes in Tiger Beetles

In most holometabolous insects, sex determination involves differentiated sex chromosomes of the XY type (ZW, with heterogametic females in Lepidoptera). In Coleoptera (beetles) a general karyotype with meioformula 2n = 9 + XY is prevalent in the approximately 3,000 species studied to date (Smith and Virkki 1978 ; Petitpierre 1996 ), but the genetic system of Cicindelidae (tiger beetles) is unusually diverse, with many species exhibiting multiple sex chromosomes, XnY, where n varies between 2 and 4. During meiosis these heterosomes form a characteristic rosette-like multivalent linked by telomeric connections without forming chiasmata between the various X chromosomes (Giers 1977 ). The multiple sex chromosome system is widespread in cicindelids and is found in both the recognized subfamilies, including the Cicindelinae with a multiple system, which has been described for 55 species mostly in the species rich genus Cicindela (sensu lato), and the Collyrinae with three representatives studied to date. Other cicindelids, including two species of Megacephala (Serrano, Galián, and Ortiz 1986 ; Galián and Hudson 1999 ) and two species in the genus Cicindela, were found to have XY or X0 systems. Multiple sex chromosome systems in insects appear to be evolutionarily short-lived because they are generally confined to small taxonomic groups, where close relatives exhibit simple genetic systems. The unusually broad taxonomic distribution of multiple systems in Cicindelidae may suggest either a common ancestry of this trait and evolutionary stability (single-origin hypothesis) or a tendency to generate multiple systems repeatedly in this lineage (multiple origin). It is possible that lineage-specific properties canalize the evolution of structural features of the karyotype, and a repeated origin of a multiple system may represent the outcome of “karyotypic orthoselection” (White 1973 ). In either case the wide distribution of the multiple system in Cicindelidae and the comparison with related species exhibiting simple sex chromosome systems lend themselves to investigation of the evolution and the functional significance of this intriguing phenomenon. Here, we analyzed the evolutionary origin of multiple sex chromosomes in Cicindelidae by establishing the distribution of the multiple systems in major taxonomic groups in the context of a phylogenetic tree derived from 18S rRNA gene sequences. Chromosome counts in mitotic metaphases (diploid number) and the comparison of chromosomes paired during the first meiotic division and haploid counts in the second meiotic metaphases were used to infer the autosome and sex chromosome content (meioformula). Three species of Omini, Omus californicus, O. dejeani, and Amblycheila baroni, and a Mantichorini, Mantichora amigdaloides, were inferred to have a simple sex chromosome system of the XY type (table 1 ; see supplementary information available at www.smbe.org). In these species the corresponding heteromorphic bivalent of diakinesis of the first meiotic division was observed as expected in an XY system. This was further confirmed in the second metaphase plates, where two types of cells were apparent, exhibiting the autosomes plus either the X or the Y chromosome. The karyotypes in these species were generally similar in the overall morphology, with the largest pairs characteristically asymmetric (subtelocentric) and the majority of intermediate and small autosomes mediocentric (meta- or submetacentric), as is the rule in Carabidae and Cicindelidae (Serrano and Galián 1998 ). Species in the subfamily Collyrinae and the cicindeline subtribes, Prothymina and Cicindelina, revealed the presence of a multiple sex chromosome system with three X chromosomes (Prothymia sp., n = 10 + XXXY; Neocollyris sp., n = 12 + XXXY; and Therates sp., n = 10 + XXY) (table 1 ; supplementary material). The general appearance of the karyotypes was different from that of the karyotypes observed in the single system, with most chromosomes being mediocentric. Finally, Pentacomia sp., a member of Cicindelini, had a single sex chromosome system with n = 10 + X0, but an autosome morphology resembling the mediocentric type was observed in the species with multiple-X systems. Similarly, reports of the presence of a single-X system in C. germanica (Giers 1977 ) were confirmed, but with n = 7 + XY (originally reported as n = 8 + XY). The simple systems in Omini and Megacephalini differ in other features from the karyotype of the other taxa. First, meiosis showed a chiasmatic pattern in the sex chromosomes with one or two chiasmata per bivalent, whereas there was no evidence of chiasmata formation in the multiple-X chromosomes. Second, autosome numbers differed widely but were generally higher in the Omini (n = 17 to n = 21) of intermediate range (n = 12) in Neocollyris and were lowest in Therates and Cicindelini (n = 9 or n = 10) (table 1 ). Finally, fluorescent in situ hybridization using a ribosomal DNA probe (Galián et al. 1995 ; Galián and Hudson 1999 ) showed the presence of eight rRNA loci in Amblycheila and Mantichora, and six loci in Omus, whereas rDNA was limited to two or four chromosomes in the other groups. A further difference is that in the latter the rRNA loci are located on the heterosomes (the Y and in one of the X), whereas in species exhibiting the single-X system they were found in the autosomes. Our findings are generally consistent with data from the literature, which also described a higher number of autosomes (12–15) and a higher number of rDNA clusters (six) in Megacephalini, located on the autosomes (Galián et al. 1995 ; Galián and Hudson 1999 ; Proença, Serrano, and Collares-Pereira 2000 ), than in Cicindela, which shows 7–11 autosome pairs and two, three, or four rRNA clusters located either on the autosomes or on the heterosomes (Galián et al. 1995 ). A close relative of Pentacomia, Odontocheila confusa, has a low number of autosomes (n = 10) and a single sex chromosome system, but in this case it is of the XY type (n = 10 + XY) (Proença, Serrano, and Collares-Pereira 2000 ). Both species have two rDNA loci located on the autosomes. But the report of a multiple system in Megacephala megacephala with n = 12 + XXY (XX XX in females) (Proença, Serrano, and Collares-Pereira 1999 ) may need to be reconsidered because the study did not provide the critical male meiotic stages, such as diplotene or metaphase I, where the multiple sex chromosomes pair would be recognizable as a distinct vesicle. An alternative, more parsimonious interpretation is a meioformula of n = 13 + X0 (males) and n = 13 + XX (females), which fits with male and female mitotic numbers and the observations on the male second metaphase and would be similar to the findings in other Megacephala. To place the karyotype information in a phylogenetic context, basal relationships in Cicindelidae were established from full-length 18S rRNA sequences for 26 species, selected to give a broad representation of major groups, and including some partial sequences from earlier studies (Vogler and Pearson 1996 ; Vogler, Welsh and Hancock 1997 ). A set of 16 sequences of Carabidae and Trachypachidae, which together with Cicindelidae are the Geadephaga (Shull et al. 2001 ), were used as outgroups. Because of extensive length variation in the 18S rRNA gene of Cicindelidae, ranging from 2095 bp in Cheiloxya sp. to 2403 bp in Picnochyla (excluding the 5′ and 3′ termini used as primer annealing sites), the effect of differences in alignment parameters on tree topology was tested using POY (Gladstein and Wheeler 1997 ). Under the POY default options of gap cost = 2, three shortest tree alignments of cost 2403, consistency index = 0.77, and retention index = 0.82 were found (fig. 1 ). They show the monophyly of major taxonomic groups, including Collyrinae, Omina, and Cicindelini, the sister relation of Iresina and Theratina, and the paraphyly of Prothymina with respect to Cicindelina. The analysis also narrows down the sister relationships of Cicindelina, near the African genera Dromica and Prothyma. Similar to what has been established in recent studies (Vogler and Pearson 1996 ; Arndt and Putchkov 1 997), the 18S rRNA data revealed two major deviations from the traditional taxonomy: the position of Collyrinae nested within Cicindelinae and the polyphyly of Megacephalini, which are separated in the Megacephala-Aniara and Oxycheila (and relatives) clades. Most nodes were well supported, as indicated by high Bremer Support values. But changes in gap costs resulted in substantial changes in topology. Under high gap costs, basal groups were included within the derived taxa (e.g., Megacephala within Cicindelini), but this effect is exclusively due to the length-variable middle regions of the V4 and V6 expansion segments. Removing these parts from the tree resulted in a set of nodes (labeled with asterisks in fig. 1 ) universally supported under all gap costs. Inclusion of 14 carabid outgroups recovered the Cicindelidae as monophyletic under all the gap costs tested, in an unexpected derived position within the Harpalini and Scaritini but consistent with more extensive studies of carabid phylogeny based on this gene (Maddison, Baker, and Ober 1999 ). The inclusion of outgroups had a moderate effect on cicindelid relationships, in particular with respect to the Collyrinae, which appear in a more basal position under some alignment schemes. But because of the great length differences of the outgroups, we use the outgroup information only as the basis for rooting the ingroup tree at the node separating the Mantichorini plus Omini from all the others (fig. 1 ). This is the basal node in all the analyses with or without the inclusion of carabid outgroups, and it is always strongly supported. Using the new karyotype data and all existing information from the literature (table 1 ), major karyotypic features were optimized on the phylogenetic tree. This analysis revealed a single inferred gain of the multiple sex chromosome system at the node that groups Collyrinae and Cicindelini (fig. 1 ). Two or possibly three independent losses of the multiple system can be inferred within Cicindela to accommodate C. paludosa and C. germanica, and in Pentacomia and Odontocheila. Because the latter two taxa group together in some of the most parsimonious trees, it is possible that their karyotypic similarity is due to common ancestry and, hence, the loss of the multiple system may be a trait of the wider Pentacomia-Odontocheila clade (a large group of South American tiger beetles associated with tropical forest). Further, this analysis revealed a continuous reduction of chromosome numbers from the basal Mantichora and Omini to the derived taxa in the genus Cicindela with 9, 10, or 11 (or as few as 7, in the case of C. paludosa and C. germanica) autosome pairs. This coincides with a reduction in the number of rRNA gene clusters from the basal to the derived groups and a switch in localization from autosomes to heterosomes. But the position of rDNA appears very unstable and involves repeated switches between the autosomal and heterosomal locations in Cicindelina and appears confined to the autosomes in taxa, which are secondarily single-X. In summary, the phylogenetic analysis favors a single origin of the multiple sex chromosome system in a common ancestor of Collyrinae and Cicindelini. Several species with a single system contained in the multiple-system clade likely represent reversals to the original state, but the gross karyotype morphology resembles the derived type (lower autosome counts, symmetrical karyotype, only two rDNA clusters). Denser taxon sampling still has to confirm that no species with multiple systems exist in the basal groups of Cicindelidae. Accepting the single-origin hypothesis, the multiple-X system is remarkable within beetles for its evolutionary antiquity. Although the age of the common ancestor leading to the split of Megacephala and Collyrinae + Cicindelini can only be estimated very roughly, biogeographical evidence dates this node to a time before the split of the Gondwanian continent >100 MYA (Pearson and Vogler 2001 ). Multiple systems in related groups of Carabidae, including Scarites buparius (n = 17 + XXY) and S. occidentalis (n = 19 + XXY) (Serrano 1981 ; Galián et al. 1999 ), Brachinus sclopeta (n = 14 + XXY) (Galián, Ortiz, and Serrano 1990 ), Ceroglossus chilensis (n = 19 + XXY) (Galián et al. 1996 ), and distantly related Blaps (Tenebrionidae; Vitturi et al. 1996 ; Palmer and Petitpierre 1997 ) are taxonomically restricted and, therefore, of recent origin. These systems are chiasmatic and resemble the classical X1X2neo-Y type whose origin is easily explained as the results of fusions or translocations between the heterosomes and an autosomal pair (White 1973 ). It is unclear why the chiasmatic systems in other Coleoptera are evolutionarily less stable than the achiasmatic system in tiger beetles. Clearly, the cicindelid karyotype is not static, and several features, including the exact numbers of X chromosomes, are highly variable. If indeed the multiple systems in Cicindelidae are derived from a single common ancestor, their widespread presence does not require invoking of the processes of karyotypic orthoselection to promote the repeated evolution of this system. But some kind of selection may still be needed for maintaining the morphological similarities of all multiple-X systems, including the mediocentric appearance of autosomes, which are conserved even if there are obvious chromosome rearrangements, changes in the numbers of autosomes or heterosomes (or both), or even the secondary loss of the multiple system. Whether the primary diversification of the basal and the derived groups was a shift in these orthoselective regimes or a switch from single- to multiple-X systems is still unclear. This regime, however, might have favored the achiasmatic multiple-X system, and once that was in place, changes in chromosome numbers (independently in autosomes and heterosomes) were permissible. The selective forces that maintain the multiple-X system and the underlying molecular mechanisms remain entirely unknown. Whereas knowledge about the evolutionary origin of the sex chromosome system does not permit direct conclusions on the mechanisms of its origination, nor its function in sex determination, the phylogenetic analysis now permits appropriate comparisons to be made. Comparative genome analysis will involve species exhibiting multiple systems with those that exhibit the ancestral and the derived types of single systems, evaluating their commonalties and differences. Given the diversity of sex determination mechanisms and the limited knowledge about sex determination outside a few model organisms, tiger beetles lend themselves very well to these analyses. An important requirement for comparative studies in the future is to establish which genes have accumulated in the sex chromosomes (Wang et al. 2001 ), their relative rates of sequence evolution (Montell, Fridolfsson, and Ellegren 2001 ), and what are the evolutionary consequences of the genome organization for sex determination and differential gene expression in males and females (Hurst 2001 ). A list of GenBank accession numbers and specimens used, and karyotype images of mitotic and meiotic chromosomes of the species analyzed in this study are provided at the SMBE website. Ross Crozier, Reviewing Editor Keywords: chromosome evolution Cicindelidae Coleoptera multiple sex chromosomes rDNA localization Address for correspondence and reprints: Alfried P. Vogler, Department of Entomology, The Natural History Museum, Cromwell Road, London SW7 5BD. a.vogler@nhm.ac.uk Table 1 Chromosome Numbers and Localization of rDNA in Tiger Beetles (Coleoptera: Cicindelidae). Species Numbers in Tribes and Subtribes are from Wiesner (1992). Subtribes Prothymina, Theratina, and Cicindelina are Members of the Tribe Cicindelini. Images of Mitotic and Meiotic Chromosomes of the Species Analyzed in This Study are Available at www.smbe.org Table 1 Chromosome Numbers and Localization of rDNA in Tiger Beetles (Coleoptera: Cicindelidae). Species Numbers in Tribes and Subtribes are from Wiesner (1992). Subtribes Prothymina, Theratina, and Cicindelina are Members of the Tribe Cicindelini. Images of Mitotic and Meiotic Chromosomes of the Species Analyzed in This Study are Available at www.smbe.org Fig. 1.—A phylogenetic hypothesis of Cicindelidae from 18S rRNA. This tree represents the most parsimonious tree alignment derived using POY version 2.7 (Gladstein and Wheeler 1997 ) based on all data (conserved and hypervariable regions), with gap cost = 2 and 20 random addition replicates. Bremer support was calculated in POY using the “-constrain -bremer” command. Asterisks mark the nodes supported under all the alignment conditions tested (conserved regions only). Note the single inferred origin of the multiple sex system. Reversals occurred in derived groups of Cicindela and in Odontocheila-Pentacomia (not shown). All sequences were newly deposited under GenBank accession numbers AF423036–AF423059, except the sequences for Cicindela sedecimpunctata (AF012518) (Maddison, Baker, and Ober 1999 ), and Omus californicus (AF201392) and Oxycheila nigroaenea (AF201393) (Shull et al. 2001 ) We are grateful to J. Serrano for advice and comments, to E. Arndt, D. Brzoska, J. Buestan, D. Maddison, A. Oesterle, D. Pearson, L. E. Peña G., T. Shivashankar, D. Sumlin, and K. Werner for providing specimens, and to S. Proença and E. Martínez-Navarro for technical assistance. We thank Ross Crozier and two anonymous reviewers for their helpful comments. J.G. was supported by Acciones Integradas Hispano Britanicas/The British Council 193A, the Spanish DGICYT Project PB95-1005, and the Large Scale Facility Program which enabled him to visit the NHM, and A.P.V. was supported by the Leverhulme Trust (F/696/H) and NERC (NER/A/S/2000/00489).