Application of ubiquitin SSCP analysis in taxonomic studies within the subgenus Orinocarabus ( Coleóptera : Carabidae : Carabus )

SSCP (single-strand conformation polymorphism) analyses of ubiquitin genes were used to investigate evolutionary rela­ tionships within the subgenus Orinocarabus of the genus Carabus. After SSCP electrophoresis of PCR-amplified ubiquitin copies, population-specific band patterns were obtained. Ubiquitin-SSCP-analyses of the six central European Orinocarabus species, including three subspecies and thirteen populations, resulted in a dendrogram that differed from that based on morphology. Phyloge­ netic analysis of mitochondrial DNA (mtDNA) did not support the SSCP dendrogram, but was in good accordance with the tax­ onomy based on morphological characters. The reason for the discrepancies seems to be evolutionary conservation of the ubiquitin genes. The time that elapsed since the evolution of the closely related Orinocarabus species is too short for concerted evolution of the ubiquitin genes.


INTRODUCTION
The subgenus Orinocarabus shows a disjunct distribu tion in the European Alps.The relatively high number of species, subspecies, microraces and hybrids within the genus Carabus has resulted in a diversity of views on the subgeneric and subspecific classification.The most recent revision of the genus (Deuve, 1991) is based on the system suggested by Ishikawa (1973Ishikawa ( , 1978Ishikawa ( , 1979)), which uses endophallic structures as the main criteria for infra generic classification.In recent years an analysis of DNA sequences was used for taxonomic studies (Pruser, 1996;Pruser & Mossakowski, 1998).
The classification at the species and subspecies level within the subgenus Orinocarabus is based on morpho logical data.The occurrence of many subspecies and races is controversial.In this study molecular techniques were used to obtain more information about the taxo nomic relationships within this group.
Ubiquitin is a small 228 bp multifunction protein found in all eukaryotes.It is involved in the nonlysosomal pro teolysis of proteins (Hershko & Ciechanover, 1986).Fur thermore, it is important in other cellular processes, such as DNA repair (Jentsch et al., 1987), cell response to stress (Bond & Schlesinger, 1985), programmed cell death (Schwartz et al., 1990) and kinetochore function (Kopski & Huffaker, 1997).In the nucleus, ubiquitin is bound to histone 2A and 2B (Bonner et al., 1987), impli cating its role in the regulation of gene expression.
Two different classes of ubiquitin genes are present in the eukaryotic genome.One class comprises two types of fusion genes, encoding a single ubiquitinjoined to a ribosomal protein (Swindle et al., 1988;Cabrera et al., 1992;Barrio et al., 1994).The other class consists of polyubiquitin genes, which contain ubiquitin repeats in tandem (Wiborg et al., 1985).The number of repeats within such polyubiquitin genes varies from one in Giardia lamblia (Krebber et al., 1994) up to 52 in Trypanosoma cruzi (Swindle et al., 1988).Examples of insect species with a relatively high number of repeats in polyubiquitin genes are Manduca sexta, with 15 repeats (Myer & Schwartz, 1996) and Drosophila melanogaster, with 18 repeats (Lee et al., 1988).Thus, SSCP analysis of ubiquitin genes yields information about numerous loci.
To investigate polymorphisms associated with ubiquitin genes we used SSCP (single strand conformation poly morphism) electrophoresis.This technique enables one to detect a polymorphism in a fragment of DNA due to as little as a single base substitution (Orita et al., 1989a, b).The differentiation of DNA fragments of the same length differing in their sequences is based on a sequencedependent mobility shift of single-stranded DNA during electrophoresis.Sheffield et al. (1993) found that single base substitutions are detectable by SSCP analysis but that the sensitivity depends on the fragment length.Their data revealed a high sensitivity in a range between 95 and 212 bp (up to 76% of all single point mutations were detected among 29 different 212 bp fragments) and a strong decrease in mutation detection with increasing length above these values.We used a 210 bp PCR frag ment of the whole 228 bp ubiquitin gene, which should enable a high resolution in SSCP analysis.
Although the amino acid sequence of the ubiquitin pro tein is highly conserved the degenerated genetic code enables variation on DNA sequence level.Therefore, the multicopy gene ubiquitin with its several repeats seems to have the potential for SSCP based analysis of population differentiation and in taxonomic studies.
Only a few examples of the use of SSCP analysis in taxonomic investigations have been reported.In most of these studies (Hiss et al., 1994;Tokue et al., 1995;Travis & Keim, 1995;Walsh et al., 1995;Stothard et al., 1998;Koekemoer et al., 1999) single gene loci were used, which limited the information content of the SSCP pat terns.Rarely have multiple loci been investigated (Ohsako et al., 1996, Nakamura et al., 1998).
The advantage of SSCP analysis of the multicopy gene ubiquitin is that the simultaneous PCR amplification of several loci gives SSCP banding patterns with numerous bands.Thus, the information content of these patterns is greater than that obtained from single-copy genes.A first attempt to analyse the multicopy gene ubiquitin by SSCP was made by Boge et al. (1994).SSCP analysis of ubiq uitin genes of four Carabus species showed numerous bands for each specimen, which differed between the spe cies.But no further analysis of the ubiquitin SSCP banding pattern was made.
In this study we examined the suitability of SSCP analysis of the ubiquitin genes, using new beetle-specific primers, for taxonomic research.Our aim was to improve the identification and differentiation of Orinocarabus subspecies and populations (races) using SSCP electro phoresis, and to obtain further relevant data for taxo nomic studies.

Samples
Six species of ground beetles of the subgenus Orinocarabus, including three conspecific subspecies were collected at twelve locations in the central European Alps (Fig.

DNA extraction
Total DNA was extracted from the thorax and femora of the beetles as described by Boge et al. (1994).

Sequencing of an ubiquitin dimer
To obtain sequence information from beetle specific ubiquitin genes, an ubiquitin dimer of C. alpestris dolomitanus was sequenced.For this purpose ubiquitin genes were amplified by PCR as described by Boge et al. (1994), using degenerate ubiquitin primers because no corresponding sequence data were available for Coleoptera.The PCR products were separated using a 1.5% agarose gel.Due to the tandem organisation of the polyubiquitin repeats, which are not separated by introns, ubiquitin PCR results in monomeric as well as multimeric ubiquitin fragments.The dimeric ubiquitin copies were extracted from the gel using the USBioclean MP-Kit (USB).These extracted dimeric PCR products were cloned using the pCR Script SK(+) Cloning Kit (STRATAGENE) according to the supplier's proto cols.Plasmids containing the ubiquitin insert were isolated using a QIAprep Spin Plasmid Miniprep-Kit (QUIAGEN) and a dimeric ubiquitin insert was sequenced commercially in both directions (Sequiserve, Germany).The sequence (Fig. 2) has been deposited at the EMBL data library under the accession number X94621.

Ubiquitin SSCP analysis
PCR of ubiquitin was performed with primers designed according to the sequenced ubiquitin dimer of C. alpestris dolo mitanus (Fig. 2).Sequences of the commercially (Pharmacia) synthesised primers, which enclose a 210 bp section of the whole 228 bp ubiquitin gene, were: 5'-TCT TCG TCA AGA CCC TCA CT-3' and: 5'-GAC GGA GGA CCA ACT GAA GA-3'.
PCR reactions were set up in 25 pl reaction volumes con taining 0.5 pg genomic DNA, 5 nmol of each dNTP, 25 pmol of each primer, 2.5 pl 10x reaction buffer and 0.8 units DNA polymerase (DynaZyme, Biometra).The thermocycling profile consisted of an initial step at 94°C for 2 min, 9 cycles of 30 sec at 94°C and 45 sec at 50°C, followed by 27 cycles of 30 sec at 94°C and 45 sec at 55°C and was completed by a final step at 72°C for 5 min.-c--------a--t--t---c----g After separation of resulting PCR products on a 1.5% agarose gel the band containing the monomeric ubiquitin fragments (210 bp) was cut out of each lane.The gel slices were homogenised in 20 pl 1* PCR buffer and centrifuged for 10 min at 12,000 rpm.To reamplify the monomeric ubiquitin fragments, 3 pl of the liquid phase was introduced into a second PCR for 2 min at 94°C and 24 cycles of 30 sec at 94°C and 45 sec at 55°C fol lowed by 5 min at 72°C.The isolation of the monomeric ubiq uitin fragments from other PCR products after the first PCR minimises the occurrence of non specific products during SSCP electrophoresis, which could have been amplified during the first PCR from genomic DNA.
SSCP electrophoresis of the reamplified monomeric ubiquitin fragments was carried out using the Multiphor II system (Phar macia).4.5 pl PCR product was mixed with an equal volume of formamide and denaturated at 95°C for 5 min.After denatura tion, samples were loaded immediately onto a 15% poly acrylamide gel (MiniCleanGel 15% DNA-HP; ETC, Germany).Running conditions were 20 min, 120 V, 23 mA, 5 W and 60 min, 600 V, 30 mA, 18 W using the DNA Disc Buffer Kit (ETC).After electrophoresis, gels were silver-stained according to the supplier's protocol.
SSCP banding patterns were digitised using the Image Master video system (Pharmacia) and analysed using ONE-Dscan (Scanalytics) to obtain the corresponding Rf-values.The bands were matched on a basis of maximum divergence of 1% between two compared bands.Band sharing rates of Nei & Li (1979) between the SSCP patterns and the corresponding dis tances were calculated.The resulting distance matrix was ana lysed with the RESTSITE package (Miller, 1990) using UPGMA (Sneath & Sokal, 1973).

Sequencing of mtDNA
A fragment of 559 bp of mtDNA (Fig. 5), consisting of the 3' end of NADH-dehydrogenase subunit 1 (ND1), a tRNA for Leucine and the 5' end of the 16 S rRNA was amplified from at least one specimen from each location representing all detected ubiquitin banding patterns.The following primers were used: 5'-TAG AAT TAG AAG ATC AAC CAG C-3' (Weller & Pashley, 1995) and 5'-ACA TGA TCT GAG TTC AAA CCG G-3' (Vogler & DeSalle, 1993).PCR products were sequenced commercially (TOPLAB) by direct sequencing in both direc tions.
The 3' end of the ND1 gene, which is the most variable part of this sequence, consisting of 350 bp, was analysed using PAUP 3.1.1(Swofford, 1993).A maximum parsimony cladogram was generated after 1000 bootstrap replications using the branch and bound search option.C. carinthiacus was used as the outgroup to root the tree, because this taxon has clearly more base substitutions than the other taxa (data not shown).
Furthermore, the same data set was analysed using the PHYLIP package 3.573 (Felsenstein, 1993) and the maximum likelihood method.The ratio of transitions to transversions (T/V) was set to 2.57 as this is the calculated average T/V ratio of the sequenced ND1 fragments of the relevant taxa (data not shown).A phylogenetic tree was generated after 1,000 bootstrap replications.

Ubiquitin
Sequencing of a cloned ubiquitin dimer of Orinocarabus alpestris dolomitanus resulted in a sequence of 417 bp (Fig. 2), consisting of two joined ubiquitin repeats.The end of the first repeat and the beginning of the second were utilised for primer design to optimise PCR of the ubiquitin fragments of the beetles under investigation.The new primers enclosed 210 bp of a 228 bp-ubiquitin unit, rather than the 189 bp of the primers used initially (Boge et al., 1994).SSCP-analysis of ubiquitin genes of six central Euro pean Orinocarabus species, including three conspecific subspecies and 13 populations, resulted in 14 different banding patterns (Fig. 3).The number of bands varied from 28 to 37, with an average of 31 bands.Within popu lations there was little divergence between banding pat terns.Only C. concolor concolor specimens from both locations (Fig. 4) and those of C. sylvestris nivosus from one (Berninapaß) of three locations showed two distinct banding patterns.In the case of C. concolor concolor two fingerprints (type 1 and type 2) were found in habitats that were close to one another.Of the seven Furkapaß specimens, two showed type 1 and five type 2 finger prints.Of the three Grimselpaß specimens two showed type 1 and one type 2 fingerprints.

Mitochondrial DNA
C. concolor concolor at both locations showed two dis tinct different SSCP patterns (Fig. 4).An explanation for this could be that two species coexist at each location, which are indistinguishable morphologically.To eluci date this, 559 bp of mtDNA of one individual of each SSCP type of C. concolor concolor from both locations were sequenced.This revealed identical sequence data for the two SSCP types within each habitat.Comparison of specimens from the two habitats shows sequence differ ences of 2.0% (Fig. 5).Hence, the coexistence of two species at Furkapaß and Grimselpaß was refuted.Simi larly, the two SSCP types from the Berninapaß population of C. sylvestris nivosus showed identical mtDNA sequences (data not shown).

Taxonomic classification Ubiquitin-SSCP-analysis
Analyses of all 14 different SSCP patterns resulted in a dendrogram (Fig. 6) that differs from the classification based on morphological data.This is illustrated, for example, by the distant branching of the two SSCP types of C. sylvestris nivosus from Berninapaß (B1, B2) and of the two types of C. concolor concolor (T1, T2), respec tively.Furthermore, the two populations of C. linnei linnei and all three subspecies of C. alpestris are clearly separated.
On the other hand, the combining position of both populations of C. alpestris hoppei and of two populations of C. sylvestris nivosus (A, F), respectively, accords with the morphological classification.In general, however, the tree reveals obvious inconsistencies with the conventional taxonomic classification.

DNA sequencing
Maximum parsimony analysis of 350 bp of the mito chondrial ND1 gene was performed using PAUP 3.1.1(Swofford, 1993).The resulting tree (Fig. 7a) is clearly different from the dendrogram based on ubiquitin-SSCPanalysis, but is in good agreement with the taxonomy based on morphological characters.All populations of C. alpestris are grouped together as well as both the popula tions of C. linnaei linnaei and two of the three popula tions of C. sylvestris nivosus.The third one is at the pre vious node.Only the two populations of C. concolor con color are clearly separated.
The maximum likelihood tree (Fig. 7b), calculated with PHYLIP 3.573 (Felsenstein, 1993) is similar to the maximum parsimony tree.Its topology is even more similar to the taxonomy based on morphology, because the two populations of C. concolor concolor are not sepa rated but grouped together in the same subtree with the C. sylvestris nivosus populations.(Swofford, 1993).Bootstrap values were obtained from 1,000 replicates using the branch and bound search option and are indicated on the branches.All nodes with bootstrap values above 25% are shown.7b (right) -maximum likelihood tree using PHYLIP 3.573 (Fel senstein, 1993).T/V ratio was set to 2.57.Bootstrap values of 1,000 replications are indicated on the branches.All nodes with boot strap values above 25% are shown.

DISCUSSION
SSCP analysis of ubiquitin genes of six species from thirteen populations resulted in 14 different banding pat terns.The number of bands in each pattern varied between 28 to 37. One reason for this variation could be a different number of ubiquitin repeats in these beetles.The minimum and maximum number of bands are represented by the two SSCP types of C. concolor concolor.Since these extremes are found in specimens from the same population, specific variation in the number of ubiquitin copies could be excluded.However, the variation could be due to allelic polymorphism.Such a polymorphism was found in Trypanosoma brucei (Wong et al., 1992), where two polyubiquitin alleles of the same locus consist of 13 and 30 repeats, respectively.Alternatively it could be due to sequence differences between polyubiquitin repeats.The sequences of the two repeats of the sequenced ubiquitin dimer (Fig. 2) differ by 25.2%.It can not be excluded that primer annealing during PCR is influenced by sequence variability within the different ubiquitin repeats, resulting in different numbers of SSCP bands.
The observed difference of 25.2% is very high com pared with that between conspecific ubiquitin repeats in other species.Sequence data from 20 polyubiquitin genes from 14 eukaryotic species (Tan et al., 1993) revealed a maximum difference of 25.0% between polyubiquitin repeats within a species.In this context, the two insect species, Drosophila melanogaster and Manduca sexta, exhibited a maximum difference of 8.3% and 17.5%, respectively (Tan et al., 1993).
In this study the ubiquitin SSCP patterns are almost identical, with two exceptions.The Berninapaß popula tion of C. sylvestris nivosus and the two populations of C. concolor concolor (Furkapaß and Grimselpaß) show two distinct different SSCP patterns.In C. concolor concolor the two patterns exist in each population (Fig. 4).Sequences of 559 bp of mtDNA reveal no differences between specimens from the same C. concolor concolor population but with different ubiquitin-SSCP types.How ever, the sequences for individuals from the two popula tions of C. concolor concolor differ by 2.0%.This is a high value for variation in mtDNA between populations compared with other insect species.Sequence divergence between the C. concolor populations is in the range of values reported for populations of Drosophila silvestris (DeSalle & Templeton, 1992), but greater than that reported for several other insect populations (DeSalle et al., 1987;Satta & Takahata, 1990;Düring & Mossakowski, 1995) and between insect species (Vogler et al., 1993(Vogler et al., , 1998;;Sperling & Hickey, 1994).The studies cited above used a combination of sequences from mito chondrial protein and RNA genes (rRNA and tRNA).Our data suggest that the gene pools of the two populations of C. concolor concolor have been separated for a relatively long time.The occurrence of the same two ubiquitin-SSCP patterns in both populations reveals that these two ubiquitin types of C. concolor concolor existed before the populations were separated.Hence, the relevant ubiquitin genotypes have been conserved, at least during the period of population separation.
Comparison of different SSCP patterns was based on band sharing rates of Nei & Li (1979).Due to the separa tion by sequence differences during SSCP electro phoresis, fragments from different specimens at the same position in the gel should share the same sequence.Therefore, the matched band positions, which are used for calculating these similarity indices, should belong to homologous gene loci.Analysis of all the ubiquitin-SSCP patterns resulted in a dendrogram that differed greatly from that resulting from morphological systematics.The discrepancies are most probably caused by the conserva tion of the ubiquitin genes.Therefore the reliability of a dendrogram based on ubiquitin-SSCP types depends on the evolutionary age of the genotypes.In the UPGMA dendrogram (Fig. 6) SSCP types from conspecific popu lations (C.linnaei linnaei) are well separated as are the SSCP types from the same population (B1, B2; T1, T2).From the close relationship of these types to SSCP types of other species it is concluded, that these genotypes were already present before the recent split of these species from their ancestors.In contrast, the genotypes of the SSCP types of both populations of C. alpestris hoppei and of two populations of C. sylvestris nivosus (A, F), which are united in the dendrogram in each case, may have been generated in recent times after these species separated.As the ubiquitin genes are conserved the SSCP electrophoresis of these genes did not reflect the speciation process within the subgenus Orinocarabus.
Although no conclusions can be drawn from analyses of the ubiquitin SSCP data about the phylogeny of Orino carabus, something can be concluded about the evolution of the ubiquitin genes.Sharp & Li (1987) postulated con certed evolution within polyubiquitin loci to explain the higher level of homogenisation between repeats within a polyubiquitin gene than between repeats compared across species.They assumed that concerted evolution between ubiquitin loci is much less effective than within polyubiq uitin genes.Tan et al. (1993) analysed additional polyu biquitin genes of other species and included ubiquitin fusion protein loci in their study.They concluded that concerted evolution is an effective force for homoge nising ubiquitin repeats not only within but also between loci.As a result of concerted evolution, polyubiquitin genes should evolve in accordance with the organisms in which they exist.Hence polyubiquitin genes of organisms belonging to the same taxon should differ less in their DNA sequences than the genes of different taxa.This effect of concerted evolution of ubiquitin genes was rec ognised in two other studies, dealing with higher verte brates (Vrana & Wheeler, 1996;Nenoi et al., 1998).
This study shows that within Orinocarabus it is likely that concerted evolution has not occurred in general.Only the similarity of the ubiquitin SSCP patterns of indi viduals from both populations of C. alpestris hoppei and from two populations of C. sylvestris nivosus (A, F) implies concerted evolution of the corresponding ubiq uitin genes.In all other cases, where conspecific popula tions and subspecies were investigated, no evidence for concerted evolution of the ubiquitin genes was found.
The absence of concerted evolution between the ubiq uitin genes in the taxa studied is supported by the phylo genetic analysis of mtDNA (Fig. 7a, b).In contrast to those obtained from the ubiquitin data, these cladograms group most of the conspecific populations together and are thus in great accordance with the taxonomy based on morphology.Some of the nodes in these trees are charac terised by weak bootstrap values below 50%.However, as the main focus of this work was on the SSCP analysis of ubiquitin, these nodes are shown to illustrate the close relations of the conspecific populations in these analyses of mtDNA.
The main reason for the absence of concerted evolution across the ubiquitin loci is probably the close relationship of the species studied compared to previous investigations where the species belonged to different higher taxonomic levels, like different orders or phyla.Due to the highly conserved character of the ubiquitin genes, and the recent splitting of these species from one another, there has been too little time for concerted evolution to affect the homogenisation of the ubiquitin repeats within these spe cies.To obtain more detailed knowledge about concerted evolution of ubiquitin genes in insects, it is necessary to study more species at different taxonomic levels, like genera, families or orders.
Ubiquitin SSCP analysis was not suitable for taxonomic studies of Orinocarabus species.Nevertheless, it could be useful for investigations of phylogenetic relations of higher taxa within a wide range of different species as ubiquitin is present in all eukaryotes.

Fig. 3 .
Fig. 3. SSCP patterns of all different ubiquitin types.The let ters in parentheses refer to different locations.K -Kühtai; TH -Turracher Hohe; KT -Kohlbachtal; R -RollepaB; B -Bernina-paB; F -FlüelapaB; A -Ammerwald.The numbers indicate the SSCP patterns within O. concolor concolor and within the BerninapaB population of O. sylvestris nivosus.The first and the last lane (M) show alOObp ladder (Pharmacia)

Fig. 4 .
Fig. 4. The two different uhiquitin SSCP types of O. concolor concolor (four of the six patterns of type 2 are shown).The uhiquitin SSCP pattern of individuals from the two locations Furkapaß (F) and Grimselpaß (G), of two distinct types, with little variation within type 1 and no detectahle variation within type 2. Both types occur at hoth locations.M -100 hp ladder (Pharmacia).

Fig. 7 .
Fig. 7. Phylogenetic trees based on 350 bp of ND1.Abbreviations for the sampling locations given in parentheses are explained in Fig 3, with two exceptions: Fu -FurkapaB; G -GrimselpaB.7a (left) -maximum parsimony tree using PAUP 3.1.1(Swofford, 1993).Bootstrap values were obtained from 1,000 replicates using the branch and bound search option and are indicated on the branches.All nodes with bootstrap values above 25% are shown.7b (right) -maximum likelihood tree using PHYLIP 3.573(Fel  senstein, 1993).T/V ratio was set to 2.57.Bootstrap values of 1,000 replications are indicated on the branches.All nodes with boot strap values above 25% are shown.