The developmental and the tissue-specific expression of glucosephosphate isomerase (GPI), lactate dehydrogenase (LDH) and malate dehydrogenase (MDH) multilocus isozymes were analyzed in samples of Leuciscus cephalus and the adult patterns compared with those of 8 additional Italian cyprinid species: Alburnus alburnus alborella, Chondrostoma genei, L. lucumonis, L. souffia, Rutilus rubilio, R. erythrophthalmus, Scardinius erythrophthalmus and Tinca tinca, the taxonomic status of many of them being uncertain and highly debated. The spatial and temporal patterns of expression obtained generally agree with literature data. Main exceptions are the single expression of GPI-A* and MDH-A* loci of the liver in L. cephalus and the GPI pattern of the eye in all species examined. Since delayed appearence of the subunits coded by the GPI-B* locus and the very early ontogenetic expression of the sMDH-B* locus were found in L. cephalus, the onset of expression of orthologous loci can vary in related species. Genetic structure comparisons support a high genetic divergence of T. tinca from all other species.
INTRODUCTION
In natural fish populations the range of variability and the genetic control of various enzyme systems have been clarified by resolving electrophoretic variants and interpreting the ensuing phenotypes in terms of Mendelian gene loci.
In vertebrates glucosephosphate isomerase (GPI), lac-tate dehydrogenase (LDH) and malate dehydrogenase (MDH) enzymes are known to exist in multiple molecular forms or different isozymes, with significant variations in spatial and temporal patterns of expression (see for a review Basaglia, 1989). The three isozymes are coded by a different number of genetic loci, only some of them being homologous between tetrapods and teleosts (Fisher et al., 1980). Within the latter, mitochondrial and supernatant GPI*, LDH* and MDH* loci have been found to be duplicated, owing to tetraploidization events that independently occurred in different phyletic lineages and were sometimes followed by functional diploidization, revealed as gene silencing (Schmidtke et al., 1975; Ferris and Whitt, 1977). This is a well known phenomenon in the fish family Cyprinidae and in Cypriniformes in general, in which spontaneous tandem duplications of a GPI* locus and the existence of two mAAT* and FH* loci were reported (Buth, 1983; Woods and Buth, 1984; Agnese et al., 1990; Berrebi et al., 1990; Manaresi and Mantovani, 1997).
Moreover, in Cypriniformes the LDH-C* locus has a very specific tissue expression being active only in the liver, as it happens in the order Siluriformes and in some Gadiformes (Basaglia, 1989).
In the present study the developmental and the tissue-specific expression of GPI, LDH and MDH have been determined in samples of the chub Leuciscus cephalus L. and the adult patterns compared with those of other northern-central Italian cyprinid species: Alburnus alburnus alborella De Filippi, 1844 (white bleak), Chondrostoma genei Bonaparte, 1839, Leuciscus souffia Risso, 1826, Rutilus rubilio Bonaparte, 1837 (Adriatic roach), Scardinius erythrophthalmus L. (rudd) and Tinca tinca L. (tench). L. lucumonis Bianco, 1982 (Etruscan chub) and R. erythrophthalmus Zerunian, 1982 (northern Italian roach), two taxa only recently erected as distinct species, have been also analyzed. According to Howes (1991), the majority of these taxa pertain to the subfamily Leuciscinae, with the exceptions of A. alburnus (Alburninae) and T. tinca (incertae sedis). An alternative classification includes the genus Alburnus in the Leuciscinae and T. tinca in the Cyprininae (Banarescu and Coad, 1991). The taxonomic status of most European cyprinid species is unresolved, since species definition is up to now mainly based on morphological characters, which cannot often unambiguosly be used with success. Main debated points on Italian taxa are the taxonomic position of T. tinca within the family Cyprinidae, the generic status of Rutilus and Scardinius, which are considered synonymous by some authors and the relationships among species actually included in the genera Leuciscus and Rutilus (Gandolfi et al., 1991; Howes, 1991; Bianco, 1995).
Considering the usefulness of the electrophoretic analyses applied to the clarification of several taxonomic and phylogenetic questions about American cyprinids (Buth et al., 1991), the necessity of the biochemical approach to the systematic revision of European taxa has been pointed out (Crivelli and Dupont, 1987; Gandolfi et al., 1991).
The aims of the present analysis are therefore to define the expression patterns of three multilocus enzymes in 9 representative cyprinids from Northern Italy and to throw some light on the relationships among these Italian freshwater fishes from a biochemical-genetic perspective.
MATERIALS AND METHODS
Sampling localities and species abbreviations are reported in Table 1. All samples were collected in northern-central Italy from wild population, with the exception of T. tinca obtained from Gavello aqua-culture station.
Table 1
Sampling localities of the analyzed species of Cyprinidae

Specimens were placed on dry ice upon capture and stored at −80°C until utilized for electrophoretic analysis. Twenty individuals of L. cephalus were maintained in a fish tank untill reproduction. After spawning, about 100 eggs were reared in aerated aquaria (10 l) and, at hatching, the larvae fed on natural plancton or commercial foodstuffs. Samples of fertilized eggs and larvae at the time of endogenous feeding (9–11 mm total length, 7 days) were stored at −80°C until singly analyzed. Developmental stages were scored following Economou et al. (1991).
A preliminary survey to check isozyme expression was carried out on fertilized eggs, larvae and samples of adult tissues (white muscle, liver, heart and eye) of L. cephalus; samples of white muscle or eye of adults of the following species were thereafter analyzed for genetic comparisons: A. alburnus alborella, L. lucumonis, L. souffia, C. genei, R. rubilio, R. erythrophthalmus, S. erythrophthalmus, T. tinca.
Samples were analyzed for glucosephosphate isomerase (D-glucose-6-phosphate ketol-isomerase, GPI, EC 5.3.1.9), lactate dehydrogenase (lactate: NAD-oxidoreductase, LDH, EC 1.1.1.27) and malate dehydrogenase (L-malate: NAD+- oxidoreductase, MDH, EC 1.1.1.37). Homogenates were mechanically prepared from equal volumes of tissue and 0.1 M Tris-HCl, pH 7.5 and centrifuged at 12000 ×g for 10 min at room temperature; 3 μl of the supernatant were loaded on cellulose acetate membranes (Cellogel, 17 × 17 cm) previously equilibrated with TEC 0.075 run buffer (Meera Khan, 1971) for horizontal electrophoresis. Runs were carried at constant voltage (220V) for 4 hr (GPI), or 5 (LDH), or 3 (MDH). Staining procedures were after Ayala et al. (1972; LDH), Meera Khan et al. (1982; MDH) and Van Someren et al. (1974; GPI).
Enzyme and locus nomenclature follows Shaklee et al. (1990) and Buth et al. (1991). At each locus, the commonest allele of L. cephalus was reported as 100, while the other alleles were given respective figures on the basis of the relative mobility of electromorphs, adding or substracting from the 100 figure the corresponding millimeters for faster or slower electromorphs, respectively. For polymorphic loci, the 99% criterion was adopted.
RESULTS
Glucosephosphate isomerase
As for the two GPI* loci described in fishes, the commonest electrophoretic pattern observed in the adult specimens of L. cephalus is represented by three equidistant bands, whose relative staining intensity varies with the analyzed tissue. The heart electromorph is the result of the binomial combination of the GPI-A* and GPI-B* subunits nearly equally expressed, while in the eye a predominant activity of the former is found; in the white muscle the reverse is true. Moreover, the phenotype found in eye homogenates shows that the AB heterodimer is less evident than expected on the basis of the free assembly of the A and B subunits. Finally, in liver the GPI-A homodimeric band is the only one present, as it occurs in eggs and larvae (Fig. 1a).
Fig. 1
Schematic representation of differential expression of GPI (a), LDH (b) and MDH (c) isozymes in adult tissues (Mu, white muscle; Li, liver; Ey, eye; He, heart) and larvae (La) of L. cephalus. Egg pattern is identical to the larval one. Differences in line thickness refer to different staining intensities.

The analysis of eye homogenates shows that both loci are polymorphic in the majority of species, but none codes for completely diagnostic alleles (Table 2). In detail, at the GPI-A* locus A. alburnus alborella, L. cephalus and L. lucumonis are monomorphic for the same allele, which represents the most frequent one also in L. souffia, R. erythrophthalmus and R. rubilio. The two congeneric species of Rutilus share the same two alleles. L. souffia shows three alleles, two of them being shared with the Rutilus taxa and the third one with Chondrostoma genei, where it is present at a low frequency. On the other hand, the commonest allele of C. genei is clearly species-specific (Fig. 2a). Furthermore, T. tinca is monomorphic for the fastest allele also observed in S. erythrophthalmus, the latter species showing a second diagnostic allele in addition. Also GPI-B* is polymorphic in all species but three, which show the same allele (A. alburnus alborella, L. cephalus and S. erythrophthalmus). In L. lucumonis, L. souffia, C. genei, R. erythrophthalmus and T. tinca GPI-B* is diallelic, while in R. rubilio is triallelic. Moreover the commonest allele in the tench is species-specific, while C. genei and L. souffia, one side, and L. lucumonis and R. erythrophthalmus, the other, show the same allelic constitution. It is also interesting to note that R. rubilio differs from the congeneric species, R. erythrophthalmus, in having an additional allele (GPI-B*104).
Table 2
Allelic frequencies at the GPI*, LDH* and MDH* loci in 9 cyprinid species. Abbreviations as in Table 1 (n° = sample size).

Fig. 2
GPI (a) and LDH (b) multilocus enzyme expression in eye. (a) 1-2, ALA; 3, LEL; 4-5, LEC; 6-7, CHG; 8-9, LES; 10, RUE; 11, RUR; all electromorphs are GPI-A*100/100 and GPI-B*100/100, with the exception of specimens 3 (GPI-B*95/100), 6 (GPI-A*94/94), 7 (GPI-A*91/94), 8 (GPI-A*91/104), 9 (GPI-A*100/104 and GPI-B*100/106), 10 (GPI-A*100/104), 11 (GPI-B*95/104). (b) 1, ALA; 2, LEC; 3, LEL; 4, SCE; 5-6, LES; 7, CHG; 8-9, RUE; 10-11, RUR; all electromorphs are LDH-A*100/100 and LDH-B*100/100, with the exception of specimens 1 (LDH-B*94/100); 5-10 (LDH-B*85/85) and 11 (LDH-A*100/103 and LDH-B*85/85). Abbreviations as in Table 1.

Lactate dehydrogenase
Tissue analysis in L. cephalus shows that eye and heart electrophoretic patterns consist of five isozymes derived by the contemporary expression of the two isoloci LDH-A* and LDH-B*, with a comparable homo- and heterotetrameric activity, except for the most anodal band, which is less intense in the heart. It should be noted that, when the run is prolonged, the central heterotetrameric band A2B2 appears to be split, as commonly observed in fishes, owing to the formation of conformational isomers (Frankel, 1987). In the muscle, the pattern is characterized by bands of decreasing staining intensity with the slowest one of the A4 homopolymer more expressed; the most anodal isozymes could not be detected. In the liver we found the expression of a third locus (LDH-C*), whose product is the slowest (Fig. 1b). It is important to note that the most intense bands in liver phenotypes are the homotetramer C4 and A4; the heterotetramers from the LDH-C* and LDH-A* loci are also expressed, but they show a decreasing intensity from the C3A1 to the C1A3 heteropolymer. Finally, the heteropolymer A3B1 and A2B2, with a decreasing intensity in this order, are the only evidence on the activity of LDH-B*.
Eggs and larvae electromorphs in L. cephalus show a decreasing staining intensity from the most anodal bands, thus indicating that the LDH-B* locus is the most active, while the A4 homopolymer is clearly missing (Fig. 1b).
For population analysis, only the LDH-A* and LDH-B* were studied extensively in eye homogenates (Fig. 2b and Table 2). The former locus is monomorphic for the same allele (LDH-A*100) in all species but R. rubilio. Only T. tinca can be discriminated on the basis of LDH-A*. A quite different situation emerges from the LDH-B* locus. L. souffia, C. genei, R. rubilio and T. tinca, one side, and L. lucumonis and S. erythrophthalmus, the other, show fixed alternative alleles. The commonest alleles of R. erythrophthalmus and of A. alburnus alborella and L. cephalus match those found in the former and in the latter group, respectively.
Malate dehydrogenase
Three loci are known to exist in fishes; the product from the mitochondrial locus (mMDH*) forms a homodimer and never assembles with those from the two supernatant loci (sMDH-A* and sMDH-B*). On the contrary, the heterodimer is always formed when both sMDH-A* and sMDH-B* loci are active. The mitochondrial and the supernatant electrophoretic zone of MDH activity are therefore easily recognized.
The mMDH* appears almost equally expressed in all analyzed adult tissues and in eggs and larvae of L. cephalus, whereas the products of the two cytoplasmic loci are differently expressed during the developmental stages and in the adult tissues. The only sMDH-A* locus is expressed in the liver and it is also predominant in the eye and heart. Both sMDH-A* and sMDH-B* loci are expressed in the muscle, but the homodimeric activity of the latter is predominant. It should be noted that in developing eggs and larvae electromorphs expressing both supernatant genes are found; they show a clearcut difference in activity, no B2 band being detectable (Fig. 1c).
The three MDH* loci are monomorphic for the same allele in muscle homogenates of all analyzed species, with the exceptions of T. tinca, which has a fixed diagnostic allele at each MDH* locus, and of R. erythrophthalmus, which is polymorphic at the sMDH-B* locus with a low frequency allele (sMDH-B*86) (Table 2).
DISCUSSION
The presence of two GPI*, three LDH* and three MDH* loci supports, for all taxa including L. lucumonis and R. erythrophthalmus, a diploid constitution in line with their karyotypic characterization (Fontana et al., 1970; Cataudella et al., 1977; Manaresi, 1996).
Present data in L. cephalus show that, as in other cyprinids and advanced teleosts, GPI-A* is the only locus expressed during the earliest stages of development and significantly expressed in all tissues with the exception of the white muscle, where GPI-B* predominates (Schmidtke et al., 1975; Basaglia, 1989; Basaglia et al., 1990). The appearence of the subunits coded by the GPI-B* locus seems to be delayed when compared to the condition described in other cypriniforms (Shaklee et al., 1974), a noticeable level of AB isozyme being reached only some days after hatching. Our results support the observation that the onset of the developmental expression of orthologous loci can vary in related fish species (Philipp et al., 1983).
At variance with literature data (Buth, 1984), we observed the expression of the only GPI-A* locus in the liver of L. cephalus and the unbinomial GPI isozymes pattern in the eye of all species examined.
The predominant activity of the LDH-B* locus during early developmental stage has been ascertained in other fishes and in other vertebrate classes (Shaklee et al., 1974), the only exception so far evidenced being two Micropterus species (Philipp et al., 1979). The presence of the heterotetrameric bands indicates that in larvae of L. cephalus, the gene product of LDH-A* locus is also expressed during early ontogenetic stages, even if to a low extent. The appearence of the homotetramer A4 could be related to the differentiation of the muscle cells (Frankel, 1987; Basaglia, 1989). A correlation between isozyme patterns and morphogenetic events, leading to the differentiation of specific tissues and organs, is also suggested by present observation concerning the lack of LDHC isozymes during the early stages of development and their expression in the adult liver of L. cephalus (Shaklee et al., 1974; Basaglia, 1989).
The tissue-specific expression of LDH loci in adults of L. cephalus supports the existence of the cyprinid liver-specific LDH-C* locus, with the slowest migrating C4 homotetramer (Shaklee et al., 1974, Frankel, 1987; Basaglia, 1989). While the simultaneous presence of A, B and C subunits in liver extracts is not unusual in the family Cyprinidae (Frankel, 1987; Basaglia, 1989), it should be noted that LDH-C* appears here the most active, followed by LDH-A* and LDH-B*. The non-binomial staining intensity of the C and A heterotetrames is probably due to the instability of the A-rich heterotetramers in the hepatic tissue, as in AB heterotetramers in the liver of other teleosts (Basaglia, 1989).
The presence of A3B1 and A2B2 heterotetramers in the white skeletal muscle of A. alburnus alborella and S. erythrophthalmus significantly differs from literature data (Basaglia, 1989). Moreover, differently from several cyprinids (Rainboth and Whitt, 1974; Frankel, 1987), all species examined completely lack the fastest muscle isozymes. The polymorphism found at LDH-B* in A. alburnus alborella well agrees with literature data (Callegarini and Basaglia, 1982), whereas in T. tinca we did not found any variation at LDH-A* locus (Basaglia and Callegarini, 1981). The lower variability of LDH-A* locus appears to be a general feature of cyprinids and teleosts (Rainboth and Whitt, 1974).
Comparing the expression patterns of the sMDH-A* and sMDH-B* loci with those in literature data, the A2 homodimer is more anodal (and consequently the fastest MDH form) than the B2 band in this study and vice versa in literature. Otherwise, the MDH patterns observed during development and in adult tissues well agree with previous data on cypriniformes and other teleosts (Philipp et al., 1979; Buth, 1984; Basaglia, 1989), with the exception of the liver in which only the activity of the sMDH-A* locus is evident. A further interesting finding is the existence of the AB heterodimer in L. cephalus eggs a few hours after fertilization, indicating a very early ontogenetic expression of the sMDH-B* locus. This is an additional example of variation in the timing of expression of homologous loci among different fish species (Philipp et al., 1983).
It is evident that the three MDH* gene loci are evolutionally more conservative than GPI* and LDH* loci, 8 species being virtually identical and only T. tinca showing a diagnostic allele at each locus. Intraspecific polymorphism is also very low, R. erythrophthalmus only being diallelic. T. tinca is well separated from other species at the LDH-A* locus as well. On the other hand, the LDH-B* locus defines two groups: the former, sharing the fastest migrating allele, includes A. alburnus alborella, L. cephalus, L. lucumonis and S. erythrophthalmus, while the latter, with the slowest migrating allele, is formed by C. genei, L. souffia, R. rubilio, R. erythrophthalmus and T. tinca. Considering present data together, the tench is clearly differentiated from the other species, supporting the hypothesis that T. tinca should be considered taxonomically apart from the other cyprinid species belonging to the subfamily Leuciscinae (Kryzhanovskii, 1947; Bogutskaya, 1986; Banarescu and Coad, 1991). On the contrary, A. alburnus alborella, which is assigned to the subfamily Alburninae by some authors (Howes, 1991), is not so different from Leuciscinae species by preliminary pairwise comparison of present data. In particular, L. cephalus and L. lucumonis appear very similar to this species. From present analysis, therefore, the retention of A. alburnus alborella in the subfamily Leuciscinae seems reliable, in accordance with Banarescu and Coad (1991).
Acknowledgments
We are grateful to Dott. M. Rizzoli (Ufficio Tutela e Sviluppo Fauna, Provincia di Bologna), Prof. G. Giovinazzo and Dott. M. Lorenzoni (Istituto di Idrobiologia e Pescicoltura, Università di Perugia) for enabling us to collect specimens. This work was supported by Canziani funds.