Skeletal muscle cells fall into several specialized classes, termed fiber types, which show variations in contractile and metabolic properties. The isoforms of the myosin heavy chain (MHC) molecule represent the best markers of muscle fiber diversity (Pette and Staron, 1990). It is thought that both MHC composition and skeletal muscle fiber types are two of the most important determinants in meat quality and meat products (Xiong, 1994; Picard et al., 1998a). Different combinations of MHC isoforms may occur within the same fiber, but the predominant isoform is the main determinant of the fiber's functional properties such as the speed of contraction and fatigue resistance (Schiaffino and Reggiani, 1996). The molecular diversity of adult skeletal muscle fibers is species specific (Pette and Staron, 1990), and important physiological differences between species related to body size have already been reported (Rome et al., 1990). In adult muscle of certain small species of mammals (i.e., rat, mouse, and guinea pig) three fast MHC isoforms—termed IIA, IIX, or IID (henceforth IIX), and IIB—have been identified (Bär and Pette 1988; Schiaffino et al., 1989; Gorza, 1990). The relative abundance of the MHC-IIB isoform decreases among mammals with increasing body size and it is not expressed in humans (Smerdu et al., 1994), carnivores (Talmadge et al., 1996), ruminants (Tanabe et al., 1998), and horses (Rivero et al., 1999). Some recent studies in adult pigs (Lefaucheur et al., 1998) and llamas, Lama glama (Graziotti et al., 2001), have shown evidence for the existence of all three fast MHC isoforms, including the MHC-IIB, and suggest that the lack of expression of the MHC-IIB is not only a matter of body size.
The differential distribution of MHCs defines a number of pure fibers containing a single isoform and some hybrid fiber populations containing two or even more isoforms. Most work in muscle research has relied upon a histochemical classification system, where stainings for acid or alkaline labile myofibrillar ATPase (mATPase) activity (Guth and Samaha, 1970; Brooke and Kaiser, 1970) enable muscle fibers to be categorized in three main phenotypes, one slow-twitch (type I) and two fast-twitch (IIA and IIB). Some refined mATPase histochemical methods have been developed to identify hybrid fibers (Staron and Pette, 1993). Quantitative protein characterization of such histochemically defined fiber types has demonstrated a close relationship between specific mATPase staining profiles and the expression of distinct MHC isoforms (Sant'Ana Pereira et al., 1995). However, the main disadvantage of the mATPase histochemical classification system is the subjective delineation among fiber types (Rivero et al., 1996a). This imprecision becomes critical in some ruminant species in which both acid (Suzuki, 1989; Maier et al., 1992; Ibenbujo et al., 1996; Picard et al., 1998a) and alkaline (Spurway et al., 1996; Picard et al., 1998a) stabilities of their mATPase systems are very close. For this reason, skeletal muscle fiber types in ruminants have frequently been classified according to the metabolic scheme proposed by Peter et al. (1972) (e.g. Suzuki and Tamate, 1988). Nevertheless, this scheme underestimates the full diversity in the expression of MHCs. Furthermore, both oxidative and glycolytic potentials of muscle fiber types with homogeneous MHC content vary greatly (Rivero et al., 1998).
An alternative, and probably much more objective, approach to identify the MHC isoform (s) expressed in individual muscle fibers is to apply immunohistochemistry with specific monoclonal antibodies (MAb). In recent years, the specificity of these antibodies for the identification of orthologous myosins in a number of mammals has been documented (Table 1). This is possible because the structure of each MHC isotype is generally conserved between species, whereas greater sequence divergence can be found between different isoforms within the same species (Schiaffino and Reggiani, 1996). Nevertheless, a significant functional diversity between orthologous sarcomeric myosins has recently been documented (Canepari et al., 2000). The full cellular diversity of bovine skeletal muscle fiber types has been reported by immunohistochemistry (Picard et al., 1998a). In contrast, the complete range of muscle fiber phenotypes was not achieved in two previous studies of small ruminants (Maier et al., 1992; Manabe et al., 1996). The major problem related either to the identification of hybrid fibers containing two MHC isoforms (Maier et al., 1992) or to the objective resolution of the fast-twitch fiber subtypes (Manabe et al., 1996).
Table 1. Specificity of monoclonal antibodies used in the study against adult mammalian skeletal myosin heavy chain (MHC) isoform*
UO = unpublished observation.
This isoform (MHC-IIB) has recently been identified as MHC-IIX in both bovine (Tanabe et al., 1998) and equine (Rivero et al., 1999) skeletal muscles.
Therefore, reliable methods for classifying fiber types need to be developed in small ruminants. Here we present a novel approach, based on the combined immunohistochemical analyses of MHC isoforms in serial frozen sections of caprine skeletal muscle, which allows identification of the main five muscle fiber phenotypes (I, I+IIA, IIA, IIAX, and IIX). Antibodies used in the study were characterized in tissue homogenates by immunoblottings. Acid and alkaline stabilities of mATPase systems, oxidative and glycolytic capacities, and size of skeletal muscle fiber types were measured by quantitative histochemistry and correlated on a fiber-to-fiber basis. In this paper, the nomenclature MHC-I, MHC-IIA, and MHC-IIX has been adopted when referring to the three isoforms obtained in adult caprine skeletal muscle. However, our results are not conclusive with regard to the identity of the second fast-twitch MHC isoform expressed in limb goat muscles.
MATERIALS AND METHODS
Four adult (2–3.5 years old) female and clinically healthy Murciano-Granadina (Spanish breed) goats were used for the study after approval by the Animal Care Committee of the University of Cordoba. Once the goats were sacrificed with an overdose of pentobarbital, right M. semitendinosus was dissected and cleaned of connective tissue. The semitendinosus was chosen because it is frequently sampled in caprine meat quality studies (Suzuki, 1989) and due to its propulsive nature a high number and full diversity of fast-twitch fibers in this muscle can be expected. Muscle samples (1×1×0.5 cm) were removed from the midbelly region of the muscle, frozen by immersion in isopentane kept in liquid nitrogen, and then stored at −80°C until analyzed.
In order to characterize the specificity of the MAbs, used for the first time on caprine skeletal muscle, MHC electrophoresis was performed following the protocol for sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) described by Talmadge and Roy (1993). The original method was modified by: (1) heating the myofibrillar protein at 60°C for 19 min, (2) increasing the glycerol content of the separating and stacking gels to 30%, (3) increasing the gel thickness to 1 mm, and (4) decreasing the voltage to 130 V for 24 hr. Aliquots of diluted myofibrillar protein were electrophoresed in a large-gel apparatus (SE 600, 18 × 16 cm; Hoefer Pharmacia Biotech. Inc., San Francisco, CA) and separating gels were stained with Coomassie blue. Three bands were clearly separated in the gels (see Results).
In separate unstained gels, the MHC isoforms were electrophoretically transferred to nitrocellulose sheets (Towbin et al., 1979) for immunoblot analyses (Rossini et al., 1995). Nitrocellulose filters were stained for specific MHCs using primary antibodies listed in Table 1. The primary antibody was visualized with a peroxidase-labeled secondary antibody (goat anti-mouse IgG, code No. E0433, Dako A/S, Glostrup, Denmark) followed by development with diaminobenzidine in the presence of imidazole.
Frozen samples were transferred to a cryostat at −20°C, serially sectioned at 10 μm thickness and mounted on poly-L-lysine-coated glass slides for immunohistochemistry and histochemistry. Serial sections were reacted with a panel of MAbs specific to MHC isoform (Table 1). The avidin-biotin complex (ABC, PK-6100, Vector Laboratories, Inc. Burlingame, CA) immunohistochemical procedure was used for the localization of primary antibody binding as previously described (Rivero et al., 1996b). In brief, sections were preincubated in a blocking solution of stock goat serum. The primary MAb was then applied and allowed to incubate overnight in a humid chamber at 4°C. An additional section was incubated without specific primary MAb and used as blank tissue to demonstrate the non-specific reactivity. On the second day the sections were washed and then reacted with a secondary antibody (biotinylated goat antimouse IgG). Sections were washed again and reacted in ABC reagent. Diaminobenzidine tetrahydrochloride was used as a chromogen to localize peroxidase. The optimum working dilution of primary antibodies was searched individually for each MAb (Table 1) in order to distinguish only two staining intensities (positive or negative).
Additional sections were stained for myofibrillar ATPase (mATPase) activity after acid (pH 4.3 to 4.6, 2 min) and alkaline (pH 10.3 to 10.6, 10 min) preincubations by using a modification (Nwoye et al., 1982) of the Brooke and Kaiser (1970) method for the acid pretreatment and of the Guth and Samaha (1970) method for the alkaline preincubation. The optimum pH of the preincubation solution was searched carefully to distinguish at least three levels of staining intensities after both acid and alkaline preincubations. Additional serial sections were stained for quantitative succinic dehydrogenase (SDH; Blanco et al., 1988) and α-glycerophosphate dehydrogenase (GPDH, Martin et al., 1985), and used to assess oxidative and glycolytic capabilities, respectively, of individual muscle fibers. For each quantitative enzyme assay (SDH and GPDH), four consecutive cross-sections were cut; two sections were incubated in a medium containing substrate and two in a medium that lacked substrate. Sections incubated without the substrate were used as tissue blanks to correct for non-specific staining occurring during the reaction. The activity of SDH was determined in 10-μm-thick sections according to the quantitative histochemical procedure previously described (Blanco et al., 1988). The activity of GPDH was determined on 14-μm-thick sections according to Martin et al. (1985), except that the time of incubation was 25 min. After staining, all sections (SDH and GPDH) were digitized within 2–3 hr with a computerized image processing system (see below). Since SDH and GPDH activities were linear for 15 and 40 min, respectively, termination of the reactions before these points allowed enzymatic activities to be expressed as steady-state enzyme activity rates. These activities were expressed as optical density (OD)/min (Martin et al. 1985).
A region of the cross-sections containing apporximately125 fibers was selected for further analyses. The sections stained for immunohistochemistry, mATPase, SDH, and GPDH histochemistry were surveyed to find regions free of artifacts. Serial sections were visualized and analyzed using a Leica DMLS microscope (Leica Microsistemas, Barcelona, Spain), a Leica high-resolution color charge-coupled device camera (Leica Microsistemas, Barcelona, Spain), an eight-bit Matrox meteor frame-grabber (Matrox Electronic Systems, Barcelona, Spain), combined with an image-analyzing software (Visilog 5, Noemi, Microptic, Barcelona, Spain). With the use of the mATPase staining after acid preincubation, a fiber mask was drawn along the cell borders of the desired number of fibers. Images of the remaining sections were then fitted to the fiber mask. Single fibers were subsequently identified and visually classified as positive vs negative for immunohistochemistry, and as light, moderate, or dark for histochemistry.
Furthermore, for each fiber analyzed the mean optical density (OD) was determined for acid and alkaline mATPase, SDH, and GPDH reactions. The cross-sectional areas (CSA) of the same individual muscle fibers were also measured in the SDH histochemical reaction. Measurements for each fiber were made by duplicate in two consecutive serial sections, and the mean of both was used to quantify the intensity of the reaction and size of individual fibers. For each individual fiber, the mean of the two measurements with substrate minus the mean of the two tissue blank sections was used to calculate the enzyme activity. Since a number of factors can influence the reliability of histochemical enzyme activity determination, we examined the measurement precision associated with several sources of variability (thickness of the tissue, image analysis system, etc.) by repeated measurements on the same single fibers in two consecutive serial sections. The coefficients of variation for duplicate measurement of ODs and CSAs in all histochemical assays were always below 4%.
Quantitative information was averaged according to fiber type and differences among mean values analyzed by a one-way ANOVA. In the presence of a significant F ratio, post hoc comparisons of means were provided by a Fisher's least significant difference test. Statistical significance was accepted at P < 0.05.
Characterization of MHC Markers
A typical MHC isoform separation of the goat M. semitendinosus obtained by using a large gel system and Coomassie blue staining is shown in Figure 1 (lane 1). Identification of these MHC bands was examined by immunoblotting of MHCs after SDS-PAGE (Fig. 1A–F). The fastest migrating MHC band was labeled by the MAbs BA-D5, BF-35, and S5-8H2 (Fig. 1A,D,E). The slowest migrating band was labeled by MAbs SC-75, SC-71, and BF-35 (Fig. 1B–D). The middle band was labeled by MAbs SC-75, SC-71, S5-8H2, and BF-G6 (Fig. 1B,C,E,F). Based on the specificity of these MAbs (Tables 1 and 2), the three electrophoretical bands were tentatively identified as types IIA, IIX, and I, going from the slowest to the fastest-migrating band (Fig. 1).
Table 2. Visual immunohistochemical and histochemical characterization of skeletal muscle fiber types in the goat
The number of each fiber type (1 to 5) corresponds to those shown in Figure 2.
For monoclonal antibodies: + and − = positive and negative reaction, respectively.
For histochemistry: +, ++, and +++ = light, moderate, and dark stainings, respectively. Hybrid fibers 2 and 4 had intermediate (Interm) staining intensities in between their respective pure phenotypes.
Immunohistochemistry allowed subdivision of three different muscle fiber populations containing a unique MHC isoform: one slow-twitch (type I), two fast-twitch (IIA and IIX), and two hybrid populations, one containing MHCs I+IIA and another with both fast-twitch MHCs (IIAX) (Fig. 2A–F and Table 2). Hybrid IIAX fibers were clearly discriminated from pure IIX fibers because they were positive with the MAb BF-35, specific to MHC-I and MHC-IIA, (e.g., fiber labeled 4 in Fig. 2D), whereas pure IIX fibers did not react with this MAb (e.g., fiber 5 in Fig. 2D). Similarly, pure IIA and hybrid IIAX fiber phenotypes were clearly separated by their differential reaction against the MAb S5-8H2, specific to MHC-I and MHC-IIX. Thus, pure IIA fibers were not labeled with this antibody (e.g., fiber 3 in Fig. 2E), whereas this reaction was positive for IIAX fibers (e.g., fiber 4 in Fig. 2E). The hybrid type I+IIA fibers—a fiber phenotype equivalent to C fibers (Staron and Hikida, 1992)—were also well discriminated using the present immunohistochemical approach. These fibers were labeled with all MAbs used (e.g., fiber 2 in Fig. 2A–E), except the MAb BF-G6. After analyzing the MHC composition of a population of 500 fibers in the goat M. semitendinosus, 40, 2, 23, 21, and 14% were identified as types I, I+IIA, IIA, IIAX, and IIX, respectively. These figures indicate the prominence of hybrid IIAX fibers.
Acid and Alkaline mATPase Stabilities
Based on combined mATPase reactions after acid (pH range, 4.4–4.5) and alkaline (pH range 10.4–10.5) preincubations, caprine skeletal muscle fibers could be divided into three main categories (Fig. 2G–I and Table 2). Type I fibers were acid-stable and alkaline-labile (e.g., fiber 1 in Fig. 2G,H). The reverse was true for type IIA fibers (e.g., fiber 3 in Fig. 2G,H). Type IIX fibers were partially acid and alkaline stable (e.g., fiber 5 in Fig. 2G,H). Quantitative differences in the staining intensity for mATPase activity after acid and alkaline preincubations of these three pure fiber types were statistically significant (Fig. 3A). Consequently, these three major fiber phenotypes could be objectively discriminated with only two sections stained for mATPase histochemistry (Fig. 4A). In addition, intermediate hybrid fibers containing MHC-I plus MHC-IIA, were also delineated with these two mATPase histochemical reactions (Fig. 4A), as they stained medium-to-dark after both acid and alkaline preincubations (e.g., fiber 2 in Fig. 2G,H). Conversely, fast-twitch hybrid fibers (IIAX) had mATPase activities intermediate to those of their respective pure MHC fiber types (e.g., fiber 4 in Fig. 2G,H). Quantitatively, the acid mATPase stability of hybrid IIAX fibers was not statistically different when compared with pure IIA and IIX fibers (Fig. 3A). Alkaline mATPase stability of IIAX and IIX fibers was not statistically different, but the alkaline ATPase activity of IIAX fibers was significantly lower than that of pure IIA fibers (Fig. 3A). As a consequence, all fast-twitch hybrid fibers had overlapping mATPase activities and they could not be objectively divided into discrete categories with this mATPase histochemical scheme (Fig. 4A).
Metabolic and Fiber Size Properties
The staining intensities of SDH (Fig. 2I) and GPDH (not shown) activities formed a continuum, but significant differences in mean values were observed between the three pure fiber types (Fig. 3B). Type I fibers had the highest SDH activity and the lowest GPDH activity. The reverse was true for type IIX fibers. Type IIA fibers had intermediate SDH and GPDH activities. Once again, hybrid fibers coexpressing two MHC isoforms showed intermediate mean SDH and GPDH values, lying in between their respective pure phenotypes. When the SDH and GPDH activities were plotted on a bi-dimensional basis, there was an inverse relationship (r = −0.78, P < 0.01) between SDH and GPDH activities, but discrete groups of fibers were not objectively discriminated by these variables (Fig. 4B).
Pure MHC muscle fiber types were statistically different with regard to their fiber CSA (Fig. 3C). Type I fibers were the smallest fibers, type IIX fibers were the largest, and type IIA were intermediate. Hybrid fibers were of intermediate size in between their respective pure phenotypes. There was an inverse relationship (r = −0.85, P < 0.001) between CSA and SDH activity (Fig. 4C), but this relationship did not allow a clear grouping of muscle fibers. Nevertheless, the smaller type I and type IIA fibers had the highest SDH activities, while the largest type IIX fibers had the lowest SDH activity. A positive correlation between fiber CSA and GPDH activity (r = 0.81, P < 0.01) was also recorded (Fig. 4D). The values of GPDH activity were the highest in the larger type IIX and IIA fibers, and lowest in the smaller type I fibers.
The primary focus of this study was the accurate classification of adult skeletal muscle fiber types in a small ruminant (Capra hircus) according to the MHC isoprotein they express. The application of the anti-MHC MAbs panel used resulted in a satisfactory method for this goal, being particularly useful to delineate fast-twitch hybrid fibers. The results also showed evidence that both acid and alkaline stabilities of mATPase systems of the two fast MHCs are very close (but significantly different within a narrow range of pHs) in this species. Fast-twitch hybrid fibers, however, could not be objectively delineated by our mATPase histochemical methods. Phenotypic differences in metabolic and size properties observed in goat myofibers were linked to the MHC expression, but these variations were deeply overlapped and did not allow discrimination of myofibers according to their MHC content into a system of discrete types.
Three MHC isoproteins (one slow-twitch and two fast-twitch) were demonstrated to be expressed in adult caprine skeletal muscle. Similar findings were observed in ovine (Maier et al., 1992) and bovine (Tanabe et al., 1998) skeletal muscles. Whereas the identity of one of the two fast MHC isoforms seems to be clearly a type IIA-MHC isoform, the present results are not conclusive regarding the identity of the second fast MHC isoform. Nevertheless, several lines of evidence support the notion that the MHC-IIB isoform is not expressed in limb skeletal muscles of small and large ruminants. First, these three isoforms have recently been sequenced and identified as types I, IIA, and IIX in cattle muscles (Tanabe et al., 1998). Second, in our study the same notion is based on the homologies of the electrophoretical mobility and immunoreactivity of the MHC-IIX in the goat, with the same isoform of rat (Schiaffino et al., 1989; Gorza, 1990; Talmadge and Roy, 1993; Rossini et al., 1995) and other big mammals (Talmadge et al., 1996; Rivero et al., 1999). Nevertheless, some evidence for structural divergence between orthologous MHC-IIX isoforms can be suspected from the variations in the immunoreactivity of this isoform between goat (present results) and rat (Schiaffino et al., 1986, 1989; Gorza, 1990) muscles. And third, the gene order in the direction of mRNA transcription of adult sarcomeric MHC family is IIA → IIX → IIB (Shrager et al., 2000) and this physical organization is highly conserved among mammalian species (Weiss et al., 1999). This linear order has been related to regulation of temporal expression patterns and physiological properties of these MHC isoforms (Weiss et al., 1999; Shrager et al., 2000). Thus, one cannot expect the temporal expression of the fastest MHC-IIB without the previous expression of the MHC-IIX, which has a slower shortening velocity. Nevertheless, further molecular analyses are necessary to confirm if the second fast-twitch MHC isoform expressed in limb muscles of the goat is really a type IIX-MHC isoform.
The differential distribution of the three MHCs expressed in the goat muscle defines three pure fiber types containing a single MHC (I, IIA, and IIX), and two intermediate hybrid fiber populations containing two MHCs (I+IIA and IIAX fibers). The hybrid IIAX fiber phenotype totaled 21% of the M. semitendinosus fiber population in the goat samples. In cattle, these fibers represented 11% of M. longissimus thoracis and 8% of M. semitendinosus (Picard et al., 1998a). These data show that a considerable proportion of hybrid IIAX fibers exists in skeletal muscles of sedentary animals, confirming that they are not necessarily the expression of transitional status, but are in dynamic equilibrium as a stable fiber population. Nevertheless, since the animals used in the present study were relatively young adults (2–3.5 years old), the high percentage of hybrid IIAX fibers can be related with the maturation process of caprine skeletal muscle, since the numbers of this fiber increase with age upon the basis of a fiber transition in the order IIX → IIA (Rivero et al., 1993).
Our data confirm a high correlation between MHC content and mATPase activity in goat pure muscle fiber types, but not in hybrid fibers. A similar situation has already been reported in other species (Staron and Pette, 1983; Rivero et al., 1996a). An objective delineation of type II fiber subpopulations can be achieved with the simultaneous application of two mATPase histochemical reactions after acid and alkaline denaturations (Fig. 4A). Type IIA fibers were more acid-labile than IIXs, but only within a very narrow range of pHs (from 4.4 to 4.5). Below this range, both fiber types showed identical mATPase activities. This pattern of the acid mATPse reaction is common in many other species, but the range of pHs for optimum discrimination between II fiber subtypes is usually more expanded (Gorza, 1990; Pette and Staron, 1990). The alkaline mATPase reaction of caprine type II fibers was higher in IIA fibers than in IIXs. This response, which is standard in rodent muscle (e.g., Spurway, 1981; Gorza, 1990) and has many analogies in lower vertebrates (e.g., Rowlerson and Spurway, 1988) is not commonly noted in large mammals, where IIX fibers are more alkaline-stable than IIAs (e.g., Picard et al., 1998a). An alkaline mATPase reaction of type II fibers similar to that seen in the goat (current study) was reported in the hartebeest (Alcelaphus buslaphus), an East African ruminant (Spurway et al., 1996), as well as in the llama (Lama glama), a South American cud-chewing mammal related to camels (Graziotti et al., 2001). In the light of these similarities, it can be speculated that the consistent “rodent” pattern of the alkaline mATPase method seen in these larger mammals may well be a zoological trait of Artiodactyla (ruminants and llama) species. Otherwise (and/or additionally), this apparent inconsistency of the alkaline mATPase reaction across different species can also be connected with technical variations of mATPase histochemical methods used, as demonstrated in dog skeletal muscle fiber types (Latorre et al., 1993).
Differences in oxidative capacity, glycolytic capacity, and size observed among goat muscle fiber types were, in general, associated with the MHC content, and very similar (in relative terms) to those previously reported in other mammalian species (Spurway et al., 1996; Rivero et al., 1998). The physiological background of these variations and the functional impact of their interrelationships have been extensively discussed in these latter studies. The current data in this subject confirm that all fibers in the goat formed a continuum in staining intensities for quantitative SDH and GPDH activities, and CSA values. As a consequence, myofibers could not be satisfactorily classified according to these characteristics into discrete groups. Nevertheless, SDH, GPDH, and CSA values of individual muscle fibers are not randomly distributed, but they are closely correlated on a fiber-to-fiber basis.
In conclusion, the present study clearly shows the existence of three MHC isoforms in the M. semitendinosus of the adult goat: one slow and two fast. Whereas the identity of two of these isoforms seems to be clearly types I and IIA MHCs, the second fast MHC isoform remains to be elucidated. Nevertheless, this isoforms has been tentatively designated as MHC-IIX in the present study based on the homologies of its electrophoretical mobility and immunoreactivity with the same isoform of other mammals. The differential distribution of these MHCs defines three major fiber types containing a single MHC (I, IIA, and IIX) and two intermediate hybrid fiber populations containing both slow and fast IIA-MHCs and the two fast MHC isoforms (type IIAX fibers). These data demonstrate that skeletal muscle fiber types of domestic small ruminants have not been accurately classified in previous studies. The prominence of fast-twitch hybrid fibers and the metabolic diversity of type II fibers have probably been underestimated in all these publications. The improvement in accuracy of muscle fiber typing is of practical importance to better understand the involvement of fiber types in growth and meat quality traits of these meat-producing species.
The authors thank Dr. Stefano Schiaffino (University of Padova, Italy) for his generous gift of monoclonal antibodies. The antibody S5-8H2 is another generous gift of Dr. Eric Barrey (INRA, France). The authors also acknowledge Linda Linnane for supervising the grammar and style of the manuscript, and Dr. Brigitte Picard for critically reading the manuscript. Jose-Luis L. Rivero was supported by the Spanish DGESIC (PB98-1016).