Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Foschino Dipartimento di Scienze e Tecnologie Alimentari e Microbiologiche, via Celoria 2-20133 Milano, Italy (e-mail: Roberto.Foschino@unimi.it).
Aims: To investigate the basic properties of six temperate and three virulent phages, active on Lactobacillus fermentum, on the basis of morphology, host ranges, protein composition and genome characterization.
Methods and Results: All phages belonged to the Siphoviridae family; two of them showed prolate heads. The host ranges of seven phages contained a common group of strains. SDS-PAGE protein profiles, restriction analysis of DNA and Southern blot hybridization revealed a high degree of homology between four temperate phages; partial homologies were also detected among virulent and temperate phages. Clustering derived from host range analysis was not related to the results of the DNA hybridizations.
Conclusions: The phages investigated have common characteristics with other known phages active on the genus Lactobacillus. Sensitivity to viral infection is apparently enhanced by the presence of a resident prophage.
Significance and Impact of the Study: These relationships contribute to the explanation for the origin of phage infection in food processes where Lact. fermentum is involved, such as sourdough fermentation.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Lactobacillus fermentum is an obligately heterofermentative lactic acid bacterium involved in sourdough fermentation for the production of some wheat and rye breads (Spicher and Schröder 1978; Gobbetti et al. 1994; Onno and Roussel 1994). It is well established that the metabolic activities of lactobacilli significantly contribute to the sensory qualities of these types of baked products, such as the formation of flavour compounds, the rheological characteristics of the crumb, and shelf life (Vogel et al. 1996; Hammes and Gânzle 1997). Most small bakeries use natural cultures in which the typical microflora propagates from a sourdough to the following one (Onno and Roussel 1994; Ottogalli et al. 1996), whereas industrial bakeries prefer selected starter cultures or dried fermented doughs to control and shorten the fermentative process (Stolz and Bocker 1996; Vogel et al. 1996). Continuous systems for sourdough making have already been introduced in bakery plants (Vollmar and Meuser 1992). Lactobacillus fermentum is also used in the fermentation of soy milk products (Poullain 1994), and it is present as a minor component in the natural microflora of whey starter cultures for the manufacture of hard cheeses such as Grana and Gruyère de Comté. Laboratory and pilot tests are in progress to develop an industrial process for making ethanol from sugar-cane juice using Lact. fermentum (Finguerut 2000). One of the greatest hazards of any technological process based on bacterial fermentation is bacteriophage infection capable of slowing or even stopping production. The behaviour of these viruses and their relationships with the host strains requires a thorough understanding in order to exert control, particularly if a fermentation in continuous system is carried out. Little information is available on the occurrence of bacteriophages in Lact. fermentum (De Klerk et al. 1965; Foschino et al. 1995, 1998). The aim of the present work was to investigate the basic properties of six temperate and three virulent bacteriophages active on Lact. fermentum by comparing morphology, host range, structural proteins and genome characteristics.
MATERIALS AND METHODS
Bacterial strains and culture conditions
The lactobacilli used in this study are listed in Table 1. Each strain was confirmed as belonging to Lact. fermentum species (Kandler and Weiss 1986) by phenotypic analysis: cell morphology, Gram stain, catalase reaction, growth at temperature limits, CO2 production from glucose and gluconate, arginine deamination, type of lactic acid isomer, and sugar fermentation patterns with API CHL 50 (BioMérieux, Marcy L’Etoile, France). Selection of strains was carried out according to physiological characteristics and to the different specimen from which the strains were isolated. Lactobacilli were propagated in MRS medium (De Man et al. 1960; Difco) and where specified, the medium was supplemented with CaCl2 10 mmol l–1 (MRS-Ca2+). Frozen stock cultures were maintained at − 30°C in MRS broth containing 20% v/v glycerol.
Table 1. Lactobacillus fermentum strains used and their sensitivity to phages
Induction and isolation of bacteriophages
The phages used in this study are listed in Table 2. Six of them (017, 064, 0209, 0BU130, 0FE129 and FEM) were induced, respectively, from strains CNRZ17, CNRZ64, CNRZ209, BU130, FE129 and FE3, while the remaining three (BU77-B1, Z63-B2 and Z63-B3) were isolated from different environments of food manufacturing. Induction of temperate phages was performed by addition of Mitomycin C at a final concentration of 0·2 μg ml–1 to cultures in the exponential phase of growth (O.D.600nm=0·15) in MRS-Ca2+ broth, and subsequent incubation at 37°C for 8 h. Virulent phages BU77-B1 and Z63-B2 were detected, in Italy, in natural sourdough cultures for wheat bread; virulent phage Z63-B3 was isolated in France from a cheese whey sample. Lysates derived from induction experiments, or from liquid mixtures from food specimens suspected of containing phages, were centrifuged at 3500 g for 30 min to remove bacterial cells and solid materials; the supernatant fluids were filtered through a 0·45 μm membrane and stored at 4°C. In order to find an appropriate indicator strain as propagative culture for each phage, 0·1 ml filtrate was tested against 0·1 ml of 31 different strains of Lact. fermentum in the exponential phase of growth (O.D.600nm=0·15), added to 10 ml fresh MRS-Ca2+ broth and then incubated at 30°C for 12 h until lysis. Three consecutive subcultures were performed to detect delayed lyses due to low phage concentration (Séchaud et al. 1988). The isolation of viruses was carried out by the double-layer technique (Svensson and Christiansson 1991) using MRS-Ca2+ agar inoculated with a selected propagative culture; the plates were incubated at 30°C for 16 h. A single plaque was picked up and mixed with a fresh culture (O.D.600nm=0·15) of the indicator strain; after lysis it was centrifuged at 3500 g for 15 min and then filtered through a 0·45 μm membrane. This procedure was repeated twice. The titre of the phage cultures was defined as plaque-forming units per millilitre.
Table 2. Bacteriophages used and their host ranges
Concentration and purification of bacteriophages
To produce a high number of viral particles, 1 litre of the propagative strain (O.D.600nm=0·15) in MRS-Ca2+ broth was infected with the respective phage at a multiplicity of infection (m.o.i.) of 0·01. After 6–8 h of incubation at 37°C, the lysate was chilled at 4°C, centrifuged at 2000 g for 30 min and then the bacteriophages concentrated by precipitation, according to Yamamoto et al. (1970). Pellets resuspended in buffer TM (Tris-HCl 10 mmol l–1, pH 7·4; MgSO4 10 mmol l–1) were centrifuged at 9000 g for 20 min to eliminate residual polyethylene glycol, and the supernatant fluids containing viruses were treated with RNase A and DNase I (Boehringer Mannheim) to degrade nucleic acid residuals from the bacterial cells. After centrifugation at 100 000 g for 30 min, the bacteriophages, resuspended in TM buffer at 4°C for 48 h, were purified by centrifugation at 100 000 g for 1 h in a discontinuous CsCl gradient (d=1·3–1·7 g cm–3); the blue band of phage particles obtained at 1·5 g cm–3 was then removed and dialysed against buffer Tris-HCl 10 mmol l–1 pH 7·4, MgSO4 10 mmol l–1 and NaCl 0·5 mol l–1 (Séchaud et al. 1992). Viral suspensions were stored at 4 or –30°C.
Phage morphology was observed by transmission electron microscopy (Philips 201, 80 kV, Philips International B.V., Eindhoven, The Netherlands). Suitable dilutions of purified viral suspensions were placed on 300 mesh copper specimen grids coated with carbon film; the grids were then dipped in uranyl acetate (2% w/v, pH 4·5) for 1 min to negatively stain the viral particles. After drying, the preparations were observed at different magnitudes. Data are averages of 10 measurements carried out on at least two different microscopic preparations. Results were submitted to analysis of variance (ANOVA) using Fisher’s method with the application of Duncan’s test (Camussi et al. 1986).
Phage host ranges
Each phage isolated was tested against 31 different strains of Lact. fermentum to establish the relative host spectrum. A 10 ml sample of fresh culture in MRS-Ca2+ broth (O.D.600nm=0·15) was infected with aliquots of phage suspension at a m.o.i. of 0·1. The mixtures were incubated at 30°C and observed periodically. A control test was prepared with every non-infected strain to verify bacterial lysis.
Analysis of phage structural proteins
Aliquots of purified viral suspension were treated with 12% w/v trichloroacetic acid to precipitate phage proteins. After boiling for 10 min in a denaturant solution (Tris-HCl 0·5 mol l–1, pH 9·0; SDS 4% w/v; glycerol 10% v/v; β mercaptoethanol 10% v/v; bromophenol blue 10% w/v), protein preparations were subjected to electrophoresis on a 14% w/v polyacrylamide gel at 15 V cm–1 for 100 min in buffer (Tris-HCl 0·25 mol l–1; glycine 0·19 mol l–1; SDS 0·1% w/v, pH 8·65), as described by Laemmli (1970). Bands were stained with Coomassie brilliant blue R250. Low molecular weight protein markers (Amersham Pharmacia Biotech, Uppsala, Sweden) were used as mass standards to calculate the size of phage proteins. The inaccuracy of this method is ± 10%. Data are averages of measurements calculated on two different gels.
Analysis of phage genome
Nucleic acid was extracted following the procedures of Sambrook et al. (1989). The DNA was digested with four different restriction enzymes (BamHI, EcoRI, EcoRV and HindIII) according to the recommendations of the manufacturer (Amersham Pharmacia Biotech). DNA fragments were subjected to electrophoresis on a 1% w/v agarose gel at 5 V cm–1 in Tris-Acetate-EDTA buffer. Gels were stained with ethidium bromide and photographed under u.v. illumination. DNA standard fragments of HindIII/phage Lambda marker (Amersham Pharmacia Biotech) or EcoRI and HindIII/phage Lambda marker (Boehringer, Mannheim) were used to calculate the size of DNA fragments. The genome sizes were estimated by the sum of molecular weights of fragments generated by digestion. Data are averages of six measurements carried out on at least two different gels.
DNA–DNA hybridization experiments
EcoRV generated fragments of DNA from each of the nine phages were separated by electrophoresis on a 1% w/v agarose gel. The Southern blot technique (Southern 1975) was used to transfer DNA fragments to positively-charged nylon filters. Probes derived from the genome of phages 064, 0FE129, FEM, Z63-B2 and Z63-B3 were used to investigate homology between the viruses; DNAs were labelled with digoxigenin-dUTP by the Random Primed technique, as recommended by the manufacturer (Boehringer Mannheim 1995). Hybridization reactions were performed at 65°C for 24 h. The filters were then submitted to the following sequential washes: 0·1% w/v SDS and 2 × SSC (1 × SSC is NaCl 0·15 mol l–1 plus sodium citrate 0·015 mol l–1) for 10 min, three times, at room temperature; 0·1% w/v SDS and 0·1 × SSC for 10 min, three times, at 55°C. Colorimetric detections with an anti-DIG-alkaline phosphatase conjugate were performed as reported by Boehringer Mannheim (1995).
Induction of lysogenic strains and plaque formation
Treatment with Mitomycin C caused induction of phages 017, 064, 0209, 0BU130, 0FE129 and FEM from strains CNRZ17, CNRZ64, CNRZ209, BU130, FE129 and FE3, respectively. Propagating strains were selected according to non-lysogenic state and ability to form plaques. Strain CNRZ63 was selected as propagating culture for the temperate phages, 017, 064 and 0209, and for the virulent phages, Z63-B2 and Z63-B3, which all formed clear plaques. Strain FE81 was the propagating host for the temperate phages, 0FE129 and 0BU130, that produced small and turbid plaques. Strain BU77 was the propagating host for the lytic phage BU77-B1, which formed clear plaques, and for the temperate phage, FEM, which did not form plaques with any strain tested under the conditions used. Phage FEM was then purified on the above-mentioned strain by the limiting dilution method (Svensson and Christiansson 1991) in successive multiplication steps. Only for the temperate phages, 0FE129 and 0BU130, was MRS medium supplemented with CaCl2 10 mmol l–1 necessary for plaque formation.
Electron micrographs show that all phages belonged to the Siphoviridae family of the International Committee on Taxonomy of Viruses (Matthews 1982), as they had a binary structure with heads and non-contractile flexible tails. ANOVA revealed that there were no significant differences between total lengths of phages 0BU130 and 0FE129, and between those of phages 064 and 0209. Seven phages were ascribed to the B1 group of Ackermann and Du Bow (1986), showing isometric heads of approximately 50–59 nm in diameter. Phage Z63-B2 had a tail 169 nm long; tails of phages 0BU130 and 0FE129 were nearly 185 nm long, whereas those of phages 017, FEM, BU77-B1 and Z63-B3 were around 200 nm long. Base plates 14–20 nm wide were visible only in phages FEM, BU77-B1, Z63-B2 and Z63-B3. For phage Z63-B3, a single fibre at the tip of the tail was present (Fig. 1). The other two phages, 064 and 0209, were ascribed to the B3 group since they showed prolate heads of approximately 120 × 41 nm (Fig. 1). The tail length of phages 064 and 0209 measured 289 nm and 293 nm, respectively. Averages of measurements and relative standard error are reported in Table 3.
Table 3. Morphology measurements, molecular weights of structural proteins and genome sizes of Lactobacillus fermentum bacteriophages
Phage host ranges
The ability of the nine phages to propagate on 31 strains of Lact. fermentum was tested. Patterns obtained from these tests (Table 2) enabled the delimitation of two clusters. The first consisted of temperate phages 017, 064, 0209, 0BU130 and 0FE129, and virulent phages Z63-B2 and Z63-B3, which shared the ability to multiply on a common group of strains. In particular, phages 0BU130 and 0FE129 had nearly identical host ranges. The prolate-headed phage, 064, was active on the same strains as the prolate-headed phage, 0209, plus two additional strains; only virulent phages Z63-B2 and Z63-B3 were able to attack strain CNRZ64, but they failed to lyse strain FE81. The second cluster consisted of temperate phage FEM and virulent phage BU77-B1, which had a coincident spectrum of lytic activity, quite different from the former group of phages. Seventeen out of 31 strains were not attacked by any phages.
Protein composition of phage particles
Protein profiles obtained by SDS-polyacrylamide gel electrophoresis showed one or two intense bands and up to seven minor bands for each phage (Table 3). Some similarities were observed when comparing protein composition. Major proteins with estimated molecular weights ranging from 24 000 to 38 000 were observed in all phages. Very close analogies between protein patterns of phages 064 and 0209, and phages 0BU130 and 0FE129, were found.
Characterization of phage genome
All phages contained DNA. Genomes of the temperate and virulent phages were compared by restriction endonuclease analysis. Phage DNAs were treated with four enzymes: BamHI, EcoRI, EcoRV and HindIII. Electrophoresis revealed coincident profiles for phages 064 and 0209 (Fig. 2a), and for phages 0BU130 and 0FE129 (Fig. 2b), with all the restriction enzymes tested. Digested DNA of the other phages showed different patterns. Lengths of DNA ranged from 31·7 to 49·6 kbp (Table 3).
Each labelled genome of phages 064, 0FE129, FEM, Z63-B2 and Z63-B3 was used as a probe to hybridize the EcoRV digested DNAs of all phages blotted onto nylon membranes. It is possible that differences in signal intensity were due to different amounts of DNA being loaded onto the gel. The probe from temperate phage 064 showed homology only with most of the restriction fragments of phage 0209 (Fig. 3b). The probe from temperate phage 0FE129 revealed a high homology with DNA of temperate phage 0BU130, since signals were detected in all the bands, and also a partial homology with the genome of temperate phage FEM, located in one fragment of 3·6 kb, and with the genome of virulent phage Z63-B2 located in two fragments of 7·4 and 4·5 kb (Fig. 3c). The probe from temperate phage FEM was able to hybridize with one fragment of 7·2 kb of the genome of temperate phages 0BU130 and 0FE129 (Fig. 3d). The probe from virulent phage Z63-B2 confirmed the ability to hybridize with DNAs from temperate phages 0BU130 and 0FE129 (signals located in the same three fragments of 7·2, 6·3 and 4·2 kb), and also with DNA from virulent phage BU77-B1 (signals located in two fragments of 4·1 and 1·3 kb) (Fig. 3e). Conversely, no hybridization was detected between labelled DNA from virulent phage Z63-B3 and the genome of the other phages.
In order to verify whether the high level of similarity found between temperate phages 064 and 0209 was due to the fact that strains CNRZ64 and CNRZ209 could be two isolates of the same micro-organism, RAPD PCR with bacterial DNA was performed using eight different oligonucleotides, 0·5 μmol l–1, as primers (Operon Technologies Inc., Alameda, CA, USA). The same investigation was carried out for strains BU130 and FE129 harbouring the temperate phages 0BU130 and 0FE129. Bacterial DNA was amplified in a thermal cycler (PCR Sprint, Hybaid Ltd, Ashford, Middlesex, UK) with the following temperature profile: denaturation step at 94°C for 1 min; annealing step at 32°C for 45 s through 5 cycles, then 40°C for 45 s through 25 cycles; extension step at 72°C for 1 min. The results of amplification with OPI 4 primer (5′-AATCGGGCTG-3′) (data not shown) showed that CNRZ 64 and CNRZ 209 were really two different strains, while none of the primers used was able to distinguish strains 0BU130 and 0FE129.
There are few reports in the literature concerning Lact. fermentum phages. De Klerk et al. (1965), who first studied the fine structure of viral particles active on this species, described eight lytic phages belonging to the Myoviridae family and only one temperate phage ascribed to the Siphoviridae family, having an unsheathed tail 182 nm long with an icosahedral head 50 nm wide. Recently, Foschino et al. (1995) isolated two virulent Lact. fermentum phages from a sourdough sample for wheat bread production. The first, FE5-B1, presented the typical morphology of Myoviridae, with a sheathed contractile tail 170 nm long and an icosahedral head 83 nm wide; the second, Z63-B1, belonged to the Siphoviridae family, having a non-contractile tail 160 nm long and an isometric head 60 nm wide. Another virulent phage, Z63-B2 (Foschino et al. 1998), active on the same species but isolated from sourdoughs in a different bakery plant, was shown to be very similar to phage Z63-B1. All phages examined in the present study were ascribed to the Siphoviridae family; seven out of nine belonged to the B1 morphotype, while the other two phages revealed uncommon shapes and were attributed to the B3 morphotype according to Ackermann’s classification (Ackermann and Du Bow 1986). Phages of lactobacilli with prolate heads have rarely been isolated. The first, described by Séchaud et al. (1988), was the temperate phage 0235 harboured in Lact. delbrueckii subsp. lactis strain CNRZ 235; the second, active on the same species of bacterium, was the lytic phage JLC 1032 characterized by Forsman (1993). Another prolate-headed phage, named Φy8 and spontaneously released from Lact. acidophilus strain Y8, was isolated from fermented milk by Kiliçet al. (1996); the morphology of phages 064 and 0209 was similar to this one. All had unusual elongated flexible tails.
Data on host ranges showed two separated clusters containing both temperate and virulent phages. Viruses of the first group attacked a set of strains where, remarkably, six out of eight were lysogenic. Because of the effect of immunity, none of the temperate phages caused lysis of the respective harbouring strain. Phage 064 was unable to lyse strain CNRZ 209, which is lysogenic for phage 0209 and vice versa; the same observation applied to temperate phages 0BU130 and 0FE129. The second group consisted of the temperate phage FEM and the virulent phage BU77-B1, which both propagated only on four strains isolated in Italy from whey samples. They were completely inactive on strains that were sensitive to the viruses of the first cluster. Most of the cultures were insensitive to the attack of any phage; 14 out of 17 of these strains appeared to be non-lysogenic. Furthermore, analysis of the sensitivity patterns of the 31 strains revealed that cultures that proved to be sensitive to any phage could be found more frequently among lysogenic (six out of nine) than non-lysogenic strains (eight out of 22). Similar observations were made in the comparative study of Séchaud et al. (1992), carried out on 35 bacteriophages of Lact. helveticus; in this case, there were 77% sensitive strains among lysogenic cultures and 58% among non-lysogenic strains. In the same way, Fayard et al. (1993) reported that lysogenic strains of Streptococcus salivarius subsp. thermophilus were good, or sole, indicators for temperate phages which did not belong to the same immunity group of the harboured phage. Moreover, they observed that the passage of a native temperate phage in a lysogenic indicator strain expanded the host range and triggered a DNA rearrangement of the viral particle. Therefore, contrary to what has generally been described, these results suggest that sensitivity to viral infection is apparently enhanced by the presence of a resident prophage. However, phage resistance of strains is not necessarily affected by lysogeny, since the mechanism of immunity works only with closely-related phages (Davidson et al. 1990; Hill 1993). It is thus likely that the sensitivity pattern of a lysogenic strain is concerned more with its genetic properties and prophage than with the molecular characteristics of fibres or base plates of an outer infecting viral particle. In fact, grouping of phages obtained by host ranges is not related to the results of the DNA hybridization analysis. Viral particles which showed equal or narrow sensitive strain patterns, such as phages FEM and BU77-B1 or phages 017 and 0209, showed no evidence of genetic homology; on the other hand, the virulent phage Z63-B3, whose genome did not hybridize with any other, shared sensitive strains with six other phages.
Protein profiles obtained by SDS-PAGE of phages 064 and 0209 appeared identical to those of phages 0BU130 and 0FE129, while patterns of the remaining viral particles were all different. Numbers and values of molecular masses of the major proteins were similar to those reported in other studies concerning the characterization of phages active on lactobacilli (Mata et al. 1986; Séchaud et al. 1988; Davidson et al. 1990; Forsman 1993; Foschino et al. 1995).
Restriction analysis of genomes showed coincident patterns between DNAs of phages 064 and 0209, and between those of phages 0BU130 and 0FE129. All the other phages exhibited individual profiles of restriction. The calculated sizes of genomes agree with those reported in the literature for other bacteriophages of lactobacilli (Mata et al. 1986; Séchaud et al. 1988; Davidson et al. 1990; Forsman 1993; Foschino et al. 1995). Strict relationships between phages Z63-B2 and Z63-B1 have already been found at the physiological, structural and genetic level, confirming them as variants of one another (Foschino et al. 1995, 1998).
Southern hybridization experiments confirmed a high degree of homology between the genomes of temperate phages 064 and 0209, and those of temperate phages 0BU130 and 0FE129. Results of RAPD PCR with DNAs of strains CNRZ64 and CNRZ209 suggested that the respective phages, 064 and 0209, are identical viral particles harboured in two strains isolated from different specimens at different places and times. With regard to strains BU130 and FE129, the amplification patterns were coincident, so the two harboured phages, 0BU130 and 0FE129, are the same viral particle present in two isolates of a unique culture found in a sample of Grana cheese whey. DNAs from prolate-headed phages 064 and 0209 did not show any homology with genomic DNA of the other isometric-headed phages, which is in disagreement with Forsman (1993), who found few, but highly homologous, DNA regions between the genomes of these two morphotypes.
Despite the differences found in restriction enzyme digestion patterns, partial homologies were also detected among DNAs from virulent phages, Z63-B2 and BU77-B1, and temperate phages, 0BU130, 0FE129 and FEM. In a comparative study on Lact. casei phages reported by Forsman et al. (1993), similar results were obtained; lytic phages, ΦFSW and LC-Nu, and temperate phage PL-1 showed different structural proteins and DNA restriction profiles, but one-third of each viral genome proved to be highly homologous (> 85%) with those of the other phages. A significant relationship was found by Vasala et al. (1993) between the virulent phage, LL-H, and the temperate phage, mv4, both active on Lact. delbrueckii subsp. lactis. Although they were isolated in different places at different times, these phages revealed a high level of homology in the structure and organization of the gene cluster encoding structural proteins.
The demonstration of lysogeny is valuable at the applied level, as temperate phages are a potential source of virulent phages in food processing. Similarities in morphology, host range, structural proteins and DNA compositions between temperate and lytic phages reported in several studies (Mata et al. 1986; Séchaud et al. 1988; Davidson et al. 1990) support this thesis. Until now, phage infections in small bakeries and industries have not caused serious problems because most sourdough processes are still carried out in a batch system using mixed natural starters. These mixtures contain several strains, often belonging to different bacterial species and not all sensitive to a given phage strain (Ottogalli et al. 1996). The risk of a phage infection increases when a continuous system or selected culture is adopted. Information about the genetic characteristics of starter cultures and their interactions with bacteriophages can contribute to correct control procedures being applied during the fermentative process.
This work is dedicated to the memory of Jean-Pierre Accolas, esteemed researcher and beloved teacher, who passed away in January 2001. The authors are grateful to Dr F. Fauro and Dr D. Mora for their advice and assistance. The study was supported by grants from Ministero dell’Università e della Ricerca Scientifica e Tecnologica, Quota ex-60% 1998 and 1999.