OUP user menu

Characterization of the cellulolytic complex (cellulosome) of Clostridium acetobutylicum

Fabrice Sabathé , Anne Bélaïch , Philippe Soucaille
DOI: http://dx.doi.org/10.1111/j.1574-6968.2002.tb11450.x 15-22 First published online: 1 November 2002


A large cellulosomal gene cluster was identified in the recently sequenced genome of Clostridium acetobutylicum ATCC 824. Sequence analysis revealed that this cluster contains the genes for the scaffolding protein CipA, the processive endocellulase Cel48A, several endoglucanases of families 5 and 9, the mannanase Man5G, and a hydrophobic protein, OrfXp. Surprisingly, genetic organization of this large cluster is very similar to that of Clostridium cellulolyticum, the model of mesophilic clostridial cellulosomes. As C. acetobutylicum is unable to grow on cellulosic substrates, the existence of a cellulosomal gene cluster in the genome raises questions about its expression, function and evolution. Biochemical evidence for the expression of a cellulosomal protein complex was investigated. The results of sodium dodecyl sulfate–polyacrylamide gel electrophoresis, N-terminal sequencing and Western blotting with antibodies against specific components of the C. cellulolyticum cellulosome suggest that at least four major cellulosomal proteins are present. In addition, despite the fact that no cellulolytic activities were detected, we report here the evidence for the production of a high molecular mass cellulosomal complex in C. acetobutylicum.

  • Cellulosome
  • Dockerin domain
  • Cohesin domain
  • Modular protein
  • Clostridium acetobutylicum

1 Introduction

Multienzyme complexes having high activity against crystalline cellulose, known as the cellulosome, have been identified and characterized in various microorganisms [13]. The best-characterized cellulosome systems are those from Clostridia species such as Clostridium cellulolyticum [4], Clostridium cellulovorans [5], Clostridium thermocellum [6] and Clostridium josui [7]. A common feature of the clostridial cellulosomes is that they consist of a large number of catalytic components arranged around non-catalytic scaffolding proteins. The following scaffolding proteins, CipC in C. cellulolyticum [8,9], CbpA in C. cellulovorans [10], CipA in C. thermocellum [11] and CbpJ in C. josui [7] have been identified and characterized from a molecular point of view. These non-catalytic proteins have a cellulose-binding domain (CBD) [12,13] and several hydrophobic domains termed cohesin. Cellulosomal enzymes have a C-terminal 7-kDa dockerin domain which is involved in the adhesion to the cohesin domains of the scaffolding protein.

Clostridium acetobutylicum ATCC 824 is a Gram-positive, spore-forming, anaerobic bacterium that converts sugars and polysaccharides (e.g. starch) into acids (acetate and butyrate) and solvents (acetone, butanol and ethanol). It is one of the best-studied solventogenic bacteria and a closely related strain has been used extensively for the industrial production of solvent. Although this bacterium is no longer used in industry, C. acetobutylicum continues to be the subject of numerous studies, including recent efforts to apply the tools of molecular biology and genetic engineering to alter substrate specificity and improve the efficiency of solvent production.

Recently, the entire sequence of the C. acetobutylicum genome has revealed the presence of a large cellulosomal gene cluster [14], showing strong similarity with that of C. cellulolyticum. It is well known that C. acetobutylicum is not a cellulolytic bacterium since this strain cannot grow on cellulose as the sole carbon source [15]. It is thus intriguing that while this microorganism apparently possesses all the necessary genes to produce cellulose-degrading enzymes, it is actually unable to hydrolyze this substrate.

In the present study, we have clearly established a link between the gene sequences and the existence/activity of a cellulolytic complex in C. acetobutylicum. Sequence alignments between the cellulases from C. acetobutylicum and C. cellulolyticum showed a high degree of similarity between the two species. For this reason, we used antibodies raised against the cellulases from C. cellulolyticum (Cel5A, Cel8C, Cel9E, Cel48F and the CBD of Cel9G) to obtain immunochemical evidence of the production of homologous cellulases by C. acetobutylicum. Finally, the identification of the main components was confirmed by N-terminal sequencing after separation by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and the scaffolding protein was identified by immunodetection using antibodies raised against the last cohesin domain.

2 Materials and methods

2.1 Bacterial strain and media

C. acetobutylicum was obtained from the American Type Culture Collection (ATCC 824) and was grown anaerobically at 37°C in the previously described synthetic medium [16] with cellobiose (20 g l−1) and crystalline cellulose (10 g l−1) as the carbon source.

2.2 Cellulosome purification

C. acetobutylicum was grown for 16 h in 50 ml serum bottles with shaking at low speed (100 rpm). Before total consumption of cellobiose (5 g l−1 residual), cellulose was harvested by decantation of the culture fluids and washed two times in 50 ml of 100 mM potassium phosphate buffer (PBB), pH 7.0, to remove cells. Then the cellulose pellet was filtered through a 3-µm pore-size glass filter and extensively washed two times with 50 ml of 100 mM PBB, and one time with 50 ml of 25 mM PBB. The cellulosome was then eluted with 10 ml 1% triethylamine. The eluted fraction was centrifuged at 15 000×g for 15 min to remove all of the insoluble material, and the supernatant was concentrated by diafiltration against water on a Millipore PBGC (10-kDa cutoff) membrane. The 10 ml diafiltrate that was obtained was concentrated to a final volume of 1 ml by centrifugation on a Biomax-30K NMWL membrane (Millipore).

2.3 Cloning of a cohesin domain

The DNA fragment encoding the fifth cohesin domain of CipA was amplified by polymerase chain reaction (PCR) from C. acetobutylicum genomic DNA. The following forward 5′-CGCCACGCATATGGCAGGAAATGCAGGTACATTT-3′ and reverse 5′-CCCGCTCGAGTTAGTGGTGGTGGTGGTGTTCAACAGTTATTTTTCCATT-3′ primers, which possess partial homology with the 5′ and 3′ ends respectively of the DNA region encoding the cohesin 5, were used for the amplification and introduction of Nde I (5′) and Xho I (3′) restriction sites. The reverse primer was also designed to add five histidines at the C-terminus of the cohesin 5. The amplified fragment was digested by Nde I and Xho I and ligated into the pET22b(+) vector linearized with the same endonucleases. The resulting plasmid, pETC5, was used to transform Escherichia coli BL21(DE3) strains containing an inducible T7 polymerase under the control of the lac promoter.

2.4 Purification of a cohesin domain

Cells were grown aerobically in Luria–Bertani medium supplemented with glycerol (12 g l−1) and ampicillin (200 µg l−1), until the optical density at 600 nm (OD600) reached a value of 1.9. The culture was then cooled to 25°C and isopropyl-β-d-thiogalactopyranoside (IPTG) was added to a final concentration of 400 µM. The culture was further shaken at 25°C until the OD600 of the culture reached a value of 7. The cells were cooled to 5°C, harvested by centrifugation, resuspended in cold 30 mM Tris–HCl (pH 8.0) buffer (RB), and broken in a French pressure cell. The crude extract was centrifuged at 26 000×g for 15 min, and the supernatant was loaded onto a 3 ml Ni-NTA resin (Qiagen) equilibrated with RB. After washing the column with RB supplemented with 5, 10 and 20 mM imidazole, the protein was eluted with RB supplemented with 50 mM imidazole. The eluate was immediately dialyzed against RB and concentrated on an Amicon concentrator (Millipore, cut-off 10 kDa). SDS–PAGE analysis indicated that the apparent size of the purified recombinant protein was 18 000 Da, which is in good agreement with the theoretical value (17 153 Da). The protein concentration was estimated by measuring the absorbance at 280 nm in 6 M guanidium chloride using a molar extinction coefficient of 2560 M−1 cm−1.

2.5 Antibody preparation

Polyclonal antibodies against the cohesin 5 of the C. acetobutylicum scaffolding protein were raised in rabbits by subcutaneous injection of pure protein (0.4 mg) mixed with 1 ml of Freund's complete adjuvant. Booster injections were performed after 10 and 15 days, using the same amount of protein in incomplete adjuvant. The serum was collected, kept for 2 h at room temperature, and centrifuged at 4000×g. The supernatant was stored at 4°C with 0.3% NaN3.

2.6 Western blotting

Proteins separated by SDS–PAGE [17] were transferred onto a nitrocellulose membrane (Amersham Pharmacia Biotech) by standard procedures. The membrane was incubated overnight at 4°C with shaking in blocking buffer [10% milk powder, 50 mM Tris–HCl buffer (pH 7.5) containing 150 mM NaCl and 0.3% Tween 20]. The dilution used for the primary antibody was 1/1000. Anti-rabbit immunoglobulin G conjugated with alkaline phosphatase (Sigma) was used as the secondary antibody, according to the manufacturer's instructions. The membrane was then incubated with the primary antibody for 1 h at room temperature in blocking buffer and washed three times with Wash Buffer (50 mM Tris–HCl buffer, pH 7.5, containing 150 mM NaCl and 0.3% Tween 20). The membrane was incubated at least for 1 h at room temperature with alkaline phosphatase-conjugated antibody in blocking buffer and washed three times with Wash Buffer. The bands were visualized using 5-bromo-4-chloro-3-indolylphosphate/nitro blue tetrazolium (BCPIP/NBT) liquid substrate system (Sigma).

2.7 Enzyme assays

Cellulolytic activity was assayed by mixing the appropriate concentration of cellulosomal fraction with carboxymethyl cellulose (CMC), Avicel, bacterial cellulose (BC) and phosphoric acid swollen cellulose (PASC) at a final concentration of 0.8%, in potassium phosphate buffer (PPB, 25 mM, pH 7.0) at 37°C. BC was a generous gift from H.P. Fierobe (BIP, Marseille, France) and PASC was prepared from Avicel as described by Walseth [18]. Aliquots were collected at specific intervals and centrifuged at 5000×g for 15 min at 4°C to remove the insoluble substrate and the reducing sugar content was determined by the 3,5-dinitrosalicylic acid method [19] on supernatant. Specific activities were determined in the linear range of the reaction and expressed in U mg−1 protein (1 U of enzyme liberates 1 µmol glucose equivalent min−1). The protein concentration was determined as previously described by Lowry et al. [20].

3 Results and discussion

3.1 Amino acid sequence analysis of the proteins encoded by the gene cluster

The potential 11 cellulosomal subunits, deduced from the analysis of C. acetobutylicum genome sequence, are summarized in Table 1. Nine of the encoding genes form a large cluster and the last two are scattered across the chromosome. Three families of cellulases are present in the gene cluster, families 5, 9, and 48. Cel48A is the only member of family 48, while five cellulases from family 9 are found, Cel9C, Cel9F, Cel9E, Cel9H and Cel9X. Except for Cel9F, all the ‘family 9’ cellulases contain a CBD. Cel9C, Cel9E and Cel9H contain a subfamily 3c CBD, while Cel9X, the cellulase that is not part of the gene cluster, has a family 4 CBD.

View this table:

Cellulosomal components of C. acetobutylicum

Cellulosomal componentsFunctionMolecular mass (kDa)Modular structureHomolog
in Ccelin Ctmin Ccl
OrfXpAnchoring protein18PTS-CohIOrfXpOlpAHbpA
  • The function, molecular mass and structure of the cellulosomal proteins of C. acetobutylicum are presented. Homologous components of C. cellulolyticum (Ccel), C. thermocellum (Ctm) and C. cellulovorans (Ccl) are indicated. CBD3a, cellulose binding domain family 3a; CohI, cohesin type I; DDI, dockerin domain type I; GH9, glycosyl hydrolase family 9; Ig, immunoglobulin-like module; X, hydrophilic module.

As shown in Fig. 1, there is a high level of similarity between the cellulosomal gene clusters of C. acetobutylicum and C. cellulolyticum, suggesting a close taxonomic relatedness. It is interesting to note that the highest amino acid sequence homologies between the two species are obtained with Cel48F and Cel9E, the two major cellulases of the C. cellulolyticum cellulosome [21,22], and respectively Cel48A and Cel9X from C. acetobutylicum. Cel48A shows 52% sequence identity with Cel48F, a processive endocellulase with hydrolytic activity toward Avicel [21]. Generally, family 48 enzymes are known to play an important role in clostridial cellulolytic systems. Cel9X also shows 52% similarity with Cel9E from C. cellulolyticum. Like Cel9X, Cel9E is a multidomain protein containing a CBD4 domain associated with an Ig domain, a GH9 catalytic core domain, and a dockerin domain. A similar organization is found in Cel9K from C. thermocellum [23]. Cel9E is known to be a key enzyme in the cellulosome of C. cellulolyticum since it has a synergistic effect on the activities of other cellulosomal enzymes [22]. In addition to these cellulases, Cel9C from C. acetobutylicum possesses 60% identity with Cel9G from C. cellulolyticum [24]. Both proteins contain a subfamily 3c CBD domain that, in Cel9G, is known to act as a helper CBD.


Cellulosomal gene clusters from C. cellulolyticum and C. acetobutylicum, and the corresponding encoded cellulase families.

The presence of dockerins at the C-terminus of each cellulase is one of the characteristics of cellulases belonging to a cellulosome. Alignment of the dockerins of the cellulosomal subunits of C. acetobutylicum is shown in Fig. 2. Like all clostridial dockerins described so far, the C. acetobutylicum dockerins contain a duplicated sequence of about 22 amino acid residues, the first 12 of which are homologous to a known structure, the calcium-binding loop in the EF-hand motif [25]. Thus, it is highly probable that the cohesin–dockerin interaction in the C. acetobutylicum cellulosome is calcium-dependent.


Alignment of dockerin domains of cellulosomal subunits of C. acetobutylicum, Cel5A of C. cellulolyticum [31], Cel48S of C. thermocellum [32] and Exg48S of C. cellulovorans [33]. Asterisks indicate amino acid residues involved in calcium binding. Amino acids that are conserved in at least seven of 13 sequences are shaded. Conserved amino acids in the whole sequence are highlighted. Dashes indicate gaps left to improve alignment.

Recent studies have shown that the cohesin–dockerin interaction in Clostridium is species-specific. Moreover, positions 10 and 11 of the conserved motifs appear to exhibit species specificity [26]. Thus, the conserved AL or AI motifs of the dockerins from C. cellulolyticum are replaced by the GR motif in the C. acetobutylicum dockerins.

OrfXp from C. acetobutylicum, is similar (34.4% identity) to OrfXp from C. cellulolyticum, for which a role in the extracellular assembly of the cellulosome has been proposed [9].

The molecular architecture of the C. acetobutylicum CipA is similar to that of the scaffolding proteins reported so far, such as C. thermocellum CipA [11], C. cellulovorans CbpA [10], C. cellulolyticum CipC [9] and C. josui CipA [7]. These scaffolding proteins are composed of a CBD, multiple repeats of cohesin domains and several hydrophilic domains (Fig. 3A). The cohesin domains are known to bind strongly to the complementary dockerin domain of the cellulosomal enzymes, allowing the formation of a cellulosome complex around the scaffolding protein [27].


A: Schematic representation of CipC from C. cellulolyticum and CipA from C. acetobutylicum. B: Phylogenetic relationship of C. acetobutylicum CipA cohesin domains with other type I cohesins. The sequences of the scaffolding cohesins were obtained from GenBank with the following accession numbers: CipA from C. thermocellum, L08665; CbpA from C. cellulovorans, M73817 and CipC from C. cellulolyticum, U40345. The triangle represents the weighted centroid of the tree. Arrows indicate the more divergent cohesin of each strain.

The CipA scaffolding protein of C. acetobutylicum is organized into five type I cohesin domains and a NH2-terminal family 3 CBD. The internal family 3 CBD exhibits several features consistent with the subfamily 3a CBD, like the scaffolding protein CBDs from C. cellulolyticum, C. thermocellum and C. cellulovorans [1]. The sequence identities between cohesin domain 4 and the others were determined. This domain is 90–77% identical with cohesins 2–5. Cohesin 1 possesses only 57% identity with cohesin 4. A similar observation was made for CipC from C. cellulolyticum [9]. As illustrated in the phylogenetic tree (Fig. 3B), we identified a large group of highly homologous cohesins, and at least one cohesin domain, always located at one end of the protein, that shows significant sequence differences. Although the C. acetobutylicum cohesins are highly similar, it appears that they are relatively divergent from the other clostridial cohesins, as they show little overall similarity (less than 25%). Moreover, a phylogenetic comparison of the known cohesins reveals that the C. acetobutylicum cohesins represent a distinctive branch from the ones previously identified in the other clostridia, especially the C. thermocellum, C. cellulolyticum and C. cellulovorans cohesins which emanate from the same branch (Fig. 3B).

The cohesin domains are separated by distinctly hydrophilic domains of approximately 90 aa. As shown in Fig. 4, these six domains are highly homologous, especially the last three domains, which are 84% homologous. These domains share 40% identity with the hydrophilic domain of CbpA from C. cellulovorans. A lower degree of identity, only 33%, was found with HD1 and HD2 of CipC from C. cellulolyticum. The role of these domains remains unknown.


Alignment of the hydrophilic modules of the CipA scaffolding protein from C. acetobutylicum. Amino acids that are conserved in at least four of six sequences are shaded. Conserved amino acids in the whole sequence are highlighted. Dashes indicate gaps left to improve alignment.

CipA from C. acetobutylicum, with an estimated molecular mass of 154 kDa, is one of the smallest scaffolding proteins characterized to date. Moreover, the CipA scaffolding protein is unique for two reasons. First, it contains only five cohesin domains, which is the smallest number of cohesin domains reported for any clostridial scaffolding protein to date [28]. Interestingly, the newly described scaffolding protein ScaA, from the rumen bacterium Ruminococcus flavefaciens contains only three cohesin domains [29]. Secondly, CipA is the only scaffolding protein where each cohesin domain is systematically separated by a long hydrophilic domain (Fig. 3A). For all the other scaffolding proteins each cohesin domain is separated by short, proline–threonine-rich linkers. As an example, linkers of CipC from C. cellulolyticum or CbpA from C. cellulovorans contain nine and seven amino acids, respectively, while linkers of the C. thermocellum CipA contain 25 amino acids [30]. These features clearly distinguish the C. acetobutylicum scaffolding protein from those of other Clostridia.

3.2 Protein composition and detection of cellulosomal components

As C. acetobutylicum was unable to grow on microcrystalline cellulose, cultures were done on synthetic medium with cellobiose (20 g l−1) as the sole carbon source, and microcrystalline cellulose (10 g l−1) for adsorption of the putative cellulosomal components. The purification procedure of the adsorbed protein is described in Section 2. When subjected to SDS–PAGE, the fraction eluted from cellulose was found to consist of at least six proteins with molecular masses ranging from 160 to 50 kDa (Fig. 5A, lane 1). Other minor proteins seem to be present in the low molecular mass range.


A: Electrophoretic characterization of components of the cellulosome from C. acetobutylicum. The purified cellulosome was analyzed by SDS–PAGE under the conditions described in Section 2. Lanes: 1, molecular mass marker; 2, Coomassie blue-stained subunits of the cellulosome (seven bands were clearly visible and named S1–S7); 3, Western blots showing patterns obtained with specific Coh5 antibody; 4–6, Western blots obtained with C. cellulolyticum Cel9E, Cel48F and CBD–Cel9G specific antibodies. B: Immunological detection of the cellulolytic complex purified from C. acetobutylicum. Detection was done using the anti-Coh5 antiserum.

3.3 Western blotting experiments

The purified Coh5 cohesin domain from C. acetobutylicum was used to elicit antibodies in rabbits. Anti-Coh5, in combination with C. cellulolyticum antibodies obtained by using purified recombinant protein [21,22,24], were used for a Western blot analysis after the fractions eluted from cellulose were separated on SDS–PAGE and transferred onto a nitrocellulose membrane (Fig. 5A, lanes 2–6).

The 180 kDa protein (S1) was recognized specifically by Coh5 antibodies, suggesting that the S1 fraction corresponds to the 154 kDa scaffolding protein CipA. Differences in size might be due to glycosylation of CipA as previously shown for other scaffolding protein [5]. This is in agreement with the expected molecular mass of 154 kDa. The S2, S3 and S4 fractions respectively cross-react with Cel9E, Cel48F and CBD-Cel9G antibodies from C. cellulolyticum. Only one band was revealed on the film for each cellulase, indicating the specificity of the cross-reaction obtained. Based on the percentage of homology previously discussed, the 90 kDa (S2) protein specifically recognized by Cel9E antibodies should be Cel9X from C. acetobutylicum that has a theoretical molecular mass of 96 kDa. Similarly, the 75 kDa (S3) protein specifically recognized by Cel48F antibodies should be Cel48A from C. acetobutylicum that has a theoretical molecular mass of 80.9 kDa. Concerning the CBD–Cel9G experiment, a signal was obtained with the 74 kDa (S4) protein that could correspond to the Cel9C or Cel9E cellulases from C. acetobutylicum with theoretical molecular masses of 79.7 kDa and 74 kDa respectively. These two cellulases possess the same modular structure, which is composed of a family 9 catalytic domain associated with a subfamily 3c CBD, and share 46% identity. As is shown in Fig. 5, bands corresponding to Cel48A and Cel9C or Cel9E are very close on the gel. By increasing the acrylamide concentration from 8% to 12% in the SDS–PAGE gel, the two proteins were clearly separated. Western results (data not shown) revealed that the larger and more intense band corresponded to Cel48A.

3.4 N-terminal sequencing

N-terminal sequencing was carried out on fractions S1, S3, S4, S6 and S7. N-terminal sequencing on S1 and S4 gave no results due to the low amount of these components. S3, the major component of the protein mixture, gave the N-terminal sequence (ATTTDSSLK) of Cel48A, which confirmed the results obtained by Western blotting. The N-terminal sequence (TTTGEP) of the S6 fraction corresponds to Cac2988, a non-cellulosomal cellulase of C. acetobutylicum. Cac2988, with a molecular mass of 55.6 kDa, is the endoglucanase homologous to Cel26H of C. thermocellum. Cac2988 contains a GHF 26 catalytic module. No dockerin domain is present but this protein possesses a SLH domain. Because of the presence of at least two proteins in fraction S7 it was not possible to obtain an N-terminal sequence.

3.5 Can C. acetobutylicum produce a cellulosome?

The fraction previously eluted from microcrystalline cellulose (Fig. 5A, lane 2) was analyzed on a native polyacrylamide gel to reveal the presence of a cellulolytic complex. Coomassie blue staining showed a low intensity band in the high molecular mass region (higher than 665 kDa). To confirm that this complex was a cellulosome, an immunological analysis was done using the Coh5 antibody. After optimizing the transfer conditions of the putative cellulolytic complex onto nitrocellulose, Western blotting (Fig. 5B) revealed the presence of a positive signal in the high molecular mass region. These results indicated that the CipA scaffolding protein resided in a large protein complex. Presence of the cel48A in the cellulosome complex was demonstrated using antibodies against cel48F from C. cellulolyticum (data not shown).

3.6 Cellulolytic complex activities

Enzyme activities of the purified cellulolytic complex were measured on CMC, PASC, Avicel and BC. Activities were also done on a concentrated culture supernatant from cells grown only on cellobiose as a carbon source. Results indicated that no activity was detected on Avicel and BC, in agreement with the fact that C. acetobutylicum was unable to grow on crystalline cellulose. Low levels of activity were detected on CMC and PASC, both for the fraction eluted from crystalline cellulose (0.011 and 0.002 UI mg−1, respectively) and the concentrated culture supernatant (0.115 and 0.037 UI mg−1, respectively). As demonstrated here, CMCase and PASCase activities were at least 10 times higher from culture supernatants than from the purified cellulolytic complex, suggesting that not all of the cellulases present in the culture supernatant were able to form a complex. Compared to CMCase or PASCase specific activity (respectively 2.6 and 0.5 UI mg−1) reported for the purified cellulosome of C. cellulolyticum [4], the activities of the C. acetobutylicum cellulosome were very low.

3.7 Conclusions

Experiments done with antibodies against the main components of the C. cellulolyticum cellulosome showed that the respective homologous cellulases are expressed in C. acetobutylicum. The fact that Cel48A is much more abundant than the other catalytic subunits could indicate that this protein plays a peculiar role in the degradation of cellulose. Although C. acetobutylicum can produce a cellulosome, this bacterium is unable to hydrolyze cellulose. It will now be interesting to analyze the activity of the purified fractions of each of the main cellulosome cellulases on various cellulosic substrates to identify those that are defective. In this regard, work is currently underway in our laboratory to clone, over-express, purify and characterize from a biochemical point of view the Cel48A, Cel9X and Cel9C cellulases. Maintenance of the ATCC824 strain in laboratory conditions for many years, without a selective advantage for cellulose hydrolysis, might be responsible for the cellulosome defect. For the development of an economical process for the direct conversion of biomass to solvent, it will be very important to understand what component(s) is affected and repair it.


The authors would like to thank Maggie Cervin for the critical reading of the manuscript. This work was financially supported by a grant from the AGRICE Program (CNRS-ADEME) no. 94N80/0168. F.S. was the recipient of a predoctoral fellowship from ADEME.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
  24. [24].
  25. [25].
  26. [26].
  27. [27].
  28. [28].
  29. [29].
  30. [30].
  31. [31].
  32. [32].
  33. [33].
View Abstract