The Acetivibrio cellulolyticus CD2 genome 22 is the largest among the known cellulolytic bacteria (6.1 Mb). Analysis of its recent genome sequence revealed 41 putative cohesin modules, distributed in 16 scaffoldins, some of which have both cohesins and dockerins in the same polypeptide chain (Figure 1 and Additional file 1: Table S1). These include the four genes of the scaffoldin cluster (scaA, scaB, scaC and scaD), which were originally identified, sequenced and characterized in A. cellulolyticus ATCC 33288 171819.

The previous publications have indicated that this mesophilic bacterium harbors an intricate cellulosome system, which is characterized by several unique properties that distinguish A. cellulolyticus from the archetypical C. thermocellum cellulosome: The progression of the ScaA primary scaffoldin, the ScaB adaptor scaffoldin and the ScaC anchoring scaffoldin, with their resident cohesins (7, 4 and 3, respectively), suggests that the resultant fully occupied cellulosome complex would include up to 84 dockerin- containing proteins (enzymes) in addition to the intrinsic ScaA cellulase. The second type of cellulosome complex comprises a divergent anchoring scaffoldin, ScaD, which contains different cohesin specificities: two type-II cohesins that incorporate two ScaA subunits with their complement of dockerin-containing enzymes and a single type-I cohesin that binds a lone dockerin-containing protein.

Comparison of the original A. cellulolyticus sca genes which were individually sequenced by conventional methodology 171819 to those of the newly sequenced genome shows only a few differences (two nucleotide substitutions out of 2601 in the ScaB gene [GenBank: ZP_09464032]).

In the present work, the sequenced A. cellulolyticus genome revealed 12 cohesin- containing proteins in addition to the previously known four major scaffoldins encoded by the sca gene cluster. Figure 1 presents their modular architecture. All of the proteins listed in the figure, except for ScaI, contain a credible signal peptide, suggesting that these proteins would be secreted.

The cohesin modules exhibit a variety of intriguing sequence features. Like C. thermocellum, the 41 cohesins of A. cellulolyticus are classified into two types: type I (26 modules) and type II (15 modules). We examined the conservation of the cohesin sequences and compared copies of the various cohesin modules within a given scaffoldin protein, and among the different scaffoldins. The overall sequence similarity among the A. cellulolyticus cohesin modules ranges from 41 to 97%. Some scaffoldins contain similar repeats of the same type of cohesin module, whereas others bear a single cohesin. ScaD alone contains a combination of two heterogeneous cohesin types on the same polypeptide chain. As has been experimentally documented 171819, the cohesin type (i.e., type I or type II cohesin) does not necessarily indicate its binding specificity to a given dockerin. For example, the cohesins from ScaA and ScaC (Figure 1) are all type I according to their sequences, but they bind to different dockerins – the ScaA cohesins bind to the dockerin-bearing enzymes, and the ScaC cohesins bind to the ScaB dockerin.

The combination of S-layer homology (SLH) modules with cohesin modules on the same polypeptide suggests a role for such proteins in anchoring the cellulosome assemblies or specific enzymes to the cell wall of the gram-positive bacterium 2728.

In addition to the previously described anchoring scaffoldins, ScaC and ScaD, three more proteins which contain SLH modules are now revealed, i.e. ScaF [GenBank: ZP_09464236], ScaJ [ZP_09462222] and ScaK [ZP_09464725]. Of the 37 SLH- containing proteins encoded in the A. cellulolyticus genome, ScaK was identified with an SLH module, two dockerins and two cohesin modules. This is the first example of such an architectural arrangement of a cell-surface anchoring scaffoldin that contains both types of cellulosome-related modules.

Uniquely, one cohesin-containing protein also contains two family 2 CBMs, interspacing its type-I cohesins (ScaM, [ZP_09463433]). To our knowledge, this is the first description of a scaffoldin-borne CBM2; all previous CBMs located on scaffoldins have been from family 3. CBM2s have been described as ancillary modules of enzymes and were shown to bind efficiently to cellulose and/or xylan. Thus, their appearance on a scaffoldin may serve to enhance the substrate-binding function of the dockerin- containing enzymes, which bind to this scaffoldin protein via its type-I cohesins. Other cohesins were identified in novel types of scaffoldins which bear FN3 (Fibronectin type III) repeats, PA14 (protective antigen) domain, peptidase or other extracellular modules.

The A. cellulolyticus genome is particularly enriched with dockerin-containing genes, and 143 genes that contain putative dockerin modules were identified. Therefore, A. cellulolyticus contains almost twice the number of dockerins as other Clostridial bacteria, such as Clostridium cellulolyticum (>60 dockerins) or Clostridium thermocellum (>70 dockerins) 233132. Only the genome of Ruminococcus flavefaciens FD-1 is known currently to contain more dockerin-containing genes (>220) 2633. Unlike the R. flavefaciens dockerins, which are classified into 6 major groups and 11 subgroups 33, the A. cellulolyticus dockerins are highly similar, with the exception of six dockerins located downstream of an X module. These latter dockerins have distinctive sequence features compared to the rest of the A. cellulolyticus dockerins. Their X-modules are of family X60 34, which display significant sequence similarity with the X-module at the C-terminus of the C. thermocellum CipA scaffoldin. Indeed, several of these X-dockerin pairs are found at the C-terminus of A. cellulolyticus scaffoldins (ScaA, ScaP and ScaI). Interestingly, ScaI protein contains an X-dockerin modular dyad with a truncated type-II dockerin at its C-terminus.

The characteristic sequence conservation profile 353637 of the A. cellulolyticus dockerin module is shown in Figure 3. The sequence similarity among A. cellulolyticus dockerin modules is 53% on average (73% for the most similar dockerins pairs, with no two identical dockerins). Like the dockerins in C. thermocellum and unlike those of R. flavefaciens, each A. cellulolyticus dockerin module contains two canonical Ca + 2 binding repeats, followed by putative helices and linkers. Examination of the putative “recognition” residues of the dockerins, which may participate in their tight binding interface with cohesins, shows a conserved pattern of the two repeated segments wherein S(I/L) residues occupy positions 10 and 11, R(X) positions 17 and 18, and a highly conserved G in position 22 (Figure 3, in yellow). The corresponding positions in the C.thermocellum dockerins are S(T/S), K(R/K) and K/R/G, respectively. Position 18 is much less conserved in the A. cellulolyticus dockerins than those of C. thermocellum, whereas the reverse is true for position 22. Some modifications are evident in position 11 of the A. cellulolyticus dockerin sequences. For example, the ScaK scaffoldin contains an N- terminal dockerin with an Asn residue in position 11 of its first dockerin repeat. ScaB dockerin contains Asn residues in both repeats, and instead of the conserved Asn in position 9 it contains a positively charged Lys or Arg residue. In the case of ScaB, these modifications lead to different specificity characteristics, as the dockerin binds selectively to the cohesins of ScaC 18.

A. cellulolyticus grows on amorphous and crystalline forms of cellulose, xylans, and cellobiose 3839; the bacterium can also be adapted to grow on glucose and xylose 1340. Consequently, it was presumed in these early works that the bacterium produces endoglucanases, exoglucanases, β-glucosidases and xylanase activities. Indeed, the present study reveals an intricate array of cellulolytic and hemicellulolytic enzymes in the A. cellulolyticus genome, capable of hydrolyzing diverse cellulosic substrates to reducing sugars.

The sequence features of the dockerin-containing enzymes of A. cellulolyticus were assessed using the following approach: (i) Like the cohesin-bearing proteins, the dockerin-containing proteins are multimodular in nature, composed of more than one type of module (catalytic, structural, etc.) and sometimes more than one repeat of the same module. The different modular types were therefore enumerated, in order to determine their general distribution among the A. cellulolyticus proteins. (ii) Where appropriate, we distinguished between cellulosomal (i.e., those that harbor a dockerin) and non-cellulosomal (without a dockerin) proteins. (iii) We compared the A. cellulolyticus proteins with those of C. thermocellum.

Among the 143 dockerin-containing proteins, about half (63 proteins) contain one or more known carbohydrate-active CAZyme module(s) 41, and their composition is presented in Table 1 and Additional file 1: Table S1. Because of the multimodular nature of the proteins, some of them contain more than one type of catalytic module, therefore the total sum of catalytic modules in the 63 enzymes is 80 in Table 1 (62 GH-, 13 CE- and/or 5 PL- containing enzymes). Of the 92 GHs, about two-thirds are equipped with dockerins, suggesting that they are recruited to the cellulosome and may thus play a critical role in biomass degradation. Interestingly, the percentage of dockerin-containing GHs in the A. cellulolyticus genome is almost identical to that of C. thermocellum. The 62 dockerin-containing GHs belong to 19 different families according to the CAZy database (Table 1). As in all known cellulosomes produced by other species, the A. cellulolyticus cellulosome contains a single distinctive GH48 enzyme. As in C. thermocellum, the A. cellulolyticus genome also codes for a second, non-cellulosomal GH48-containing cellulase, as opposed to other characterized cellulosome-producing species that possess only one cellulosomal enzyme. The most abundant GH family is represented by the GH9 enzymes, again like in the C. thermocellum cellulosome. This is followed by the GH5 enzymes which are also numerous in both cellulosome-producing species. Of the 21 GH9 enzymes, 10 exhibit a GH9-CBM3 motif that would potentially modulate the activity as in C. thermocellum and other cellulolytic bacteria 42434445. In addition, there are three enzymes that show an extended GH9-CBM3-CBM3 motif, compared to two such enzymes in C. thermocellum46.

Numbers represent proteins which contain one or more modules of the different protein families (glycoside hydrolases, polysaccharide lyases and carbohydrate esterases) as were identified by CAZy. The number of proteins are compared between cellulosomal and non-cellulosomal (genome-wide) proteins. Data are provided for both species. Numbers of modules which appear more than once in the same protein are shown parenthetically.

In one third of the dockerin-containing proteins (46 proteins) we identified modules which are predicted to be associated with extra-cellular proteins (i.e., FN3 modules, Leu-rich repeats, RhsA and PKD domains, see Table 2). Some of these modules are conserved in sequence, but their function is still unknown; some may represent a yet undiscovered enzyme. In this regard, a C. thermocellum dockerin-containing protein of previously unknown function was recently demonstrated to be a cellulase 47. The dockerin-containing proteins of A. cellulolyticus are more enriched with such structural and unknown modules than those of C. thermocellum (Table 2).

^a Numbers in parenthesis indicate the number of X60-dockerin modular pairs in the given species.

^b Catalytic modules, such as: GH, PL, CE, and CBMs according to CAZy (http://www.cazy.org/).

^c Structural domains are defined in Pfam (http://pfam.sanger.ac.uk/), such as: FN3, GDSL, SNGH, CotH, TRX-like, Kelch-like, SLH, RshA, LRR, PKD.

^d Others include Pfam domains such as peptidases, serpins, DUF303, DUF1565, DUF3237.

Many of the GH or CE catalytic modules in the multi-modular proteins are associated with CBMs. In the case of a non-cellulosomal protein, a CBM may serve to deliver the parent catalytic module to a preferred site on the polysaccharide substrate.

Otherwise, an appended CBM may serve to modulate directly the hydrolytic properties of the catalytic module. Table 3 shows the number and distribution of such proteins in the genomes of both bacteria, A. cellulolyticus and C. thermocellum. Interestingly, 38 of the dockerin-containing enzymes in A. cellulolyticus consist of both a catalytic module and a CBM, most of the latter mostly families 3 and 6 (Table 3). In addition, another 12 non- cellulosomal enzymes contain an appended CBM. Although A. cellulolyticus contains approximately double the number of dockerin-containing proteins as C. thermocellum, the two species have the same number of CBM-appended enzymes (Table 2), and their distribution into different CAZy families largely overlaps.

The number of proteins with the combination of the specified modules is noted in the genomes of A. cellulolyticus (left) vs. C. thermocellum (right).

Even more intriguing are the 10 multi-functional enzymes of A. cellulolyticus, which harbor a combination of at least two catalytic modules, including one or two GHs, CEs, PLs and/or glycosyl transferases (GTs), on the same polypeptide (Table 4). In A. cellulolyticus, some of these enzymes do not contain a dockerin module. In contrast, C. thermocellum codes for 8 multi-functional dockerin-containing enzymes, and Ruminococcus flavefaciens FD-1 codes for 18 dockerin-containing multi-functional enzymes. As stated in an earlier section, both genomes encode for two GH48 enzymes – one cellulosomal and another non-cellulosomal. In C. thermocellum, there are two separate non-cellulosomal enzymes – Cel48Y (GH48-CBM3b) and the other Cel9I (GH9-CBM3c-CBM3b), whereas in A. cellulolyticus the two catalytic modules are fused together into a single polypeptide chain that share a single cellulose-binding CBM3b, thus forming a multi-functional non-cellulosomal enzyme (GH48-GH9-CBM3c-CBM3b, [GenBank:ZP_09464448]).

Domain architectures and their corresponding accession numbers or names are listed above. Catalytic modules are marked in bold. GH, glycoside hydrolase; PL, polysaccharide lyase; CE, carbohydrate esterase; CBM, carbohydrate-binding module; GT, glycosyl transferase; Doc, dockerin; numbers indicate family of the indicated module.

Draft genome sequences of Acetivibrio cellulolyticus CD2 (DSM 1870, ATCC 33288) (30 Dec. 2011), and Clostridium thermocellum ATCC 27405 (16 Feb. 2007) were obtained from GenBank (accession: AEDB00000000 and CP000568, respectively). Assembly of A. cellulolyticus genome was approached by a combination of sequencing methods, using Sanger, 454-Titanium, 454 Titanium Paired-end and Solexa Paired-end technologies, as detailed in Hemme et al. 22. The genome assembled into 112 contigs with an average coverage depth of x71.9 +/− 6.3 (interval of depths 9 – 111). Protocols of the A. cellulolyticus sequencing methods, assemblies and annotation are detailed in Land et al. 52.

BLAST 53 searches were applied on A. cellulolyticus DNA contigs and predicted proteins, using sequences of known cohesin and dockerin modules as queries. All hits above E-value of 10–4 were retrieved and inspected individually, by examining their characteristic sequence features. Obvious dockerin modules were expected to contain two Ca + 2-binding repeats, putative helices and linker regions. Low-scoring hits of dockerins and cohesins were examined by comparing them against known dockerin or cohesin sequences, respectively. Sequence logos of dockerins were created with Weblogo v.2.8.2 (http://weblogo.berkeley.edu/) 54. Multiple sequence alignment was obtained using CLASTALW 55, with manual corrections when needed.

The scaffoldin genes from A. cellulolyticus ATCC 33288 which were manually sequenced 171819 are ScaA, [GenBank: AF155197]; ScaB, [GenBank: AY221112]; ScaC, [GenBank: AY221113], ScaD, [GenBank: AY221114]). The cohesin dendrogram was generated using PhyML algorithms (with LG substitution model, and default parameters of the Approximate Likelihood-Ratio test) 56 and visualized using TreeView 57.

Dockerin-containing proteins of A. cellulolyticus CD2 and C. thermocellum ATCC 27405 were annotated by CAZy database (http://www.cazy.org) 41, in order to bioinformatically analyze their catalytic modules. This includes identification of the catalytic modules and their classification into family types, according to sequence conservation, for glycoside hydrolases, carbohydrate esterases, polysaccharide lyases, carbohydrate-binding modules and glycosyl transferases. Additional conserved domains of the proteins were analyzed using the CD-search website (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) and the Pfam database (http://pfam.sanger.ac.uk/). Putative cellulosome-related regulatory elements were identified by BLAST searches and sequence similarity using known elements from C. thermocellum as queries 484950.