Strains and plasmids used in this study are listed in Table 1. The ΔX2-N and ΔX2-C mutants with deletions of the conserved motif in the N-terminal and C-terminal X2 respectively, were constructed in our previous study (Xu et al., 2015). The dual X2 modules mutant (ΔX2-NC) was constructed using the pCas9n-X2-N-delete-donor to further delete the conserved motif of the X2-N module in ΔX2-C (Li et al., 2012). In brief, for each X2 module, its corresponding variant was created by deleting the 12-bp DNA sequence coding for the conserved Asn-Gly-Asn-Thr motif. To do so, the 23-bp target site for Cas9 nickase was partially overlapped with the deletion region such that the customized donor template could direct the nick repair in the genome to make intended changes (Figure 1A). Finally, the N-terminal and C-terminal X2 modules (X2-N and X2-C) were mutated precisely in the chromosome, generating ΔX2-N and ΔX2-C variants respectively (Xu et al., 2015). Sequential mutation of the N-terminal X2 module in ΔX2-C variant created a dual mutated X2 variant, named ΔX2-NC (Figure 1A). All mutations were verified by amplicon sequencing (Figure 1B).

E. coli DH5α strain (Invitrogen, Carlsbad, CA, United States) was used for cloning. Rosetta™(DE3) (Invitrogen, Carlsbad, CA, United States) was used for protein expression. Sequences of X2-C module and CBM3a, amplified from the genome of WT strain, were cloned into pET-28a vectors with a C-terminal 6 × His tag, yielding pET-X2-C and pET- CBM3a, respectively. The pET-ΔX2-C vector was generated from the pET-X2-C by Q5 Site-Directed Mutagenesis Kit (New England Biolabs, Ipswich, MA, United States).

LB medium with 35 μg/ml chloramphenicol or 50 μg/ml kanamycin was used for pCas9n-X2-N-delete-donor cloning or pET-X2C/pET-ΔX2C/pET-CBM3a cloning. Complex modified VM medium supplemented with 2.0 g/L yeast extract was used for reviving and transformation of ΔX2-C (Li et al., 2012). Defined modified VM medium containing necessary vitamin solution and mineral solution was used for growth determination and omics experiments (Higashide et al., 2011). All strains (ΔX2-N, ΔX2-C, ΔX2-NC, and WT) were cultured with 5 g/L cellobiose or 10 g/L Avicel PH101 crystalline cellulose (Sigma-Aldrich, St. Louis, MO, United States) at 34°C anaerobically. Solid VM medium with 1.0% (w/v) of Bacto agar (VWR) and 15 μg/ml thiamphenicol was used for selecting the ΔX2-NC mutant.

The pCas9n-X2N-donor was firstly methylated by MspI Methyltransferase (New England Biolabs, Ipswich, MA, United States) and then transformed into ΔX2-C by electroporation as described previously (Li et al., 2012). The pET-X2N, pET-ΔX2N, and pET-CBM3a were transformed into Rosetta™(DE3) according to the manufacturer’s instructions. For each strain, the cellulosome fraction was isolated from 200 ml cellulose-grown culture at mid-exponential growth phase as previously reported (Tao et al., 2020).

Microbial growth (ΔX2-N, ΔX2-C, ΔX2-NC, and WT) on cellobiose was determined by optical density (OD) at 600 nm; on cellulose, cell growth was estimated by total protein measurement as previously described (Li et al., 2012). Each strain had three biological replicates. The remaining cellulose and released soluble sugar in the medium were measured by the phenol-sulfuric acid method (Hemme et al., 2011).

All strains (ΔX2-N, ΔX2-C, ΔX2-NC, and WT) were grown on 5 g/L cellobiose to the same OD₆₀₀. Cells were incubated with or without Whatman filter paper (cellulose) on a tube rotator for 1 h. Then, the OD₆₀₀ of the planktonic phase was determined to reversely infer the binding affinity of cells to cellulose. Each group contained three biological replicates. The relative cell adhesion capability for each strain was normalized to WT.

All strains (ΔX2-N, ΔX2-C, ΔX2-NC, and WT) were grown in the defined VM medium with 10 g/L cellulose. Each strain was collected at the mid-exponential growth phase. RNA extraction, microarray hybridization, and microarray data analysis were performed as previously described (Tao et al., 2020). For each strain, four biological replicates were performed.

For the expression of X2-N or ΔX2-N module, Rosetta™(DE3) containing the corresponding vector was grown to OD₆₀₀ 1.0–1.2 and then induced with 0.5 mM isopropyl-d-1-thiogalactopyranoside (IPTG) for 8 h at 37°C. Then, cells pellets were resuspended in the lysis buffer containing 20 mM Tris-HCl pH 8.0, 100 mM NaCl, 10 mM imidazole. After sonication and centrifugation at 12, 000 g for 30 min, the supernatant was loaded to Ni²⁺-nitrilotriacetate affinity resin (Qiagen, Hilden, Germany) equilibrated with 20 mM Tris−HCl pH 8.0, 150 mM NaCl. The X2-N or ΔX2-N protein was eluted with 20 mM Tris−HCl pH 8.0, 150 mM NaCl, 350 mM imidazole and further purified by buffer exchange. For CBM3a expression, Rosetta™(DE3) with pET-CBM3a was grown to OD₆₀₀ 0.6–0.8 and then induced with 0.2 mM IPTG for 20 h at 16°C. The remaining steps were the same as above. The structure of X2-N (PDB: 1EHX) was used as the template to construct the structure of X2-C by homology modeling in SWISS-MODEL¹.

A 20 ml of E. coli grown in LB medium, Clostridium thermocellum and C. cellulolyticum grown on 5 g/L cellobiose in VM medium were collected and centrifuged respectively. For each strain, the cell pellets were treated with NaN₃/Ca²⁺ and washed three times with 50 mM Phosphate-buffered saline (PBS) buffer (Sigma-Aldrich, St. Louis, MO, United States). Then, 50 μg purified X2-N, ΔX2-N, and CBM3a proteins were incubated with collected cell pellets respectively in the PBS buffer at 4°C for 12 h. After centrifugation, the pellet from each incubation was resuspended in the SDS-PAGE Protein Loading Buffer (Thermo Fisher, Waltham, MA, United States) and boiled for 10 min. Finally, the 6x-His Tag antibody (R930-25, Thermo Fisher, Waltham, MA, United States) was used to detect the binding between proteins and cell wall by Western blotting.

A 50 μg of X2-N, CBM3a, and BSA (Thermo Fisher, Waltham, MA, United States) were separately incubated with 20 mg Avicel PH101 crystalline cellulose in the reaction buffer (20 mM Tris-HCl pH 8.0) at 4°C for 12 h. After centrifugation, the pellet from each incubation was washed with the reaction buffer three times. Then, the supernatant and pellet were analyzed by SDS-PAGE to detect binding between proteins and cellulose.

Titration calorimetry measurement was performed with a VP-Isothermal Titration Calorimetry (ITC) calorimeter (MicroCal, Northampton, MA, United States) as previously described (Duff et al., 2011; Brown et al., 2013). The buffer containing 20 mM Tris-HCl pH 8.0 and 100 mM NaCl was used for the assay. In brief, titration with buffer alone (background) and titration from X2-N protein into the buffer (control), were firstly performed as background and control groups. Then, 70 μM X2-N protein was titrated into 10 μM CBM3a protein with an injection volume of 70 μL and constant stirring at 25°C. Finally, a one-site binding model from the Origin 8.0 software (MicroCal) was used for data analysis.

We first evaluated the prevalence of X2 modules (PF03442) in assembled bacterial genomes using HMMER tools² and found 1243 species that genetically encode proteins containing X2 modules including bacteria (1024), eukaryote (214), archaea (4), and unclassified (1). Among bacteria, more than 60% of identified X2 modules were found in the Firmicutes phylum including the classes Bacilli and Clostridia. Interestingly, at the protein level, more than 50% of X2 modules were found in multi-modular cellulases or cellulosomes, adjacent to the CBM module (X2-CBM). These architectural features may indicate an important role of X2 in insoluble carbon-associated microbial metabolism.

To test this hypothesis, we selected X2 modules found in the cellulosomal scaffolding protein CipC of C. cellulolyticum to interrogate further. These two X2 modules were named as X2-N (near the N terminus of CipC) and X2-C (near the C terminus of CipC) based on their loci in the CipC protein (Supplementary Figure 1). Since a truncated CipC led to the failure of cellulosome assembly (Maamar et al., 2004), a precise genome editing tool is required to mutate the X2 modules without sacrificing the overall architecture of CipC. It is interesting that an alignment of X2-like modules from different bacteria revealed a highly conserved short motif (Asn-Gly-Asn-Thr) (Supplementary Figure 2). This motif is structurally located in an exposed loop on X2-N (PDB: 1EHX) and X2-C (Figure 2). Deletion of this short motif would shorten the loop region and form a wider groove as shown by structure modeling (Figure 2), which would allow for investigation of X2 function in CipC with a minimal impact on CipC structure.

We precisely deleted the conserved motif in the X2 modules of CipC by Cas9 nickase-based genome editing, yielding single (ΔX2-N and ΔX2-C) and double X2 mutants (ΔX2-NC) (Figures 1A,B). SDS-PAGE analysis of cellulosome fractions from cellulose-grown WT and mutant strains demonstrated no significant changes in gel patterns and band intensities (Figure 1C), which suggested the cellulosome architecture was not impeded by the X2 mutations.

To test if X2 was involved in insoluble carbon-dependent microbial metabolism, we monitored the growth of X2 mutants on different carbon sources. On cellobiose, all mutants (ΔX2-N, ΔX2-C, and ΔX2-NC) presented similar growth profiles as WT (Figure 3A). However, when grown on 10 g/L cellulose, ΔX2-N, ΔX2-C, and ΔX2-NC had a much longer adaptation stage than WT before achieving a similar growth rate and final biomass level as WT (Figure 3B), suggesting the importance of X2 modules in cell growth on insoluble cellulose. In addition, compared to WT, the release of soluble sugars in ΔX2-N, ΔX2-C, and ΔX2-NC was decreased by 28, 40, and 63%, respectively. The amount of residual cellulose in ΔX2-NC was 15% higher than WT and approximately 10% higher than ΔX2-N and ΔX2-C. These results concordantly demonstrate that genetic mutation of X2 in CipC can reduce cellulose hydrolysis of the cellulosome (Figures 3C,D and Supplementary Figure 3). This is consistent with previous in vitro studies on X2 in C. cellulovorans (Kosugi et al., 2004). These data also solidify the importance of the conserved motifs in the X2 module for the first time.

To understand if a polar effect was introduced to downstream genes or the expressions of other genes were influenced by the X2 mutations (Maamar et al., 2004), we performed microarray-based transcriptomic analysis for the whole transcriptome of C. cellulolyticum. We found there were no differentially expressed genes present between WT and X2 mutants grown on cellulose during the exponential phase. Therefore, the disruption of the X2 modules seemed to have influenced cellulose degradation not by regulating gene expressions.

We then speculated that X2 modules might be key factors responsible for (i) increasing the binding affinity of cellulosomes to cellulose, (ii) increasing the localization and adhesion of cellulosomes to the cell surfaces, or (iii) increasing both binding affinity to cellulose and adhesion to cell surfaces. To determine whether the X2 mutations influenced the binding affinity between cells and cellulose, a cell-cellulose adhesion assay was performed. With cells collected at the early exponential stage, we found ΔX2-NC mutant decreased cell adhesion to cellulose by 50% when compared to WT. The ΔX2-C presented a more profound effect than ΔX2-N (Figure 4A). The decrease in cell adhesion was even more obvious when cells were at the late exponential stage (Figure 4B). Therefore, X2 modules, especially the X2-C module in CipC, play an important role in cell adhesion to cellulose.

To understand how X2 modules influence cell adhesion to cellulose, we performed in vitro X2 module-cell wall binding assay to test if the X2 modules mediate cellulosome localization on the cell surface. As ΔX2-C generated more significant physiological changes than ΔX2-N on cellulose, we expressed and purified X2-C module, CBM3a and ΔX2-C module (deletion of the conserved short motif) for in vitro assays (If not specified, we used X2 to represent X2-C in the following in vitro assays). In this binding assay, we separately mixed above-purified module proteins with C. cellulolyticum cells or E. coli (gram-negative) and C. thermocellum (Gram-positive) as controls. Western blotting did not detect the presence of X2 on any of these three strains post incubation with purified X2, indicating that X2 cannot directly bind to cell surfaces (Figure 4C). This agrees well with a recent study (Tarraran et al., 2021).

However, purified CBM3a was found in all three strains, indicating it is able to bind bacterial cell walls (Figure 4C). More interesting, we found a very weak signal of ΔX2 protein in the group of C. thermocellum and C. cellulolyticum. It is possible that ΔX2 could enhance the localization of cellulosomes on the surface of these cellulose degrading strains.

Since the X2 module could not directly bind to cell wall, here we performed in vitro X2 module-cellulose binding assays to test if X2 modules mediate the binding affinity between cellulosomes and cellulose. Using BSA as the negative control and purified CBM3a as the positive control for cellulose binding assays, we found that purified X2 only present in the supernatant instead of cellulose pellets, indicating that the X2 protein cannot directly bind to the cellulose (Figure 4D). Meanwhile, the ITC assay indicated that there was a weak interaction between the X2 module and the CBM3a (Supplementary Figure 4).

In summary, the X2 module in CipC cannot directly bind to cellulose or the bacterial cell surface. However, removing the conserved motif in the X2 modules decreased cellulose utilization and severely reduced cell attachment on cellulose.