Role of the Sortase A in the Release of Cell-Wall Proteinase PrtS in the Growth Medium of Streptococcus thermophilus 4F44

Growth of the lactic acid bacterium Streptococcus thermophilus in milk depends on its capacity to hydrolyze proteins of this medium through its surface proteolytic activity. Thus, strains exhibiting the cell envelope proteinase (CEP) PrtS are able to grow in milk at high cellular density. Due to its LPNTG motif, which is possibly the substrate of the sortase A (SrtA), PrtS is anchored to the cell wall in most S. thermophilus strains. Conversely, a soluble extracellular PrtS activity has been reported in the strain 4F44. It corresponds, in fact, to a certain proportion of PrtS that is not anchored to the cell wall but rather is released in the growth medium. The main difference between PrtS of strain 4F44 (PrtS4F44) and other PrtS concerns the absence of a 32-residue imperfect duplication in the prodomain of the CEP, postulated as being required for the maturation and correct subsequent anchoring of PrtS. In fact, both mature (without the prodomain at the N-terminal extremity) and immature (with the prodomain) forms are found in the soluble PrtS4F44 form along with an intact LPNTG at their C-terminal extremity. Investigations we present in this work show that (i) the imperfect duplication is not implied in PrtS maturation; (ii) the maturase PrtM is irrelevant in PrtS maturation which is probably automaturated; and (iii) SrtA allows for the PrtS anchoring in S. thermophilus but the SrtA of strain 4F44 (SrtA4F44) displays an altered activity.


Introduction
The lactic acid bacterium S. thermophilus is widely used in the manufacturing of domestic and industrial fermented dairy products [1], and has obtained the "Qualified Presumption of Safety" and "Generally Recognized as Safe" (GRAS) designations. It belongs to the Streptococcus genus which is mainly pathogenic or commensal [2,3].
S. thermophilus is auxotrophic for certain amino acids and needs to get them from its environment [4,5], whereas milk, the only known habitat of this bacterium, mainly contains proteins and very few immediately assimilable peptides and amino acids. Hence, S. thermophilus needs them for its growth to reach a high cell density of a functional surface proteolytic system [6,7]. It consists of the CEP PrtS, able to break down caseins into peptides, which are then internalized through specific transporters and hydrolyzed by

Bacterial Strains and Culture Conditions
The strains and plasmids used in this study are presented in the Table 1. Strains were stored in reconstituted skim milk 10% (m/v) at −80 • C. They were precultured in reconstituted skim milk and then introduced at 1% into M17 medium with lactose 20 g L −1 (LM17) [25]. For the transformation experiments, the strains were first cultivated in LM17 before being inoculated in a chemically defined medium with 20 g L −1 of lactose [26]. The incubation temperature was 42 • C. For the LMD-9 ∆srtA , LMD-9 ∆prtS , and LMD-9 ∆prtM mutants, 5 µg mL −1 of erythromycin was added to the LM17 medium, while for the mutants LMD-9 srtALMD-9 , LMD-9 srtA4F44 , LMD-9 prtS4F44 , and LMD-9 srtA:Ile218→Val218 , the medium was supplemented with 20 µg mL −1 of streptomycin and 300 µg mL −1 of spectinomycin. The growth of the strains was assessed for measuring the pH of the medium and the optical density (OD) at 600 nm.

DNA Extraction, PCR Amplification, Electrophoresis, and Sequencing Conditions
Plasmid DNAs were isolated from Escherichia coli using the Miniprep Kit (Fermentas, Villebon-sur-Yvette, France) according to the manufacturer's instructions. Genomic DNAs were extracted as previously described [27]. Primer3plus software was used to design primers, which were then synthesized by Eurogentec (Seraing, Belgium). Sequences of primers and the sizes of amplicons are reported in Tables S1-S3 (Supplementary Data). Polymerase chain reactions (PCRs) were achieved according to the supplier's recommendations (Fermentas, Villebon-sur-Yvette, France) in a Mastercycler proS thermocycler (Eppendorf, Hambourg, Germany). Cycle conditions were: 95 • C for 5 min, 35 cycles of 3 steps (95 • C for 30 s; hybridization at appropriate temperatures (Tables S1-S3); 72 • C for 1 min kb −1 ), and 10 min at 72 • C. For mutant construction, high fidelity Phusion DNA polymerase (Fermentas, Saint Rémy-lès-Chevreuse, France) was used for the amplification of each fragment (final DNA concentration of 5 µg mL −1 ; extension time 30 s kb −1 ). For overlapping (OL) PCR, DNA fragments required for the constructions were pooled in equal amount (final concentration of 5 µg mL −1 ). The mixture also contained 0.5 µmol mL −1 of each primer (forward complementary with 5 end of the first fragment and reverse complementary with 3 end of the last fragment); high fidelity Phusion DNA polymerase 4U, each dNTP of 0.2 µmol mL −1 ; and 5X Phusion HF buffer of 4 µL. The program used was: 2 min at 95 • C, 35 cycles of 3 steps (95 • C 30 s, hybridization for 30 s at the annealing temperatures, and 72 • C 30 s kb −1 ), and finally 72 • C for 10 min.
The high pure PCR product purification kit (Roche Applied Science, Meylan, France) was used to purify PCR products, taking as an eluent either the elution buffer of the kit (for OL PCRs) or ultra-pure water (for sequencing reactions). PCR products were separated by electrophoresis on 1% (w/v) agarose gel in 0.5× TAE buffer [28] at 100 V. The molecular weight markers used were 1 kb and 100 bp DNA ladders (Fermentas). Sequencing was performed by Beckman Coulter Genomics (Essex, UK) with the Sanger method [29].
Molecular modeling simulations were run on a bi-processor AMD Dual Core 280 with 2.4 GHz. Docking and scoring simulations were performed using the LibDock algorithm [34] and the Consensus Score modules of the program-package Discovery Studio version 3.5 (Accelrys, Inc., San Diego, CA, USA), respectively. All molecular mechanics calculations were performed with the CHARMm force field [35]. The Protein Data Bank entry 3FN5 corresponding to the SrtA (Spy1154) of the S. pyogenes serotype M1 strain SF370 was used as the input structure [24]. This was made of two chains per unit cell (length: a-39.8 Å, b-59.46 Å, and c-65.11 Å; angles: α-90 • , β-101.96 • , and γ-90 • ). These two chains contained 187 residues corresponding to the catalytic domain. The first 18 residues corresponded to the His-tag region expediting the purification of the protein.
The chain A included a 4-(2-Hydroxyethyl)-1-Piperazine Ethanesulfonic Acid (HEPES) entity as an inhibitor. The chain B was picked as the input structure for all simulations.
In our study, the first 18 residues due to the crystallization step were withdrawn from chain B. The resulting structure Sp-SrtA SF370∆86 was used as a template to design homology models for St-SrtA 4F44∆90 and St-SrtA LMD-9∆90 . In order to investigate the proteinase PrtS "sorting pattern" binding modes within the SrtA active site, two ligands were designed, namely LPNTG and Ace-QLPNTGEND-NMe. Their possible binding modes within the active site of the SrtA systems (Sp-SrtA SF370∆86 , St-SrtA 4F44∆90 , and St-SrtA LMD-9∆90 ) were studied through docking simulations using the LibDock algorithm. This method allows the identification of low-energy binding modes of ligands based on polar and apolar interactions sites (hotspots). The SrtA hotspots were defined as the catalytic site which encloses the catalytic residues. To prevent potential interactions, water molecules were removed from the cavity. The structure of the ligand complies following polar and apolar interaction sites of the receptor. Using this docking procedure guarantees that only the highest scoring poses (30 to 100 poses) are kept. The Libdock scoring functions based on simple pair-wise score calculations were used in all simulations. Finally, poses were assessed considering the position and orientation of the sorting pattern within the catalytic cavity and its proximity to the catalytic triad. Delorme et al. [11] postulated that the imperfect duplication of 32 amino acid residues (residues 63 to 90) in the prodomain PP of PrtS of the strain LMD-9 (PrtS LMD-9 ; locus: STER_RS04165; NCBI reference sequence: NC_008532.1) is responsible for the anchoring and/or maturation of this CEP. To evaluate this, the LMD-9 prtS4F44 mutant strain was constructed by integrating the prtS allele of the 4F44 strain (prtS 4F44 (taxon: 1308; GenBank: GU459009.1)), encoding a CEP devoid of such duplication, in the LMD-9 ∆prtS mutant strain [31]. The proteolytic activity was evaluated at the cells' surface and in the growth supernatant of this mutant as well as the mutant LMD-9 ∆prtS , served as a negative control. Followed by the growth of both strains in LM17 to an OD 600nm of 1, cells were harvested by centrifugation and both the filtered supernatants and the cells were subsequently incubated with the synthetic substrate Suc-Ala-Ala-Pro-Phe-pNA. As anticipated, a strong proteolytic activity was noticed at the surface of the cells of LMD-9 prtS4F44 but not in the filtered supernatant, whereas no such activity was reported for the negative control (Table 1). Hence, it is concluded that any mutation in the sequence of PrtS 4F44 , particularly the absence of the imperfect duplication of 32 amino acid residues in its prodomain PP, is not responsible for the liberation of PrtS 4F44 into the extracellular medium of strain 4F44.

Role of Maturases in the Maturation of PrtS in S. thermophilus LMD-9
An exploration of data banks allowed for identification of four genes that encode PPIases (prtM, tig or ropA, ppiA, and pplB) in sequenced genomes of S. thermophilus. Among them, two maturases, PplB and PpiA, belong to the cyclophilin family, whereas PrtM and RopA belong to the parvulin and FKBP (FK506 binding protein) families, respectively. An analysis of these four genes in the 4F44 strain revealed that their deduced proteins are identical with the corresponding ones of the LMD-9 strain, except the PrtM protein (taxon: 1308; locus: AMM43147; GenBank: KT809299.1), which displays four differences (D30N, V75A, A78V, and A242T). Hence, to determine its implication in PrtS maturation, the LMD-9 ∆prtM mutant strain was constructed by deleting the prtM gene in the LMD-9 strain.
The LMD-9, LMD-9 ∆prtS , and LMD-9 ∆prtM strains were grown in milk to evaluate the effect of prtM deletion on the growth performance of the mutant. This approach is an indirect measure of the PrtS proteolytic activity considering the relationship between the PrtS activity and S. thermophilus capacity to grow in milk [6,7]. The results revealed that the LMD-9 ∆prtS mutant showed a delayed growth both in LM17 and milk media contrary to strains LMD-9 and LMD-9 ∆prtM , which showed a similar growth behavior (Figure 1). In a similar manner, the proteolytic activity was observed only at the cells' surface of LMD-9 and LMD-9 ∆prtM (Table 1), and not in the culture supernatants. maturation, the LMD-9∆prtM mutant strain was constructed by deleting the prtM gene in the LMD-9 strain.
The LMD-9, LMD-9∆prtS, and LMD-9∆prtM strains were grown in milk to evaluate the effect of prtM deletion on the growth performance of the mutant. This approach is an indirect measure of the PrtS proteolytic activity considering the relationship between the PrtS activity and S. thermophilus capacity to grow in milk [6,7]. The results revealed that the LMD-9∆prtS mutant showed a delayed growth both in LM17 and milk media contrary to strains LMD-9 and LMD-9∆prtM, which showed a similar growth behavior (Figure 1). In a similar manner, the proteolytic activity was observed only at the cells' surface of LMD-9 and LMD-9∆prtM (Table 1), and not in the culture supernatants.  Consequently, the lack of the putative PrtM maturase in the LMD-9 ∆prtM strain did not result in any growth retardation, nor in a decrease in PrtS activity and/or its release into the extracellular medium; it is most probable that the PrtM maturase was not responsible for the PrtS maturation. Several factors support the hypothesis that PrtS could undergo automaturation. First, unlike thee prsA of S. pyogenes and prtM of L. lactis, which are located upstream of the speB and prtP genes, respectively and co-transcribed, the prtM of S. thermophilus is not located near PrtS [9,12,14,36]. Second, when Chang et al. [23] established the N-terminal sequence of the soluble PrtS form of the 4F44 strain, they detected the N-terminal sequence corresponding to the proenzyme (with prodomain PP) as well as the N-terminal sequence of the mature form (without prodomain PP) of PrtS. The same situation was also detected for anchored PrtS forms present at the cells' surface of the LMD-9 strain [31]. It was also noticed that the PrtS proenzyme form disappeared progressively in favor of the mature form (unpublished results) preceding a prolonged incubation or high concentration preservation. The fact that PrtS could be automaturated is not inimitable in LAB, e.g., regarding CEP PrtB of Lb. bulgaricus [37]. Finally, the fact that the inactivation of prtM did not lead to any release of PrtS in the growth medium, associated with the absence of difference between the three other known maturases (PpiA, PplB, and RopA), ruled out the hypothesis of a link between maturation via PrtM especially, correct folding, and anchoring.

SrtA Is Responsible for the Anchoring of PrtS to the Cell Wall of S. thermophilus and Is Deficient in Strain 4F44
The hypothesis that, despite few numbers of substitutions found between SrtA 4F44 and SrtA LMD-9, the extracellular liberation of PrtS 4F44 could result from a partial deficiency of SrtA 4F44 implies first providing evidence that SrtA is actually responsible for the PrtS anchoring to the cell wall of S. thermophilus as in other Gram-positive bacteria. Hence, the LMD-9 ∆srtA mutant strain was constructed by replacing the srtA gene by an erythromycin resistance gene. Afterwards, a complemented mutant was constructed by reintegrating the srtA LMD-9 gene to the LMD-9 ∆srtA mutant to verify whether the phenotype of wild type LMD-9 strain could be restored (LMD-9 srtALMD-9 ).
The synthetic substrate Suc-Ala-Ala-Pro-Phe-pNA was used to search proteolytic activity in the filtered growth medium of the LMD-9 ∆srtA mutant, complemented mutant, and WT strain (Table 1). It was detected only in the LMD-9 ∆srtA mutant (1020 mAU of PrtS activity; Table 1). The same supernatants were then analyzed by a casein-zymogram to detect PrtS (Figure 2A). Three caseinolytic bands (170, 154, and 115 kDa) previously shown to correspond to PrtS [23] were observed (Figure 2A) in the supernatant of mutant LMD-9 ∆srtA contrary to WT and LMD-9 srtALMD-9. This corroborates the results obtained using synthetic substrate Suc-Ala-Ala-Pro-Phe-pNA. Unexpectedly, the LMD-9 ∆srtA strain displayed a surface PrtS activity (Table 1) despite of srtA deletion. To determine whether this resulted from electrostatic or other low-force interactions of PrtS with the cell surface, cells of this mutant and of strains 4F44 and LMD-9 srtALMD-9 , used as controls, were suspended in Tris HCl (100 mmol/L, pH 7) buffer and incubated from 0 to 70 h at 4 • C. The cells' surface-bound PrtS activity was determined at t0 and t70 using a casein-zymogram ( Figure 2) and the synthetic substrate Suc-Ala-Ala-Pro-Phe-pNA. After 70 h of incubation, the surface PrtS activity of strains 4F44, LMD-9, and LMD-9 srtALMD-9 was close to that obtained at t0, while that of the LMD-9 ∆srtA strain appeared to be undetectable. The PrtS activity was exclusively found in the incubation buffer of this strain ( Figure 2C). These last results proved that the SrtA of S. thermophilus is responsible for the anchoring of PrtS to the cell wall and probably the other proteins possessing a LPXTG motif. Such a phenotype has already been described in a St. aureus mutant that was defective in the anchoring of surface LPXTG proteins because of a mutation in the srtA gene. The deletion of this gene resulted in the liberation of surface LPXTG proteins, thereby leading to a decreased virulence of the bacterium [38]. In a similar manner, in the Streptococcus genus, another study demonstrated that SrtA is a key player responsible for the anchoring of surface LPXTG proteins of S. pyogenes [39].
To further investigate whether SrtA 4F44 is partially defective, the LMD-9 srtA4F44 mutant was constructed by introducing the srtA 4F44 allele into the genome of the LMD-9 ∆srtA strain. Hereafter, the soluble PrtS activity was examined in filtered growth supernatants of the strains LMD-9 srtA4F44 and 4F44 by using both the Suc-Ala-Ala-Pro-Phe-pNA substrate (Table 1) and a casein-zymogram ( Figure 2B). Through analyses being realized in the conditions, levels of PrtS activity could be compared (see Section 2). Twenty-six percentages of total PrtS activity of the mutant LMD-9 srtA4F44 were found in its growth supernatant, i.e., a proportion similar to that of strain 4F44 (37%). In addition, the incubation of the cells of this mutant in Tris-HCl (100 mmol L −1 , pH 7) buffer during 70 h did not result in a significant increase of extracellular PrtS activity. Therefore, despite the few numbers of substitutions found between the SrtA of strains LMD-9 and 4F44, our results provide genetic proof of the dysfunctioning of SrtA 4F44 , which leads to the anchoring to the cell wall of the majority of PrtS molecules and to the release of a fraction of PrtS in the growth medium of the 4F44 strain, as postulated by Chang et al. [23]. Indeed, the fact that after 70 h of incubation of the cells of 4F44 and LMD-9 srtA4F44 strains in Tris-HCl buffer the surface PrtS activity remained similar to the initial one strongly suggests that it corresponds to PrtS molecules correctly anchored to the cell wall. Besides, even if the total PrtS activity (bound plus free) appeared in our assays to be higher in strain 4F44 than in strain LMD-9 (8860 mAU against 4194, Table 1), the PrtS release cannot be attributed to a higher expression of its gene in strain 4F44, leading to the saturation of SrtA activity and ultimately to a leakage of non-anchored PrtS molecules in the external environment. Indeed, in the mutant LMD-9 srtA4F44 , the gene prtS undergoes the same regulation like in the wild-type LMD-9 strain, as suggested by the PrtS activity levels observed in this mutant (Table 1). Therefore, no saturation of the anchoring activity of sortase SrtA is expected and SrtA 4F44 , which is expressed in this mutant, should anchor all PrtS molecules.
Hereafter, the soluble PrtS activity was examined in filtered growth supernatants of the strains LMD-9srtA4F44 and 4F44 by using both the Suc-Ala-Ala-Pro-Phe-pNA substrate (Table 1) and a casein-zymogram ( Figure 2B). Through analyses being realized in the conditions, levels of PrtS activity could be compared (see Materials and Methods section). Twenty-six percentages of total PrtS activity of the mutant LMD-9srtA4F44 were found in its growth supernatant, i.e., a proportion similar to that of strain 4F44 (37%). In addition, the incubation of the cells of this mutant in Tris-HCl (100 mmol L −1 , pH 7) buffer during 70 h did not result in a significant increase of extracellular PrtS activity. Therefore, despite the few numbers of substitutions found between the SrtA of strains LMD-9 and 4F44, our results provide genetic proof of the dysfunctioning of SrtA4F44, which leads to the anchoring to the cell wall of the majority of PrtS molecules and to the release of a fraction of PrtS in the growth medium of the 4F44 strain, as postulated by Chang et al. [23]. Indeed, the fact that after 70 h of incubation of the cells of 4F44 and LMD-9srtA4F44 strains in Tris-HCl buffer the surface PrtS activity remained similar to the initial one strongly suggests that it corresponds to PrtS molecules correctly anchored to the cell wall. Besides, even if the total PrtS activity (bound plus free) appeared in our assays to be higher in strain 4F44 than in strain LMD-9 (8860 mAU against 4194, Table 1), the PrtS release cannot be attributed to a higher expression of its gene in strain 4F44, leading to the saturation of SrtA activity and ultimately to a leakage of non-anchored PrtS molecules in the external environment. Indeed, in the mutant LMD-9srtA4F44, the gene prtS undergoes the same regulation like in the wild-type LMD-9 strain, as suggested by the PrtS activity levels observed in this mutant (Table 1). Therefore, no saturation of the anchoring activity of sortase SrtA is expected and SrtA4F44, which is expressed in this mutant, should anchor all PrtS molecules.

Substitution of the Ile 218 Residue Is Not Responsible for the Deficiency of SrtA 4F44
As we knew the LPNTG motif was present at the C-terminal extremity of the extracellular soluble form of PrtS 4F44 [23], molecular modeling simulations were performed to determine whether the binding mechanism of the LPNTG motif to the SrtA catalytic site was at least partially altered in SrtA 4F44 .
Since S. pyogenes and S. thermophilus belong to the same genus, their respective sortases A were assumed to have similar mechanisms. Thus, the structural models of SrtA LMD-9 (St-SrtA LMD-9∆90 ) and SrtA 4F44 (St-SrtA 4F44∆90 ) were built from the already defined structure of S. pyogenes SF370 SrtA (Sp-SrtA SF370∆86 ) [24]. In order to build the models with the same amino acid residues as in the Sp-SrtA SF370∆86 , it was necessary to delete the first 90 residues of St-SrtA LMD-9∆90 and SrtA 4F44 (St-SrtA 4F44∆90 ). The percentage of identity/similarity between the C-terminal domain of Sp-SrtA SF370∆86 and St-SrtA 4F44∆90 , and Sp-SrtA SF370∆86 and St-SrtA LMD-9∆90 were found to be sufficient (between 70.6% to 97.5% and 87.1% to 99.4%, respectively) to use Sp-SrtA SF370∆86 as a structural pattern for constructing the St-SrtA 4F44∆90 and St-SrtA LMD-9∆90 ones. Hence, surimposition of the three structures led to RMSD values below 0.16 Å considering all the 169 C α atoms of the residues 87 to 249. No significant structural difference was observed between the structures, as St-SrtA 4F44∆90 and St-SrtA LMD-9∆90 models displayed the characteristic structure of sortases, i.e., the eightstranded β-barrel fold and a long hydrophobic cleft corresponding to the catalytic cavity located at the center of the protein (Figure 3). It was assumed that the residues Cys 208 , His 142 , and Arg 216 compose the catalytic triad ( Figure 3). The orientation of these residues is consistent with the model of reverse protonation that has been proposed in biochemical studies of the sortase A of St. aureus and S. pyogenes [24,40].
was at least partially altered in SrtA4F44.
Since S. pyogenes and S. thermophilus belong to the same genus, their respective sortases A were assumed to have similar mechanisms. Thus, the structural models of SrtALMD-9 (St-SrtALMD-9∆90) and SrtA4F44 (St-SrtA4F44∆90) were built from the already defined structure of S. pyogenes SF370 SrtA (Sp-SrtASF370∆86) [24]. In order to build the models with the same amino acid residues as in the Sp-SrtASF370∆86, it was necessary to delete the first 90 residues of St-SrtALMD-9∆90 and SrtA4F44 (St-SrtA4F44∆90). The percentage of identity/similarity between the C-terminal domain of Sp-SrtASF370∆86 and St-SrtA4F44∆90, and Sp-SrtASF370∆86 and St-SrtALMD-9∆90 were found to be sufficient (between 70.6% to 97.5% and 87.1% to 99.4%, respectively) to use Sp-SrtASF370∆86 as a structural pattern for constructing the St-SrtA4F44∆90 and St-SrtALMD-9∆90 ones. Hence, surimposition of the three structures led to RMSD values below 0.16 Å considering all the 169 Cα atoms of the residues 87 to 249. No significant structural difference was observed between the structures, as St-SrtA4F44∆90 and St-SrtALMD-9∆90 models displayed the characteristic structure of sortases, i.e., the eight-stranded β-barrel fold and a long hydrophobic cleft corresponding to the catalytic cavity located at the center of the protein (Figure 3). It was assumed that the residues Cys208, His142, and Arg216 compose the catalytic triad ( Figure 3). The orientation of these residues is consistent with the model of reverse protonation that has been proposed in biochemical studies of the sortase A of St. aureus and S. pyogenes [24,40].  These homology models were then used to study possible binding modes of the LPNTG pattern within the sortase catalytic cavity. Enzyme/substrate complexes were generated using docking simulations with two substrates: the LPNTG pattern and a longer pattern, specifically the Ace-QLPNTGEND-Nme pattern. Unfortunately, no significant difference was observed between the complexes Sp-SrtA SF370∆86  To definitively rule out the hypothesis of an eventual role of the substitution of Ile 218 by Val 218 in the deficiency of StrA 4F44 and to validate the modeling prediction, the residue Ile 218 of strain LMD-9 srtALMD-9 was replaced by a valine residue ( Figure A1). The absence of extracellular PrtS activity (searched using the Suc-Ala-Ala-Pro-Phe-pNA substrate) in the supernatant of this strain confirms the non-involvement of Ile 218 substitution by Val 218 in the deficiency of StrA 4F44 .
To conclude we showed that (i) the 32 amino acid residues' imperfect duplication located in the prodomain of certain PrtS, such as PrtS LMD-9 , was not essential for the correct maturation and subsequent anchoring of PrtS; (ii) the maturase PrtM, homologous to maturases of other CEP, was not responsible for the maturation of PrtS and neither for its correct anchoring to the cell surface; and (iii) SrtA was responsible for the anchoring of PrtS to the cell wall and (iv) SrtA of strain 4F44 was partially defective despite the low number of dissimilar residues (six substitutions), which differentiates it from that of the LMD-9 strain and probably through a subtle mechanism not yet elucidated perhaps because of a lack of a structural model, including the N-terminal part of SrtA.   Appendix A Figure A1. Strategy to replace the ATT codon specifying the Ile218 residue of SrtALMD-9 by a GTT codon corresponding to a Valine residue. Two PCRs were carried out using genomic DNA of the LMD-9srtALMD-9 (LMD-9srtALMD-9-g DNA) mutant to obtain (i) an upstream DNA fragment of the srtA gene (UpsrtA) and the first 675 nucleotides of the same gene, and (ii) a DNA fragment containing the second part of the srtA gene (87 nucleotides), the spec gene (confers resistance to spectinomycin), and a downstream sequence of the srtA gene (DownsrtA) (A). These two overlapped PCR fragments were then used to make the recombinant DNA fragment by overlapping PCR (B). The recombinant DNA amplicon was introduced into competent cells of the LMD-9∆srtA mutant, thereby permitting obtainment after its integration through a double crossing-over event, specifically the mutant LMD-9srtA:Ile218  Val218 (C). Primer sequences used to amplify different fragments are presented in Table S1. Figure A1. Strategy to replace the ATT codon specifying the Ile 218 residue of SrtA LMD-9 by a GTT codon corresponding to a Valine residue. Two PCRs were carried out using genomic DNA of the LMD-9 srtALMD-9 (LMD-9 srtALMD-9 -g DNA) mutant to obtain (i) an upstream DNA fragment of the srtA gene (UpsrtA) and the first 675 nucleotides of the same gene, and (ii) a DNA fragment containing the second part of the srtA gene (87 nucleotides), the spec gene (confers resistance to spectinomycin), and a downstream sequence of the srtA gene (DownsrtA) (A). These two overlapped PCR fragments were then used to make the recombinant DNA fragment by overlapping PCR (B). The recombinant DNA amplicon was introduced into competent cells of the LMD-9 ∆srtA mutant, thereby permitting obtainment after its integration through a double crossing-over event, specifically the mutant LMD-9 srtA:Ile218→Val218 (C). Primer sequences used to amplify different fragments are presented in Table S1. Figure A2. Interaction between the LPNTG ligand and amino acid residues of St-SrtA4F44∆90. The numbers represent the distance between atoms of the ligand and residues of the cavity of sortase A. Red letters indicate residues of the catalytic triad. The residues Gln187, Glu190, Val194, Ala211, and Val218 correspond to the residues Ala187, Arg190, Ile194, Ile211, and Ile218 of Sp-SrtASF370∆86, while residue Val218 corresponds to the residue Ile218 of St-SrtALMD-9.  Figure A3. Superimposition of the main binding modes of ligand Ace-QLPNTGEND-Nme within the sortase A (here SrtA 4F44∆90 ) catalytic cavity. Ligands denoted here correspond to the best ligand obtained for each complex (Sp-SrtA SF370∆86 :Ace-QLPNTGEND-Nme; SrtA LMD-9∆90 :Ace-QLPNTGEND-Nme; and SrtA 4F44∆90 :Ace-QLPNTGEND-Nme) according to (i) the distance between the S atom of the corresponding sortase A Cys 208 residue and the alpha carbon of the threonine residue of the ligand, and (ii) the position of the ligand within the cavity. Thus, the brown ligand corresponds to the ligand of complex Sp-SrtA SF370∆86 :Ace-QLPNTGEND-Nme, the purple one to SrtA LMD-9∆90 :Ace-QLPNTGEND-Nme, and the blue one to SrtA 4F44∆90 :Ace-QLPNTGEND-Nme.