Inhibition of Candida albicans morphogenesis by chitinase from Lactobacillus rhamnosus GG

Lactobacilli have been evaluated as probiotics against Candida infections in several clinical trials, but with variable results. Predicting and understanding the clinical efficacy of Lactobacillus strains is hampered by an overall lack of insights into their modes of action. In this study, we aimed to unravel molecular mechanisms underlying the inhibitory effects of lactobacilli on hyphal morphogenesis, which is a crucial step in C. albicans virulence. Based on a screening of different Lactobacillus strains, we found that the closely related taxa L. rhamnosus, L. casei and L. paracasei showed stronger activity against Candida hyphae formation compared to other Lactobacillus species tested. By exploring the activity of purified compounds and mutants of the model strain L. rhamnosus GG, the major peptidoglycan hydrolase Msp1, conserved in the three closely related taxa, was identified as a key effector molecule. We could show that this activity of Msp1 was due to its ability to break down chitin, the main polymer in the hyphal cell wall of C. albicans. This identification of a Lactobacillus-specific protein with chitinase activity having anti-hyphal activity will assist in better strain selection and improved application in future clinical trials for Lactobacillus-based Candida-management strategies.


Results
selected Lactobacillus strains show strong hyphae-inhibitory activity. First, we aimed to compare the anti-Candida activity between different Lactobacillus taxa. Since hyphal morphogenesis is the most important virulence factor of C. albicans 6 , we focused on the effect of lactobacilli on serum-induced hyphal morphogenesis. We selected twenty strains available in-house or in the Belgian Co-ordinated Collections of Micro-organisms, representing the different taxa/phylogenetic groups that have been recently described as being mainly nomadic or vertebrate-adapted 38 . Strains were thus selected from the L. casei group, L. plantarum group, L. reuteri, L. fermentum, L. gasseri, L. jensenii and L. crispatus. The inhibition rates showed large variation among the tested strains, ranging from 91% (L. casei AMBR2) to 14% (L. plantarum WCFS1) (Fig. 1a).
Lactic acid has been described as key bioactive metabolite of Lactobacillus, and is also reported to affect C. albicans 39,40 . Therefore we next measured the concentration of D-lactic acid and L-lactic acid in the supernatant of these strains, after growth into stationary phase. All strains were able to produce lactic acid from glucose, although in different ratios of D-and L-lactic acid (Fig. 1b). The level of inhibitory activity of the tested lactobacilli did not increase with an increasing concentration of either isomer, in fact, the inhibitory activity actually showed a negative correlation with the concentration of D-lactic acid (based on Pearson correlation, p-value < 0.0001 for D-lactic acid).
The five best performing strains in our tests all belonged to the L. casei group (L. rhamnosus, L. casei and L. paracasei, based on a comparative genome analysis-defined taxonomy as proposed in 41 ), suggesting an effector molecule that is shared among these taxa. the major peptidoglycan hydrolase of L. rhamnosus GG and lactic acid jointly mediate C. albicans hyphae inhibition. To further elucidate how Lactobacillus can impact hyphal morphogenesis, we first explored whether the contributing L. (para)casei/rhamnosus factors are surface-bound, secreted, or both. L. rhamnosus GG was chosen as model, since this strain is well-characterized at genetic and molecular level 42 . We first compared the effect of live L. rhamnosus GG cells on serum-induced hyphal formation to its cell-free culture supernatant, containing solely secreted molecules, and to UV-inactivated or heat-killed L. rhamnosus GG cells. Cells treated in both ways should no longer secrete molecules, but in contrast to the heat-killed cells, the surface proteins of the UV-inactivated cells should not be denatured. We showed that the supernatant from L. rhamnosus GG inhibited hyphal formation almost completely (97%), whereas the UV-inactivated L. rhamnosus GG cells inhibited hyphal formation of C. albicans to the same extent as live cells (57% and 51%, respectively) (Fig. 2a). The heat-killed cells, on the other hand, were no longer able to inhibit C. albicans hyphal formation. These results thus indicate that the main core L. rhamnosus-specific effectors molecules are secreted, but can also be surface-bound or are supplemented by a heat-sensitive cell-bound effector.
Next, we explored the activity of the major documented L. rhamnosus GG surface molecules that could have putative hyphae-binding properties due to lectin-sugar interactions. Key candidates for hyphae-binding include the lectin-like protein 1 (Llp1) and 2 (Llp2) 43 , the galactose-rich exopolysaccharides (EPS) 44 and the major secreted protein 1 (Msp1), which is mannosylated 45 . Llp1 and Llp2 have been shown to bind to D-mannose and the complex sugar mannan by sepharose-binding and glycan array screening 43 , both of which are present in the outer layer of C. albicans cell wall 3,6 . We therefore aimed to explore whether this sugar-binding capacity could also result in interference with hyphal morphogenesis. Treatment with Llp1 and Llp2 did not, however, show a reduction of Candida hyphal formation at 50 µg/ml (Fig. 2b), a previously documented active antibacterial concentration 43 . Proteins with lectin-like properties can also be found on the hyphal surface 46,47 , rendering the glycoconjugates on the lactobacillary surface potential interaction partners as well. In agreement with previous results 48 , isolated EPS from L. rhamnosus GG was able to inhibit hyphal morphogenesis, but only at a rather high concentration of 200 µg/mL (Fig. 2b). In contrast, the peptidoglycan hydrolase Msp1 from L. rhamnosus GG tested here demonstrated a remarkably strong inhibitory activity (Fig. 2c), reducing hyphal morphogenesis with more than 50%, at concentrations as low as 5 µg/mL. To check whether Msp1 was only inhibiting hyphal morphogenesis and not the viability of C. albicans, we determined the growth capacity of the C. albicans cells after three hours and six hours of hyphal induction in presence of Msp1. This showed that the viability of the C. albicans was not affected during the treatment with Msp1 ( Supplementary Fig. S1).
Although the production of lactic acid by the lactobacilli could not really explain the observed variation in anti-hyphal activity between different Lactobacillus strains ( Fig. 1), we also exogenously added lactic acid in this screening to quantify its contribution to the antihyphal activity of L. rhamnosus GG. Lactic acid as such, at naturally occurring culture supernatant concentrations (1%, a combination of D-and L-lactic acid in a 1:1 ratio), also reduced morphogenesis by approximately 50% (Fig. 2d).
Since the supernatant showed very strong activity and since Msp1 and lactic acid are major components of the supernatant, we next investigated whether Msp1 could act synergistically with lactic acid. The combination of lactic acid at a lower concentration than present in the supernatant (0.5%) and Msp1 (2 µg/ml) was shown to decrease hyphal formation more than 94%, a level of inhibition comparable to the cell-free supernatant, indicating this combination contained the main effectors conferring the anti-hyphal activity to L. rhamnosus GG (Fig. 2e). Since we observed a negative correlation between D-lactic acid production and hyphal inhibition and since the best-performing strains mainly produced L-lactic acid, we compared the synergistic effect of mixed lactic acid on Msp1 activity to both isomers separately. Remarkably, this comparison showed no differences between the isomers (Fig. 2e).
Hyphal morphogenesis is tightly linked to biofilm regulation of C. albicans 49 , we therefore next investigated whether L. rhamnosus GG could also inhibit C. albicans biofilm formation. This experimental set-up revealed that the supernatant of L. rhamnosus GG was able to decrease biofilm formation of C. albicans. The two main components of the supernatant, lactic acid and Msp1, separately also showed anti-biofilm activity, however no clear synergistic effect was observed with the concentrations of lactic acid and Msp1 tested (Fig. 2f).

Mutant analysis of L. rhamnosus GG supports key role for Msp1. As mutual interactions between
the individual molecules on the lactobacillary surface might strengthen or attenuate the anti-hyphal activity of individual purified molecules, we performed additional experiments with specific L. rhamnosus GG isogenic mutants available from our previous research (see Materials and methods section) 42 . This complementary approach also allowed us to study molecules that could not be purified to a sufficient level.
Mutant analysis confirmed that the presence or absence of the EPS layer and lectins does not play a crucial role in the anti-hyphal activity of L. rhamnosus GG cells, as shown in Fig. 3a. Previous research showed the importance of the SpaCBA pili and their fucose and mannose residues in L. rhamnosus GG interactions with host cells and glycoconjugates, such as intestinal mucus 50,51 , of which structural homologs might be present on the hyphal surface. These complex, heteropolymeric SpaCBA pili themselves are difficult to purify 50-52 , therefore we included the isogenic spaCBA mutant of L. rhamnosus GG in the mutant analysis. This showed that the presence or absence of these SpaCBA pili did not play a significant role in the anti-hyphal activity of L. rhamnosus GG (Fig. 3a).
Due to the central role of Msp1 in bacterial growth and cell separation, an isogenic knock-out mutant is not available in L. rhamnosus GG 53 . However, the dltD mutant is an interesting generic surface mutant of L. rhamnosus GG, because the lipoteichoic acids are no longer D-alanylated, resulting in dramatic shifts in surface charge and association with surface proteins and other molecules 54 . Remarkably, the hyphal morphogenesis of C. albicans was almost completely abolished by L. rhamnosus GG dltD mutant cells. To explore whether this could also be explained by the activity of Msp1, we checked whether Msp1 stayed more associated with the surface of dltD mutant cells after secretion than in the wild-type cells. Fluorescently labelled anti-Msp1 antibodies showed that Msp1 was indeed a twofold less secreted in the supernatant of the dltD mutant (Fig. 3b) and appeared to be present in higher concentration on the surface of these mutant cells (Fig. 3c). This thus probably resulted in a higher bio-availability of Msp1 in experiments using the dltD mutant cells as compared to wild-type cells. The consequential comparison between the effects of the supernatant from L. rhamnosus GG wild-type and dltD mutant on hyphal morphogenesis showed that the lower secretion of Msp1 in the supernatant of the dltD mutant indeed resulted in a significantly lower inhibition (p = 0.0001) (Fig. 3d).
The combination of the approach using either mutants or isolated molecules thus further demonstrated the key role for Msp1 in the anti-hyphal activity of L. rhamnosus GG. This finding is in agreement with the fact that the other tested strains from the L. casei group showed strong activity (Fig. 1a), since Msp1 has been shown to be conserved among -at least a part of -the L. casei group 45 , while the other studied molecules are rather specific for the strain L. rhamnosus GG.
Msp1 shows chitinase activity, independent of its glycosylation state. We subsequently aimed to explore the interaction between Msp1 and Candida cells in more detail. First, we compared the binding to hyphal cells between L. rhamnosus GG, as a strong anti-hyphal strain, and L. plantarum WCFS1, being one of the least effective strains tested previously (Fig. 1a). These strains belong to the limited number of Lactobacillus strains whose main peptidoglycan hydrolases have been thoroughly characterized 53,55 . Both their major peptidoglycan hydrolases have been shown to be localized at the poles of the Lactobacillus cells, but they differ in hydrolytic activity and glycosylation state: Msp1 has documented γ-D-glutamyl-L-lysyl-endopeptidase activity 53 and appears to be glycosylated with mannose residues 45 , while Acm2 from L. plantarum WCFS1 was identified as an endo-β-acetylglucosaminidase 55 and appears to glycosylated with N-acetylglucosamine residues 56 . We first explored whether these dissimilarities are reflected in a different interaction of the Lactobacillus strains with the hyphae. Microscopic inspection of C. albicans hyphae after induction in presence of L. rhamnosus GG revealed that the Lactobacillus poles appeared to be the main contact point with the hyphal cells (Fig. 4a, right panel). In contrast to L. rhamnosus GG, L. plantarum WCFS1 cells did not appear to closely interact with the hyphae (Fig. 4a, left panel), suggesting that the close binding of L. rhamnosus GG poles to the hyphae is important for its (2019) 9:2900 | https://doi.org/10.1038/s41598-019-39625-0 www.nature.com/scientificreports www.nature.com/scientificreports/ anti-hyphal activity. Counting the attached and unattached Lactobacillus cells in three different repeats showed that 60 ± 6% of the L. rhamnosus GG bound to the hyphae, while none of L. plantarum WCFS1 did (data not shown).
To explore whether the binding between Msp1 and C. albicans hyphae could indeed be due to their sugar-lectin interactions, as suggested above, we next investigated the activity of non-glycosylated Msp1. After chemical deglycosylation, the level of hyphal inhibition showed to be similar to native (glycosylated) Msp1 (Fig. 4b), indicating that another mechanism probably underlies the anti-hyphal activity of Msp1.
Despite their different origin, chitin from C. albicans and peptidoglycan from L. rhamnosus GG show some structural similarities due to the presence of N-acetylglucosamine residues in both their backbones. Because of this, and because of the close contact between the Lactobacillus poles and the hyphae, we hypothesized that Msp1 might be able to use chitin, the main polymer of the hyphal cell wall, as a substrate. Based on assays with chitin-azure, we found that Msp1 is indeed able to break down chitin, to the same extent as a commercially available chitinase from Streptomyces griseus (Fig. 4c). Finally, we determined whether a chitinase inhibitor would be www.nature.com/scientificreports www.nature.com/scientificreports/ able to restore C. albicans hyphal morphogenesis. Bisdionine C, a known chitinase inhibitor, partially reversed the inhibitory effects of Msp1 on hyphal morphogenesis (Fig. 4d), further substantiating the chitinase activity as basis for the anti-hyphal capacity of Msp1.

Discussion
In the present study, we showed that certain Lactobacillus taxa can inhibit hyphal morphogenesis of C. albicans more efficiently than others. More specifically, we demonstrated that the major secreted protein and main peptidoglycan hydrolase of L. rhamnosus GG, Msp1, is the key effector and can reduce hyphal formation by its chitinase activity, especially in combination with lactic acid, another important metabolite of lactobacilli.
Our findings on the complete inhibition of hyphal formation by the supernatant from L. rhamnosus GG is in line with previous observations on the effect of L. rhamnosus LR32 supernatant on hyphae density in C. albicans biofilms 35 . Moreover, a comparison between live cells and both UV-inactivated and heat-killed cells provided novel insights into the underlying molecular mechanism, as the effectors need to be structurally intact, but not necessarily actively secreted during the hyphal induction.
The multilayered cell wall of C. albicans, existing of an inner layer of chitin and β-glucans and an outer layer of mannans and (glycosylated) proteins 6 , offers several potential target sites for the binding with lactobacillary factors. We tested different secreted and surface-bound molecules from L. rhamnosus GG, which are represented in the schematic overview in Fig. 5. By combining the results on anti-hyphal activity of purified molecules with these of different mutant strains, we found the combination of Msp1 and lactic acid to be the key effectors and synergistically abolish hyphal morphogenesis.
The degrading effects of chitinases on a yeast cell wall were first described for the fungus Trichoderma viride 57 , but they have to the best of our knowledge not yet been described for bacterial peptidoglycan hydrolases nor for any lactobacillary protein. Although chitin and peptidoglycan share some structural similarities, namely the presence of N-acetylglucosamine residues in their backbones, the previously studied peptidoglycan hydrolase activity of Msp1 was not shown to involve these bounds. Msp1 was shown to carry γ-D-glutamyl-L-lysyl-endopeptidase activity 53 , cutting between the D-glutamine and L-lysine residues in the peptide stem in bacterial peptidoglycan. This peptidoglycan hydrolase activity was also shown to be conserved among a number of the tested L. casei group strains 45 and was also found in L. casei BL23 58 . Unfortunately, structural information on peptidoglycan www.nature.com/scientificreports www.nature.com/scientificreports/ composition and accompanying hydrolase activity in Lactobacillus is quite limited. In L. plantarum WCFS1, the hydrolase responsible for cell septation (Acm2) was identified as an endo-β-acetylglucosaminidase 55 , and in L. gasseri DSM 20243, the major peptidoglycan hydrolase was shown to have N-acetylmuramidase activity 59 . Microscopic examination and anti-hyphal experiments with L. plantarum WCFS1 indicate that Msp1 does not share its chitinase activity with other types of peptidoglycan hydrolases. In L. fermentum, however, D-glutamine and lysine were also observed in the peptide stem 60 . This indicates that L. fermentum strains might have a similar peptidoglycan structure and possibly a similar main peptidoglycan hydrolase activity as L. rhamnosus GG, which could explain that the inhibition level of L. fermentum AMBV1 was almost to the same extent as some L. casei strains. Yet, the exact enzymatic activity remains to be substantiated in follow-up studies.
In light of the observed chitinase activity of Msp1, a number of factors could explain the synergistic effects with lactic acid. Firstly, Msp1, as a hydrolase, has an acidic pH optimum 53 . Secondly, while C. albicans is known for its acid tolerance, the proportion of chitin in the hyphal cell wall has been shown to be even more increased in an acid environment 39 . Thirdly, although we could not observe a difference in synergistic effect with Msp1 between D-and L-lactic acid at the tested concentrations, the presence of L-lactic acid, the main isomer in the supernatant of L. casei group strains, has been shown to result in glucan masking on the hyphal surface 40 , potentially making the hyphae more sensitive/accessible to the chitinase activity of Msp1.
The chitinase activity of Msp1 might also be relevant for non-hyphae producing Candida species, as it has been described that the chitin levels are elevated in C. glabrata during infections in a murine colitis model 61 . In addition, in Crohn's disease patients, both an increase in C. glabrata and a decrease in Firmicutes was found to characterize the gut microbiota 62 .
Lactobacilli and their specific properties are often evaluated at strain-level. Thanks to the approach in this study, we could suggest that this C. albicans hyphae inhibitory activity is possibly present in most strains belonging to the L. casei group due to their specific peptidoglycan structure and accompanying hydrolases. In this way, the study gives additional indications that probiotic mechanistic research should not only be performed on strain-level to find core properties 36 . The effects of Lactobacillus strains expressing this specific type of peptidoglycan hydrolase and whether they show stronger potential as anti-C. albicans strategy than others, should of course still be substantiated with in vivo evidence. Although it is difficult to explain the mixed results of clinical trials based on our findings in hindsight, since often the applied Lactobacillus strains are not specified to strain-or species level or a probiotic mixture and different formulations were often used, L. casei group strains were used in a number of clinical trials with positive outcomes. Treatment with L. rhamnosus GR-1 (formulated in gelatin capsules) improved symptoms in women suffering from vulvovaginal candidosis 15 and L. rhamnosus HS111 (formulated as dry powder in capsules) contributed to a significant reduction of Candida infection in the oral cavity 63 . In contrast, a clinical trial assessing L. casei Shirota on Candida with negative results actually investigated Candida viability rather than virulence 64 , which would not be affected by the hyphal-inhibitory activity of the lactobacilli. Of course, when evaluating a Lactobacillus strain or species for its anti-Candida potential, other factors than inhibition of hyphae should be considered. For example, L. plantarum CMPG5300 did not show high hyphal formation inhibition rates (30%) but has previously been shown to co-aggregate with C. albicans and may in this way inhibit C. albicans adhesion and contribute to disease prevention 65 . Depending on the niche, other factors may play a role for applying lactobacilli as an anti-Candida therapy, such as the epithelial adhesion of L. rhamnosus GG to the gastro-intestinal tract by its SpaCBA pili 51 and L. rhamnosus GR-1 to the vaginal mucosa by its Llp1 lectin 66 . Additional aspects of clinical trials will also influence the outcome, such as the production or formulation, including encapsulation, of the probiotics 67 and organisation of clinical trials, including randomisation and the inclusion of control groups 68,69 .
In conclusion, our data demonstrate that selected Lactobacillus taxa show stronger C. albicans hyphae inhibition activity than others, especially the taxa belonging to the L. casei group. These taxa appear to owe this inhibitory activity to their major peptidoglycan hydrolase, breaking down the main polymer of the hyphal cell wall, chitin. The identification of the peptidoglycan hydrolase as a core probiotic property helps to unravel the complex interactions between probiotic bacteria and Candida species, and can assist in the selection of proper probiotic strains for use as potential probiotics in patients with Candida infections or at risk for frequent recurrences of it.

Materials and Methods
Microbial strains and culture conditions. Lactobacillus strains (Table 1) were grown at 37 °C without agitation in de Man, Rogosa and Sharpe (MRS) broth (Difco, Erembodegem, Belgium). C. albicans SC5314 was grown in yeast extract peptone dextrose (YPD) broth (Carl Roth, Karlsruhe, Germany) at 37 °C and with continuous shaking 70 .
The in-house Lactobacillus isolates were taxonomically characterized to the species level by sequencing the 16S ribosomal RNA gene. Briefly, the complete 16S rRNA gene (1.5 kb) was amplified with the universal 27 F and 1492 R primers and sequenced. The obtained sequences were compared with reference 16S rRNA gene sequences by BLAST analysis at the National Center for Biotechnology Information (NCBI) website (https://blast.ncbi.nlm. nih.gov/Blast.cgi).
The in-house Lactobacillus isolates were collected during a clinical study (Nr 20040719) that was reviewed and approved by the ethical committee of regional hospital of Tienen (Belgium) and all patients gave their explicit consent before sampling.

Inhibition of hyphal formation in C. albicans.
Hyphal growth of C. albicans was induced by supplementing YPD broth with 10% heat inactivated fetal bovine serum (FBS) (Thermo Fischer, Asse, Belgium), while incubated with or without lactobacilli (10 8 CFU/ml) or purified molecules. After 3 hours of incubation, at least a hundred yeast cells and/or hyphae in four biological repeats were counted microscopically and the ratio of hyphae to yeast cells was calculated.

Viability of C. albicans.
The viability of C. albicans during hyphae formation and hyphae-inhibitory treatments was checked by quantifying the viable plate count at 3 and 6 hours of incubation with the macrodilution method on YPD agar.

Inhibition of C. albicans biofilm development.
The inhibiting effects on C. albicans biofilms were assessed as described previously by 71 . Briefly, 8 × 10 4 C. albicans cells were added to the wells of a 96 well plate, together with the samples (supernatant, lactic acid, Msp1) or controls (MRS or H 2 O). After incubation for 24 h at 37 °C, the biofilms were washed twice and then 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-2H -tetrazolium-5-carboxanilide (90 µl, 1 mg/ml) (Sigma Aldrich) and phenazine methosulphate (10 µl, 0.2 mg/ml) (Sigma Aldrich) were added to the wells. After a second incubation (37 °C, 30 minutes, in the dark), the absorbance at 492 nm was measured using a Synergy HTX multi-mode reader (Biotek, Drogenbos, Belgium). Isolation of Llp1 and Llp2 from L. rhamnosus GG. The Llp1 and Llp2 proteins from L. rhamnosus GG were isolated as described before 43 . Briefly, the production of the recombinant protein was induced with 1 mM isopropyl β-D-thiogalactopyranoside (IPTG) in recombinant E. coli BL21 cells expressing the lectins (CMPG10708 and CMPG10709). After incubation (25 °C, shaking), the pellets were suspended in non-denaturing lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl and 20 mM imidazole) and sonicated to release the soluble recombinant lectins from the cells. Afterwards, the lectins were purified using affinity chromatography with a HisTrap ™ HP column (GE Healthcare) and size exclusion chromatography with a Highload ™ 16/60 column packed with a matrix of Superdex ™ prep grade (GE Healthcare).

UV-inactivation and heat
Isolation of Msp1 from L. rhamnosus GG. Msp1 was purified by cationic exchange chromatography as described previously 45,72 . Briefly, the culture supernatant was loaded onto SP Sepharose High Performance (GE Healthcare), equilibrated with 60 mM lactate buffer (pH 4.0). Lactate buffer containing ascending NaCl