Voice Prosthetic Biofilm Formation and Candida Morphogenic Conversions in Absence and Presence of Different Bacterial Strains and Species on Silicone-Rubber

Morphogenic conversion of Candida from a yeast to hyphal morphology plays a pivotal role in the pathogenicity of Candida species. Both Candida albicans and Candida tropicalis, in combination with a variety of different bacterial strains and species, appear in biofilms on silicone-rubber voice prostheses used in laryngectomized patients. Here we study biofilm formation on silicone-rubber by C. albicans or C. tropicalis in combination with different commensal bacterial strains and lactobacillus strains. In addition, hyphal formation in C. albicans and C. tropicalis, as stimulated by Rothia dentocariosa and lactobacilli was evaluated, as clinical studies outlined that these bacterial strains have opposite results on the clinical life-time of silicone-rubber voice prostheses. Biofilms were grown during eight days in a silicone-rubber tube, while passing the biofilms through episodes of nutritional feast and famine. Biofilms consisting of combinations of C. albicans and a bacterial strain comprised significantly less viable organisms than combinations comprising C. tropicalis. High percentages of Candida were found in biofilms grown in combination with lactobacilli. Interestingly, L. casei, with demonstrated favorable effects on the clinical life-time of voice prostheses, reduced the percentage hyphal formation in Candida biofilms as compared with Candida biofilms grown in absence of bacteria or grown in combination with R. dentocariosa, a bacterial strain whose presence is associated with short clinical life-times of voice prostheses.


Introduction
Verbal communication in patients after laryngectomy can be restored (Figure 1) by placing a silicone-rubber voice prosthesis into a surgically created puncture between the trachea and esophagus [1]. By closing the stoma with a finger, air from the lungs can be forced through the silicone-rubber valve to produce tissue vibration and accompanying voice. In the non-sterile environment of the esophagus, valves sides of voice prostheses become rapidly colonized by microorganisms, leading to increased airflow resistance or leakage of food and liquid [2], which results in frequent replacements of prostheses. Candida albicans and Candida tropicalis are regarded as the main fungal species in voice prosthetic biofilms, and seldom exist alone in a biofilm. Next to Candida species, several other bacterial members of the commensal oral and skin flora of the host have been detected in voice prosthetic biofilms [1,3], mainly comprising streptococci, staphylococci and lactobacilli. Clearly, these commensal have easy access to prosthesis site. Recently it was shown that the diversity (i.e. number of bacterial species detected) of pathogens detected on voice prostheses correlated positively with the diversity of pathogens in the oral cavity [4].
An important aspect of Candida pathogenicity [5] is the morphogenic conversion between yeast and filamentous forms of hyphal growth [6]. Hyphae have been associated with the invasive characteristics of Candida species and their growth into siliconerubber, when exposed to cycles of feast and famine [7]. A mature biofilm on a voice prosthetic surface consists of densely packed Candida and bacteria, embedded within a matrix of exopolysaccharides [8] with a high degree of interspecies interaction [9] and co-adhesion occurring on a spatio-temporal basis. The presence of bacteria is essential for the integrity of voice prosthetic biofilms [10] and like in the oral cavity, metabolic cooperation among the organisms may be of importance to the establishment of stable biofilm communities [11]. Interestingly, combinations of different bacterial strains, like Rothia dentocariosa with C. albicans have been associated with a clinical life-time of voice prostheses lower than average, while combinations with lactobacilli have been suggested to prevail in prostheses with extended clinical life-times [1,3]. It is currently unknown why certain combinations of bacteria with Candida are more or less harmful than others.
The aim of this study was to investigate early biofilm formation on silicone-rubber by combinations of Candida (C. albicans or C. tropicalis) with different commensal bacterial strains and strains of lactobacilli. In addition, hyphal formation in C. albicans and C. tropicalis, as stimulated by R. dentocariosa or lactobacilli was evaluated, as clinical studies outlined that these bacterial strains have opposite results on the life-time of silicone-rubber voice prostheses in vivo [1,3]. All strains and species used in this study, except the lactobacilli, were originally isolated from biofilms on explanted voice prostheses.

Silicone-rubber
Commercially available silicone-rubber tubes (inner diameter 4.0 mm) were used (Rubber B.V., Hilversum, The Netherlands). Silicone-rubber surfaces were characterized by water contact angles, taken at 25uC using the sessile drop (3 mL) technique and a homemade counter monitor. This monitor registers the contour of a liquid droplet based on grey value thresholding, after which contact angles are calculated from the height and base width of a droplet.
Biofilms were grown in silicone-rubber tubes and maintained at a temperature between 36uC and 37uC during an experiment [1]. To grow biofilms, the silicone-rubber tubes were filled for 5 h with the dual species suspension.
After 5 h, adhering bacteria and yeast were allowed to grow into a biofilm on the silicone-rubber tubes during three days, by filling the tubes with growth medium. From day four till day seven, the tubes were perfused three times a day with 250 mL phosphate buffered saline (10 mM potassium phosphate and 150 mM sodium chloride, pH 6.8). Subsequently, the tubes were left in the moist environment. At the end of each day, the tubes were filled with growth medium during 30 min and left overnight in the moist environment of the drained tubes (see Figure 2A). This cycle of feast and famine was performed in order to mimic the pattern of eating and drinking of a patient and found necessary to stimulate ingrowth of hyphae into the silicone rubber (see also Figure 2B).

Evaluation of Biofilms: Amount of Biofilm
On day eight of the experiment, the silicone-rubber tubes were removed to assess the number of colony forming yeast and bacteria [1]. To this end, tubes were cut open and biofilms were removed by cotton swabs and sonication in Reduced Transport Fluid (per liter: NaCl 0.9 g, (NH 4 ) 2 SO 4 0.9 g, KH 2 PO 4 0.45 g, Mg 2 SO 4 0.19 g, K 2 HPO 4 0.45 g, EDTA 0.37 g, L-Cysteine HCl 0.2 g, pH 6.8), after which the resulting suspension was serially diluted and plated on MRS (de Man, Rogosa and Sharpe) agar for yeast and blood agar for bacteria. Plates were incubated at 37uC in an aerobic incubator for 3 days prior to enumeration. In case of combinations with lactobacilli, the resulting suspension was plated on MRS plates with incubation in 5% carbon dioxide incubator, while Candida strains were plated on Sabouraud agar plates (incubation at 26uC).
In order to demonstrate the presence of a biofilm in the siliconerubber tubes, biofilms were imaged using optical coherence tomography (OCT) and confocal laser scanning microscopy (CLSM). OCT allowed in situ imaging of biofilms within the silicone-rubber tubes without additional staining or sectioning using a spectral domain OCT (Ganymade, Thorlabs Inc., Munich, Germany) with an axial resolution of 5.8 mm and a lateral resolution of 8 mm. CLSM images were collected using a Leica TCS-SP2 CSLM (Leica Microsystems Heidelberg GmbH, Heidelberg, Germany) after taking transversal sections from the silicone-rubber tube. Sections were placed on a microscope glass slide and biofilm stained by live/dead stain (Baclight TM , Molecular probes Europe BV) for 30 min in the dark, after which images were taken.

Evaluation of Biofilms: Hyphal Induction by Selected Bacterial Strains
Biofilms grown with combinations of Candida and R. dentocariosa or lactobacilli, were further examined immediately after suspending of the biofilm in RTF. The resulting microbial suspension was analyzed using phase-contrast microscopy in order to determine the percentage of fungi in a yeast or hyphal morphology after observing 500 Candida cells in total in one experiment.

Statistical Analysis
All experiments were done in triplicate and data were compared using a Student t-test, accepting p,0.05 as statistically significant.

Results
Water contact angles on the silicone-rubber tubing amounted 11061 degrees, attesting to the hydrophobic nature of the material and corresponding with water contact angles on medical-grade silicone-rubber voice prostheses [12].
Examples of OCT and CLSM images of biofilms grown with different combinations of yeast and bacteria are shown in Figure 3. Note that biofilms consisting of a combination of C. tropicalis with L. crispatus are thicker than observed for a combination of C. albicans with R. dentocariosa regardless of the technique applied, although biofilms thicknesses obtained using OCT do not match numerically with those from CLSM. This is most likely related with the fact that tubes with biofilm were sectioned for CLSM, while OCT was done on in situ biofilms. Moreover, CLSM requires staining. Table 1 summarizes the total numbers of CFU/cm 2 of the mixed species biofilm on the silicone-rubber tubes. In general, biofilms consisting of combinations of C. albicans with a bacterial strain comprised significantly less viable organisms than C. albicans biofilms grown in the absence of bacteria. C. tropicalis on the other hand, showed a higher number of viable organisms when grown in combination with a bacterial strain, except when grown in combination with S. salivarius. The percentage of Candida present in biofilms grown in combination with lactobacilli, was systematically higher than when grown in combination with other commensal bacterial strains. In general, biofilms comprising C. tropicalis harvested higher numbers of viable Table 1. Total number of CFU/cm 2 (Candida and bacteria) grown on silicone-rubber in an eight day time period after inoculation with a combination of C. albicans or C. tropicalis and a commensal bacterial strain or a lactobacillus strain, together with the percentage prevalence of the yeast in the final biofilm.

C. albicans combined with
Total CFUs (10 5 /cm 2 ) C. albicans (%) organisms than when C. albicans was involved, in line with the biofilm thicknesses observed using OCT and CLSM (compare Figure 3 and Table 1). In C. albicans biofilms however, growth in presence of R. dentocariosa significantly stimulated hyphal formation (see Table 2), while especially L. casei significantly suppressed morphogenic conversion of C. albicans to its hyphal form. In C. tropicalis biofilms, we did not see a significant stimulation by R. dentocariosa on hyphal formation, while L. casei reduced hyphal formation.

Discussion
The clinical life-time of silicone-rubber voice prostheses varies over extremely wide ranges within different laryngectomized patients. Sub-division of 692 failed voice prostheses into a short lifetime group (implantation-period less than 4 months) and an extended lifetime group (implantation-period over 9 months) revealed that R. dentocariosa and C. albicans and C. tropicalis were predominant strains in the short lifetime group. In the extended lifetime group R. dentocariosa was found with a fourfold Influence Bacteria on Candida Morphology in Biofilms PLOS ONE | www.plosone.org lower isolation frequency and C. albicans was found with a twofold lower isolation frequency [3]. In a separate study, it was shown that daily consumption of a fermented milk containing L. casei yielded an increased prosthesis lifetime by a factor of 3.76 as compared to a historical control in the same group of patients [13].
In the present study we show that these observations on the clinical life-time of voice prostheses cannot be explained on the basis of numbers of microorganisms found in dual-species biofilms on silicone-rubber, or for that matter: the biofilm thickness. Combinations of lactobacillus strains with Candida yield less biofilm, but with an increased prevalence of Candida as compared with combinations involving other bacterial strains. At the same time, we show that L. casei, with demonstrated favorable effects on the clinical life-time of voice prostheses [13] and adhering in full grown clinical biofilms in close association with yeast (see Figure 4), reduced the percentage hyphal formation in Candida biofilms as compared with Candida biofilms grown in absence of bacteria by a factor of three. When compared with the percentage hyphal formation in Candida biofilms grown in combination with R. dentocariosa, a bacterial strain whose presence is associated with short clinical life-times of voice prostheses [3], this factor becomes ten. Therewith we here demonstrate that in voice prosthetic biofilms, morphogenic conversions in Candida biofilms depend on the bacterial strain present, corresponding with effects on the clinical life-time of the prostheses.
The percentage hyphae in Candida biofilms grown in absence of bacteria amounts 0.7% to 3.0% (see Table 2). Other studies, in which Candida biofilms were grown in serum demonstrate much higher percentages of hyphae up to 100%, which is due to the hyphal transforming activity in serum associated with the signaling molecules prostaglandin and thromboxane [15]. Also the periods of nutritional feast and famine applied in the current study to mimic the conditions in the oropharynx more closely, may have a reducing effect on morphogenic conversions in Candida. More importantly, interactions between Candida and lactobacilli have been described to regulate Candida morphogenesis and therewith its virulence and invasiveness. Short chain fatty acids produced by different strains of lactobacilli (L. casei, Lactobacillus paracasei and Lactobacillus rhamnosus) inhibited hyphal formation, as well as culture supernatants of lactobacilli and live lactobacilli [15]. In the present study, we only find a significant reduction in C. albicans hyphal formation for L. casei with respect to Candida growth in absence of bacteria, while for C. tropicalis reductions observed are not statistically significant. For C. albicans however, the percentage hyphal formation when grown in presence of L. casei is tenfold lower (statistically significant at p,0.05) than when grown in presence of R. dentocariosa. A similar comparison exists for C. tropicalis but is not statistically significant.
Thus we conclude that R. dentocariosa stimulates morphogenic conversion in Candida toward a hyphal morphology, coinciding with short clinical life-times of voice prostheses from which R. dentocariosa was isolated [3]. In addition it is concluded that especially L. casei reduces hyphal formation in Candida, which coincides with clinical observations that the consumption of a fermented milk with L. casei extends the clinical life-time of voice prostheses in laryngectomized patients [13]. These conclusions are based on relatively small differences in hyphal formation. However, these differences were established over a time period of only eight days, whereas the above clinical observations were made over a period of several months to years. Likely, larger effects will develop over time. Moreover, a small number of hyphae may be sufficient to grow into the silicone-rubber and the close proximity of Candida and bacteria in pockets created in silicone-rubber [16] will enhance the effects of bacterial presence on morphogenic conversion of Candida. Since hyphal ingrowth is a major factor in the strong binding of initial Candida biofilms to silicone-rubber [16,17], biofilm maturation after initial adhesion does not necessarily have to involve Candida in its hyphal morphology. This is in line with the high percentages of Candida  in biofilms grown in the presence of lactobacilli, despite low conversions to the hyphal morphology (compare Tables 1 and 2). The high magnification panel on the right clearly shows the association of lactobacilli (red) and yeasts (green). Bars equal 20 mm and 5 mm for the left and right panels, respectively. Taken with permission from [14]. doi:10.1371/journal.pone.0104508.g004