Candida albicans biofilm formation in a new in vivo rat model

The human fungal pathogen Candida albicans, which can cause both superficial and systemic infections, is commonly diagnosed in biofilms, defined as structured cell communities attached to a support (Kojic & Darouiche, 2004; Ramage et al., 2006). Fungal infectious biofilms can develop on a variety of frequently implanted biomaterials, including urinary and vascular catheters, but also on joint and voice prostheses, and ocular lenses as a few examples. All these devices can serve as an ideal niche for biofilm structures, which can then become a potential systemic infection. Increased use of medical implant devices and high resistance of biofilm to antimicrobial treatment are two important factors contributing to biomaterial-related biofilm infections (Douglas, 2003; Chandra et al., 2001). In vitro biofilm assays are commonly developed on substrates such as 96-well polystyrene plates or silicone discs. Such assays have provided many insights into the finding that biofilms are less susceptible to antimicrobial agents than free-living cells (Chandra et al., 2001; Ramage et al., 2001). Models in vivo, where the biofilms are formed on catheters implanted in the central venous system of rodents, have also been developed (Schinabeck et al., 2004; Andes et al., 2004; Lazzell et al., 2009). The benefit of these models is that they also take into account the host immune system and infection site, and therefore they are indispensable for our understanding of biofilms associated with clinical devices.

Candida biofilms are formed by a basal monolayer of yeast cells attached to the substrate, which then grow as pseudohyphal and hyphal cells. The whole cell population is finally embedded in extracellular material, forming a very rigid attached sessile community (Seneviratne et al., 2008). As well as being extremely difficult to eradicate from the substrate mechanically, biofilms are highly resistant to most antifungal drugs. Echinocandins and lipid formulations of amphotericin B have, however, been shown to remain potent against fungal biofilms in both in vitro and in vivo assays (Martinez & Casadevall, 2006; Kuhn & Ghannoum, 2004; Mukherjee et al., 2009; Lazzell et al., 2009; Ferreira et al., 2009).

Genes playing a role in biofilm formation in C. albicans have been identified by genetic approaches, and they are involved in processes varying from cell-surface recognition to quorum sensing and adherence (reviewed by Nobile & Mitchell, 2006). Little is known about the gene products involved in the early attachment of the biofilm. Loss of adhesion in early-stage biofilm is not orchestrated by the loss of regulation of one particular adhesin-encoding gene, or at least not in the whole cell population (Sellam et al., 2009). However, members of the ALS family, as well as Hwp1 and Eap1 adhesins, are all implicated in the adherence of hyphal cells, and differentially contribute to biofilm formation (Zhao et al., 2006; Nobile & Mitchell, 2006; Li et al., 2007). Bcr1 transcription factor is a positive regulator of adherence as proposed by Nobile & Mitchell (2005). It governs the expression of glycophosphatidylinositol- (GPI)-anchored protein-encoding genes, in particular of ALS3, ALS1 and HWP1, to activate biofilm development and yet is not required for hyphal morphogenesis. Gene expression analyses of biofilms grown in vitro revealed a progressive regulation of gene expression, notably of the sulfur amino acid biosynthesis and salvage pathways (Murillo et al., 2005; García-Sánchez et al., 2004). Transcript profiling data from in vitro biofilm resembled poorly the recent findings obtained from in vivo transcriptional data (Nett et al., 2009) with the exception of a few but important genes, including ALS1 and ALS4 (adhesins), ECE1 and SAP5 (cell-wall metabolism), MET3, MET10, CYS3 and CYS4 (sulfur metabolism), and CGT1, ICL1, MLS1, PCK1 and PDK1 (carbohydrate and general metabolism).

The present study describes a novel in vivo biofilm model, which can be used for C. albicans. Small pieces of catheter challenged with Candida cells were implanted under the skin of rats and mature biofilm structures were obtained after 2 days. Our results show that the avascular subcutaneous model provides a useful and simple tool to study Candida biofilms in vivo.

Animals.
The animals used for the in vivo experiments were 200 g specific-pathogen-free female Sprague–Dawley rats (Janvier). All animals were given standard ad libitum diet and were immunosuppressed with 1 mg dexamethasone (Organon) per litre of their drinking water up to 72 h before and during the whole experimental procedure (unless otherwise stated). Tetracycline (1 g l^–1) was added to the water to minimize bacterial infections. All animal experiments were maintained in accordance with European regulations regarding the protection and well-being of laboratory animals and were approved by the animal ethical committee of the Katholieke Universiteit Leuven.

Strains and media.
C. albicans strains used in this study were wild-type strains SC5314 (Gillum et al., 1984), and DAY185 (ura3Δ : : λimm434/ura3Δ : : λimm434 HIS1 : : his1 : : hisG/his1 : : hisG ARG4 : URA3 : : arg4 : : hisG/arg4 : : hisG; Davis et al., 2000). All other strains were knockout strains: CAYF178U (ura3Δ : : λimm434 : : URA3-IRO1/ura3Δ : : λimm434 arg4 : : hisG/arg4 : : hisG his1 : : hisG/his1 : : hisG als3 : : ARG4/als3 : : HIS1; Nobile et al., 2006), CJN702 (ura3Δ : : λimm434/ura3Δ : : λimm434 arg4 : : hisG/arg4 : : hisG his1 : : hisG/his1 : : hisG : : pHIS1 bcr1 : : ARG4/bcr1 : : URA3; Nobile & Mitchell, 2005) and HLC54 (ura3Δ : : λimm434/ura3Δ : : λimm434 cph1 : : hisG/cph1 : : hisG efg1 : : hisG/efg1 : : hisG-URA3-hisG; Lo et al., 1997). All strains were grown on YPD medium (1 % yeast extract, 2 % Bacto-peptone and 2 % glucose) at 37 °C.

In vitro biofilm assays.
Strains were grown overnight on YPD plates at 37 °C, washed and resuspended in PBS. Polyurethane substrates were incubated overnight with bovine serum at 37 °C (F7524, Sigma). A suspension of 5x10⁴ cells ml^–1 was prepared in RPMI medium (Applichem; RPMI 1640 with L-glutamine and without sodium carbonate buffered with MOPS, pH 7). The cell suspension was added to the biomaterials and incubated for 90 min at 37 °C for adhesion. Devices were washed twice with PBS and submerged in fresh RPMI medium for 2–6 days at 37 °C.

In vivo biofilm model.
C. albicans cells were grown overnight at 37 °C on YPD plates, washed and resuspended in PBS. A suspension of 5x10⁴ cells ml^–1 was prepared in RPMI medium by counting. Polyurethane triple-lumen intravenous catheters (2.4 mm diameter) cut into segments of 1 cm (Arrow International) were incubated overnight in bovine serum at 37 °C. Serum-coated catheters were incubated for 90 min at 37 °C in 1 ml cell suspension. After incubation, catheters were washed twice with PBS before being implanted under the skin of rats as described previously (Van Wijngaerden et al., 1999; Massonet et al., 2006). Anaesthesia was performed by a short inhalation period of enflurane gas (Alyrane, Pharmacia). Rats were kept asleep during the implantation procedure by a gaseous mix of enflurane (20 %) and oxygen (80 %). The lower back of the rat was shaved and disinfected with 0.5 % chlorhexidine in 70 % alcohol. A 10 mm incision was made longitudinally and the subcutis was carefully dissected to create three subcutaneous tunnels. Up to ten catheter fragments were implanted. The incision was closed with surgical staples (Precise), and disinfected with 0.5 % chlorhexidine in 70 % alcohol. For catheter explantation, rats were euthanized by CO₂ inhalation. The skin was disinfected, and catheter fragments were removed from under the subcutaneous tissue and washed twice with PBS.

Biomass quantification.
In vitro substrates and explanted catheters were sonicated for 10 min at 40 000 Hz in a water bath sonicator (Branson 2210) and vortexed for 30 s in PBS. Original samples and a 1 : 10 dilution were plated on YPD agar in duplicate. C.f.u. were counted after 2 days at 37 °C. They are represented as mean of the counts of both dilutions per substrate or per six substrates (specified in figure legends accordingly).

Fluorescence microscopy.
Catheters (longitudinally cut) were incubated in PBS with 50 µg calcofluor white (Sigma) ml^–1 for 10 min and observed with a Zeiss Axioplan 2 fluorescence microscope. Images were acquired by a Zeiss Axiocam HRm camera using Axiovision 3.0 software.

Scanning electron microscopy.
Catheters (longitudinally cut) were fixed in 3 % glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) for at least 2 h. Catheters were removed from the fixation solution and dried overnight. Mounted samples were sputter-coated with gold and viewed in a XL30 ESEM FEG scanning electron microscope (Philips).

Confocal laser scanning microscopy.
Fixed catheters were incubated with 50 µg concanavalin A ml^–1 for 1 h at 37 °C (concanavalin A, Alexa Fluor488 conjugate, Invitrogen). Confocal images were acquired and analysed with an LSM510/ConfoCor2 system (Carl Zeiss). The argon laser (6 A, acousto-optical tunable filter adjusted to 50 %) provided the excitation line of 488 nm (for concanavalin A-Alexa488 fluorescence). The excitation light was reflected by a dichroic mirror (HFT 488) and focused through a Plan-NeoFluar 20x NA0.5 objective. The fluorescence emission light passed through a 505 nm longpass filter and a 1-Airy unit pinhole.

Quantitative real-time PCR.
Total RNA from biofilms was extracted using the RiboPure-Yeast kit (# AM1926, Ambion). Up to 10 catheter pieces were pooled to obtain sufficient fungal biomass in vitro and in vivo. Cells growing in the RPMI medium surrounding the catheters in vitro were collected, and are referred to as the planktonic sample. Complementary DNA was prepared from DNase-treated RNA samples with the Reverse Transcription kit A3500 (Promega). Quantitative PCR was performed on a StepOnePlus real-time PCR system (Applied Biosystems) using the Kapa SYBR Fast kit (Kapabiosystems) according to the manufacturer's instructions. The fold regulation of each target gene was calculated using the comparative C_t method, using ACT1 C_t to normalize the data, and the planktonic cell expression data as the reference sample to determine the ΔΔC_t values. The following primers were used: ACT1-Fw (5'-CTCTTCTGGTAGAACCACCGGTAT-3'), ACT1-Rev (5'-TAAAGAGAAACCAGCGTAAATTGGA-3'), EAP1-Fw (5'-CTGCTCACTCAACTTCAATTGTCG-3'), EAP1-Rev (5'-GAACACATCCACCTTCGGGA-3'), HWP1-Fw (5'-TCAGTTCCACT CATGCAACCA-3'), HWP1-Rev (5'-AATCTCATGTTGTTACCAGCACCTT-3'). ALS3 and ALS1 primers were designed as described by Green et al. (2005).

Wild-type C. albicans biofilm formation in vivo in a subcutaneous rat model system
The present in vivo Candida biofilm model is based on the implantation of 1 cm pieces of triple lumen Candida-infected polyurethane catheters underneath the skin of rats, and was adapted from a model described by Van Wijngaerden et al. (1999) for Staphylococcus epidermidis biofilm formation. Bacterial biofilms developed in this subcutaneous in vivo model grew better in Fisher rats (ex-germ-free Fisher rat strain, inbred line raised at the Katholieke Universiteit Leuven) than in other types of rats tested. Although we could obtain Candida biofilm growth in Fisher rats (data not shown), we adapted the model to commercially available and commonly used Sprague–Dawley rats (much more docile animals).

In the initial development of the subcutaneous model, immunocompetent female rats were used. However, the yield of cell biomass recovered from the polyurethane fragments implanted was very variable (Fig. 1), ranging from very few cells to thousands of cells. The inflammatory response following surgery, together with the low cell density within the catheter at time of implantation, could contribute to the inhibition of biofilm development in immunocompetent animals. As it was not possible to infect catheters after their implantation, at a time where the inflammatory response would be reduced, it seemed adequate to impair the animal's immune response before surgery. It appeared that indeed the reproducibility and the yield of the biofilms formed increased greatly when animals were partly immunosuppressed, with a mean of 3–4 log₁₀ units per device after 6 days (Fig. 1). We were therefore able to reproducibly obtain up to 5x10⁴ viable cells from ten biofilms grown in one animal when treated with the anti-inflammatory glucocorticoid dexamethasone.

(9K): Fig. 1. Increased reproducibility and higher biomass recovered from biofilms developed in dexamethasone-treated animals. Dexamethasone was added to the drinking water for the treated group at a concentration of 1 mg l–1 (•, dexamethasone-treated animals; , non-treated animals). C.f.u. were obtained from six implanted devices, which were extracted from two animals for each of the conditions. In the non-treated group, two catheters had very low cell count and are shown as one overlapping data point. Biofilms were grown for 6 days.

The avascular location of the catheters eased the experimental animal procedure, yet still allowing the generation of wild-type biofilms as visualized by scanning electron microscopy (Fig. 2a). The dense populations of hyphal cells seldom covered the whole surface of the catheter; instead they appeared in patches along the length of the catheter. To differentiate between simple cell growth and biofilm formation within the catheters implanted in the absence of vascular flow, catheters were challenged with efg1Δ/efg1Δ cph1Δ/cph1Δ, a strain well known for its deficiency in hyphal switch and biofilm formation (Ramage et al., 2002). As expected, this strain failed to form biofilm, illustrated by fluorescence microscopy shown in Fig. 2(b). Scattered yeast cells were observed instead of the structured biofilm seen in catheters challenged with wild-type cells.

(129K): Fig. 2. Scanning electron (a) and fluorescence micrographs (b) of C. albicans biofilms formed in vivo inside catheters implanted in the subcutaneous model. Biofilms were grown for 6 days, and catheters were cut longitudinally and fixed before microscopy. The left panels show biofilms obtained with strain SC5314, and the right panels show catheters challenged with strain efg1Δ/efg1Δ cph1Δ/cph1Δ. In (a) magnifications are indicated to the left of the top panels (scale bars, 1 mm and 500 µm for SC5314 and HLC54 respectively), and of the bottom panel (bars, 20 µm). In (b), biofilms were stained with calcofluor, which illustrated the three-dimensional hyphal growth of the wild-type (indicated by the indistinct appearance of the cells towards the lumen side), contrasting with the single layer of yeast cells observed with the mutant.

In addition to the immune status of the host mentioned above, another important factor influencing the reproducibility of biofilm development in the subcutaneous model was the cell density at time of implantation. The size of fungal inoculum during adhesion was limited to 5x10⁴ cells, which resulted in less than 5x10² cells to adhere (the remaining cells being washed off during the experimental procedure before implantation). This was done to avoid clutter of the cells within the lumen of the catheters before implantation, and to begin the biofilm development from a cell number mimicking the minimal threshold of recognized infection used in clinical diagnostic guidelines (Mermel et al., 2001).

Kinetics of biofilm development in the subcutaneous model
Fully mature biofilms were produced after 24 h, and were fully grown by 72 h in the central venous catheter (CVC) model described by Andes et al. (2004). To gain insights into the kinetics of biofilm growth and cell morphology within the biofilm in the subcutaneous model, c.f.u. and scanning electron microscopy images were collected over a period of 6 days. The data from in vivo biofilms were compared to biofilms developed on a similar substrate in vitro, as shown in Fig. 3. After the adhesion phase of 90 min (time point zero on Fig. 3a), most cells were forming germ tubes and a few hundreds of cells were attached to the devices (data not shown).

(79K): Fig. 3. Kinetics and architecture of the biofilms formed on polyurethane devices in vitro and in vivo. Serum-treated substrates were challenged with SC5314, and biofilms were grown in RPMI at 37 °C in vitro, and in the subcutaneous model in vivo for up to 6 days. C.f.u. counts were recovered from biofilms at five time points (a). Six devices were collected in vitro (black columns) or explanted in vivo (white columns) from two independent biological experiments at each time. (b, c) Scanning electron micrographs of in vitro biofilms (b) and in vivo biofilms (c). Images were captured after 2 days (left panel) and 6 days (right panel). Magnifications are indicated on the left side of each panel (scale bars, 20 µm). Mature biofilms were observed after 2 days in vitro and in vivo as shown by the presence of mainly hyphal cells and of extracellular matrix (*).

In vitro biofilms in RPMI medium proliferated over the first 24 h. They remained stable and viable for up to 6 days. They consisted of a mixed population of yeast and hyphal cells as shown in Fig. 3(b). The extracellular matrix, which was not measured, was clearly visualized after 6 days in vitro. In the subcutaneous model, biofilms grew more slowly than in vitro. The number of cells significantly decreased in the first few hours following implantation of the catheters, as less biomass was recovered after 5 h than before implantation (Fig. 3a). The first hours after implantation seemed crucial to the biofilm establishment and determined the persistence of the biofilm at the host site. In accordance with this, the 24 h time point of in vivo biofilm development always showed the highest variation in cell number (variation between catheter pieces in the same animal and between animals). This high variability of the 24 h time point was confirmed by microscopy, revealing that after 24 h and from the same animal some infected devices contained almost no biofilm, most showed patches of biofilms and a few were already covered with mature biofilms (data not shown). In vivo biofilms grew by almost 1 log₁₀ unit between 24 and 48 h, and then remained stable for up to 6 days (longer periods of time were not studied).

Andes et al., (2004) described that biofilms grew more slowly and less in the CVC model than in the kidneys, a distant site of infection. Dissemination to other organs, including kidneys and lungs, was never observed in the subcutaneous model (data not shown). In vivo biofilms in the subcutaneous model were composed of a dense network of hyphal cells (Fig. 3c), similarly to biofilms in vitro. Previous data illustrated the lower production of extracellular matrix in biofilms grown in static conditions than those grown in constant liquid flow in vitro (Hawser et al., 1998). Importantly, extracellular matrix was present in biofilms in vivo, clearly visible after 2 days as shown in Fig. 3(c), despite the absence of strong biological flow.

Biofilm formation in the subcutaneous model requires Bcr1 but not Als3
To further validate the present in vivo model, two very well-characterized mutants, namely bcr1Δ/bcr1Δ and als3Δ/als3Δ, were tested for their ability to make biofilm structures. Previous studies showed that the Bcr1 transcription factor plays an important role in biofilm formation and that the surface protein Als3 is one of the main targets of Bcr1. Deletion of BCR1 resulted in a strain that was unable to form biofilm in both in vitro and in vivo conditions (Nobile & Mitchell, 2005). Deletion of ALS3 was reported to have a severe suppressing effect on biofilm formation in vitro but not in vivo (Nobile et al., 2006). It therefore seemed important to establish the ability of both bcr1Δ/bcr1Δ and als3Δ/als3Δ to form biofilm in the subcutaneous model. Results in Fig. 4(a) supported those obtained previously in the CVC model. Low cell counts illustrated that bcr1Δ/bcr1Δ failed to form biofilm in the polyurethane pieces of catheter placed in 24-well plates in RPMI in vitro or subcutaneously in rats in vivo, whereas als3Δ/als3Δ was able to proliferate as an attached structure in vivo but not in vitro. Indeed, als3Δ/als3Δ produced up to six times more biofilm in vivo than it did in vitro. It is not clear yet how als3Δ/als3Δ can regain the ability to form biofilm in vivo, but it underlines the influence of the host–pathogen interaction. The lack of biofilm observed in bcr1Δ/bcr1Δ mainly resided in the function of Bcr1 in regulating adherence during biofilm development, notably in the hyphal layers of the biofilm (Nobile et al., 2006), as indicated in Fig. 4(b). After the first 90 min of adherence, the bcr1 mutant showed much less adherence capability than the DAY185 and als3 strains. Both SC5314 and DAY185 behaved similarly in adhering and forming biofilm in vitro and in vivo (Fig. 4b, and data not shown).

(12K): Fig. 4. Biofilm and adherence properties of als3Δ/als3Δ and bcr1Δ/bcr1Δ strains. (a) Biomass recovered from catheters is shown as the mean c.f.u. per device. Biofilms were grown for 6 days in RPMI () or in the subcutaneous model (). Standard deviations were calculated from six challenged devices for each strain in vitro and from two animals in vivo. Wild-type was DAY185. (b) C.f.u. counts after 90 min of adhesion to catheters. Substrates were incubated in RPMI at 37 °C, and challenged with 5x104 cells ml–1. Substrates were washed before counting. On average, 2 log10 units were obtained with wild-type strains.

Confocal imaging was employed to assess the three-dimensional structure of the biofilms formed in vivo. It was somewhat difficult to obtain reliable and reproducible data regarding the depth of the biofilms because of the technical procedure followed. Catheters were cut through the largest lumen into two parts, consequently damaging the biofilms formed inside. However, some of the remaining biofilms observed inside the cut catheters retrieved from the animals could be visualized as shown in Fig. 5. Sequential capture of biofilm sections from the bottom layer attached to the inside wall of the catheter towards the lumen illustrated the layered structure of the wild-type biofilms (Fig. 5a). Yeast and hyphal cells formed the adherent layer, and sparsely covered the catheter interior surface. The three-dimensional sections through the wild-type biofilm covered 140 µm, which represented the biofilm thickness. They also illustrated the dense network of hyphal cells that became visible, together with the appearance of the matrix (represented by the out-of-focus areas on the confocal images). In the final section, a lot of the cell biomass was not clearly visible because of the matrix covering the biofilm. Biofilms formed by als3Δ/als3Δ were also multilayered structures, as shown by the three-dimensional section in Fig. 5(b). The cross-sections through the biofilm indicated the presence of more than one layer of cells, together with the presence of extracellular matrix. These results indicated that the biofilms formed in the subcutaneous model in vivo were multilayered structures, as previously described, and therefore showed the reliability of such a model for studying biofilm formation.

(67K): Fig. 5. Confocal laser scanning visualization of in vivo biofilms developed in the subcutaneous model. (a) Sequential imaging of wild-type biofilm. Successive z-images were captured through the biofilm structure from the basal layer attached to the inside wall of the catheter to the top layers of the biofilm towards the lumen. (b) Biofilm obtained with strain als3Δ/als3Δ and captured as a three-dimensional view (z-view). The vertical and horizontal z-stacks were composed of 245 sections of 1 µm each. Biofilms were stained with concanavalin A conjugate before imaging. According to the fluorescence signal through the z-sections, biofilms shown here of SC5314 and als3Δ/als3Δ had a thickness of 140 µm and 135 µm respectively. Yeast and hyphal forms are highlighted, together with the extracellular matrix.

Biofilm-dependent expression of adhesin genes
Adherence to the substrate and cell–cell adhesion are processes fundamental to biofilm formation. Despite the fact that adhesins are not all essential to biofilm formation in vitro or in vivo, their transcriptional expression is regulated during the development of wild-type biofilms as described by García-Sánchez et al. (2004) in vitro and by Nett et al. (2009) in the CVC model.

The expression of four genes encoding cell-wall proteins, ALS1, ALS3, HWP1 and EAP1, was quantified in biofilms developed in the subcutaneous model and in catheters in vitro (Fig. 6). The expression of these genes in sessile cell populations in both biofilm models was compared to their respective expression in free-living cells. As previously described, EAP1 was found more expressed in biofilms than in planktonic cells (Li et al., 2007). In contrast, the hyphal-specific gene HWP1 was not differentially regulated, as expected, as the planktonic cells were mostly hyphal cells. The increased expression of ALS3 in wild-type biofilms grown in vivo but not in vitro is paradoxical to the fact that als3Δ/als3Δ can form biofilms in vivo, but not in vitro. A global expression profile of als3Δ/als3Δ grown in vivo would likely show an alternative biofilm profile, with changes in the expression of genes involved in adherence, potentially other Bcr1 targets as suggested by Nobile et al. (2006). In contrast to ALS3, ALS1 expression was upregulated in both biofilm conditions. Although it would be inappropriate to directly compare expression data obtained by others to those described here, because of the great differences in growth conditions, including the host–pathogen interactions at the infection sites, it is noteworthy that ALS1 was identified previously in microarray data as greatly upregulated in biofilms in vitro and in vivo (García-Sánchez et al., 2004; Nett et al., 2009).

(11K): Fig. 6. Differential expression of adhesin genes in biofilms. Fold changes of expression were calculated using the comparative Ct method. Each gene was normalized to ACT1. For each target gene, the data are shown as fold regulation in biofilm in vitro (black columns) or in vivo (white columns) compared to planktonic cells (free-living cells in the catheter-surrounding media in vitro). Error bars were calculated from two independent biological replicates.

In this study, we have demonstrated the reliable use of a novel in vivo C. albicans biofilm rat model, based on the subcutaneous implantation of small pieces of biomaterials infected with Candida cells. Biofilms in this infection model persisted for up to 9 days (data not shown; longer periods were not examined), permitting the study of biofilm evolution and maintenance in vivo. The increased yield and reproducibility of the biofilms following immunosuppressive treatment of the rats was striking but perhaps not surprising if one considers the immune response of the host to the implanted devices themselves rather than to the fungal biomass. The host response to implanted biomaterials, also known as the foreign body response, is responsible for the promotion of immune modulators but also enzymes, which can be detrimental to the catheters (Higgins et al., 2009). Macrophages, which are the dominant cell type in the foreign body response (Anderson et al., 2008), have a key role in the innate immunity against C. albicans (Vasquez-Torres & Balish, 1997). Although fungal cells were never recovered from kidneys, nor from blood of the implanted rats, the serum of the animals did show the presence of antibodies against C. albicans protein extracts (data not shown), suggesting the recognition and/or killing of the foreign organism by the immune system without detectable dissemination, even in the dexamethasone-treated animals. In the early hours of biofilm establishment in vivo, cell numbers decreased. It is possible that cells may be detached more easily during that period because of their encounter with the host. These cells would then be removed during the experimental procedure following removal of the catheters, resulting in lower biomass at these time points. Another explanation is that a certain proportion of the cell populations (up to a fifth) does not survive inside the implanted material, because of the growing conditions inside the host. Debris and contents of C. albicans cells could therefore promote antibody production. Further analyses are needed to determine the actual immune state and response to the fungal burden as it develops inside the catheters.

C. albicans growth on catheters consistently yielded between 3 and 4 log₁₀ c.f.u. in each implanted device. It is noteworthy that despite the fact that most cells from wild-type biofilms were hyphal cells, c.f.u. counting after sonication and vortexing did correlate with genomic copy numbers (data obtained by a quantitative genomic PCR approach; data not shown). For a genuine comparison, biofilms formed in the rat CVC model generated more than 1 log₁₀ unit more biomass than in the subcutaneous model (Andes et al., 2004), with indwelling devices twice the size and a higher cell density at time of infection. Despite a reduction in size of the biofilms obtained in the subcutaneous system, we could demonstrate that mature multilayered wild-type biofilm structures established and persisted inside the host. Another important result was obtained with bcr1Δ/bcr1Δ and als3Δ/als3Δ strains. Deletion of BCR1 did impair the ability of the cells to form a biofilm, whereas ALS3 was not essential to the process in vivo, as previously demonstrated (Nobile et al., 2006).

Cell morphogenesis and hyphal development are important factors contributing to biofilm formation in C. albicans, and yet not essential in other species such as Candida glabrata (Paulitsch et al., 2009). Expression of hyphal-specific genes, such as the genes encoding adhesins, is tightly regulated by a complex network of signal-transduction pathways (Biswas et al., 2007). Interestingly, it has recently been shown that Tor1 kinase, a protein central to cellular nutrient responses, also plays a role in regulating cellular adhesion in nutrient-limited conditions (Bastidas et al., 2009). A genome-wide transcriptional profile would certainly give clues into the potential role of Tor1 in regulating gene expression in biofilms grown in the subcutaneous model. Increased expression of ALS3 in the present model could be a characteristic of the model itself, despite the fact that ALS3 is not essential to biofilm formation in that same model. Expression of BCR1 was also quantified (data not shown) and was upregulated in biofilms after 6 days growth (earlier time points were not studied). Recent microarray data obtained from biofilms grown in the rat CVC model did not reveal an increased expression of either BCR1 or ALS3 when compared to planktonic cells, at least within the first 24 h of growth (Nett et al., 2009). Among the other genes identified as being regulated during the whole process of biofilm formation, the genes involved in sulfur metabolism are of particular interest (García-Sánchez et al., 2004; Murillo et al., 2005). The regulation of some of the MET genes in biofilm did not correlate with the classical auxotrophic requirement of the growth medium, but seemed to be one of the characteristics of cells growing as a biofilm structure. MET3 was also found to be upregulated in biofilms grown in vitro in polyurethane catheters and in vivo in the subcutaneous model (data not shown). The observed invariance of transcriptome between multiple in vitro biofilm assays (García-Sánchez et al., 2004) remains to be established for the in vivo models.

The host sites at the subcutaneous location and in the central venous system presumably differ greatly. Whereas in the latter, biofilms are subjected to the blood flow, biofilms developing in catheters implanted under the skin of rats are not exposed to a physiological flow. The access to nutrients will also differ, and be greater in the blood. The subcutaneous model is somewhat more related to biofilm infections that develop in joint prostheses and voice prostheses for example, and may reflect better these host infection sites, in term of environmental conditions and nutrient supplies.

One important feature of biofilm populations is their high resistance to antifungal agents. However, a recent study by Lazzell et al. (2009) illustrated the biofilm's susceptibility to high doses of caspofungin in vivo by a lock therapy approach. In the subcutaneous model, pre-treatment of the polyurethane pieces of catheters with caspofungin before challenge with C. albicans cells and implantation eliminated the fungal biomass when high doses were used, and reduced the biofilm biomass by 90 % when lower doses were employed (data not shown). The efficiency of treatments with antifungal agents relies on multiple factors including drug permeability and diffusion through the tissues, as there is no direct access of the drug to the fungal burden in the present model. However, such an approach promises to be a great challenge in the subcutaneous model.

To conclude, we have developed a simple tool for the analysis of in vivo biofilm formation by C. albicans. The results obtained clearly show similar results to those obtained with the more sophisticated central venous model when analysing two mutants previously shown to be important for biofilm formation. For future applications, the model can be used to evaluate the biocompatibility of implanted devices (reduction of the foreign body response), the preventive treatment efficacy of new antimicrobial reagents and the potency of chemotherapeutic agents to eliminate biofilms.

We thank Aaron Mitchell from the University of Pittsburg for the als3Δ/als3Δ, bcr1Δ/bcr1Δ and DAY185 strains. We are grateful to Rudy de Vos from the Department of Metallurgy and Materials Engineering at K.U. Leuven for SEM imaging and Valerie Pintens and Caroline Massonet from the Experimental Laboratory of Microbiology at K. U. Leuven for their help with initial animal experiments. We also thank Deborah Seys for technical assistance with animal experiments, and Nico Vangoethem for his assistance in the preparation of the figures. This work was supported by the FWO (G.0242.04 and W. O. G.) and by the Marie Curie Research Training Network (MCRTN-CT-2004-512481). M. R. received a grant from the FWO (GP.011.06N).

Edited by: K. Kuchler

Footnotes

^,†, Soa Kucharíková^{1,2,3 ,†}, Hélène Tournu^1,2 †These authors contributed equally to this work.