GENETIC CONTROL OF CANDIDA ALBICANS BIOFILM DEVELOPMENT

Preface

Candida species cause frequent infections due to their ability to form biofilms – surface-associated microbial communities – primarily on implanted medical devices. Increasingly, mechanistic studies have identified the gene products that participate directly in Candida albicans biofilm formation, as well as the regulatory circuitry and networks that control their expression and activity. These studies have revealed new mechanisms and signals that govern C. albicans biofilm formation and associated drug resistance, thus providing biological insight and therapeutic foresight.

Introduction

The medically relevant Candida species ¹ are mainly commensal fungi that reside on mucosal surfaces and in the gastrointestinal (GI) and genitourinary (GU) tracts. Although these organisms are generally benign, they can cause infection if immune function is impaired or if an environmental niche becomes available ². Many Candida infections arise as a result of the organism’s ability to grow as a biofilm on implanted medical devices ^3–6. The use of these devices, such as venous catheters, urinary catheters, and artificial joints, is routine, with well over 10 million recipients per year ⁷. Device-associated Candida infections have mortality rates as high as 30% ^{7, 8}, and the annual cost of antifungal therapies in the United States alone is estimated at $2.6 billion ⁹. Like the biofilms of bacterial pathogens, Candida biofilms are resistant to many antimicrobial agents, so treatment may require surgical removal and later replacement of the infected device ^{5, 7}. Here, we review Candida biofilm formation with a focus on Candida albicans, the most frequently isolated Candida pathogen ¹⁰.

Overview of C. albicans Biofilm Formation

C. albicans biofilms comprise mainly two kinds of cells: small oval yeast-form cells (also called blastospores), and long tubular hyphal cells. Biofilms grown in vitro often have a foundation of yeast cells, from which a hyphal layer emanates (Fig. 1A) ⁵. Extracellular matrix material is clearly evident as well, bound to both yeast and hyphal cells. It is typically interspersed throughout the biofilm, though it is apparent primarily at the top of this sample. Biofilms from in vivo catheter infection models appear more complex, with yeast and hyphae interspersed (Fig. 1B) ¹¹. Genetic analysis indicates that both yeast cells and hyphae are critical for biofilm formation, which suggests that each cell type has unique roles in the process ⁵.

Figure 1. C. albicans biofilm structure in vitro and in vivo.

A) Scanning electron micrograph (SEM) of an in vitro biofilm. The biofilm sample was sliced to reveal three layers in a cross-sectional view (J. S. Finkel, J. Suhan, and A. P. Mitchell, unpublished results). The basal layer includes primarily yeast cells, as evident in the lower enlarged inset. The central layer is mainly hyphae. The upper layer has yeast cells budding from the hyphae. The upper enlarged inset shows extracellular matrix material, which appears fibrous in this preparation. B) SEM of an in vivo biofilm from the rat catheter model ¹¹. Yeast cells, hyphae, and some pseudohyphal cells are evident, along with extracellular matrix material. (This image was provided by J. Nett and D. Andes.)

In vitro experiments allow C. albicans biofilm formation to be viewed as a series of sequential steps (Fig. 2) ^{5, 12}. Biofilm formation begins with adherence of yeast cells to a substrate (adherence step) (Fig. 2). Soon afterwards, the yeast cells proliferate across the surface and produce elongated projections that grow into filamentous forms, including hyphae or and pseudohyphae (initiation step). Extracellular matrix accumulates as the biofilm matures, and high-level drug resistance is also acquired (maturation step). Finally, non-adherent yeast cells are released from the biofilm into the surrounding medium (dispersal step) (Fig. 2). While these steps may occur concurrently rather than sequentially during natural biofilm formation in vivo, they provide a useful framework to guide mechanistic analysis.

Figure 2. Gene function in biofilm formation.

In the adherence step, yeast-form cells adhere to the substrate. In the initiation step, the cells propagate to form microcolonies and germ tubes form to yield hyphae. In the maturation step, the biofilm biomass expands, the extracellular matrix accumulates and drug resistance increases. In the dispersal step, yeast-form cells are released to colonize the surrounding environment. The upper half of the diagram depicts several known pathway relationships. The bottom half includes additional genes that function in a specific step, but may not be connected to a known pathway. For simplicity, some known pathway relationships have been omitted. A few genes are presented more than once if they have roles in more then one step of biofilm formation. Arrows represent positive relationships; T-shaped bars represent negative relationships. A “+” indicates that the upstream gene/signal stimulates expression of the downstream target; a “−” indicates that the upstream gene/signal inhibits expression of the downstream target. Dashed lines indicate repression by an indirect mechanism.

Simple Inferences from Biofilm Genetics

Recent strides in expression profiling and genetic manipulation have driven our understanding of the regulatory pathways and mechanisms that govern biofilm formation and biofilm-based drug resistance. In addition, such analyses have pointed to an intriguing relationship between biofilm formation and mating. Of course, genes can have net positive or negative roles in biofilm formation, based on known mutant phenotypes. This distinction is useful in thinking about gene relationships because, for example, a negative gene product can act through inhibition of a positive gene product. On the other hand, it is worth bearing in mind that “dispersal” is an unraveling of the steps in biofilm formation. Therefore, for example, the negative function of Ywp1 in the adherence step that leads to biofilm formation might go hand in hand with a positive function in biofilm dispersal.

The genes that govern C. albicans biofilm formation (Table 1) fit into several broad functional categories. Many of these genes are required for production of hyphae (filamentation). Some of the first C. albicans biofilm genetic studies indicated that hyphae are required for stable biofilm formation ^{13, 14}. In addition, several biofilm genes are involved in the response to the quorum-sensing molecule farnesol ^{15, 16}. Farnesol is an inhibitor of filamentation ^{15, 16} and, as one would thus expect, inhibits biofilm formation in vitro ¹⁷. In fact, several additional quorum sensing molecules accumulate in mature biofilms (see Box 1), and addition of such molecules to biofilm cultures in vitro suggests that they may promote biofilm dispersal ^18–20.

Table 1.

Molecular function of gene producta	Role of gene product	Gene	Reference
Transcription Factor	Positive	ACE2c, BCR1, CPH1, CZF1c, EFG1c, FLO8c, GCN4, TEC1c, UME6c, NRG1c	14, 27, 28, 90, 98, 99, 118–122
	Negative	ZAP1	29
Cell Wall Related Protein	Positive	ALS1, ALS2c, ALS3, ALS4, ALS5, ALS7, ALS9, CSA1, EAP1, FKS1, HWP1, HWP2, OCH1, PGA1, PGA10c, PMT1c, PMT2c, PMT4, PMT6, RBT1, RBT5, SUN41c	22, 23, 28, 38, 56, 67, 71, 73–76, 97, 123–125
	Negative	YWP1	126
Alcohol dehydrogenase	Positive	ADH5	127
	Negative	ADH1, CSH1, IFD6	29, 30
Protein Kinase	Positive	CBK1c, GIN4c, IRE1c, MKC1, YAK1c	64, 78, 128
	Negative	CHK1, TOR1,	129, 130
Drug efflux pump	Positive	CDR1, CDR2, MDR1	89
Gluco-amylase	Positive	GCA1, GCA2	29
Otherb	Positive	CAT2, ECE1, KEM1c, MDS3c, NDH51, NUP85c, PBR1, PES1, PDX1, RIX7, SUV3c, VAM3c, VPS1c	28, 68, 98, 131–136

^aMolecular functions have been inferred from protein sequence homology in most cases.

^bOther refers to gene product functions that does not fit into any of the above categories. For specific functions refer to supplemental table 1.

^cIndicates a regulator of filamentation

Box 1. Quorum sensing.

Quorum sensing phenomena are those in which microbial behaviors or responses are governed by cell density. Such community behaviors are generally determined by secreted signaling molecules, whose accumulation is a measure of cell density ¹⁰⁰. Quorum sensing plays a pivotal role in biofilms of all kinds ^{101, 102}. The most well studied quorum sensing molecule in C. albicans is E,E-farnesol, an inhibitor of hyphal formation. Exogenous farnesol inhibits biofilm formation if provided early, during the time of adherence ^{17, 18}. The limited biofilms that form in the presence of farnesol comprise mainly yeast and pseudohyphal cells, rather than hyphae. Farnesol accumulates in supernatants of mature biofilms ²⁰, where stimulation of yeast cell production may promote biofilm dispersal. Tyrosol, an alcohol derived from tyrosine, has an activity opposite that of farnesol: it stimulates hyphal formation. Exogenous tyrosol addition does not have a measurable effect on overall biofilm formation, but can partially overcome biofilm inhibition by exogenous farnesol ¹⁸. Tyrosol also accumulates in mature biofilm supernatants ^{18, 20}, and the overall inhibition of hyphal formation by such supernatants ^{17, 18, 20} seems to reflect the dominant activity of farnesol ¹⁸. Several additional small molecules are detectable in biofilm supernatants, including phenylethyl alcohol, dodecanol, and nerolidol ²⁰. Each isolated molecule can inhibit hyphal formation, and thus all may aid in biofilm dispersal by promoting yeast cell formation. It will be of interest to block synthesis of or response to individual molecules in order to assess their biological functions, and to test their roles in biofilms formed in vivo.

Several noteworthy classes of gene products govern biofilm properties. Many biofilm genes encode known or predicted cell wall proteins. These proteins are of special interest because they may play a direct role in cell-substrate or cell-cell adherence. Indeed, heterologous expression studies indicate that Eap1, Hwp1, Als1, and Als3 have such roles ^21–23. Surface proteins are of further interest as accessible therapeutic targets. Finally, it has become increasingly evident that cell heterogeneity is a critical feature of biofilms ²⁴. This attribute is obvious from the different cell types evident in C. albicans biofilms (Fig. 1), and fungal cell wall protein genes are subject to both genetic and epigenetic mechanisms that further contribute to cell heterogeneity ^{25, 26}.

Many of the C. albicans genes involved in biofilm formation encode predicted transcription factors or protein kinases. These regulatory proteins must act indirectly to control biofilm properties, but can be informative indicators of the internal and external signals that influence biofilm development. For example, the transcription factor Bcr1 is required for biofilm formation and is upregulated in hyphae, thus suggesting that Bcr1-dependent gene products may be the hyphal components that are required for biofilm formation ^{27, 28}. Similarly, the zinc-responsive transcription factor Zap1 is a regulator of extracellular matrix accumulation, suggesting that ambient zinc levels may alter matrix levels ²⁹. The signals that influence the activity of many of the other biofilm regulators listed in Table 1 are not well understood, thus presenting an opportunity for future study.

Several alcohol dehydrogenase and aryl-alcohol dehydrogenase genes have an impact on biofilm formation. The fact that both positive and negative roles for these genes have been deduced may suggest that substrate specificity is critical for their biological function ^{29, 30}. Although it is possible that their substrates and products act primarily through effects on intermediary metabolism, we note that the aryl alcohol dehydrogenases in particular have been implicated in synthesis of amino acid-derived alcohols that may function in quorum sensing ^{19, 31, 32}.

We have assembled a model that connects steps in biofilm formation, biofilm genes, and regulatory pathways (Fig. 2). The regulatory relationships diagrammed are based on diverse lines of evidence, including mutant phenotypes, genetic epistasis tests, microarray analysis, gene overexpression phenotypes, and chromatin immunoprecipitation. Some pathway relationships are thus indirect or tentative. The model is intended as a framework for the identification of new areas of inquiry and for interpretation of future studies.

Perhaps the greatest utility of a model for gene function in biofilm development is the application to in vivo models for biofilm formation on implanted devices ^{11, 33–35}. In vivo models are critically important because the nature of the device surface, the presence of host derived conditioning film, the levels of oxygen and carbon dioxide, and liquid flow all affect biofilm development ^{12, 36–38}. It is thus impossible to duplicate all relevant in vivo conditions in vitro.

Diversity of Biofilm Environments and Cohabitants

Many studies have focused on C. albicans biofilm formation on implanted vascular catheters, a major source of infection ⁷. However, biofilm formation occurs on many other devices. A rat denture biofilm model has been described recently that recapitulates features of denture stomatitis ³³. Microscopic and microbiological analysis revealed the signature features of biofilm formation – adherent cells, presence of extracellular matrix material, and high-level drug resistance. One other recently described rat model, in which biofilms form on catheters implanted subcutaneously, also has characteristic matrix material and an abundant hyphal population ³⁴. The latter provides both ethical and technical advantages in that a single animal can be used to culture numerous biofilms.

Biofilms also form on tissue surfaces, such as infections of the oral and vaginal mucosa. In such infection models, C. albicans produces dense three-dimensional biofilms embedded in extracellular matrix material ^{39, 40}. The level of biofilm drug resistance has not been tested directly, though Noverr and colleagues point out that drug resistance is seldom a clinical problem with vaginal candidiasis ⁴⁰.

Is biofilm formation in these novel environments mechanistically distinct from that the more commonly studied in vitro and in vivo models? There are certainly some conserved features. For example, biofilm formation in all of the models described above depends upon transcription factor Bcr1 (Filler, personal communication, Dongari-Bagtzoglou, personal communication) ^{33, 34, 40}. Where tested, filamentation-defective strains were biofilm-defective as well ^{34, 40}. However, it is likely that distinct genetic requirements and mechanisms will emerge for each system, based on precedents from systematic manipulation of in vitro biofilm environments ³⁸ and the pronounced gene expression responses of C. albicans to distinct host niches ⁴¹.

For any one type of biofilm, the environment may be altered by presence of co-infecting microbes. The overall frequency of mixed species biofilm infections has not been reported, but over 20% of Candida bloodstream infections are polymicrobial ⁴². A recent analysis of 24 implanted device-associated endocarditis cases reported that ~25% of those infections were polymicrobial ⁴³. Within biofilms grown in vitro, the interactions between bacteria and C. albicans are diverse ⁴⁴. Symbiotic interactions include augmented adherence or antibiotic resistance ^{45, 46}. However, most known interactions are inhibitory. Among the most intriguing examples are those that arise from transkingdom quorum sensing molecule responses. For example, farnesol produced by fungi inhibits biofilm formation of Staphylococcus aureus, and increases its antibiotic susceptibility ^{47, 48}. Bacteria fight back though; for example, the Pseudomonas aeruginosa quorum sensing molecule homoserine lactone mimics farnesol and inhibits C. albicans filamentation, thus preventing biofilm formation by C. albicans ⁴⁹. Other inhibitory interactions arise from broader environmental manipulation: vaginal bacteria inhibit C. albicans growth and virulence by producing H₂O₂ or lactic acid ^{44, 50}. The significance of further study in this area is demonstrated by the fact that the presence of combined infection by both bacteria and C. albicans can result in increased mortality ^{51, 52}.

Adherence and Attachment Responses

Gene products assigned to the adherence step (Fig. 2) have been shown through either null mutant analysis or heterologous expression studies to affect binding to a plastic or protein-coated substrate. One of the most clearly defined biofilm adhesins that mediates surface binding is Eap1 ^{21, 53}. Eap1 has sequence features commonly found among fungal cell surface proteins ^{54, 55}, including a signal sequence and a composition rich in the prospective glycosylation acceptors serine and threonine. It also contains internal repeats of a peptide motif, WPCL, that is found in numerous fungal cell surface proteins. Finally, it has a short C-terminal sequence that directs attachment of a glycosylphosphatidylinositol (GPI) anchor. GPI-linked proteins are found in many eukaryotes, where the GPI moiety tethers proteins to the plasma membrane. However, C. albicans and many other fungi can cleave this anchor, and then transfer the cleavage product and attached protein to a covalent linkage with β glucan in the cell wall ^{54, 55}. Several approaches indicate that Eap1 is indeed a GPI-linked cell wall protein ⁵³. Eap1 functions directly in biofilm adherence, as indicated by three observations: its expression in a non-adherent Saccharomyces cerevisiae strain confers adherence to polystyrene; a C. albicans eap1Δ/Δ deletion mutant has reduced adherence to polystyrene; and a C. albicans eap1Δ/Δ deletion mutant is defective in biofilm formation, as assayed both in vitro and in an in vivo catheter model ^{53, 56}.

The highly related cell wall proteins Als1 and Als3 also may function in biofilm surface attachment ⁵⁷. Expression of Als1 or Als3 in S. cerevisiae promotes binding to several different protein-coated substrates ⁵⁸, which may resemble the conditioned surface of an implanted device. In addition, a C. albicans als1Δ/Δ als3Δ/Δ double mutant is defective in biofilm formation in vitro and in vivo ²². In particular, catheter surfaces inoculated with the double mutant are virtually devoid of cells after incubation in vivo ²², as expected if the mutant has a severe substrate adherence defect.

The idea that Eap1 and Als1 might function in the initial adherence step is consistent with the fact that expression of both genes is detectable in cells grown as either yeast or hyphal cell types ^{56, 59}. This is not the case for Als3, which is expressed primarily or exclusively in hyphae ⁵⁹. It is possible that the initial adherence step leading to biofilm formation in vivo can be carried out by either yeast-form cells, which express Als1, or by hyphae, which express Als3.

Adherence itself may activate a gene expression response. Murillo and colleagues conducted a microarray comparison of planktonic cells and substrate-adherent cells ⁶⁰. Interestingly, as little as 30 min after adherence, a change in gene expression was established that, for several genes, was maintained for hours. In addition, Mateus and colleagues found that GFP fusions to the drug efflux proteins Cdr1 and Mdr1 were up-regulated within a few minutes after adherence to a glass slide ⁶¹. Thus C. albicans may sense and respond to surface contact. The regulators that promote attachment responses are unknown, but the transmembrane protein Dfi1 and MAP kinases Mkc1 and Cek1 are mediators of other surface-dependent responses ^62–64 and are thus excellent candidates. Mkc1 is required for normal biofilm formation ⁶⁴, and it would be interesting if this requirement reflected effects on adherence-induced gene expression.

Adherence is also highly regulated through a novel mating factor-response pathway (Box 2). Interestingly, the pathway operates in cells that do not mate, but rather assist in mating through formation of a biofilm ^65–68. The responsive cells are of genotype MTLa/MTLa, and thus have the potential to mate with MTLα/MTLα cells. They are unable to do so because they have not made the epigenetic transition from the “white” mating-incompetent state to the “opaque” mating-competent state ⁶⁹. They nonetheless respond to α mating factor through a newly evolved hybrid signal transduction pathway ⁶⁶ to create a biofilm. Four white cell α factor-induced genes are required for full adherence of this biofilm, those encoding the cell surface proteins Eap1 and Pga10, the predicted secreted protein Pbr1, and the putative aryl-alcohol dehydrogenase Csh1 ⁶⁸. (Csh1 has been detected on the C. albicans cell surface ⁷⁰ so the nature of its role in adherence is uncertain.) Most C. albicans isolates are MTLa/MTLα and do not secrete or respond to mating factor; MTLa/MTLα strains have been used for the bulk of the biofilm studies described in this review. However, the fact that Eap1, Pga10, and Csh1 all have roles in MTLa/MTLα biofilm formation (discussed above and below) argues that both kinds of C. albicans biofilms may employ similar gene products ⁶⁸.

Box 2. C. albicans mating.

One of the most exciting C. albicans biological discoveries in recent years is the finding that this organism, long considered asexual, can mate. A mating type locus called MTL determines sexual identity through regulatory relationships with some similarity to those of S. cerevisiae: MTLa/MTLacells can mate with MTLa/MTLα cells, and MTLa/MTLα cells cannot mate ¹⁰³. The mating response is induced by secreted mating pheromones: MTLa/MTLacells secrete a factor; MTLa/MTLα cells secrete α factor ¹⁰⁴. Mating involves more than just MTL-specified sexual identity, though; cells must switch from the mating-incompetent “white” cell type to the mating-competent “opaque” cell type ^105–109. The epigenetic white-opaque switch responds to numerous genetic and environmental signals, but is not regulated by mating factors as far as we know now^{110, 111}.

Mating of two C. albicans diploid cells yields a tetraploid that breaks down through chromosome loss to yield recombinant diploid progeny ^112–115. Normal functioning of the chromosome loss pathway depends upon Spo11 ¹¹⁶, whose orthologs function in meiosis in other organisms, but there is no evidence that C. albicans has a complete meiotic pathway.

Although white cells do not mate, they do respond to mating pheromone. The response can be assayed through changes in gene expression as well as increases cell-cell and cell-substrate adherence, yielding biofilm formation ^{65–68, 104, 117} Interestingly, only a portion of the opaque pheromone-response pathway is utilized in white cells; they employ a hybrid pathway with novel downstream components ^66–68. Opaque cells may be rare in many niches, and the biological role of this white cell biofilm seems to be to facilitate mating among disperse opaque cells ^{65, 110, 115}.

Biofilm Initiation, Filamentation and Cell-Cell Adhesion

Gene products assigned to the initiation step (Fig. 2) reflect a range of functions. This diversity reflects in part our broad definition: these are genes in which mutations cause production of only a small, rudimentary biofilm in vitro. A few additional gene products in this group (Rbt5, Als9 and Ece1) are those whose overexpression improves biofilm formation in a bcr1Δ/Δ mutant, which is defective in biofilm initiation (see below). We note that a mutant with a partial adherence defect might be categorized as an initiation mutant; thus our assignment of gene products to this step is tentative.

The production of hyphae is a hallmark of initiation, and many initiation-defective mutants grow solely as yeast cells under biofilm conditions (see Table 1). What is the role of hyphae? Insight has come from the transcription factor Bcr1, expression of which is up-regulated in hyphae ^{27, 28}. Bcr1 is required for biofilm initiation, but not for production of morphologically normal hyphae ^{27, 28}. Rather, it is required for full expression of a number of cell-surface proteins, several of which are hyphally induced. The failure to express hyphally induced surface proteins, such as Als3 and Hwp1, is the cause of the bcr1Δ/Δ mutant biofilm defect, because increased expression of ALS3 or HWP1 in the bcr1Δ/Δ mutant restores biofilm formation ability, both in vitro and in vivo (Fig. 3) ^{23, 28}. Moreover, expression of Bcr1 or its target genes can even permit biofilm formation by mutants defective in hyphal morphogenesis: Specifically, increased expression of BCR1 in a hyphal-defective tec1-/- mutant permits in vitro formation of a biofilm, albeit a fragile one ²⁸. In addition, expression of a surface-directed Als3 fusion protein permits biofilm formation in vitro by a hyphal-defective efg1Δ/Δ mutant ⁷¹. Therefore, the major way that hyphae promote biofilm formation is through expression of their surface protein complement.

Figure 3. Restoration of biofilm formation in a bcr1Δ/Δ mutant background by overexpression of surface protein gene ALS3.

These panels are confocal scanning laser micrographs of concanavalin A-alexafluor stained biofilms, grown under standard in vitro conditions ²⁸. The top panels are side views; the bottom panels are pseudocolor depth views, in which blue color represents cells closest to the substrate and red color represents cells farthest from the substrate. The wild type biofilm has a dense mixture of yeast cells and hyphae, which gradually becomes predominantly hyphae at the top of the biofilm. The bcr1Δ/Δ biofilm forms a basal layer of yeast cells attached to the substrate with little to no hyphae. Increased expression of ALS3 in the bcr1Δ/Δ strain permits substantial biofilm formation.

Interestingly, the Bcr1 ortholog in biofilm-forming Candida parapsilosis is also required for biofilm formation ⁷². Because C. parapsilosis does not form hyphae, the regulatory pathway upstream of Bcr1 may be divergent. Nonetheless, this finding points toward the exciting possibility that Bcr1 orthologs in other species may also govern biofilm formation.

What do these hyphal surface proteins do? Hyphae are extremely “sticky”, and both Als3 and Hwp1 are adhesins in some contexts ⁵⁴, so it seems reasonable that they may promote cell-cell or cell-substrate binding. In fact, Als3 (along with the closely related Als1) and Hwp1 seem to function as complementary cell-cell adhesins, analogous to the mating agglutinins of S. cerevisiae that promote binding between a and α cells. Two main observations support this idea. First, both an hwp1Δ/Δ mutant and a double als1Δ/Δ als3Δ/Δ mutant are defective in biofilm formation, but a mixture of the two mutant strains produces a robust biofilm both in vitro and in vivo ²². This finding indicates that Hwp1 and Als1/3 have distinct and complementary roles in biofilm formation. Second, expression of HWP1 in S. cerevisiae promotes its adherence to hyphae of wild-type C. albicans ²². Adherence is diminished when tested with hyphae of an als1Δ/Δ als3Δ/Δ mutant C. albicans. These findings points toward cell-cell adherence as the function that is mediated by Hwp1 and Als1/3.

At this juncture, one can envision a minimalist pathway of biofilm formation. First, yeast cells express Eap1 and Als1, which mediate cell-substrate binding. Second, surface-bound cells propagate and express Als3 and Hwp1, which mediate cell-cell binding. Hyphal formation might provide a simple pathway that leads to Als3 and Hwp1 accumulation. Als3 would also augment cell-substrate binding, as discussed in the previous section. In addition, it has been shown that Eap1 mediates cell-cell binding as well as cell-substrate binding, so it would participate in both processes. Many gene products required for biofilm initiation are also required for hyphal formation; their functions are explained by the minimalist model as ultimately being required for ALS3 and HWP1 expression.

However, one group of biofilm initiation gene products is less readily explained by the minimalist model: the additional cell surface proteins, including Sun41, Csa1, Pga10, Rbt5, Hwp2, and Rbt1 (Table 1). Analysis has been challenging for several of these gene products because they belong to families with overlapping or compensatory functions (Csa1/Pga10/Rbt5; Hwp2/Rbt1/Hwp1; Sun41/Sun42) ^{67, 73–76}. The requisite construction of multi-gene mutants for study of these genes is non-trivial, even in this era of accelerated C. albicans genetics. In any case, current observations suggest that some of these gene products may function as adhesins. In particular, additive effects of hwp2 and rbt1 mutations with an hwp1 mutation, along with the known role of Hwp1 as an adhesin, suggests that Hwp2 and Rbt1 are adhesins⁶⁷. They may contribute to a threshold level of cell-cell binding, for example, that is required for biofilm stability. (C. parapsilosis Bcr1 promotes expression of CpRBT1, so perhaps CpRbt1, which has a major role in biofilm formation ^{72, 77}, has assumed a predominant adhesin function in that species.) A second suggestion is that some of these cell-surface proteins may have general roles in cell wall structure, and that perturbation of cell wall architecture impairs adherence through effects on either post-translational modification or expression of adhesins. We note that loss of Sun41 or Pga10 confers hypersensitivity to cell wall inhibitors, an expected consequence of a general cell wall defect ^73–76. It is also noteworthy that ALS1 RNA levels are reduced in the biofilm- and cell wall-defective protein kinase mutants gin4-/-, ire1-/-, and cbk1-/-, thus suggesting that adhesin gene expression may be regulated through cell wall regulatory pathways ⁷⁸. The mechanistic contribution of so many cell surface proteins to biofilm formation is among the major questions to be addressed, particularly because such proteins are inviting therapeutic targets.

Biofilm Maturation and the Extracellular Matrix

Biofilm maturation includes continued growth as well as accumulation of extracellular matrix material. Genes assigned to this category (Fig. 2) include those affecting matrix production or overall biofilm biomass.

The composition of the matrix that is produced in vitro includes carbohydrate, protein, hexosamine, phosphorus and uronic acid ⁷⁹. One major extracellular carbohydrate constituent is β-1,3 glucan, and increased production is associated with biofilm cells compared to planktonic cells ⁸⁰. Proteomic analysis has revealed the presence of specific proteins associated with the biofilm cell surface; these proteins may include matrix components ^{30, 81}. Finally, a recent study reports detection of extracellular DNA ⁸², as has been found in bacterial biofilms ⁸³. Addition of DNase to a mature biofilm partially disrupts the biofilm ^{79, 82} and the addition of eDNA at the beginning of biofilm development results in mature biofilms with increased biofilm biomass. Thus the matrix eDNA contributes to the structure and stability of a mature biofilm.

The transcription factor Zap1, a regulator of zinc acquisition ^{84, 85}, is a net negative regulator of biofilm matrix production (Fig. 2). A zap1Δ/Δ mutant forms a biofilm with elevated levels of matrix β-1,3 glucan in vitro and in vivo ²⁹. Zap1 activates expression of CSH1 and IFD6, which have inferred negative roles in matrix production, and represses GCA1, GCA2, and ADH5, which have inferred positive roles ²⁹. Gca1 and Gca2 may function through hydrolytic release of soluble β-1,3 glucan fragments from longer glucan chains. The precise functions of the alcohol dehydrogenase-related gene products, Csh1, Ifd6 and Adh5, are unknown, but several similar S. cerevisiae alcohol dehydrogenases function in synthesis of acyl and aryl alcohols ^{31, 32, 86, 87}. These alcohols have roles in quorum sensing and cell signaling (see Box 1), as indicated for example by effects on hyphal growth ^{15–18, 20, 88}. Thus a possible mechanistic role for these dehydrogenases is to promote biogenesis of biofilm-associated acyl and aryl alcohols that in turn would control matrix synthesis. Csh1 and Ifd6 may act preferentially to yield a matrix inhibitory signal, while Adh5 may act preferentially to yield a matrix stimulatory signal ²⁹. The known role of Zap1 in zinc-responsive gene expression suggests that ambient zinc levels may be a critical determinant of biofilm matrix levels.

Zap1 may have a broader role in biofilm maturation than simply to control matrix accumulation. The zap1Δ/Δ mutant has reduced levels of several genes that are normally up-regulated in mature biofilms, including ergosterol biosynthetic genes and putative hexose transporters ²⁹. Thus Zap1 seems to govern several aspects of biofilm maturation. It will be interesting to see if any of these mutant phenotypes reflect the postulated alteration of quorum sensing molecule levels.

A unique feature of mature biofilms, in addition to matrix accumulation, is the acquisition of high-level resistance to antifungals ⁵, notably the azoles and polyenes that target membrane sterols. The nature of biofilm drug resistance may reflect four distinct mechanisms. First, mature biofilm cells have reduced membrane sterol levels ⁸⁹ and elevated expression of several ergosterol biosynthetic genes ^{60, 90, 91}, perhaps reflecting hypoxia ^{38, 77}. The ability of mature biofilm cells to survive with low sterol levels, combined with elevated biosynthetic enzyme levels, may contribute to azole and polyene resistance. Indeed, a recent study revealed that a polyene-resistant biofilm cell subpopulation displays substantially increased ergosterol biosynthetic gene expression ⁹². Second, the azole efflux genes CDR1, CDR2, and MDR1 are induced early in biofilm formation, and may contribute to overall azole resistance. However, their phenotypic contribution is detectable only in early biofilm cells ⁸⁹. Third, like several bacterial biofilms ⁹³, C. albicans biofilms contain a subpopulation of “persister cells” that are tolerant to a variety of otherwise cidal treatments ⁹⁴. The significance of this phenomenon is highlighted by the presence of persisters in human oral C. albicans populations ⁹⁵. Persisters do not have the long-term stability of mutants, but are phenotypic variants that may arise from an epigenetic change or, perhaps, transient aneuploidy ⁹⁶. Finally, the β-1,3 glucan of biofilm matrix binds to and sequesters azole drugs ⁸⁰. The physiological significance of this mechanism has been demonstrated through analysis of a strain with reduced β-1,3 glucan biosynthetic capacity (genotype FKS1/fks1Δ) ⁹⁷. Biofilms of this strain have reduced matrix β-1,3 glucan and reduced azole resistance levels in both in vitro and in vivo models. The dramatic sensitivity of the mutant biofilm cells to azole treatment, particularly in the in vivo biofilm model, suggests that this sequestration mechanism is a major contributor to C. albicans biofilm azole resistance ⁹⁷.

Cell Dispersal

Ultimately a biofilm releases cells that can initiate formation of new biofilms or disseminate into host tissues. Recent studies examined the quantitative and qualitative properties of cells released from a C. albicans biofilm ^{98, 99}, yielding three important findings. First, the majority of dispersed cells are yeast cells, as depicted in Fig. 2 ⁹⁸. This observation suggests that the transition from yeast to hyphae that occurs during biofilm initiation may be thrown into reverse for dispersal. Second, three new regulators of dispersal, Ume6, Pes1 and Nrg1, were identified. Over-expression of UME6 reduced the release of cells from a biofilm, while over-expression of either PES1 or NRG1 increased release ^{98, 99}. Thus changes in expression or activity of Ume6 or Pes1 during biofilm maturation – perhaps in response to quorum sensing molecule accumulation – may govern cell dispersal. Finally, the study reports the fascinating observation that dispersed cells have distinct phenotypes from planktonic cells; they display elevated adherence, filamentation capacity, and increased pathogenicity in a disseminated infection model. Thus the dispersal step releases cells that are uniquely equipped to seed new biofilms and sites of infection ^{98, 99}.

Concluding Remarks

C. albicans biofilm formation on implanted devices is a major source of disseminated Candida infection. The last decade has seen major advances in the definition of C. albicans genes that govern biofilm formation. In many cases we have moved forward from gene discovery to definition of pathway relationships and, in some cases, mechanistic understanding. In addition, numerous molecules with the potential for quorum-sensing roles in biofilm maturation have been defined. Moreover, there are now several animal models for analysis of biofilm formation in vivo that have validated the significance of key biofilm genes discovered in vitro. Yet a summary of that substantial progress also reveals major gaps in our understanding: How can so many cell wall proteins participate in biofilm formation; are they all adhesins? Which small molecules are actually active in quorum sensing in vivo, and can we harness their activities for therapeutic development? Can we use our understanding of biofilm drug resistance to develop better therapeutics and more focused assays of biological activity? What are the dynamics of formation and key molecular players in mixed-species biofilms? And, perhaps most difficult to answer and most interesting to ponder, what selective pressures caused evolution of biofilm formation ability – was it for mating, or for mucosal surface adherence and persistence in the host? There has never been a more interesting time to study C. albicans biofilms.