Formation of biofilms allows microbial pathogens to create a safe sanctuary in which sessile cells remain in a protected environment. However, cells within a biofilm may be also confronted with adverse environmental conditions (i.e.
reduced nutrient availability, accumulation of toxic waste products) so that dispersion of cells would be beneficial for survival. Furthermore, this release of cells from the original biofilm community is required to generate novel communities at new locations. It follows that gaining knowledge about mechanisms regulating biofilm dispersion, at both the physiochemical and molecular levels can potentially lead to better strategies for the prevention and treatment of biofilm-associated infections. This is particularly important in the case of C. albicans
, since dispersed cells are responsible for candidemia and disseminated invasive candidiasis, which are among the gravest forms of infection and carry the highest mortality rates 
Biofilm formation by C. albicans
progresses through multiple development stages, beginning with attachment to a surface, microcolony formation, proliferation and development and maturation involving encapsulation within an exopolymeric matrix 
. Cells at each stage of the biofilm developmental process display distinct phenotypes and properties markedly different from those of the same group growing planktonically. The developmental life cycle of C. albicans
biofilms comes full circle when biofilm cells disperse and go on to colonize new surfaces again. In bacteria, loss of cells from biofilms is known to be triggered by changes in the physiochemical environment, including changes in nutrients, starvation, quorum sensing, modifications in exocellular biofilm components and regulation at the molecular level of certain gene products 
. In C. albicans
; however, very little is known about the environmental cues that initiate biofilm dispersion, about the phenotypic properties of the dispersed cells, or about the molecular mechanisms governing release of cells from biofilms. In the present study, we investigated these aspects of this understudied stage of the C. albicans
biofilm life cycle.
We took advantage of a simple in vitro
model of C. albicans
biofilm formation under conditions of flow recently described by our group 
which enabled us to develop C. albicans
biofilms under controlled conditions of media flow for prolonged periods of time. We were thus able to study the phenomenon of biofilm dispersion in a controlled manner which is not possible by using “static” biofilm models such as the widely used 96-well microtiter plate model 
. At the same time, one of the decisive advantages of this model is that it yields enough dispersed cells needed for all subsequent phenotypic and molecular analyses. Furthermore, our in vitro
flow model can simulate physiologically relevant shear stresses experienced by the biofilms in the blood vessel.
Rather than an often hypothesized end-stage process, we found that dispersion occurred continuously over the course of biofilm development. Our initial observations indicated that, regardless of the media used, the number of dispersed cells released into the flow-through liquid depended upon the growth stage of the biofilm. During the early stages of biofilm growth (0–3 hr), the extent of dispersion is lesser than that at the intermediate stage of biofilm growth (5–12 hr). As the biofilms matured (20–24 hr), the numbers of dispersed cells again decreased. Because we grow our biofilms under conditions of flow, the low levels of dispersion observed during the early stages of biofilm development may be due to passive dispersal of biofilms as a result of hydrodynamic parameters such as shear stress, a phenomenon also referred to as “erosion” 
. Overall, these observations carry important clinical implications as release of cells from an in vivo
catheter) may occur early after initial colonization of the biomaterial without the need for the formation of a fully developed mature biofilm. Maximum numbers of cells are released when the biofilm proliferates rapidly during the intermediate phase (5–12 hr), while numbers decrease again during the stationary phase of a mature biofilm. As long as biofilms were supplied with fresh nutritients, complete cessation of dispersion was never observed, even after 48 hr of growth.
Since, C. albicans
biofilms are made up of a mixture of yeast, pseudohyphae and hyphae 
, we hypothesized that the cells released from biofilms may also include all three morphological forms. To the contrary, we found that under a steady rate of media flow, the dispersed cells occurred predominantly in the yeast form. Interestingly, we observed that a large percentage of these yeast cells were unbudded and elongated in shape, depending on the nutrient content of the media used for biofilm development. Larger numbers of unbudded cells were recovered from biofilms grown in synthetic defined media (YNB and RPMI) rather than from nutrient rich medium like YPD. Although a quantitative distribution of the total dispersed cell population was not performed in this study, we have found that the majority of the dispersed cells originate from the topmost outer layers of the biofilm (not shown). The yeast form of C. albicans
is considered to be most conducive for dissemination into the blood stream 
; and the results of our study may serve as a proof of principle that release of these potentially infectious particles from biofilms may prove to be a major mechanism of candidemia and disseminated invasive infections that often occur after C. albicans
biofilms initially form on the surface of intravascular catheters.
In bacteria the nutritional status of the environment most often dictates biofilm dispersal and both decreases and increase in nutrients have been described to induce biofilm dispersion 
. In the case of C. albicans
, we also observed that composition of the media used for biofilm formation affected dispersion: in general, the richer the medium, the greater were the numbers of dispersed cells in the flow-through liquid. For example, the hierarchy of media triggering dispersion was YPD > YNB >RPMI. Detachment (rather than dispersion) of whole C. albicans
biofilms under the influence of rich medium has recently been reported by Sellam et. al 
. However for their study, the investigators used a strong hydrodynamic force (more than 17.3 dynes/cm2
), forcing displacement of the biofilms from the surface of the tubing. Our study employed a steady, laminar flow of 1 ml/min (corresponding to a shear force of 4.52 dyn/cm2
) to study biofilm dispersion. This mechanical force is physiologically relevant since the normal time-averaged levels of venous stresses range between 1–5 dyne/cm2 
. A similar correlation between biofilm dispersion and flow rate has been previously reported by Baillie and Douglas: in their study biofilms grown on cellulose membrane subjected to a flow rate of 1 ml/min of YNB yielded biofilm dispersion rates similar to those observed in our study under similar conditions of flow 
. Furthermore, we discovered that the rate and extent of biofilm dispersion was greatly influenced by the availability of carbon sources. We found that increasing amounts of glucose in the growth medium increased the levels of biofilm dispersion, so much so that 500 mM glucose initiated biofilm dissolution after 20 hr. In many bacterial species capable of forming biofilms, availability of nutrients especially high glucose concentrations is known to suppress biofilm formation by increasing biofilm dispersion 
. Our findings are consistent with earlier findings by Sauer et. al
, who demonstrated biofilm dissolution in Pseudomonas aeruginosa
by high concentrations of carbon sources 
. Massive dispersion under favorable growth conditions seem to indicate a biofilm survival strategy whereby C. albicans
cells only benefit from the biofilm mode of growth at low nutrient concentrations, but abandon this mode of growth when conditions in the environment become favorable. No alternative carbon source used in the present study (galactose or maltose) was as effective as glucose in induction of biofilm dispersion. In fact, the rate of dispersion was reduced >50-fold when sterile PBS instead of media was supplied to mature biofilms, to induce starvation. Until now, in the field of C. albicans
biofilm research, it was speculated that nutrient depletion enables production and release of cells from the biofilms thereby increasing dispersion 
. Although this may still hold true for statically grown biofilms (in fact we have observed massive detachment events as static biofilms age over 72 and 96 hr) our results using the more physiological flow model show the opposite: we found that poor nutrient sources led to significant curtailment in the release of cells from a biofilm while media rich in nutrients triggered higher rates of biofilm dispersion.
In addition to nutritional composition, we observed that the pH of the growing medium also exerts an important effect on C. albicans
biofilm dispersion, which was enhanced at acidic pH and decreased under alkaline conditions (Figure S2A
). We speculate that acidic pH conditions induce lateral yeast formation from biofilm hyphae and also increase the number of yeast cells in the biofilm, thereby resulting in greater number of cells dispersed under flow. Treatment of preformed biofilms with farnesol, a quorum sensing molecule that inhibits C. albicans
filamentation and biofilm formation 
only led to a significant increase in dispersion when supraphysiological concentrations of this compound were used (Figure S2B
). Although we cannot disregard the effect of accumulating farnesol concentrations in the detachment events observed when biofilms are grown under static conditions, it is likely that the continuous flow conditions used in this study, effectively preventing accumulation of high farnesol concentrations, are responsible for the lack of effect on biofilm dispersion observed at lower (physiological) concentrations of farnesol. The same diluting effect of continuous flow can also explain the lack of effect on biofilm dispersion observed at physiological concentrations of tyrosol (results not shown).
In order to colonize distal sites, cells released from biofilms must be able to disperse into the host environment and adhere to and damage the endothelial cells lining blood vessels before entering the tissues 
. Hence we postulated that the infectious particles released from C. albicans
biofilms may possess enhanced adhesive and invasive properties. Indeed, we found that C. albicans
yeast cells dispersed from 37°C grown, 24 hr old biofilms were at least 35–40% more adherent to polystyrene compared to age and temperature matched planktonic yeast cells. Not only were the dispersed cells more adherent, but they also developed germ tubes in numbers consistently higher than planktonic cells. This result was interesting because it meant that cells dispersed from C. albicans
biofilms indeed were infectious particles, already ready to deploy at least two of the major virulence factors considered of critical importance during the pathogenesis of candidiasis 
. This also seems to indicate that free living; highly adherent dispersed cells had prospects of colonizing and establishing biofilms at newer sites and indeed, we found that the dispersed cells could develop more robust biofilms compared to planktonic cells. Armored with properties such as better adhesion and filamentation, it was not surprising that dispersed cells from C. albicans
biofilm also display enhanced adhesion to endothelial cells and cause increased endothelial cell damage, which represent major hallmarks of the infectious process. But perhaps, the ultimate test regarding virulence is the assessment of pathogenesis in the murine model of hematogenously disseminated candidiasis. Indeed, all the above characteristics possessed by the dispersed cells translated to the in vivo
scenario, where cells dispersed from biofilms demonstrated enhanced virulence compared to their planktonic counterparts. This finding seems to indicate that dispersed cells retain their higher virulence over several generations, which raises the question whether heritable epigenetic modifications are responsible for enhanced adhesion, filamentation and virulence of dispersed cells.
Studying the mechanisms that regulate biofilm dispersion has been an area of intense activity in the field of bacterial biofilms. Several attempts have been made to identify cellular responses that contribute to the occurrence of active biofilm dispersion in pathogenic bacteria. Streptococcus mutans
enables biofilm dispersion by releasing enzymes that break down polysaccharide matrix material 
. In P. aeruginosa
, P. putida
and Shewanella oneidensis
changes in oxygen, nutrient sources and several chemical parameters have been reported to induce dispersion 
. Dispersion in Escherichia coli
is mediated largely through the collective regulation of intracellular glycogen biosynthesis and central carbon metabolism, alteration of matrix components, activation of flagellum, and motility 
. Thus it is becoming clear that dispersion from mature biofilms maybe triggered by environmental cues that coincide with specific phenotypic changes in the organism.
In this study, we demonstrate that C. albicans
biofilm dispersion is regulated to a great extent by two important morphogenetic processes; first, the ability to undergo morphogenetic conversions and second, the ability of filaments to produce lateral yeast cells. For the first study, we utilized a genetically engineered strain of C. albicans
, in which one allele of CaUME6
, a positive regulator of filamentation is placed under control of a tet-regulatable promoter 
. The extent of dispersal of biofilm cells was greatly dependent on the morphological characteristics of the biofilm. An all hyphal biofilm (medium without DOX; UME6
overexpression) released fewer number of cells compared to a mostly pseudohyphal and yeasty biofilm (medium with DOX, UME6
repression). We showed that C. albicans
biofilm dispersion could be controlled to a large extent by controlling the morphology of this fungus. This knowledge could be of great importance in designing antifungal drugs that could curtail biofilm dispersion by targeting specific morphological forms in biofilms. For example, the HSP90 inhibitor geldanamycin (GDA) induces hyperfilamentation in C. albicans
yeast cells and thereby renders the cells highly avirulent in a mouse model 
. More importantly, in combination with fluconazole, GDA is fungistatic in planktonic cells 
. Thus, development of drugs like GDA could prove extremely potent in abolishing dispersion of cells from biofilms.
Microscopic images in this study revealed that several dispersed cells appeared to be released from the upper-most hyphal layers of the C. albicans
biofilms. This observation was confirmed when we showed that increasing lateral yeast cells in biofilms by overexpression of CaPES1 
increased the extent of biofilm dispersion. SEM showed an overwhelming presence of yeast cells in the topmost layers of the biofilm, and a crack in the biofilm disclosed that these indeed were lateral yeast cells produced by the hyphae now completely covered by them. The repression of CaPES1
abolished lateral yeasts from the topmost layers of the biofilm resulting in a decrease in numbers of dispersed cells and reemergence of the hyphal mesh. Depletion of the levels of the Pes1p corresponding with a decrease in lateral yeasts is known to render C. albicans
avirulent in Galleria mellonella
. Discovery of antifungal agents targeting Pes1p has implications for C. albicans
biofilm related infectious diseases because lateral yeasts released from biofilms have the highest potential to disseminate and cause invasive diseases. In S. cerevisiae
, the pescadillo
homolog Yph1 participates in a nutritional signaling network that relays information about the cell's nutritional status to both the cell cycle machinery and the ribosomal biogenesis machinery 
. In C. albicans
, fresh media has been shown to induce Pes1p expression and subsequent increase in lateral yeast production 
. If C. albicans
Pes1p is linked to a nutritional signaling network, it could very well be a major player in orchestrating biofilm dispersion via interactions with the cAMP-dependent protein kinase (PKA) and TOR pathways. Thus, the C. albicans
pescadillo homolog may represent a link between two processes with important roles in biofilm dispersion: morphology and sensing of nutritional conditions.
In summary, we have initiated studies on the phenomenon of C. albicans biofilm dispersion, facilitated by the recent development of a simple model for biofilm formation under physiological conditions of flow. In contrast to a “massive” detachment event, our results indicate that dispersal of cells, which are mostly in the yeast form, occurs continuously throughout the biofilm developmental cycle. The dispersion process is dependent on different environmental factors, including nutrition and pH. Dispersed cells from biofilm display distinct phenotypic properties associated with increased virulence. In addition, our results point to the presence, at the molecular level, of a complex regulatory circuitry that orchestrates the phenomenon of biofilm dispersion in C. albicans.