Determination of Spatial Distributions of Zinc and Active Biomass in Microbial Biofilms by Two-Photon Laser Scanning Microscopy

The spatial distributions of zinc, a representative transitionmetal, and active biomass in bacterial biofilms were determinedusing two-photon laser scanning microscopy (2P-LSM). Applicationof 2P-LSM permits analysis of thicker biofilms than are amenableto observation with confocal laser scanning microscopy and alsoprovides selective excitation of a smaller focal volume withgreater depth localization. Thin Escherichia coli PHL628 biofilmswere grown in a minimal mineral salts medium using pyruvateas the carbon and energy source under batch conditions, andthick biofilms were grown in Luria-Bertani medium using a continuous-flowdrip system. The biofilms were visualized by 2P-LSM and shownto have heterogeneous structures with dispersed dense cell clusters,rough surfaces, and void spaces. Contrary to homogeneous biofilmmodel predictions that active biomass would be located predominantlyin the outer regions of the biofilm and inactive or dead biomass(biomass debris) in the inner regions, significant active biomassfractions were observed at all depths in biofilms (up to 350µm) using live/dead fluorescent stains. The active fractionswere dependent on biofilm thickness and are attributed to theheterogeneous characteristics of biofilm structures. A zinc-bindingfluorochrome (8-hydroxy-5-dimethylsulfoamidoquinoline) was synthesizedand used to visualize the spatial location of added Zn withinbiofilms. Zn was distributed evenly in a thin (12 µm)biofilm but was located only at the surface of thick biofilms,penetrating less than 20 µm after 1 h of exposure. Therelatively slow movement of Zn into deeper biofilm layers providesdirect evidence in support of the concept that thick biofilmsmay confer resistance to toxic metal species by binding metalsat the biofilm-bulk liquid interface, thereby retarding metaldiffusion into the biofilm (G. M. Teitzel and M. R. Park, Appl.Environ. Microbiol. 69:2313-2320, 2003).

INTRODUCTION

Top

Introduction

Surfaces in both natural and engineered systems are typicallycolonized by adherent bacteria, resulting in heterogeneous surfacecoatings (or biofilms) consisting of attached microbial populationsand extracellular polymeric substances (EPS) of bacterial origin(4, 7, 9, 13, 14, 42). Biofilm structures consist of aggregatedmicrobial communities embedded with the EPS, which may includecomplex mixtures of heteropolysaccharides, protein, and nucleicacids (9, 14), separated by a network of open water channels(34). Biofilm formation is dependent on a variety of factors,including surface properties, availability of electron donorsand acceptors, microbial species, and hydrodynamics of the surroundingfluid phase (16, 34, 40). Among these variables, microbial speciescharacteristics and the availability of growth substrates playimportant roles in determining the ultimate spatial distributionof individual organisms in a multiple-species biofilm (31).

If a homogeneous biofilm structure is assumed, it follows thatmetabolic activity within the deeper layers may be altered becauseof substrate, oxygen, and/or nutrient limitations. Model simulationsbased on the assumption of a homogeneous biofilm result in theprediction that active biomass will be located predominantlyin the outer regions of biofilms (9, 27, 30, 39). The existenceof a heterogeneous structure in which open channels allow thebulk fluid to penetrate to variable depths within biofilms makesconclusions based on the assumption of homogeneous structureof questionable value. Experimental observations are neededto better understand mass transfer of substrate, nutrients,and toxic chemicals into biofilms and to aid in the developmentof improved biofilm models for environmental and biochemicalapplications.

Trace metals are ubiquitous, and their interactions with biofilmsare of great interest because of the potential role of adherentbacterial populations in the surface chemical reactions thatcontrol the phase distribution, bioavailability, and ultimatelythe biogeochemical cycling of many metal species in naturalaquatic systems (21). The availability of both essential andtoxic trace metals can also clearly influence the metabolicactivities of microorganisms in engineered bioreactors as wellas in medical applications (19, 40).

From the physicochemical point of view, biofilms influence thefate of metals both directly, by uptake, adsorption, and complexation(with extracellular metal-binding ligands), and indirectly,through alterations in pH and oxidation/reduction potential,as well as through production of biogenic minerals such as ironand manganese oxides, which strongly bind metals (19, 21). Previousstudies of biofilms and metal interactions have demonstratedthat metals strongly adsorb to biofilms, at levels significantlygreater than those bound by planktonic bacteria of the samestrain (12, 42). Nevertheless, the sorption mechanism for bothattached and planktonic bacteria is likely the same; i.e., metalbinding with functional groups on extracellular polymers andbacterial cell surfaces, including the amino, carboxylic, hydroxyl,and phosphate groups of (phospho)lipids, proteins, and polysaccharides(5, 36).

Significant amounts of metal sorption can inhibit microbialactivity. Recent studies on relationships between metal sorptionand microbial toxicity have revealed that inhibition by metalssuch as Zn, Ni, and Cd is subject to the slow kinetics of metalinternalization, while the mechanism of Cu inhibition may involverapid loss of membrane integrity (10). As noted above, biofilmsmay be more resistant to toxic metals than planktonic cells,possibly because of retarded metal diffusion into biofilms,which is a result of the adsorption of metals to cell surfacesand metal binding to the extracellular polymers in the biofilmmatrix (35).

Recently, biofilm formation under metal stress has been subjectedto extensive investigation. While the spatial distributionsof oxygen and nitrogen species in biofilms have been successfullydetermined using microelectrodes (24, 43), few investigationshave focused on visualizing the spatial distributions of activebiomass and metals in biofilms, largely because of the lackof noninvasive techniques that allow localization over depth.Previous noninvasive techniques for the study of biofilms mainlyutilize confocal laser scanning microscopy (CLSM) (7, 35, 42)and are usually limited to penetration depths of up to 100 µm(7, 41).

Two-photon laser scanning microscopy (2P-LSM) offers clear imagesof material two to three times deeper than that with confocalmicroscopy (3). In 2P-LSM an ultrafast infrared laser is usedto deliver two low-energy photons of sufficient intensity toexcite visible fluorescence from a fluorophore. Because theinfrared photons are typically absorbed by biological systemsmuch less strongly than are visible photons, this scheme bothavoids potential damage to the cells and allows deeper penetration(17). In addition, since two-photon absorption depends quadraticallyon laser intensity, absorption occurs only at the focus of thelaser, so that the technique provides better spatial resolutionfor imaging of samples (1, 17, 45). Thus, in comparison to single-photonCLSM, two-photon excitation allows reduced phototoxicity andphotobleaching, improved spatial localization, and deeper sectioningof samples (26, 41, 45). While applications of 2P-LSM to microbialbiofilms are beginning to appear (22, 23), to our knowledge,the advantages of 2P-LSM have yet to be applied to analysisof metal distribution in microbial biofilms.

The goal of this work was to use 2P-LSM to measure and analyzethe spatial distributions of metals and active biomass fractionswithin microbial biofilms. To this end, a specific zinc complexationfluorochrome, 8-hydroxy-5-dimethylsulfoamido-quinoline (C₁₁H₁₂N₂O₃S,molecular weight 252, hereafter referred to as quinoline), wassynthesized according to previously described methods (25) andused as a marker in the visualization of zinc. A live/dead fluorescentdye mixture (Baclight bacterial viability kit, Molecular Probes,Eugene, OR) was applied to determine bacterial viability.

MATERIALS AND METHODS

Top

Materials and Methods

Culture and media.
The test bacterium was Escherichia coli PHL628, a K-12 MG1655derivative that forms biofilms as a consequence of the overexpressionof curli (37). The test strain was grown in a modified minimalmineral salts (MMS) medium (21), which contained, per liter,0.2 mM CaCl₂, 0.14 mM MgSO₄, 0.91 mM (NH₄)₂SO₄, 0.15 mM KNO₃,0.01 mM NaHCO₃, 0.37 mM KH₂PO₄, and 31 mM pyruvate, with thepH adjusted to 6.0. The MMS medium components have defined Znbinding constants, allowing computation of Zn speciation usingchemical equilibrium software such as MINEQL+ (EnvironmentalResearch Software; Hallowell, ME) (15, 32). Calculations showedthat under the conditions used for the experiments describedin this report, mineral precipitation would not occur (assumingadded phosphate was consumed by cells) and complexation of Znwas not significant (i.e., the dominant Zn species was the freeion). The solution used to rinse biofilms contained the samecomponents as MMS medium except that no phosphate or pyruvatewas added. Thick biofilms were grown in complete Luria-Bertani(LB) medium containing, per liter, 10 g tryptone, 5 g yeast,and 10 g NaCl. All media were sterilized by autoclaving at 121°Cbefore use.

Biofilms.
Thin and thick biofilms were grown under batch and continuous-flowcultivation techniques, respectively (6, 35, 43). Thin biofilmswere grown in microwell dishes with 0.17-mm glass coverslipsat the bottom (MatTek, Ashland, MA). The microwell dishes wereinoculated with 0.5-ml portions of an overnight culture grownin MMS at 37°C, diluted sixfold with the same medium, andincubated on a mechanical shaker (50 rpm) at 30°C. The temperaturewas changed from 37 to 30°C to facilitate curli overexpressionand hence biofilm formation. After 1 to 2 days of incubation,the microwell dishes were washed twice with the MMS rinse solutionprior to microscopic examination.

Thick biofilms (up to 350 µm) were grown in a drip-flowculture system (43). Inoculated microwell dishes were kept overnightbefore the flow containing LB medium was initiated. The LB mediumwas delivered with a Masterflex peristaltic pump (Cole-Parmer,Vernon Hills, IL) and dripped onto the glass coverslips in microwelldishes at a constant flow rate of 0.08 ml/min at 30°C. After2 days of continuous feeding, the microwell dishes were washedtwice with the MMS rinse solution before staining. Since anopen air gap occurred at the point where LB medium was deliveredto the dishes, it is possible that the thick biofilms did notremain pure cultures. However, no visual evidence of contaminationwas observed and, if present, would be unlikely to alter theconclusions regarding the spatial distribution of Zn and activebiomass.

Zinc sorption.
Suspended cultures of E. coli PHL628 cells in early stationaryphase (based on the growth curves shown in Fig. 1) were usedfor both kinetic and isotherm studies. The pH of cultures wasadjusted to 7.0 before use. All glassware was acid washed andsterilized prior to use.

View larger version (15K):
[in this window]
[in a new window]
FIG. 1. E. coli PHL628 grown in the presence of zinc in MMS medium at 37°C and no Zn (control, •), 0.1 mM Zn ( ${blacksquare}$ ), and 1 mM Zn ( ${square}$ ).

Sorption kinetics.
Aliquots of concentrated Zn solution (from ZnCl₂, pH < 6)were added to the cultures to yield initial concentrations of0.1 and 1 mM Zn. The cultures were kept at constant temperature(37°C) and continuously mixed on a shaker at 300 rpm. Atpredetermined intervals (1, 2, 4, 8, and 24 h), aliquots werewithdrawn and centrifuged at 20,000 x g for 30 min. SolubleZn concentrations were determined from the supernatants usingan AAanalyst 100 atomic absorption spectrophotometer (Perkin-Elmer,Norwalk, CT) and the amount of sorbed Zn was calculated fromthe difference between the initial and measured dissolved Znconcentrations after correction for background values.

Sorption isotherms.
Aliquots of Zn stock solution were added to cell cultures toachieve metal doses from 0 to 1.0 mM. Sodium azide was addedto one of the cell cultures at a final concentration of 0.15mM to inhibit microbial activity (20). All samples were incubatedunder the conditions mentioned above for 1 h (based on the resultsof the kinetic study), and Zn concentrations in the supernatantswere analyzed to determine Zn adsorption isotherms.

Fluorescence measurements.
The fluorescence of quinoline and the Zn-quinoline complex wasmeasured with an SLM-8000 fluorometer (SLM Instrument, Urbana,IL). The sample chamber of the fluorometer was maintained ata constant temperature of 25°C with a thermostat-controlledwater bath. The fluorescence spectra were measured and analyzedwith maximal intensity at excitation and emission wavelengthsof 381 and 462 nm, respectively.

To test the effect of Zn on quinoline fluorescence, solutionswith or without 50 µM Zn were prepared at pH 6, and aliquotsof quinoline were added to yield final concentrations from 0to 150 µM. Alternatively, the quinoline concentrationwas held constant at 150 µM while the Zn concentrationwas varied from 0 to 1 mM.

To evaluate the strength of Zn-quinoline complexation and theefficacy of quinoline in sensing Zn sorbed to biomass, eitherEDTA, nitrilotriacetate (NTA), citrate, pyruvate, or early-stationary-phasecell cultures (with an assumed molecular formula of C₅H₇O₂N)at concentrations ranging from 0 to 220 µM were addedas complexing agents along with 100 µM Zn. After 1 h equilibration,aliquots of quinoline were added to yield a final concentrationof 150 µM in all samples and incubated for 15 min beforefluorescence measurements. For samples with cells, the quinoline-spikedcultures were centrifuged at 20,000 x g for 1 min and the supernatantswere collected for fluorescence measurements.

Quinoline sorption to suspended and biofilm cultures was evaluatedby measuring the fluorescence of quinoline (added at 150 µM)in supernatants after 1 h equilibration in the presence andabsence of cells.

Staining.
After rinsing twice with the MMS rinse solution, the solutionwas removed. Thin biofilms on microwell dishes were exposedto 100 µl of 100 µM Zn solution for 1 h. Residualdissolved Zn was removed by washing twice with the MMS rinsesolution, and 100 µl of 150 µM quinoline was addedand left in the dark for 15 min before being visualized with2P-LSM. In a similar manner, thick biofilms on microwell disheswere rinsed twice with MMS, the rinse solution was removed,and the biofilm was then exposed to 1 ml of 300 µM Znsolution for 1 h. The cells were then washed twice with theMMS rinse solution, and 1 ml of 300 µM quinoline solutionwas added. An aliquot (4 µl) of 5 mM SYTO17 (which stainsall cells with red fluorescence) (42) was injected immediately(recorded time, 0 min). The biofilm samples were visualizedin duplicate at 20-min intervals.

Bacterial cell live/dead distribution was determined by stainingwith nucleic acid-specific fluorochromes: a green fluorescentnucleic acid stain, SYTO 9 (which stains all cells, live ordead), and a red fluorescent nucleic acid stain, propidium iodide(which stains only bacteria with damaged membranes). A reductionin the SYTO 9 fluorescent emission results when both dyes arepresent in the cells, and dead cells produce a yellow fluorescenceas a result of the combined red and green signals. After rinsingwith the MMS rinse solution, the biofilms were stained by adding3 µl of the dye mixture per ml solution according to themanufacturer's protocols. The stained samples were left in thedark for 15 min before microscopic visualization.

Microscopic imaging and data processing.
A Bio-Rad 1024 confocal/2P-LSM microscope system was used forimaging. The system for 2P-LSM is equipped with a Ti:Sapphiremode-locked laser, which provides 100-fs pulses at 80 MHz. Themicroscope system also contains three fluorescence emissiondetectors, an Olympus IX70 inverted microscope with a 100x oil-immersionobjective, a Z-motor that allows a series of sections to betaken automatically, a Marzhuaser X-Y motorized stage for imagelocation, and LaserSharp software for image acquisition. Imageswere further processed and analyzed using Confocal Assistantsoftware (downloaded from ftp.genetics.Bio-Rad.com) and Scion(from NIH Image, http://www.scioncorp.com).

The percentage of active biomass was inferred from the differencebetween the measured average gray values (AGV) of images fromlive (green) (AGV_g) and dead (red) (AGV_r) staining as follows:% active fraction = [AGV_g/(AGV_g + AGV_r)] x 100.

RESULTS

Top

Results

Bacterial growth curves.
The growth of planktonic E. coli PHL628 cells in MMS mediumat 37°C reached early stationary phase after 24 h of cultivation(Fig. 1). Based on the growth curves, an overnight cultivationwas carried out prior to each experiment to provide uniforminocula for biofilm growth. No significant lag phase was observedin the presence of 0.1 mM Zn, but cell growth was mildly retardedin the presence of 1 mM Zn, indicating that Zn had limited toxicityfor planktonic bacterial growth.

Zn sorption to planktonic bacteria.
Zn sorption kinetics and equilibria were evaluated using planktoniccells harvested in stationary phase. As was expected, Zn sorptionto planktonic cells was fast (within an hour), and sorbed Znremained constant as the incubation time increased from 1 to24 h (Fig. 2A). Zn sorption to biomass obeyed a linear isothermup to 0.07 mM Zn (Fig. 2B). Based on mass balance analysis,most of the added Zn (80 to 93%) was sorbed to biomass at acell concentration of 25.8 mg/liter. It appeared that the amountof Zn sorbed to the inactive biomass was significantly higher(t = 0.05, P < 0.04, n = 4) than that of Zn sorbed to theactive biomass.

View larger version (9K):
[in this window]
[in a new window]
FIG. 2. Zn adsorption to planktonic E. coli PHL628 cells at pH 7. (A) Zn adsorption kinetics at 30°C with 0.1 mM Zn ( ${circ}$ ) and 1 mM Zn (•). (B) Zn sorption isotherms with active ( ${blacksquare}$ ) and inactive ( ${square}$ ) (0.15 mM NaN₃-treated) cells at 37°C.

Evaluation of fluorochrome.
The synthesized quinoline fluorochrome specifically bound zinc(Fig. 3A) but not other test metals such as copper and lead(data not shown). Zn addition increased quinoline fluorescenceintensity by 9- to 20-fold (Fig. 3A). Based on the molar ratioof quinoline to added Zn, it appeared that the synthesized quinolinewas capable of forming up to 3:1 complexes with Zn²⁺ (3 molquinoline to 1 mol Zn), similar to other quinoline-based Znchelators (8, 25) (Fig. 3A and 3B). 8-Hydroxy-quinoline canform both 1:1 and 2:1 complexes with Zn²⁺ (depending on pH andsolvent). At the physiological pH of approximately 7, both 1:1and 2:1 complexes are likely to be present in solution (2).

View larger version (11K):
[in this window]
[in a new window]
FIG. 3. (A) Fluorescence intensity as a function of quinoline concentration in the presence ( ${circ}$ ) and absence (•) of 50 µM Zn. (B) Fluorescence intensity as a function of Zn concentration in the presence of 150 µM quinoline.

A series of complexing agents and E. coli PHL628 cell cultureswas selected to determine whether complexed and cell-bound Znspecies were labile with the addition of quinoline. Zn stronglycomplexes with EDTA, with a first-step Zn-EDTA formation constant(log K_f) of 18.3. As Zn²⁺ was complexed with more EDTA, lessZn-quinoline fluorescence was detected (Fig. 4). Similar resultswere obtained for NTA, although the decrease in total fluorescencewas less than that seen with EDTA. This is not surprising giventhe lower binding strength of NTA (log K_f = 12.0) relative toEDTA. Zn and citrate formed a weak complex (log K_f = 6.1) thathad no effect on the Zn-quinoline fluorescence measurements.This was also the case for solutions containing pyruvate andcell cultures, indicating that Zn adsorbed to biomass was labileand could be rapidly (<15 min) desorbed and complexed bythe quinoline.

View larger version (17K):
[in this window]
[in a new window]
FIG. 4. Effect of biomass and complexing agents on Zn-quinoline fluorescence intensity, including EDTA (•), NTA ( ${circ}$ ), citrate ( ${blacksquare}$ ), cell cultures ( ${square}$ ), and pyruvate ( ${blacktriangleup}$ ).

Spatial distribution of Zinc in biofilms.
Biofilm development was investigated using both batch and continuous-flowcultivation techniques. In general, thin biofilms were developedin the batch conditions, with typical thicknesses of 10 to 50µm after 1 to 2 days of cultivation. In comparison, thickbiofilms (up to 350 µm in this study) were obtained onlyin the continuous-flow conditions with LB medium.

After application of the SYTO 9 and propidium iodide stainsin the absence of Zn, thin biofilms fluoresced green (a fewcells close to the substratum fluoresced yellow), indicatingthat the preponderance of cells in the thin biofilms were viableand metabolically active (Fig. 5A and C). The thick biofilmscontained dispersed, brightly fluorescent cell clusters (Fig.5A) with horizontally oriented dark areas, which we presumedto be voids that allow fluid flow (Fig. 5B). As expected, thebiofilm images showed irregular edges (Fig. 5C and D), whichindicate the complex three-dimensional structure of the biofilm.

View larger version (19K):
[in this window]
[in a new window]
FIG. 5. Photomicrographs of E. coli PHL628 biofilms grown in MMS medium using pyruvate as the carbon and energy source in the absence of zinc at 30°C. (A and C) Biofilms in the absence of Zn. (A) Top-down view (4 µm away from the substratum). (C) Cross-sectional view of the biofilm from A with the substratum at the bottom. (B) Cross-sectional view of a biofilm showing fluid channels. (D) Cross-sectional view of a biofilm with nonuniform structures. Dead (inactive) cells were stained red with propidium iodide; live (active) cells were stained green with SYTO9.

Zn additions were distributed quite evenly in the thin biofilms(12 µm) (Fig. 6A and B). In contrast, Zn was located onlyat the surface of thick biofilms, with a penetration depth ofless than 20 µm (Fig. 7A). The results of intensity estimatesshowed that duplicates were similar (Fig. 7B has a mean grayvalue of 38.52 versus that of 38.33 in Fig. 7A). The signalfrom the penetration of Zn into the biofilm was significantlyreduced after 20 min of reaction with the quinoline (Fig. 7C),and the Zn distribution could not be observed after 40 min (Fig.7D). The gradual reduction in the Zn signal with time afterthe addition of quinoline was attributed to diffusion of theZn-quinoline complex out of the biofilm.

View larger version (43K):
[in this window]
[in a new window]
FIG. 6. Photomicrographs of E. coli PHL628 biofilms grown in MMS medium at 30°C after Zn treatment. An aliquot of Zn (100 µM) was added 1 h before staining. Zn in the biofilms was stained blue with 150 µM quinoline. The blue fluorescence has been replaced with the false color green to improve contrast in the image. (A) Top-down view (2.5 µm away from the substratum). (B) Side view with the substratum at the bottom.

View larger version (69K):
[in this window]
[in a new window]
FIG. 7. Zn distribution in E. coli PHL628 biofilms grown in LB medium at 30°C. An aliquot of Zn (300 µM) was added 1 h before staining. Zn in the biofilms was stained blue with 300 µM quinoline, and all cells in the biofilms were stained red with SYTO 17. (A) The blue fluorescence has been replaced with the false color green to improve contrast in the image. Top-down view with the biofilm surface on the right at time zero (immediately after the addition of both dyes). (B) Top-down view at time zero (duplicate with a few minutes delay compared to A). (C) Top-down view after 20 min. (D) Top-down view after 40 min. Bars, 10 µm.

Spatial distribution of active biomass in biofilms.
The active biomass in biofilms, as indicated by live (active)/dead(inactive) staining, was dependent upon biofilm thickness. Ina 36-µm-thick biofilm, more than 90% of the total biomasswas active and uniformly distributed (Fig. 5 and 8); but a nonuniformdistribution of active biomass was observed in a 185-µmbiofilm, where the active fractions varied between 67.7% and73.6%, with an average of 70.7 ± 1.9% throughout thebiofilm structure (Fig. 8). When the thickness of the biofilmincreased to 350 µm, 61.0 ± 3.7% of the biomasswas active at the substratum. The active biomass decreased withdistance up to 20 µm away from the substratum. Between20 and 220 µm from the substratum, active biomass wasat a relatively constant low level (50.6 ± 1.0%) andthen gradually increased back to 64.7 ± 4.9% close tothe biofilm surface.

View larger version (24K):
[in this window]
[in a new window]
FIG. 8. Distribution of active biomass in a series of E. coli PHL628 biofilms. The 36-µm biofilm was grown in MMS medium at 30°C under batch conditions. The 185-µm and 350-µm biofilms were both grown in LB medium at 30°C under continuous-drip flow conditions.

DISCUSSION

Top

Discussion

Several Zn chelators based on a quinoline core, such as 6-methoxy-(8-p-toluenesulfonamido)-quinoline,have been developed for imaging of biological samples (8, 38).These zinc-activated fluorochromes are usually capable of permeatingthe cell, with moderately strong Zn complexation (first-steplog K_f values ranged from 7 to 8.4) (8, 38). The pZn values(–log [Zn²⁺]) for EDTA and the quinoline-based fluorochromesare 14.3 and 9.3, respectively) (8). Hence, quinoline-basedfluorochromes could not readily bind Zn species that were complexedwith EDTA (Fig. 4), but they were efficient in removing Zn thatwas sorbed to biomass. The constant for Zn binding to E. coliPHL628 cell surfaces is likely to be comparable in magnitudeto that of Cd complexation with bacterial surfaces (log K_f =4.0) (44).

With the experimental protocol used in this research, the sequenceof Zn-quinoline interactions within a biofilm include the following:Zn diffusion into and sorption to the biofilm; quinoline diffusioninto the biofilm; Zn desorption in the presence of quinoline;and diffusion of the Zn-quinoline complex out of the biofilm.Quinoline was confirmed to have no sorption to suspended andbiofilm cell cultures by comparing its fluorescence in the presenceand absence of cells (data not shown). Metal sorption to biomasshas been confirmed to be a rapid process, as shown before (10,18) as well as in this work. Quinoline-induced Zn desorptionfrom biomass also happened on a rapid time scale. The diffusionof nonreacting solutes (such as quinoline and the Zn-quinolinecomplexes) within a biofilm occurs on a time scale ranging fromseconds to tens of minutes, depending on the biofilm thickness(33). However, diffusion of metal ions, such as Zn²⁺, into thebiofilm will be retarded by binding to cells and extracellularpolymers and is expected to be much slower than the other processesnoted above. Thus, observations of the spatial distributionof metals using the methods described here need to be performedquickly (within minutes) relative to the time scale for movementof Zn that has desorbed from the biofilm in response to thepresence of the quinoline ligand.

Our results are consistent with those from a recent study thatshowed significant levels of viable cells in the interior regionsof a 10-µm Pseudomonas aeruginosa biofilm after 12 h oftreatment with 1 mM copper (35). Although the research reportedhere did not allow us to quantify the Zn concentration gradientin the biofilm (in part because of the bleedthrough of SYTO17 red fluorescence into the blue channel), our results clearlyshow that after 1 h of equilibration, Zn is mainly located inthe biofilm surface with a penetration depth of less than 20µm (Fig. 7). The slow penetration (hindered diffusion)of Zn inside the biofilm structure versus the fast sorptionof Zn to the biofilm matrix will enhance the survival of cellsin a biofilm under metal stress compared to planktonic cells.

Our results further suggest that total microbial activity decreaseswith increasing depth of the biofilm (Fig. 8). The lower activefraction of cells at the surface of the 185- and 350-µmbiofilms (versus the 36-µm biofilm) may have been dueto the reduced oxygen concentration and/or the accumulationof waste products that would likely accompany the observed celldensity; direct measurements of oxygen and waste products werenot made, however. Furthermore, a U-shaped distribution of activebiomass was observed in a 350-µm biofilm, suggesting thatthe middle part of the biofilm is less active than either thetop (e.g., the biofilm-bulk liquid interface) or the bottom(substratum), perhaps as a result of limited advective transport.However, the significant amount of active biomass distributedall over the thick biofilms suggests that essential nutrients,electron donors, and electron acceptors are still availableto biomass in the inner biofilm regions. Thus, our results donot support predictions based on the concept of homogeneousbiofilm models with active biomass located predominantly nearthe interface of the biofilm and the bulk aqueous phase andinactive biomass in the inner regions (27, 30, 39).

Given our current understanding, confirmed in this study, thatbiofilms are heterogeneous spatial structures consisting ofdense cell clusters (Fig. 5) (11), with open spaces betweenthe clusters, biofilm models are required that permit mass transportthrough both diffusion within cell clusters and advection inthe open channels that penetrate the biofilms. Further workis needed to build a mechanistic model with an appropriate geometricdescription, in which the combined transport processes of advectionand diffusion are accounted for, to better predict the spatialdistribution of biochemical processes in biofilms.

Previous work and this research confirm many advantages of 2P-LSMcompared to single-photon CLSM in studies of interfacial microbialecology, including reduced photobleaching and the ability toanalyze thick films (26, 45). In single-photon CLSM, quinoline-basedfluorescence probes will require UV excitation at 350 to 380nm, and this wavelength is below the range commonly availablein single-photon instruments. More importantly, shorter wavelengthsin the UV range have higher energy and consequently can be damagingto cells (38). Photobleaching is significantly reduced in 2P-LSM,where two photons are used to excite the fluorochrome of interest(the Ti-Sapphire infrared laser used in this work provided lightat 790 nm). The biofilm penetration in this work was sufficientto permit analysis of E. coli PHL628 biofilms with thicknessesas great as 350 µm.

The fact that the amount of Zn sorbed to the inactive biomasswas slightly higher than that sorbed to the active biomass (Fig.2) indicates the possible involvement of a Zn homeostasis mechanismthat regulates the uptake and efflux of zinc in the microbialsystem. In E. coli, high concentrations of zinc are reportedto result in zinc efflux by induction of ZntA, a zinc-transportingP-type ATPase (28, 29). With 2P-LSM and a judicious choice offluorochromes, it is possible to simultaneously visualize thelocalization of intracellular and extracellular Zn and the cellactivity level. Although the results presented suggest thatthe levels of Zn used were not toxic, more work is needed todetermine the actual bioavailability of metals within the biofilmand what role cellular processes, including exopolysaccharideproduction and metal efflux, play in reducing toxicity. Researchof this nature would greatly augment our understanding of thefunction of microbial biofilms under metal stress.

ACKNOWLEDGMENTS

This research was supported, in part, by NSF grant number EAR-0311767.Z.H. is a postdoctoral fellow with support provided by the CornellBiocomplexity and Biogeochemistry Initiative.

We appreciate the assistance of Carol Bayles from the NanobiotechnologyCenter (NBTC) at Cornell University and acknowledge the useof NBTC's shared facilities (NSF STC, ECS-9876771). E. coliPHL628 was a gift from Philippe Lajeune. We also thank StefanBernhard and Jay C. Henderson for providing the quinoline compound.

FOOTNOTES

* Corresponding author. Mailing address: School of Civil and Environmental Engineering, Cornell University, Hollister Hall, Ithaca, NY 14853. Phone: (607) 255-7571. Fax: (607)-255-9004. E-mail: LWL3{at}cornell.edu.

REFERENCES

Top