Experimental Apparatus forSelection ofAdherent Microorganisms underStringent GrowthConditions

Vol. 57, No. 7APPLIED AND ENVIRONMENTAL MICROBIOLOGY, JUlY 1991, P. 1987-19960099-2240/91/071987-10$02.00/0Copyright ©3 1991, American Society for Microbiology

Experimental Apparatus for Selection of Adherent Microorganismsunder Stringent Growth Conditions

GEORGE A. MURGEL,1 LEONARD W. LION,'* CAROLYN ACHESON,2 MICHAEL L. SHULER,2DAVID EMERSON,3t AND WILLIAM C. GHIORSE3

School of Civil Environmental Engineering,' School of Chemical Engineering,2 and Section of Microbiology,3Cornell University, Ithaca, New York 14853

Received 21 December 1990/Accepted 30 April 1991

A bioreactor apparatus is described for studying bacterial attachment. A cyclic, on-off, flow regime was

imposed within the apparatus. Model calculations illustrate the utility of this flow pattern in the selection andmaintenance of slow-growing, adherent organisms. The apparatus is believed to have general utility in testingbacterial attachment influenced by many types of experimental or environmental constraints, includingvariations in fluid dynamics, presence of toxic substances (metals or organics), nature of the substratumsurface, concentrations of limiting nutrients, and competition between bacterial strains. As an exampleapplication, the apparatus was employed to test 14 bacterial strains for surface attachment in a nutrient-limitedgrowth medium. The medium was developed, using the chemical equilibrium program MINEQL, for plannedstudies of biofilms in a solution with a chemically defined composition that permits calculation of trace metalspeciation. The apparatus was used to select organisms with growth and attachment characteristics that couldnot be evaluated by conventional batch, or chemostat, culture conditions. When supplied with acetate,pyruvate, or succinate as a carbon and energy source, the gram-negative strains Pseudomonas cepacia 17616and Zoogloea sp. WG04 showed superior attachment characteristics to glass surfaces in the chemically definedmedium but only moderate fluid-phase growth. The gram-positive Arthrobacter sp. strain 9G4D and gram-

negative species P. pickettii and Zoogloea sp. WNJ8, when supplied with pyruvate as a carbon and energy

source, were capable of superior growth in the fluid phase but formed only a low to moderate biofilm surfacecoverage.

Batch or chemostat culture techniques are commonlyemployed for testing adhesion of bacterial strains underdesired experimental or environmental constraints (3, 7).Batch systems lack an advective fluid flow and, therefore,are unable to apply a selective pressure for variants within aculture that have strong adhesion characteristics. Chemostatsystems may have flow rates that are too high to permitretention of slow-growing cells in the reactor unless the cellsadhere strongly to surfaces. When an additional constraint isimposed, such as reduced nutrient concentrations or thepresence of a toxic substance, the selection of suitablestrains for study becomes more difficult.

In this paper, we describe the development of a cyclic-flow bioreactor apparatus that increases the probability ofobtaining bacterial populations that attach to surfaces underrestrictive growth conditions. Adherent populations wereselected from cultures of bacterial strains that, because ofgrowth limitations or unfavorable attachment characteris-tics, failed to form a biofilm during more commonly em-ployed batch and chemostat testing procedures.Our long-term research interest lies in studying interac-

tions between trace metal adsorption and bacterial attach-ment to inanimate surfaces. The anticipated experimentalwork for this research required development of a definedmedium with a chemical composition that permitted calcu-lation of trace metal speciation. This medium is referred to inthis paper as chemically defined and requires use of compo-nents with defined metal binding constants. Thus, organic

* Corresponding author.t Present address: Center for Microbial Ecology, Michigan State

University, East Lansing, MI 48824.

growth factors (vitamins, amino acids, etc.) were excluded.Low phosphate concentrations were required to preventprecipitate formation. In addition, it was necessary to avoidiron (or other metal) supplements because dissolved ironwould compete for available ligands in solution and ironoxide precipitates would adsorb trace metals. A conse-quence of these restrictions was that many strains of bacteriatested initially did not grow in the chemically defined me-dium. Furthermore, candidate strains that were capable ofgrowth frequently did not attach to surfaces under batch orcontinuous-flow growth conditions. The impetus for devel-opment of the reactor apparatus described in this paper wasthe need to test axenic cultures of film-forming bacteriaunder the chemically defined experimental conditions. Therestrictive growth conditions presented by the chemicallydefined medium are employed in this paper to illustrate theutility of the reactor system. Other applications for thereactor system are also briefly discussed.

MATERIALS AND METHODS

Modeling reactor operation. In theory, operation of aflowthrough reactor system by oscillating the flow couldimprove bacterial colonization of a surface by allowingsufficient time for growth and attachment to occur. Duringthe time when flow is on, washout of bacteria that either arenot capable of attachment or are not firmly attached to thesurface would occur. When the flow is off, attached bacteriacould either detach and grow in the fluid phase of the reactoror continue to spread on the surface. Some of the suspendedcells might return and attach to other sites on the surface.We reasoned that cycling of the flow would apply a selectivepressure for variants within a bacterial culture that adhere

1987

1988 MURGEL ET AL.

strongly to a surface under conditions when their growth ris stringently constrained (such as by toxic substanceslimited nutrient availability). Surface colonization;growth would be enhanced, and the magnitude of the waout dilution rate required to remove the cells fromreactor would be greatly increased (12).A simple model was developed to study the influence

various cyclic-flow regimens on growth and attachmentadherent bacteria in a theoretical chemostat system.following equations were used to describe substrateutilization and growth of suspended and attached cellsand XA, respectively).

dS Q kXsS kXaS

dt V Ks+S Ks+S

dXsdt

QXs YkXsS- + kAXs + kDXa-kdXsV Ks +S

rate TABLE 1. Composition of MMS growth mediumor Composition

and Componenttsh- mg/liter mol/liter

the CaCl2. 2H20 30.0 2.04 x 10-4MgSO4 7H20 34.5 1.40 x 10-4

- of (NH4)2SO4 120.0 9.08 x 10-4t of KNO3 14.8 1.46 x 10-4T'he NaHCO3 0.840 1.00 x 10-5(S) KH2PO4 0.136 1.00 x 10-6(XI Carbon and energya 1000.0 Variable6

NaNO3 Variablec Variableca Acetate, pyruvate, or succinate.b Acetate, 1.22 x 10-2 M; pyruvate, 9.09 x 10-3 M; succinate, 6.17 x 10-3

(1) M.c To maintain constant ionic strength of 0.05 M.

(2)

dXA YkXAS

dt-Ks +- + kAXs-kkDXA-kkdXA (3)

dt Ks + S

where Q = volumetric flow rate (length3/time) (potentiallyvarying with time), V = reactor volume (length3), So =

substrate feed concentration (mass/length3), k = maximumsubstrate utilization rate (time-'), Ks = half velocity orsaturation constant for substrate utilization (mass/length3), Y

= yield coefficient (mass cells/mass substrate), kA = attach-ment rate constant (time-'), kD = detachment rate constant(time-1), and kd = bacterial death rate (time-').For simplicity in the example calculations, growth of

attached and suspended cells is described by Monod expres-sions with the same kinetic coefficients. Similarly, for thepurpose of illustrating the general effect of variable flowcycles, a simple first-order term for surface attachment anddetachment was assumed. Calculations were performed as-suming that initial (time zero) substrate and attached cellconcentrations were zero and that the initial suspended cellconcentration (dry weight cell mass) was 0.28 mg/ml. Theseconditions approximated those ultimately used in the exper-imental reactor system (described below). The hydraulicretention time and substrate feed concentration used in thecalculations were comparable to those used in the actualexperiments (6 h and 1.0 mg/ml, respectively). For purposesof illustrating the model operation, the kinetic coefficientsfor cells were set to approximately equal those obtained inprior experiments for growth of Pseudomonas atlantica in a

chemically defined seawater medium (Y = 0.5, Ks = 0.00035mg/ml, k = 0.25/h) (13, 14).Development of minimal mineral salts (MMS) medium. A

chemically defined medium was needed to allow for plannedstudies of the influence of the bacterial films on trace metaladsorption. The chemical speciation program MINEQL (26)was employed to calculate interactions of proposed chemicalcomponents of the medium based upon known stabilityconstants (4, 5, 10, 17, 23, 24). A balanced mineral saltsmixture (1) was altered by iterative calculations of compo-nent concentrations using MINEQL to select the final me-

dium composition. The final MMS medium composition(Table 1) did not produce precipitates that would influencePb or Cd adsorption, and it allowed Pb or Cd to existpredominantly (=90%) in the free-ion form at circumneutralpH. The MMS medium was low in phosphate because of thepotential for formation of insoluble phosphate salts. A vari-able amount ofNaNO3 was added to maintain constant ionic

strength (0.05 M) with different organic acid carbon andenergy sources.

Bacterial strains. Twenty-four aerobic or facultative, het-erothrophic, bacterial strains from various sources wereinitially selected for study. Their inclusion was based upontheir reported ability to produce extracellular polymers orproduce a surface attached biofilm to bind trace metals andtheir ability to grow in a freshwater defined medium. Cul-tures were maintained on a peptone-yeast extract-glucosemedium consisting of the following (per liter of solution):peptone, 0.25 g; yeast extract, 0.25 g; glucose, 0.25 g;MgSO4 7H2O, 0.6 g; CaCl2, 0.7 g; and agar, 15 g; in 10 mMHEPES (N-hydroxyethylpiperazine-N'-2-ethanesulfonic acid)buffer with pH adjusted to 7.0.The strains were screened in Nunc tissue culture plates

(3.6-ml well volume) for their ability to grow on the de-scribed MMS medium supplemented with various carbonand energy sources. Fourteen of the selected strains (2 grampositive, 12 gram negative) were able to grow when providedwith either acetate, pyruvate, or succinate as a carbon andenergy source (Table 2). These strains were subsequentlytested for surface attachment and biofilm formation in thecyclic-flow reactor.

Selection of attachment surface. Our choice of a materialfor an attachment surface was limited by the requirementsthat the material (i) have a low binding affinity for tracemetals, (ii) be manufactured without the use of trace metals,(iii) be resistant to autoclave temperature, and (iv) be resis-tant to repeated exposure to nitric acid for cleaning (15).Five materials fitting these criteria were selected: glassmicroscope slides (a material that simulates natural siliconoxides such as sand); polypropylene (used in rotating bio-logical contactors) and high-density polyethylene (used intrickling filters) (both from Almac Plastics, Rochester,N.Y.); and Teflon and glass slides silanated with dichlorodim-ethylsilane (Eastman Kodak Co., Rochester, N.Y.) (bothemployed as hydrophobic surfaces with low metal-bindingproperties). The Teflon, polyproplylene, and high-densitypolyethylene used in the reactor were cut from larger sheetsto the same dimensions as standard microscope slides.Chemical and microscopic analyses of cell growth and

biofilm formation. A majority of bacterial strains in naturaland engineered reactor systems utilize production of fibrouspolysaccharide exopolymers as a means of mediating attach-ment to surfaces (6). The biofilm assays employed in thisresearch included both direct counts of attached cells andtechniques which quantitatively and qualitatively measuredpolymer production. Surfaces immersed in the cultures were

APPL. ENVIRON. MICROBIOL.

APPARATUS FOR SELECTION OF ADHERENT MICROORGANISMS 1989

TABLE 2. Bacterial strains showing growth on MMS mediuma

Culture Gram reaction Source

Arthrobacter globiformis + ATCCb 8010Arthrobacter sp. 9G4DC + R. Beloin, Cornell UniversityAcinetobacter calcoaceticus - ATCC 31012Gm(-) isolate (Frazer River, Calif.) - G. Geesey, Montana State UniversityLeptothrix discophora SP-6 - D. Emerson, Cornell UniversityLeptothrix discophora SS-1 - W. Ghiorse, Cornell UniversityPseudomonas cepacia 17616 - T. Lessie, University of MassachusettsPseudomonas cepacia 249-100 - T. Lessie, University of MassachusettsPseudomonas fluorescens - ATCC 13524Pseudomonas pickettii - B. Pyle, Montana State UniversityPseudomonas saccharophila - ATCC 15946Zoogloea sp. U106 - R. Unz, Pennsylvania State UniversityZoogloea sp. WG04 - R. Unz, Pennsylvania State UniversityZoogloea sp. WNJ8 - R. Unz, Pennsylvania State University

a Containing either acetate, pyruvate, or succinate as carbon and energy source.b ATCC, American Type Culture Collection, Rockville, Md.c Isolated from groundwater aquifer sediment.

examined for bacterial cell attachment and biopolymer for-mation by two methods. Slides from sterile, uninoculatedcontrol reactors were carried through all analyses as blanks.Glass slides were rinsed under a gentle stream of deionizedwater from a water bottle and allowed to air dry. A pair ofslides from each reactor was subjected to the standardchemical oxygen demand (COD) procedure (2). Results werereported as microequivalents of oxidizable material per totalslide surface area. Assuming that all oxidizable material inthe COD assay was carbon with an average oxidationnumber of zero, then 1 ,ueq corresponds to 3 ,ug of carbon.By further assuming a typical cell weight (10-12 g) and cellcarbon content (50% by weight), the difference in totalcarbon from the COD assay and conversion of each slide cellcount datum would be representative of exopolymer coverage.A second pair of slides was stained with colloidal iron-

cacodylate stain (21, 22) and examined microscopically toevaluate biopolymer production and to obtain direct countsof cells on the surface. For staining, the slides were floodedwith iron-cacodylate at pH 7.0 for 10 min, washed withdistilled water, acidified with 2 to 3 drops of 0.1 N HCI,counterstained with 2 to 3 drops of 10% potassium ferrocy-anide (Prussian blue reagent), allowed to stand for 3 to 5 min,rinsed a final time with distilled water, and blotted dry. Forthe materials that were optically clear (glass and silanatedglass), the slides were examined and cells were counted byusing a Zeiss Universal light microscope equipped withphase-contrast optics and a calibrated square eyepiece reti-cle to determine the number of cells per unit surface area

(square centimeter). For the opaque materials (Teflon, poly-propylene, and high-density polyethylene), 0.1% aqueousacridine orange was employed to stain the cells previouslystained with the colloidal iron stain before cells on thesurface were counted directly by using epifluorescence mi-croscopy. The colloidal iron stain was not sufficient to allowresolution of the cells by epifluorescence without the acri-dine orange stain. A coverslip was placed on the stained area

and sealed with Vaspar (a standard mixture by weight of 50%paraffin and 50% petroleum jelly) to prevent drying. Cellcounts were of 20 fields per sector and 2 sectors per slide.Growth of cells in suspension was monitored by optical

density at 600 nm (Spectronic 20; Bausch & Lomb Inc,Rochester, N.Y.; and Model 3600 Spectrophotometer;Beckman Instruments, Inc., Irvine, Calif.). Suspended cellswere also observed with the light microscope under phase-

contrast optics to monitor consistency in cell morphologyand for cell clumping (an indication of exopolymer produc-tion). Spread plates containing peptone-yeast extract-glu-cose agar medium were also utilized to check for mainte-nance of pure cultures.

Bioreactor design and construction. A schematic diagramof the reactor system is shown in Fig. 1. The bioreactorvessel consisted of a stoppered 250-ml graduated cylinderfilled to about 192-ml capacity. The reactor was initiallywashed with a 10% reagent grade nitric acid solution andrinsed with deionized water. This was followed by a wash ina 10% redistilled nitric acid solution and a rinse with deion-ized water prior to each use. The reactor was connected tofour separate lines: a T connecting tube (inner diameter[i.d.], 1 cm) with a rubber septum (outer diameter, 1 cm) anda butyl rubber hose as an air inlet; a 15-cm length of glasstubing (i.d., 6 mm) as an effluent draw-off; a 15-cm L-shapedglass tube (i.d., 6 mm) as an air relief vent, and a glass tube(i.d., 12 mm) containing a Pasteur pipette as an inlet feed. Apiece of glass (2.5 by 2.5 cm) was placed on the bottom of thevessel at approximately a 450 angle to keep a set of fourslides from resting on the bottom. The slides were stackedvertically in pairs. A clip fabricated from Teflon was used tohold up to four slides in place during an experiment.A 60-ml glass bottle with rubber stopper was used as a drip

trap on both the air exhaust line and the air inlet line. Glasswool plugs were placed in both lines to prevent contamina-tion by airborne bacteria. Aeration was achieved by bub-bling with a constant temperature air stream provided by anaquarium pump. Air was passed through a flow meterthrottled to 1.97 cm3/s at 25°C and 1 atm (1.0129 kPa) andthen through a 0.2-,um-pore-size disposable syringe filter, adrip trap, and finally a glass wool filter prior to entering thereactor through a gas dispersion tube (pore diameter, 25 to50 ,um).A Masterflex (Cole-Parmer Instrument Co., Chicago, Ill.)

1- to 100-rpm drive powered a multichannel pump head tosimultaneously inject feed solution from a 10-liter carboythrough silicone tubing (i.d., 0.8 mm) and to withdrawreactor effluent through silicone tubing (i.d., 1.6 mm) forcollection in the waste carboy. A 10-ml disposable glasspipette was inserted in the feed line ahead of the pump to actas an air break to prevent back contamination by microor-ganisms from the reactor into the sterile feed solutioncarboy. A Pasteur pipette at the feed inlet to the bioreactor

VOL. 57, 1991


GLASSWOOLFILTER

AIRVENT

'I;

FIG. 1. Schematic of cyclic-flow reactor system.

acted as a second air break to prevent back contamination byheadspace aerosols. A three-way polytetrafluoroethylene(PTFE) valve inserted in the effluent line between the pumpand the collection bottle allowed for draw-off of nonasepticsamples for optical density and pH measurements.

Preparation of the bioreactor for experiments. The biore-actor apparatus was assembled as two separate components:(i) the inlet tubing attached to the feed solution carboy filledwith 10 liters ofMMS medium without carbon source and (ii)the assembled bioreactor containing slides and tubing withdrip traps and the attached effluent carboy. All joints, ex-posed tubing connectors, and tubing ends were wrapped inaluminum foil to protect sterility. The two components werethen sterilized together in an autoclave at 121°C for 90 min.

After sterilization, the two components were transportedto a laminar-flow hood to mate the feed solution carboy lineaseptically to the bioreactor inlet tube. A sterile 0.2-,um-pore-size disposable syringe filter was inserted aseptically in the airinlet line and then attached by silicone tubing to the airflowmeter and air pump. Finally, 100 ml of a 0.2-,um-pore-size-filter-sterilized 10-g/liter carbon source was added asepticallyto the 10 liters of MMS medium in the feed carboy.

Sterile medium (mineral components plus carbon source)was pumped into'the bioreactor until all slides were im-mersed and the effluent draw-off began removing excess

medium ("192-ml volume mark to immerse all slides). Asterile syringe was used to inject 10 ml of inoculum throughthe septum into the bioreactor while the air pump wasoperating. The air flow pushed the injected cells through thegas dispersion tube into the medium. For each experiment,two duplicate reactors were inoculated with a test strainwhile a third reactor was not inoculated, to serve as a sterilecontrol.

After inoculation, the entire apparatus was transferred toa 25'C-constant-temperature room and allowed to operate asan aerated batch reactor for 48 h to allow the test bacteria toacclimate to the reactor environment and to initiate growth(as determined by the appearance of any measurable opticaldensity). At 48 h, the pumps were activated and subse-quently cycled on and off at 12-h intervals by using a timerswitch. The 12-h cyclic on-off flow regime was continued forthe duration of the experiment. During the flow phase, thefeed pump controlled the dilution rate (i.e., fluid residencetime) at a hydraulic residence time of approximately 6 h(e.g., flow = 32 ml/h; dilution rate = 0.167/h). The airflowprovided mixing of the reactor contents during operation.Experiments using a dye tracer verified that the reactorbehaved as a completely mixed system. Samples were takenfrom the effluent sampling tube shortly after the pumps wereactivated each day to determine the liquid-phase optical

1990 MURGEL ET AL.


Substrate 2 Suspended Cells 3 Attached Cells(mg/mL) (mg/mL) (mg/mL)

1.0 1----- I M.* w I I I a I I I I I I0.28 -

-J

EE 0.21-

0

a)C 0.14.la

) 0.07-:0

Clo0.0 -

= 0.28-a)

) 0.21 -10

a)0.0 q3 J j

1-E

o-E

a) 0.07-

Ca)4.I4 n> ,%U.u A

0

-.

cn= cm>

0

co

0.00

<on

,acn cn

-J

E-0

.0-co

cn

wA

U. I IL a----b----- b-----l---

-----------------------t -------------------- ------- ----------- ------------

5 I

::Kizz-zzzzzzzzizi

0-

72.00 144.00 216.00

.00

288.00

.00

C

1.0

--0.75--1~E_%r-

a)U..

4-

0).00.

Co

-o0

1.00

0.750

0.500

0.250

0.0

25- I

o I IO0 I. ...F--* .wI _s b- -.-p-.-.-.--. -I- -V0.0

Time (hours)FIG. 2. Concentrations of substrate (1), suspended cells (2), and attached cells (3) in a chemostat with steady flow. (A) Washout of a

nonadherent strain. (B) Washout of an adherent strain with the same growth parameters as in panel A, except kA = 0.05/h and kD = 0.175/h.(C) Selection of a more adherent strain with the same growth parameters as in panel B, except kD reduced to 0.125/h. (Note that the verticalscale for cell concentration is magnified in panels A and B).

T _ I__ I I I1

B

1

VOL. 57, 1991

--7

A 7-

VF.r-.V

72.00 144.00 21 6.00 28E1.0

1992 MURGEL ET AL.

I Substrate 2 Suspended Cells(mg/mL) (mg/mL)

3 Attached(mg/mL)

1 44.00

Time (hours)

Suspended Cells(mg/L)

4 Attached Variant Cells(mg/L)

2 Attached Cells(mg/L)

3 Suspended Variant Cells(mg/L)

0.0 72.00 144.00

Time (hours)

FIG. 3. (A) Effect of cyclic flow (12 h on, 12 h off) on the adherent cell strain shown in Fig. 2B. (B) Selection of the cell strain shown inpanel 3A under cyclic-flow conditions when it is a variant (10% of initial suspended population) (for 1 and 2, detachment rate constant kD =

0.5/h; for 3 and 4, kD = 0.175/h).

density and pH. After 6 and 12 days of cyclic operation (8and 14 days total), one of the two inoculated reactors (foreach strain) was sacrificed and the attachment surfaces were

withdrawn for the chemical and microscopic analyses ofbacterial attachment and biofilm growth.

RESULTS AND DISCUSSION

Reactor modeling calculations. Example results illustratingthe operation of simple model reactors are shown in Fig. 2and 3. The 12-day simulation period matched the maximum

length of our operation of the experimental reactor systemunder conditions of cyclic flow (see Materials and Methods).Figure 2 corresponds to normal chemostat operation withconstant flow Q. Comparison of Fig. 2A and 2B illustratesthe hypothetical washout of nonadherent (Fig. 2A) andadherent (Fig. 2B) (kD = 0.175/h, kA = 0.05/h) cell strainswith otherwise identical growth characteristics. As ex-

pected, in the model attachment acts to retard washout fromthe reactor system. A 30% reduction in the detachment rateconstant (from kD = 0.175/h to 0.125/h, with kA = 0.05/h)confirms the well-established fact that a conventional che-

Cells

o 1.00

-0co E 0.750

C 0.500

'o 0 0.25000. 0

0.0

z~~~~~~~~~~~~~~~~~~~~--A --A

-- - - -,-

288.00

b- 0.5000

*0 -..4 -J

0 0.375coE-.o %-.,

*-.

4 cX 4) 0.250

C 0.125

Q cecoo

*.o

(flen0.0


0.0 72.00 216.00

216.00 288.00


Blank ControlAcinetobacter calcoaceticus

Arthrobacter globiformisArthrobacter sp. 9G4D

Gm(-) isolate(Frazer River, CA)Leptothrix discophora SP-6(S)

Leptothrix discophora SS-1Pseudomonas cepacia 17616

P. cepacia 249-100P. fluorescens

P. pickettii (Mt. St. U. isolate)P. saccharophila

Zoogloea sp. U106Zoogloea sp. WG04Zoogloea sp. WNJ8

. .I I I I I I. . .

0 50 100 150 200g Equivalents of oxidizable material/slide

250

FIG. 4. Biofilm surface coverage (in microequivalents of oxidizable material per slide) determined from slides in the bioreactor after 6 days(8 total days) of cyclic-flow reactor operation. (Note: the Leptothrix cultures contained a Caulobacter sp. contaminant. Coverage forLeptothrix strains was reduced by a 2:1 ratio of cell counts of Caulobacter spp. and Leptothrix spp. on the slides.)

mostat can select for organisms with strong adhesive prop-erties (Fig. 2C).The simulation in Fig. 2B depicts the bacterial strain being

washed out of the reactor in spite of the fact that it was ableto attach to a surface. Figure 3A illustrates the results for thesame strain if the flow into the reactor follows on-off cycles.The model predicts retention of the strain under theseconditions. A 12-h period of oscillation is employed tosimulate the period which was ultimately chosen for ourexperiments. However, the cycle period could serve as anexperimental variable which might be optimized, for exam-


Arthrobacter globiformisArthrobactersp. 9G4D






ple, to ensure maintenance of a desired minimum substrateconcentration. As shown in Fig. 3A, during the off cycle thehigher cell density in the solution forces the system to higherlevels of cell attachment while nutrients are depleted. Duringthe on cycle, nutrients are replenished, and the suspendedpopulation declines through washout. A stable oscillation isestablished in the reactor population. In Fig. 3B, the samestrain is modeled as an attached variant which makes up 10%of the initial population. The remainder of the population isassumed to have identical kinetic constants for cell growthand death but has a higher detachment rate of 0.5/h (versus

2000

. . . . . I . . . .

400

103 Cells/cm2

600 800

FIG. 5. Biofilm cell count from slides removed after 6 days (8 total days) of cyclic-flow reactor operation.

1E Succinate

PyruvateAcetate

......... . .. . .. ..............................................I.........................................................I. Z

_fM

iii.=RWM- [2 SuccinateU Pyruvate

_13_ Acetate

=................~ ~.. .. .... ...... ... .......................... ...................... . . . 9 . . . . I . . . . I . . . . I . . . .

VOL. 57, 1991

1994 MURGEL ET AL.


Arthrobacter globiformisArthrobacter sp. 9G4D






,

Zes. , I . . .I0. . . . . . . . I .

0.00 0.05 0.10 0.15 0.20OD@ 600 nm.

FIG. 6. Fluid-phase optical density after 6 days (8 total days) of cyclic-flow reactor operation.

the variant detachment rate of 0.175/h). The model resultsshow that the reactor will operate to select the variant withsuperior adherent properties and (given the assumed kineticconstants) the variant will constitute the preponderance ofthe reactor population after 12 days.These model calculations confirmed the intuitive expecta-

tion that flow cycles can provide favorable attachmentconditions for adherent bacteria that would be washed outunder normal continuous-flow chemostat operation becauseof limitations of growth rate or attachment properties. In theexperiments described below, such limitations were imposedby the need for a chemically defined medium that lackedmany of the nutrient components normally provided tosupport bacterial growth. The experimental results obtainedconfirmed that cyclic-flow operation would yield growth andattachment that the continuous and batch culture procedureswere not capable of achieving.

Reactor experimental results. The reactor system was usedto evaluate biofilm development by each of the 14 strains onglass slides over an 8- to 14-day time period in the MMSmedium (Table 2). All 14 strains grew in MMS-pyruvatemedium. Six grew on acetate, and four grew on succinate. Inmost cases the COD results, which provide a measure ofcells plus polymer, were confirmed by direct microscopiccell counts.

Figure 4 shows the results of COD tests on slides from thereactors. The results shown are for the reactor slides de-structively processed after 6 days of cyclic operation (totalrun time, 8 days). The results from the 8-day reactors for a

majority of the bacterial strains yielded a partial monolayercell coverage greater than 1.5 x 104 cells per cm2 (Fig. 5)making accurate cell counts difficult even at a magnificationof x2,000. Coverage on slides after 14 days of reactoroperation for a majority of the bacterial strains was higherthan that found at 8 days. Cell coverage for strains at day 14had to be estimated as a percentage of the total slide surfacearea (40.74 cm2). Some representative 14-day coverage

results are as follows: Pseudomonas cepacia 17616 on

MMS-acetate, 94%; P. cepacia 249-100 on MMS-pyruvate,

89%; P. fluorescens on MMS-pyruvate, 81%; Zoogloea sp.WGO4 on MMS-pyruvate, 90%; Acinetobacter calcoace-ticus on MMS-pyruvate, 52%, and on MMS-acetate, 36%;Frazer River, Calif., isolate on MMS-pyruvate, 10%; andboth P. saccharophila and P. pickettii on MMS-pyruvate,11%.On the basis of the COD analysis in Fig. 4, P. cepacia

17616 produced the highest biofilm surface coverage withacetate and succinate as the carbon and energy source (220and 145 pLeq per slide, respectively). Only Zoogloea sp.WGO4 on MMS-pyruvate exhibited a surface coveragecomparable to that of P. cepacia 17616 on MMS-acetate(Fig. 4). Direct cell counts also showed high surface cover-

age (cells per slide) for P. cepacia 17616 as well as Zoogloeasp. WGO4 with MMS-pyruvate, Zoogloea sp. WNJ8 withMMS-acetate, P. saccharophila with MMS-pyruvate, theFrazer River isolate with MMS-pyruvate, and A. calcoace-ticus with MMS-pyruvate and acetate (Fig. 5). Most of thegram-negative strains showed significantly better biofilmdevelopment on all carbon and energy sources as judged byboth COD analysis and direct microscopic cell counts thanthe two gram-positive Arthrobacter strains (Fig. 4 and 5).However, many of the gram-negative strains displayed lessgrowth in the fluid phase than the Arthrobacter spp. as

indicated by optical density measurements (Fig. 6). Forexample, Arthrobacter sp. 9G4D grew in the fluid phase onMMS-pyruvate medium to an optical density nearly twicethat of Zoogloea sp. WNJ8 and P. fluorescens on MMS-pyruvate, P. cepacia 17616 on MMS-acetate, and Acineto-bacter calcoaceticus on MMS-succinate. Zoogloea sp.WGO4 produced large clumps or aggregates that tended tosettle out of suspension and may have reduced opticaldensity measurements. Only the gram-negative strain P.pickettii on MMS-pyruvate showed fluid-phase growth com-

parable to that of the Arthrobacter spp. The low opticaldensity and high surface coverage achieved by strains suchas P. cepacia 17616 and Zoogloea sp. WGO4 supports theconcept that attachment to a surface allows bacteria tomaintain their populations in environmental conditions that,

El SuccinatePyruvate

II] Acetate

0.25

'rI I I I I I I I I I I I I I . ..

mf00 I'llI I I I I "I'M

xxa=



TABLE 3. Comparison of biofilm development on glass slides, asindicated by colloidal iron-cacodylate stain after 14 days of

growth on MMS medium in the cyclic-flow reactor

Growth" on carbon/energyBacterial strain source

Pyruvate Succinate Acetate

Arthrobacter globiformis 1+ 1+ bArthrobacter sp. 9G4D 2+ 1+ 3+Acinetobacter calcoaceticus 5+ 5+Gm(-) isolate (Frazer River, Calif.) 2+Leptothrix discophora SP-6 3 +Leptothrix discophora SS-1 1 +Pseudomonas cepacia 17616 1+ 2+ 5+Pseudomonas cepacia 249-100 4+Pseudomonas fluorescens 5+Pseudomonas pickettii 3+Pseudomonas saccharophila 2+ 2+ 1+Zoogloea sp. U106 1+Zoogloea sp. WG04 2+Zoogloea sp. WNJ8 3+ 1+

aSymbols: 1 + to 5 +, relative rating of biofilm staining (1 + = barely visible;5+ = very intense blue color); -, not tested.

b No results: reactor became contaminated before completion of theexperiment.

otherwise, would not support their growth. Some strains(e.g., P. cepacia 249-100) exhibited both moderate growth inthe fluid phase and biofilm development, indicating theiradaptability under the stringent environmental conditionsimposed by the minimal MMS medium.

Cell coverage of the slides in the reactor was generallylimited to a partial monolayer. However, there were in-stances (e.g., P. fluorescens on MMS-pyruvate) where dis-tinct microcolonies were found on the slide surfaces takenfrom the 14-day reactors. In such cases, the cell colonieswere several layers thick with patchy monolayer coveragebetween colonies. These microcolonies were not evident onslides from the 8-day reactors.

Biofilm formation and polymer production were qualita-tively evaluated by microscopic observation of color devel-opment after staining with the colloidal iron-cacodylatestain. Results are summarized in Table 3. Biofilms of P.fluorescens (on MMS-pyruvate), P. cepacia 17616 (on MMS-acetate) and A. calcoaceticus (on MMS-pyruvate or acetate)displayed the most intense reaction with the stain. Theresults for the latter two strains correlated well with the datain Fig. 4 and 5 for biofilm and cell coverage. For most of theother strains, however, the staining data in Table 3 (obtainedafter 14 days of reactor operation) do not directly correlatewith the biofilm and cell coverage data in Fig. 4 and 5(obtained after 8 days of reactor operation). The differences

in staining results may reflect differences in the types or

amounts of acidic functional groups that the exopolymercontains as it ages. Since the Fe(II) used in the stainingprocess binds to anionic functional groups on the polymerand cell surface, low relative visibility of the stain may implya higher percentage of nonionic or cationic groups or both.The intense reaction may correspond to a larger density ofanionic groups for which the stain has the greatest speci-ficity.

Table 4 provides a comparison of results from batchculture and cyclic- and continuous-flow experiments utiliz-ing the cyclic-flow reactor apparatus. The continuous-flow,batch, and cyclic-flow values are the results of analysesperformed on glass slides 8 days after the experiment wasstarted (6 days after flow cycles were started). The strains P.cepacia 249-100 and Zoogloea sp. U106 were selected forthis experiment. Both strains, while not the best candidatesin terms of growth and attachment, showed consistent,moderate growth in the fluid phase (Fig. 6) and adequateattachment as a biofilm (Fig. 4 and 5). The resulting datashow that cyclic-flow growth of both strains generally exhib-ited a higher degree of biofilm coverage than was attainedwith the batch and continuous-flow reactor operation. Thefluid-phase optical density was also significantly lower thanin the batch reactor but generally higher than in the contin-uous-flow reactor.The experimental apparatus was also used in a preliminary

experiment to study attachment of Arthrobacter sp. 9G4D toseveral alternative surfaces. Table 5 shows that normal glassslides supported the most dense surface coverage. Thenormal glass-slide-containing reactors also produced thehighest mean liquid-phase optical density, indicating thatmore growth occurred in the reactor. After 12 days ofcyclic-flow reactor operation, silanated glass and high-den-sity polyethylene surfaces supported a less dense biofilmthan at 6 days, indicating that detachment from the hydro-phobic surface had occurred during the experiment. Theopposite effect was observed with normal glass slides, whichmay have more affinity for the gram-positive Arthrobactersp. 9G4D cell surface.By using the reactor and cyclic-flow regimen described in

this paper, we were able to study bacterial strain attachmentunder operating conditions which conventional batch cultureor continuous-flow bioreactor methods could not accom-plish. Comparison of the data collected from batch andcontinuous-flow experiments demonstrates the utility ofcyclic-flow patterns for selection of slow-growing adherentstrains. The data indicated that many strains which did notgrow well in the MMS medium (e.g., P. saccharophila,Leptothrix spp.) exhibit sufficient surface attachment toprevent the strain from being washed out of the cyclic-flowreactor system. This result was consistent with calculations

TABLE 4. Comparison of biofilm development on glass surfaces in three systems, using pyruvateas a carbon and energy source in MMS medium"

Surface coverage Direct cell count (103 celIS/CM2) Fluid-phase optical density (A600)Strain (p.eq of oxidizable material/slide)"'ietcl on 1~clsc2

Batch Continuous Cyclic Batch Continuous Cyclic Batch Continuous Cyclic

Control 3.7 2.7 0.00 17.8 17.9 17.6 0.003 0.000 0.000P. cepacia 249-100 19.7 25.3 81.8 2,100 738 24,100 0.220 0.032 0.132Zoogloea sp. U106 2.5 7.0 8.1 35.2 38.1 69.6 0.174 0.002 0.002

a All experiments were done for 72 h.b Glass slide area is 40.742 cm2.

VOL. 57, 1991

1996 MURGEL ET AL.

TABLE 5. Growth of Arthrobacter 9G4D on pyruvate in MMSmedium under cyclic-flow conditions with

different surface materials

Surface cell count Bulk fluidSubstratum (10 cells/cm2)a optical densityb

Day 6c Day 12C (at 600 nm)

Silanated glass 4.2 0.44 0.06Polypropylene 0.66 0.69 0.08PTFE (Teflon)d 0.37 0.81 0.06High-density polyethylenee 15.0 3.7 0.07Normal glass 1.9 4.9 0.36

a Mean values from two slides per reactor, two sectors per slide, 20 fieldsper sector.

I All values are the mean of 14 daily measurements.c Time of reactor operation under cyclic flow conditions (8 and 14 days

total).d The Teflon surface was uneven, and a coverslip could not be used. Counts

were performed directly in oil, without a coverslip, with a neofluoar lens.e The high-density polyethylene surface was uneven and malleable, making

focusing difficult. There were loose cells suspended in the solution under thecoverslip that were not counted because of problems with drift and focus.

obtained by using a simple model of the reactor systemillustrating the utility of the technique.The cyclic-flow apparatus was found to be advantageous

for studying 14 candidate strains to select a strain for use infurther studies of metal-binding bacterial biofilms underconditions restrictive to bacterial growth. The reactor sys-tem and cyclic-flow operation should also be useful forselection of adherent cells under a wide variety of conditionsother than those we employed. A major advantage of thissystem is its adaptability to different experimental condi-tions. The reactor system is suitable for studies of a varietyof factors that may influence bacterial attachment to inani-mate surfaces. This could include influences of the type ofattachment surface (3, 8) or microbial exopolysaccharides(25); thermodynamic, electrostatic, and kinetic consider-ations (7, 16, 18-20); and predation (9, 11) or competitionbetween two or more bacterial strains on biofilm formationand maintenance. The reactor system could also be used toevaluate the effects of fluid shear on bacterial attachment byaltering gas flow rates or hydraulic residence times, theeffects of nutrient limitations (other than carbon) on attach-ment, and the influence of toxic substances.

ACKNOWLEDGMENTS

This work was supported by National Science Foundation GrantBCS-8617408.We thank Bertha Rodriguez for assistance in collecting data from

the cyclic-flow experiments and Maria Cochran and MaryElizabethGrace for performing the direct cell counts on the bioreactor slides.The technical assistance of Sharon Best is gratefully acknowledged.

REFERENCES1. Adams, L. F., and W. C. Ghiorse. 1986. Physiology and ultra-

structure of Leptothrix discophora SS-1. Arch. Microbiol. 145:126-135.

2. American Public Health Association. 1976. Standard methods forexamination of water and wastewater, 14th ed., p. 550-554.American Public Health Association, Washington, D.C.

3. Baier, R. E. 1980. Substrata influences on adhesion of microor-ganisms and their resultant new surface properties, p. 59-104. InG. Bitton and K. C. Marshall (ed.), Adsorption of microorgan-isms to surfaces. John Wiley & Sons, Inc., New York.

4. Bjerrum, J. (compiler). 1957. Organic ligands. In Stabilityconstants of metal-ion complexes. IUPAC special publicationno. 6. The Chemical Society, London.

5. Bjerrum, J. (compiler). 1958. Inorganic ligands. In Stabilityconstants of metal-ion complexes. IUPAC special publicationno. 7. The Chemical Society, London.

6. Costerton, J. W., G. G. Geesey, and K. J. Cheng. 1978. Howbacteria stick. Sci. Am. 238:86-95.

7. Daniels, S. L. 1980. Mechanisms involved in sorption of micro-organisms to solid surfaces, p. 7-58. In G. Bitton and K. C.Marshall (ed.), Adsorption of microorganisms to surfaces. JohnWiley & Sons, Inc., New York.

8. Fletcher, M., and G. I. Loeb. 1979. Influence of substratumcharacteristics on the attachment of a marine pseudomonad tosolid surfaces. Appl. Environ. Microbiol. 37:67-72.

9. Fletcher, M., and K. C. Marshall. 1982. Are solid surfaces ofecological significance to aquatic bacteria?. Adv. Microb. Ecol.6:199-236.

10. Hogfeldt, E. (compiler). 1982. Part A: inorganic ligands. InStability constants of metal-ion complexes. IUPAC chemicaldata series no. 21. Pergamon Press, Elmsford, N.Y.

11. Hoppe, H. G. 1984. Attachment of bacteria: advantage ordisadvantage for survival in the aquatic environment, p. 283-302. In K. C. Marshall (ed.), Microbial adhesion and aggrega-tion. Springer-Verlag KG, Berlin.

12. Howell, J. A., C. T. Chi, and U. Paulowsky. 1972. Effect of wallgrowth on the scale-up problems and dynamic operating char-acteristics of the biological reactor. Biotechnol. Bioeng. 14:253-265.

13. Hsieh, K. M., L. W. Lion, and M. L. Shuler. 1985. Bioreactorfor the study of defined interactions of toxic metals and biofilms.AppI. Environ. Microbiol. 50:1155-1161.

14. Hsieh, K. M., L. W. Lion, M. L. Shuler, and W. C. Ghiorse.1988. Trace metal interactions with microbial biofilms in naturaland engineered systems. Crit. Rev. Environ. Control 17:273-306.

15. Juran, R. (ed.). 1975-1986. Modern plastics encyclopedia. Mc-Graw-Hill, Inc., New York.

16. Kjellenberg, S. 1984. Adhesion to inanimate surfaces, p. 51-70.In K. C. Marshall (ed.), Microbial adhesion and aggregation.Springer-Verlag KG, Berlin.

17. Perrin, D. D. (compiler). 1979. Part B: organic ligands. InStability constants of metal-ion complexes. IUPAC chemicaldata series no. 22. Pergamon Press, Elmsford, N.Y.

18. Roper, M. M., and K. C. Marshall. 1978. Effects of a claymineral on microbial predation and parasitism of Escherichiacoli. Microb. Ecol. 4:279-289.

19. Rutter, P. R. 1980. The physical chemistry of the adhesion ofbacteria and other cells, p. 103-135. In A. S. G. Curtis and J. D.Pitts (ed.), Cell adhesion and motility. Cambridge UniversityPress, New York.

20. Rutter, P. R., and B. Vincent. 1984. Physicochemical interac-tions of the substratum, microorganisms, and the fluid phase, p.21-38. In K. C. Marshall (ed.), Microbial adhesion and aggre-gation. Springer-Verlag KG, Berlin.

21. Seno, S., M. Abata, T. Ono, and T. Tsujii. 1985. Fine granularcationic iron colloid. Histochem. J. 82:307-312.

22. Seno, S., T. Tsujii, T. Ono, and S. Ukita. 1983. Cationiccacodylate iron colloid for the detection of anionic sites on cellsurface and the histochemical stain of acid mucopolysaccha-rides. Histochem. J. 78:27-31.

23. Sillen, L. G., and A. E. Martell (compilers). 1964. Stabilityconstants of metal-ion complexes. IUPAC special publicationno. 17. The Chemical Society, London.

24. Sillen, L. G., and A. E. Martell (compilers). 1971. Stabilityconstants of metal-ion complexes. IUPAC special publicationno. 25, supplement no. 1 to special publication no. 17. TheChemical Society, London.

25. Sutherland, I. W. 1983. Microbial expolysaccharides-their rolein microbial adhesion in aqueous systems. Crit. Rev. Microbiol.10:173-201.

26. Westall, J. C., J. L. Zachary, and F. M. M. Morel. 1976.MINEQL-a computer program for the calculation of chemicalequilibrium composition of aqueous systems. Technical noteno. 18, water quality laboratory. Department of civil engineer-ing, massachusetts institute of technology, Cambridge, Mass.


Experimental Apparatus forSelection ofAdherent Microorganisms underStringent GrowthConditions

Documents

Transcript of Experimental Apparatus forSelection ofAdherent Microorganisms underStringent GrowthConditions