Plasmid Retention and Gene Expression in Suspended and Biofilm Cultures of Recombinant Escherichia coli Ching-Tsan Huang,' Steven W. Peretti? and James D. Bryers',* 'The Center for Biochemical Engineering, Duke University, Durham, NC 27708-0287; and 2Department of Chemical Engineering, North Carolina State University, Raleigh, NC 27695 Received May 6, 1992Mccepted August 12, 1992 Differences in plasmid retention and expression are studied in both suspended and biofilm cultures of Escherichia coli DH5a(pMJR1750). An alternative mathematical model is proposed which allows the determination of plasmid loss probability in both suspended batch and continuously fed biofilm cultures. In our experiments, the average probability of plasmid loss of E. coli DH5a(pMJR1750) is 0.0022 in batch culture in the absence of antibiotic selection pres- sure and inducer. Under the induction of 0.17 mM IPTG, the maximum growth rate of plasmid-bearing cells in sus- pended batch culture dropped from 0.45 h-' to 0.35 h-', and the p-galactosidase concentration reached an experi- mental maximum of 0.32 pg/cell 4 hours after the initiation of induction. At both 0.34 and 0.51 mM IPTG, growth rates in batch cultures decreased to 0.16 h-', about 36% of that without IPTG, and the /3-galactosidase concentration reached an experimental maximum of 0.47 pg/cell 3 hours after induction. In biofilm cultures, both plasmid-bearing and plasmid-free cells increase with time reaching a plateau after 96 hours in the absence of both the inducer and any antibiotic selection pressure. Average probability of plasmid loss for biofilm- bound E. coli DH5a(pMJR1750) population was 0.017 without antibiotic selection. Once the inducer IPTG was added, the concentration of plasmid-bearing cells in bio- film dropped dramatically while plasmid-free cell numbers maintained unaffected. The p-galactosidase concentration reached a maximum in all biofilm experiments 24 hours after induction; they were 0.08, 0.1, and 0.12 pg/cell under 0.170.34, and 0.51 mM IPTG, respectively. 0 1993 John Wiley & Sons, Inc. Key words: plasmid retention gene expression biofilm p-galactosidase segregational instability Escherichia coli INTRODUCTION The applications of recombinant DNA techniques have created a new era in biotechnological research and de- velopment. Many successful examples are reported on the use of recombinant DNA cultures in agriculture, in health care, and in the commercial production of phar- maceuticals and special chemicals. Plasmids containing genes of interest can be expressed in different desired procaryotic and eucaryotic organisms.'5323 However, there are two impediments to widespread commercial * To whom all correspondence should be addressed. utilization of genetically engineered expression systems: (1) the unstable characteristics of plasmids under differ- ent culture conditions; and (2) the restrictive regulations which constrain the use of recombinant strains within natural environments. This latter restriction is mainly due to a lack of knowledge regarding the fate of recombi- nant DNA released in a natural ecosystem. Two mechanisms will lead to the loss of the original plasmid DNA and subsequent cloned gene expression: structural and segregational instability. Structural insta- bility arises from physical changes in the plasmid DNA molecule which usually involves a deletion and/or inser- tion of a segment of plasmid DNA, or the rearrangement of DNA sequences within the plasmid. Segregational in- stability is the result of random, unequal partitioning of plasmid DNA molecules between daughter cells upon division, potentially leading to the generation of a cell containing no plasmids. Tjpically, plasmid-free cells ex- hibit higher turnover rates than their plasmid-bearing counterparts, thus plasmid-free cells will dominate plasmid-bearing cells after several generation^.'^,'^^^^^^^*^^ Significant research has focused on those factors that control plasmid segregational and structural stabilities in close systems, (e.g., well-controlled suspended pure culture bioreactors) .26p36,39,45,46 The mathematical model of plasmid loss in batch culture was first proposed by Imanaka and Aiba in 1981.23 They derived an expression to describe plasmid loss in batch cultures based upon the following assumptions: (1) host cells grow exponen- tially; ( 2 ) those cells that lose all plasmids cannot regain plasmids (i.e., conjugation and transformation are ig- nored); (3) the probability of losing the plasmid per cell division is constant; and (4) the initial plasmid-free cell concentration is zero. The relation between the plasmid- bearing cell fraction, generation number and probability of plasmid loss is: However, in open environmental systems, a majority of the microbial activity occurs at an interface, within thin Biotechnology and Bioengineering, Vol. 41, Pp. 211-220 (1993) 0 1993 John Wiley & Sons, Inc. CCC 0006-3592/93/020211-010 biological layers known as biof i lm~.~,~ Release of plasmid- recombinant microorganisms, either inadvertently or in- tentionally, to an open ecosystem mandates research into those fundamental organism/plasmid processes that in- fluence plasmid retention, stability, and transfer. Immobilization and Plasmid Stability That immobilization might stabilize a plasmid-bearing population in suspension can be shown mathematically.’ Plasmid persistence in suspended cultures has been ob- served in cases where the plasmid-bearing cell was at a growth rate di~advantagel,’~; this observed persistence has been directly attributed to biofilm formation and cell sloughing from the film.14 Inloes et al.24 reported the maintenance of a plasmid-containing strain of E. coli in the absence of antibiotic selection pressure when immo- bilized in a hollow fiber membrane. The effect of immobilization on plasmid stability has been investigated by de Taxis du Poet et al.” Plasmid- bearing E. coli were immobilized in K-carrageenan beads that were subsequently fluidized in a chemostat oper- ated at a volumetric residence time of 15 minutes. Cells extracted from the beads and resuspended were shown to have the same plasmid-loss frequency as suspension- cultured cells. However, it was reported that gel immo- bilization enhanced the stability of the plasmid-bearing population. Nasri et al.32 extended this analysis to three geneti- cally differrent E. coli hosts (HB101, W3101, and B) using the same plasmid (pTG201), and again reported that, without antibiotics, the fraction of cells in the beads carrying the plasmid was greater than one would find in suspension cultures. In free suspension, all strains studied exhibited varying degrees of plasmid in- stability; when immobilized, all three strains exhibited stable plasmid maintenance for the duration of the cul- ture. When plasmid-bearing and -free cells were co- immobilized, plasmid-free cells did not overrun the culture. This last result suggests that increased plas- mid stability may not be due to plasmid transfer be- tween cells. Sayadi et a1?8 repeated Nasri’s work with just E. coli B (pTG201) but under different growth nutrient limi- tations. This third study found a decreasing specific growth rate increased plasmid copy number and cloned gene activity but decreased stability. They also found that immobilization, in the absence of antibiotic selec- tion, increased stability of the plasmid under glucose, ni- trogen, or phosphate limitations, but not for magnesium limited growth. Unfortunately, all the above reports are difficult to interpret. The gel-immobilized cultures were grown in a chemostat operated at a volumetric residence time that was much higher than possible for the suspen- sion cultures used in their comparison. None of the three studies determined the intrinsic growth rates of the cells inside the gel beads. Thus, it is difficult to as- certain whether the apparent increase in stability is due to decreased growth rate in the bead resulting from nu- trient mass transport limitations, plasmid transmission between cells, or other mechanisms. Many studies of plasmid transfer rates for suspended cells indicates that transfer can occur at significant rates under a variety of conditions and organism^,'^^'^^^^^^'^^^ including Pseudomonas grown in the presence of com- peting organisms from a natural ecosystem.39 Studies of plasmid transfer in aquatic systems show that transduc- t i ~ n , ’ ~ tran~formation>~*~~ and triparental mating” are all possible means of plasmid movement in these systems. Studies of Streptococcus f a e ~ a l i s ~ ~ ~ further indicated that conjugative plasmids that transfer at a relatively low fre- quency in suspension per donor), when immobi- lized to a surface, exhibit significantly higher transfer frequencies (> per donor). There are a number of strategies to improve plasmid stability in recombinant bacterial systems. It comes as no surprise that immobilization might be responsible for alterations in cellular behavior. Many different immobi- lized cell systems have exhibited altered metabolism. Studies done with immobilized Saccharomyces have in- dicated alterations in prod~ct ivi ty ,~~ macromolecular composition, l’ and regulation of glycolytic oscillations. l3 E. coli has been found to exhibit changes in optimal growth conditions and product yields?l while Bacillus has exhibited an altered cell morphology upon immobi- l i ~ a t i o n . ~ ~ In studies of plasmid transfer in sterile soil, Graham and I s t o ~ k ” , ~ ~ found that transformation oc- curred in B. subtilis in the absence of plasmids or trans- ducing bacteriophage. Similarly, t r a n s d ~ c t i o n ~ ~ , ~ ~ and c o n j u g a t i ~ n ~ ~ ~ ~ ~ have been observed for soil microorgan- isms in natural biofilms. Kumar and SchiigerlZ7 provide an excellent review of the observed increases in plasmid retention, cloned gene expression, and system protein productivity seen in a variety of immobilized cell (bacte- rial, yeast, and animal cell) systems. Unfortunately, no concise explanation exists for these observed improve- ments in plasmid stability. This brief section indicates the ubiquitous occurrence of plasmid movement when considering the fate of ge- netically engineered DNA in the natural environment. These transfer processes occur in both gram-positive and -negative organisms, in suspended and biofilm-bound communities. Quantification of the risks involved in the release of plasmid-bearing cells to the environment, be it inadvertent or intentional, will therefore require quan- titative, mechanistic information regarding both the sur- vival and mobility of the original host/plasmid system in these environments (suspended and immobilized) and the frequency of plasmid transfer from the original host to indigenous microorganisms. Factors Affecting Plasmid Retention and Expression in Immobilized Cultures Experimental observations of the effects of immobili- zation, within artificially formed gel bead carriers, on 212 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 41, NO. 2, JANUARY 20, 1993 recombinant plasmid stability can be summarized as follows : 1. plasmid vectors exhibit higher stability in a reactor system, even in the absence of antibiotic selection; 2. reduction or elimination of plasmid structural insta- bilities3*; and 3. plasmid copy number maintained or increased.32 Several hypotheses have been suggested to explain the effects of immobilization on plasmid stability, including: 1. structure of the immobilizing gel separates the two populations, negating competition; 2. the close proximity of immobilized cells plus less freedom of motion could promote transfer of plasmid DNA between populations by either conjugation (if a mobilizing factor is present) or transformation; and/or 3. mass transfer limitations on nutrients may create lower growth rates in the interior of a gel, thus pro- moting increased plasmid stability in those regions. Similar mechanisms may serve to affect plasmid re- tention in biofilms naturally formed by attaching bac- teria. Blenkinsopp and Costerton4 report that biofilms, formed under natural conditions, create highly hetero- geneous microstructures. Use of confocal laser-scanning microscopy,31 coupled with fluorescent probe molecules (for Eh, pH, cell viability, cell species number concen- tration, matrix permeability), clearly indicate that adja- cent microcolonies within biofilms may experience radically different environments. Localized concen- trations of calcium may promote plasmid DNA incorpo- ration by transformation. Dense glycocalyx structures surrounding established colonies may exhibit different mass transfer properties than adjacent sections of bio- film, segregating bacteria into discrete microcolonies much the same way as in gel beads. One distinction be- tween biofilm-bound cells versus artificially immobilized recombinants, is that naturally attaching bacteria must metabolically produce the encapsulating extracellular polysaccharide glycocalyx; an added burden that artifi- cially immobilized cells do not bear. Metabolic effects of cell replication and extracellular polysaccharide produc- tion may affect plasmid segregation differently than in either planktonic or gel-immobilized populations. The effects of polysaccharide synthesis on membrane integ- rity could also influence both the conjugation and trans- formation of plasmid DNA. Another factor influencing plasmid retention and transfer in a biofilm is the continual intraphase trans- port of cells at the biofilm-fluid interface. Biofilms can be reinoculated with specific community members which is not likely in gel bead systems. Rochell et al.36 studied plasmid transfer from two donor Pseudomonas spp. cul- tivated wtih two recipient Pseudornonus spp. within bio- films in a laboratory reactor system. Results indicated recipient strains survived within the biofilm but not in free suspension. Transconjugants were detected in all cases, despite the fact that donor and recipient cells were inoculated at opposing ends of the substratum used. The authors attributed this observation to cell movement from the biofilm to the liquid phase and redeposition. Clearly, altered plasmid retention in immobilized or biofilm populations of recombinant cells cannot be ex- plained by any single mechanism. Research has yet to focus on developing methods and systems to assess those environmental and biological factors governing plasmid segregational stability in biofilm populations. Therefore, it is necessary to establish both a mathematical model and experimental protocol with which to describe and quantify the rates of plasmid loss in both suspended and biofilm cultures. In this study, we propose an alternative approach to calculate plasmid loss probability, in both suspended batch and continuously-fed biofilm cultures, which will be compared to the results calculated by Eq. (1). The effects of culture growth mode (suspended versus bio- film) on plasmid retention and gene expression are in- vestigated here as a function of the level of cloned gene induct ion. MATHEMATICAL CONSIDERATIONS Batch Suspension Culture The net accumulation of freely suspended plasmid- bearing and plasmid-free cells in a batch culture can be described by two processes: cell growth and growth- related plasmid loss, -- - p + x + - p p + x + dX' dt -- - p - x - + p p + x + d X - dt (3) Let x = X - / X + , and apply the product rule of differen- tiation, yielding, dx dX' -- - x + - + x- d X - dt dt dt (4) Substituting Eqs. (2) and (3) into (4) and rearranging yields, * = ( I . - - p+ + pp+)x + pp+ dt (5 ) During the exponential growth phase, p- = p i and p+ = p i . Therefore, Eqs. (2) and (5 ) can be rewrit- ten as: d X + -- dt - (Pi - p p ; ) x + (7) Plotting dX+/dt versus X + , the slope is: HUANG, PERETTI, AND BRYERS: PLASMIDS IN E. COLl BlOFlLM CULTURES 213 Plotting dx/dt versus x, the slope and intercept are, respectively, m = p i - p; + ppi (9) b = pp; (10) Solving Eqs. (8), (9), and ( lo ) , p: = m+ + b , p i = m+ + m, andp = b/(m+ + b). Biofilm Culture Accumulation of cells within a biofilm culture is the net result of several processes, including: the deposition of suspended cells from the liquid phase, cell replication and extracellular polymer production, and biofilm de- tachment due to shear stress.637 Deposition of suspended cells can be ignored, because cells were not supplied to the reactor after inoculation and the residence time was maintained much less than the generation time of the culture. Assume the existence of plasmid will not affect the adhesion or detachment of the host cells, and the detachment rate constant is the same under same hydro- dynamics condition. Therefore, the net accumulation rates of plasmid-bearing and plasmid-free cells in the biofilm can be expressed as the combination of cell growth, plasmid loss, and biofilm detachment: Assuming the distributions of plasmid-bearing and plasmid-free cells within the biofilm are uniform with depth, then one can assume that cells are detached in the same ratio as exists in the biofilm. Defining /3 = B-/B+ = B i / B ; , and substituting into Eq. (12), dp dB+ -- - B + - + P- dB- dt dt dt Substituting Eqs. (11) and (12) into (13) and rearranging, dP - = (P- - p+ + pp+)P + ppi dt The specific growth rates for the two populations can be expressed as: Thus, Eq. (14) can be rewritten as: (15) Plotting d#?/dt versus P, the slope and intercept will be: Solving Eqs. (16) and (17) to get MATERIALS AND METHODS Bacterial Strain and Plasmid E. coli DH5a (kindly donated by Dr. Vickers Burdett, Department of Microbiology and Immunology, Duke University) was selected for this study, because it can form biofilm efficiently under low substrate concentra- tions and does not naturally produce P-galactosidase. Its genotype is 080dlacZAM15, A(1acZYA-argF) , U169, deoR, recAl, endA1, hsdRl7, supE44, thi-1, gyrA96 (nalidixic acidresistant), relAl . Plasmid pMJR1750 is a 7.5-kb plasmid consisting of an ampicillin-resistant marker, a strong promoter, tac, a repressor gene, lacZQ, and the lacZ gene which encodes P-galactosidase. The P-galactosidase promoter can be induced by various in- ducers, including isopropyl P-D-thiogalactoside (IPTG) . Hence, cells harboring the plasmid which express P-galactosidase form blue colonies on plate medium con- taining 5-bromo-4-chloro-3-indol-~-~-galato-pyranos~de (X-gal), whereas plasmid-free cells form white colonies. Batch Culture The fermentation medium used was M9 minimal me- dium containing 0.2% glucose, 0.4% casamino acid, and 0.01% thiamine. The batch culture was performed in the absence of ampicillin in a Biostat M fermentor (B. Braun, Allentown, PA) of 1-L working volume at pH 7.0 and 37°C. Dissolved oxygen was maintained above 50% of saturation. One percent (vh) of exponentially growing cells resuspended in identical medium was used to inocu- late the fermentation vessel to minimize the lag phase. Ten milliliters of cell suspension was sampled every hour for analysis. For induction experiments, IPTG was added when the absorbance at 600 nm reached about 0.8. Biofilm Formation System Biofilms of E. coli DH5a(pMJR1750) were cultivated in a parallel-plate flow cell which is constructed of optically clear polymethylmethacrylate (PMMA). The detailed reactor design is illustrated in Figure 1 and its pretreat- ment protocol can be found elsewhere." The inoculum was centrifuged from an overnight suspension culture which was selectively cultivated under 100 pg/mL ampi- cillin, then resuspended into about 10' cells/mL with sterile M9 minimal medium. The flow cell reactor was inoculated by recirculating the inoculum through the cell for 2 hours at a flow rate of 45 mL/min. After in- 214 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 41, NO. 2, JANUARY 20, 1993 Flow Out - Figure 1. Schematic drawing of the parallel-plate flow cell. oculation, the inoculum container was removed from the recycle loop and replaced with a small mixing vessel (100 mL) and the loop rinsed with sterile M9 minimal medium supplemented with 50 mg/L glucose, 100 mg/L casamino acid, and 25 mg/L thiamine, and recirculation flow resumed. Fresh nutrient solution was delivered to the mixing vessel to affect an overall system dilution rate of 4 h-'. The system was oxygenated with pure oxygen to pre- vent oxygen limitation. System dilution rate was main- tained well in excess of the growth rate of E. coli DHSa, which served to minimize cell growth in the fluid phase. The mixing vessel and connecting tubing were replaced with sterile versions every 12 hours to minimize the biofilm growth outside the flow cell. A micro flow-thru dissolved oxygen probe (Lazar Research Lab., Inc., Los Angeles, CA) was placed in the recirculation loop to monitor dissolved oxygen of medium leaving the flow cell. A slide with accumulated biofilm was removed from the flow cell reactor every 12 hours and replaced with a clean slide. Biofilm on the removed slide was scraped completely into 50 mL autoclaved M9 minimal medium and vortexed at maximum for 5 minutes to completely disrupt all bacterial aggregates. The resultant suspen- sion of biofilm material was used in the analyses de- tailed below. Measurement of Plasmid Stability Segregational instability was determined by cell growth on LB-agar plates that contained 40 pg/mL X-gal, 40 pg/mL IPTG, and 50 pg/mL nalidixic acid. Biomass samples from batch and biofilm cultures were suitably diluted with sterile M9 minimal medium and spread on the plates to form between 30 and 300 colonies per plate. The number of viable plasmid-bearing and plasmid-free bacteria were determined by averaging the blue and white colony-forming units (CFU), respectively, on three plates. The probability of plasmid loss per cell division was calculated with Eq. (1) and the alternative approach described in this study. Eq. (l), nonlinear in p , was solved by a ZREAL subroutine from the IMSL mathe- matical library. Time derivatives of the ratio of plasmid- free to plasmid-bearing cells for both suspension batch and biofilm cultures (dx/dt and dpldt, respectively) in Eqs. (7) and (15) are calculated from a least-squares, second-order polynomial fit. Structural stability of the plasmid was checked periodically throughout an experi- ment by horizontal electrophoresis. No plasmid struc- tural modification was found in any experiments. PGalactosidase Assay Samples obtained from batch and biofilm cultures were centrifuged at lOOOg for 15 minutes at 5°C. Cell pellets were resuspended in 1 mL TEP buffer (10 mM Tris; 1 mM EDTA, pH 8.0; 1 mM PMSF, phenylmethylsul- fonyl fluoride) then disrupted using two 30-second pulses by a Kontes Micro-Ultrasonic Cell Disrupter (Vineland, NJ) set at 30% output. The sonicated cellular homoge- nate was transferred to microcentrifuge tubes and chilled on ice for 10 minutes then centrifuged at 5000g for 10 minutes to pellet cell debris. The p-galactosidase activity was determined by the rate of hydrolysis of the colorless compound, o-nitrophenyl-P-D-galactoside (ONPG), to the yellow chromophore, o-nitrophenol (ONP).30 Cell extract, 200 pL, was mixed with 2.5 mL reagent A (0.1M NazHP04, adjusted to pH 7.3 with 0.1M NaH2P04), 100 pL reagent B (3.6M p-mercapto- ethanol), 100 pL reagent C (30 mMMgClZ), and 200 pL reagent D (33.2 mM ONPG in reagent A) in a dispos- able cuvette. After vortex mixing, the cuvette was placed in a Mil- ton Roy Spectronic 1201 Spectrophotometer (Rochester, NY), and the change in the absorbance at 410 nm moni- tored over a 2-minute interval. The activity was calcu- lated using Beer's law where 1 unit of P-galactosidase activity was defined as the amount of enzyme that can hydrolyze 1 pmol ONPG in 1 minute at pH 7.3 and 25°C. The amount of p-galactosidase was determined by cali- brating activities of aliquots of standard p-galactosidase solutions (5 Prime -+ 3 Prime, Inc., Boulder, CO). RESULTS Segregational Instability of Plasmids Batch Culture Three individual experiments were performed to study of the probability of plasmid pMJR1750 loss; typical ex- perimental results are presented in Figure 2. Apparent in Figure 2 is that the plasmid-free cells grow faster than plasmid-bearing cells in suspension without any anti- biotic selection. The percentage of plasmid-free cells is initially less than 0.02% of the culture, but increases to 1.3% by the end of experiment. Table I summarizes the parameters and probabilities of plasmid loss calculated by the classic model and the alternative approach pro- posed here. The average probability of plasmid loss calculated using Eq. (1) is 0.0009 and that by the alter- native approach is 0.0022. Biofilm Culture Figure 3 depicts the net accumulation of the plasmid- bearing and plasmid-free E. coli DH5a(pMJR1750) HUANG, PERETTI, AND BRYERS: PLASMIDS IN E. COLI BlOFlLM CULTURES 215 Time ( h ) Figure 2. Population dynamics of plasmid-bearing (0) and plasmid-free (0) planktonic cells of E. coli DH5a(pMJR1750) in a batch culture at 37"C, pH 7.0. cells in biofilms cultivated at a carbon-to-nitrogen ratio of 0.07. Both plasmid-bearing and plasmid-free cells in- crease slowly during the first 36 hours, then enter an apparent maximum accumulation phase, and finally reach a plateau after 96 hours. Without antibiotic selec- tion, the plasmid-free cell percentage increases from 1% to 27%. The probability of plasmid loss in biofilm was calculated by Eq. (18) using the average p ; and p i determined from batch cultures. Table I1 lists the pa- rameters and probabilities of plasmid loss for three independent biofilm experiments. The plasmid loss probabilities range from 0.013 to 0.021, with an average of 0.017 20.004. pGalactosidase Expression Batch Culture Three different IPTG concentrations were added to sus- pended batch cultures. Figure 4 shows the population dynamics of plasmid-bearing cells after induction. The growth rate of plasmid-bearing cells is 0.45 20.06 h-' with no IPTG present. Once IPTG was added, the growth rate decreased. Under 0.17 mM IPTG, the growth rate drops to 0.35 50.05 h-', about 77% of the uninduced growth rate. At both 0.34 and 0.51 m M IPTG, the growth rates decreased to 0.16 20.03 h-', about 36% of that without IPTG. The j3-galactosidase production in batch cultures of E. coli DHh(pMJR1750) is shown in Figure 5 , expressed on amount of a single plasmid- bearing cell. Under 0.17 mM IPTG, the P-galactosidase concentration reaches its maximum of 0.32 pg/cell in the experiment 4 hours after induction. At 0.34 and 0.51 mM IPTG, the P-galactosidase concentration reaches a peak specific production of 0.47 pg/cell, 3 hours after induction. Biofilm Culture Figure 6a and b shows the biofilrn net accumulation of plasmid-bearing and plasmid-free cells in response to induction. Plasmid-bearing cells can reach a plateau if no IPTG is added. But, once IPTG was added, the plasmid-bearing biofilm cell density dropped dramati- cally, because cell growth rates decrease due to induc- tion. Conversely, plasmid-free biofilm cells continued to increase and were not affected by IPTG level. Figure 7 is the P-galactosidase production in biofilm cultures of E. coli DHh(pMJR1750). The P-galactosidase concen- tration can reach its maximum in each experiment about 24 hours after induction. The p-galactosidase concen- trations are 0.08, 0.1, and 0.12 pg/cell at 0.17, 0.34, and 0.51 mM IPTG, respectively. DISCUSSION Comparison of Probability Calculation in Batch Cultures Three possible reasons exist to explain the differences in probabilities of plasmid loss calculated by these two methods. First, Imanaka and Aiba assumed the initial plasmid-free cell concentrations were zero. However, it is difficult to cultivate an inoculum of 100% plasmid- bearing cells even under a high selection pressure. The traditional way to attain 100% plasmid-bearing culture is to start with the single plasmid-bearing cell, but this leads to a long lag phase in the culture. Second, calcula- tions of p are sensitive to the amount of data collected; the smaller the generation number, the larger the resul- tant value of p. Third, although exponentially growing cells from identical medium are used as inoculum in all experiments, some initial lag phase is still inevitable. Feasibility of Probability Calculation in Biofilm Cultures The approach derived for biofilm cultures assumes the cell concentration in the fluid phase is negligible, and Table I. Parameters and probabilities for suspended batch culture of E. coli DH5a(pMJR1750). Imanaka and AibaU This study Experiment na F(n)" a Pb mf (h-') m (h-') b (h-') P A 8 0.9793 1.17 0.0023 0.41 0.07 0.0004 0.0010 B 9 0.9747 1.25 0.0017 0.52 0.13 0.0003 0.0006 C 8 0.9774 1.15 0.0027 0.41 0.06 0.0005 0.0012 Average 0.0022 -C0.0005 0.0009 _C0.0003 a Determined by the final data point of experiment. Solved by the ZREAL subroutine from the IMSL mathematical library. 216 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 41, NO. 2, JANUARY 20, 1993 the course of the experiment, dominate the upper layers of the biofilm, which is reflected in the higher numbers of plasmid-free cells re-entrained into the liquid phase. This prospect will negate the validity of certain assump- tions made in the mode derived here. Further study us- ing BIOSIM?' an interactive program for the simulation of dynamics of mixed culture biofilm systems, will be modified to allow the prediction of spatial distributions of plasmid-bearing and plasmid-free strains in biofilms, which will be discussed in a subsequent article. Time (h) Figure 3. Net accumulation of plasmid-bearing (0) and plasmid- free (0) cells of E. coli DH5a(pMJR1750) in a biofilm culture grown continuously at 37"C, pH 7.0, Re = 20, and residence time in system = 0.00625 h. hence, redeposition of detached cells can be ignored. To validate this assumption, samples from the fluid phase were plate-counted. Figure 8 shows the plasmid-bearing and plasmid-free cell concentration in the fluid phase downstream of the flow cell increases in accordance with biofilm accumulation, but the cell concentration never exceeded lo5 cells/mL. Because the dilution rate for the biofilm system is 4 h-', the only source of cells in the fluid phase is by way of the detachment from the biofilm. The redeposition of these detached cells can be examined by the sticking efficiency: which is defined as the ratio of the deposition rate to the flux of cells to the substratum. Average sticking efficiencies for E. coli DHh(pMJR1750) to glass surface under 45 mL/min is 2.5 X This means that less than 3 cells were at- tached irreversibly for every lo7 cells transported to the glass surface. Therefore, the amount of detached cells that could be redeposited onto the biofilm surface can be neglected as compared with the biofilm density. The second assumption regards the even distribution of plasmid-bearing and plasmid-free cells within biofilm. The validity of this assumption was examined by com- paring the ratios of plasmid-free to plasmid-bearing cells in both the biofilm and in cells detached from the film and re-entrained into the liquid phase (Fig. 9). The ratios are approximately the same for the first 48 hours, but after that, as the biofilm accumulation approaches its maximum, the amount of plasmid-free cells detached into the liquid phase relative to plasmid-bearing cells increases dramatically-more so than in the overall bio- film. Early in the biofilm accumulation, the plasmid- bearing cells initially dominate the surface. However, the plasmid-free cells grow at a faster rate and, during Probabilities of Plasmid Loss Between Batch and Biofilm Cultures In the experiments above, the probability of plasmid loss in a biofilm culture was greater than that observed in a suspended batch culture. The probability of plasmid loss during cell division has been reported to be affected by many factors including copy number, medium composi- tion, and growth rates.42 Some models suggest the proba- bility of plasmid loss is a function of copy number which may vary during cell division; the higher the copy num- ber, the smaller the probability. Copy number variance between batch and biofilm cultures may be a key factor affecting our estimated loss probabilities. Further deter- mination of plasmid copy numbers in batch and biofilm cultures is required in order to explain the differences in plasmid loss between batch and biofilm cultures. The media used for batch and biofilm cultures are the same, except for the glucose concentrations, which can lead to different growth rates. In addition, the production of extracellular polymer in biofilm cells may compete with plasmid maintenancelreplication for metabolic inter- mediates and energy sources - a competition suspended cells may not experience. Plasmid Retention in Response to Induction Many researches have reported that induction will re- duce specific growth rates of plasmid-bearing cells dra- mati~ally.~~~' In our experiments, the growth rates of E. coli DHh(pMJR1750) decreases 36% to 77% de- pending upon the level of induction. In batch cultures, an increase in the relative amount of plasmid-free cells resulted from the reduced growth rate of plasmid-bearing cells. Growth rates of plasmid-free cells are unaffected by induction. In biofilm cultures, upon induction, the per- centage of plasmid-free cells increases due to the re- duced growth rates of plasmid-bearing cells. While the Table 11. Parameters and probabilities for biofilm culture of E. coli DH5a(pMJR1750). Ex per iment Sticking efficiency m' (h-') b' (h-') P A 2.07 x lo-' 0.0187 0.0015 0.013 B 1.92 X 0.0058 0.0007 0.021 C 3.50 x 1 0 - ~ 0.0115 0.0012 0.018 Average (2.50 +0.87) X 0.017 20.004 HUANG, PERETTI, AND BRYERS: PLASMIDS IN E. COLl BlOFlLM CULTURES 217 5 106- 0 1 2 3 4 5 6 Time (h) Figure 4. Plasmid-bearing cell concentration profile of E. coli DH5a(pMJR1750) in response to various levels of induction by IPTG: (0) 0 mM; (A) 0.17 mM; (m) 0.34 mM; (0) 0.51 mM. 0.5 n 0.4 J 0" 0.3 . rn - C f 0.2 R 0 0.1 m 0.0 a r m - z - - O 1 2 3 4 5 6 Time (h) Figure 5. /3-Galactosidase production in batch cultures of E. coli DHSa(pMJR1750) under different induction levels: (0) 0 mM; (A) 0.17 mM, (m) 0.34 mM, (0) 0.51 mM IPTG. (a) Plasmid-bearing Cell 108 G E 107 ~ 106 2 l o 5 =I Y o_ L - a c 0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 Time (h) (b) Plasmid-free Cell 0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 Time (h) Figure 6. Biofilm net accumulation of E. coli DH5a(pMJR1750) in response to the induction by IPTG: (0) 0 mM, (A) 0.17 mM; (m) 0.34 mM, (0) 0.51 mM. 1 0 2 0 3 0 4 0 Time (h) Figure 7. /3-Galactosidase production in biofilm cultures of E. coli DHk(pMJR1750) under different induction levels: (0) 0 mM; (A) 0.17 mM; (m) 0.34 mM; (0) 0.51 mM IPTG. 5 1 0 ' 1 " 0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 Time (h) Figure 8. Concentration profile of plasmid-bearing (0) and plasmid-free (0) cells of E. coli DHSa(pMJR1750) in the liquid phase of biofilm reactor. 0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 Time (h) Figure 9. Comparison of ratios of plasmid-free to plasmid-bearing cells between biofilm and detached cell in liquid phase: (m) biofilm; (0) detached cells. biofilm community loses both plasmid-bearing cells and plasmid-free cells due to detachment during early phases of development, the plasmid-bearing cells soon become relegated to the depths of the biofilm due to the faster growing plasmid-free cells. Consequently, shear removal begins to affect plasmid-bearing cells less than plasmid-free cells. Unlike biofilm cultures, batch cultures cannot operate for a long time due to the deple- tion of substrate and, in continuous cultures, segrega- tion will eliminate all plasmid-bearing cells eventually. 218 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 41, NO. 2, JANUARY 20, 1993 &Galactosidase Expression Between Batch and Biofilm Cultures The P-galactosidase concentrations in suspended cells increase with IPTG concentration and time but drop at the end of each experiment which is consistent with other reports.3p46 Maximum 8-galactosidase concentra- tions increase from 0.32 to 0.47 pg/cell when IPTG con- centration increases from 0.17 to 0.34 mM, but higher IPTG concentrations (0.51 mM) produce no further in- crease in protein production. This suggests the repres- sor, l a d Q , has been saturated by IPTG. Optimal inducer concentration is dependent on the host, repres- sor, promoter, and cloned-gene protein. In biofilm cul- tures, the P-galactosidase concentrations are very low after the first 12 hours, because plasmid-bearing cells may take a longer time to adjust their metabolic pathway and to induce enzyme production. The different IPTG concentrations did not affect the maximum P-galactosi- dase concentration. This is because the IPTG was fed continuously and the repressor was assumed saturated 24 hours after induction. The maximum P-galactosidase concentrations in suspended cells are obviously higher than those in biofilm cells. Cell volume of suspended cells were also larger. However, this does not necessarily mean the ratio of clone-gene protein to total protein, a parameter normally used to quantify gene expression, in suspended cells is higher than in biofilm cells. Unfortu- nately, in our experiments, we cannot differentiate the total protein from plasmid-bearing or plasmid-free cells. Further study using plasmid ~TKW106,4~ which contains apurB locus, will eliminate the evolution of plasmid-free cells, allowing us to estimate P-galactosidase fractions based on total protein. Portions of these results were presented at the 1992 Mid- Atlantic Biochemical Engineering Consortium at New Brunswick, NJ. The authors appreciate the helpful com- ments from Dr. Steven Karel, Department of Chemical En- gineering, Princeton University. Financial support from an NSF grant (BCS-9020502) and a subcontract (MSU-2-F1-90) from the NSF Engineering Research Center for Interfacial Microbial Process Engineering at Montana State University are greatly appreciated. NOMENCLATURE cell density of plasmid-bearing and plasmid-free bio- film (cell L-~) cell density of detached plasmid-bearing and plasmid- free biofilm (cell L-*) intercepts from Eqs. (10) and (15) ( ! - I ) plasmid-bearing cell fraction at nth generation saturation constant (M L - ~ ) detachment rate constant ( ! - I ) slopes from Eqs. (6), (7), and (15) ( ! - I ) generation number probability of plasmid loss per cell per cell division limiting substrate concentration (M L-3) suspended plasmid-bearing and plasmid-free cell concentration (cell L-~) ratio of specific growth rates of plasmid-free to plasmid-bearing cells B pf, p- p i , p i X ratio of biofilm cell density of plasmid-free to plasmid- bearing cells specific growth rate of plasmid-bearing and plasmid- free cells ( ! - I ) maximum specific growth rate of plasmid-bearing and plasmid-free cells (t-') ratio of suspended cell concentration of plasmid-free to plasmid-bearing cells References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Adams, J., Kinney, T., Thompson, S., Rubin, L., Helling, R. E. 1979. Frequency-dependent selection for plasmid-containing cells of Escherichia coli. Genetics 91: 627-637. Bailey, J. E., Ollis, D. E. 1986. Biochemical engineering fun- damentals. 2nd edition. McGraw-Hill, New York. Bentley, W. E., Davis, R. H., Kompala, D. S. 1991. Dynamics of induced CAT expression in Escherichia coli. Biotechnol. Bio- eng. 38: 749-760. Blenkinsopp, S. A., Costerton, J.W. 1991. Understanding bac- terial biofilms. Trends Biotechnol. 9: 138-143. Bryers, J. D. 1987. Biologically active surfaces: Processes gov- erning the formation and persistence of biofilms. Biotechnol. Prog. 3: 57-68. Bryers, J. D. 1984. Biofilm formation and chemostat dynamics: Pure and mixed culture considerations. Biotechnol. Bioeng. 26: Characklis, W. G., Turakhia, M. H., Zelver, N. 1990. In: W. G. Characklis and K. C. Marshall (eds.), Biofilms. Wiley, New York, pp. 195-265. Clewell, D. B. 1981. Plasmids, drug resistance, and gene trans- fer in the genus Streptococcus. Microbiol. Rev. 45: 409-436. Clewell, D. B., Yagi, Y., Ike, R.A., Craig, B. L., Brown, F. 1982. In: D. Schlesinger (ed.), Microbiology- 1982. ASM, Washington, DC, pp. 97-100. DaSilva, N. A., Bailey, J. E. 1986. Theoretical growth yield es- timates for recombinant cells. Biotechnol. Bioeng. 28: 741-746. de Taxis du Poet, P., Dhuster, P., Barbotin, J.-N., Thomas, D. 1986. Plasmid inheritability and biomass production: Compari- son between free and immobilized cell cultures of Escherichia coli BZ18(pTG201) without selection pressure. J. Bacteriol. Doran, P. M., Bailey, J. E. 1986. Effects of immobilization on growth, fermentation properties, and macromolecular composi- tion of Saccharomyces cerevisiae attached to geletin. Biotechnol. Bioeng. 28: 73-87. Doran, P. M., Bailey, J. E. 1987. Effects of immobilization on the nature of glycolytic oscillations in yeast. Biotechnol. Bioeng. Dykhuizen, D. E., Hartl, D. L. 1983. Selection in chemostat. Microbiol. Rev. 47: 150-168. Georgiou, G. 1988. Optimizing the production of recombinant proteins in microorganisms. A1.Ch.E. J. 34: 1233-1248. Godwin, D., Slater, J.H. 1979. The influence of the growth environment on the stability of a drug resistance plasmid in Escherichia coli K12. J. Gen. Microbiol. 111: 201-207. Graham, J. B., Istock, C. A. 1978. Genetic exchange in Bacillus subtilis in soil. Mol. Gen. Gen. 166: 287-290. Graham, J. B., Istock, C. A. 1979. Gene exchange and natural selection cause Bacillus subtilis to evolve in soil culture. Science Grandi, G., Mottes, M., Sagaramella, V. 1981. Specific pattern of instability of Escherichia coli HisG gene cloned in Bacillus subtilis via the Staphylcococcus aureus plasmid pCS194. Plasmid Gealt, M.A., Chai, M.D., Alpert, K. B., Boyer, J. C. 1985. Transfer of plasmids pBR322 and pBR325 in wastewater from laboratory strains of Escherichia coli to bacteria indigenous to the waste disposal system. Appl. Environ. Microbiol. 49: 948-958. 165: 871-877. 29: 892-897. 204: 637-639. 6: 99-111. 836-841. HUANG, PERETTI, AND BRYERS: PLASMIDS IN E. COLl BlOFlLM CULTURES 219 21. Hattori, R. 1972. Growth of Escherichia coli on the surface of an anion-exchange resin in continuous flow system. J. Gen. Appl. Microbiol. 18: 319-330. 22. Huang, C.-T., Peretti, S.W., Bryers, J. D. 1992. Use of flow cell reactors to quantify biofilm formation kinetics. Biotechnol. Tech. 6: 193-198. 23. Imanaka, T., Aiba, S. 1981. A perspective on the application of genetic engineering: Stability of recombinant plasmid. Ann. NY Acad. Sci. 369: 1-14. 24. Inloes, D. S., Smith, W. J., Taylor, D. P., Cohen, S. N., Michaels, A. S., Robertson, C. R. 1983. Hollow-fiber membrane bioreac- tors using immobilized Escherichia coli for protein synthesis. Biotechnol. Bioeng. 25: 2653-2681. 25. Lauffenburger, D. A. 1987. Bacterioicin production as a method of maintaining plasmid-bearing cells in continuous culture. Trends Biotechnol. 5: 87-89. 26. Kumar, P. K., Maschke, H.-E., Friechs, K., Schiigerl, K. 1991. Strategies for improving plasmid stability in genetically modi- fied bacteria in bioreactors. Trends Biotechnol. 9: 279-284. 27. Kumar, P. K. R., Schiigerl, K. 1990. Immobilization of geneti- cally engineered cells: A new strategy for higher stability. J. Biotechnol. 14: 255-272. 28. Levin, B. R., Stewart, F. M., Rice, V. A. 1979. The kinetics of conjugative plasmid transmission: Fit of a simple mass action model. Plasmid 2: 247-260. 29. Mach, P. A., Grimes, D. J. 1982. R-plasmid transfer in a waste- water treatment plant. Appl. Environ. Microbiol. 44: 1395-1403. 30. Miller, J. H. 1972. Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, 31. Monbouquette, H. G., Ollis, D. F. 1988. Scanning microfluo- rimetry of Ca-alginate immobilized Zymornonus mubilis. Bio/ Technol. 6: 1076-1079. 32. Nasri, M., Sayadi, S., Barbotin, J.-N., Dhuster, P., Thomas, D. 1987. Influence of immobilization on the stability of pTG201 recombinant plasmid in some strains of Escherichia coli. Appl. Emiron. Microbiol. 53: 740-744. 33. Ou, L.-T., Alexander, M. 1974. Influence of glass microbeads on growth, activity and morphological changes of Bacillus megafericum. Arch. Microbiol. 101: 35-44. 34. Reanney, D. C., Roberts, W. P., Kelly, W. J. 1982. In: A.T. Bull and J. H. Slater (eds.), Microbial interactions and communities, vol. I. Academic Press, New York. 35. Reichert, P., Ruchti, J., Wanner, 0. 1989. BIOSIM: Interactive program for the simulation of the dynamics of mixed culture biofilm systems on a personal computer, version 1.1. Swiss Fed- eral Institute of Technology, CH-8600 Diibendorf, Switzerland. pp. 352-355. 36. Rochelle, P.A., Fry, J.C., Day, M. J. 1989. Plasmid transfer between Pseudomonas ssp. within epilithic films in a rotating disc microcosm. FEMS Microbiol. Ecol. 62: 127-136. 37. Ryan, W., Parulekar, S. J. 1991. Recombinant protein synthesis and plasmid instability in continuous cultures of Escherichia coli JM103 harboring a high copy number plasmid. Biotechnol. Bioeng. 37: 415-429. 38. Sayadi, S., Nasri, M., Barbotin, J. N., Thomas, D. 1989. Effect of environmental growth conditions on plasmid stability, plas- mid copy number, and catechol2,3-dioxygenase activity in free and immobilized Escherichia coli cells. Biotechnol. Bioeng. 33: 39. Saye, D. J., Ogunseiten, O., Sayler, G. S., Miller, R.V. 1987. Potential for transduction of plasmids in a natural freshwater environment: Effect of plasmid donor concentration and a natural microbial community on transduction in Pseudomonus aeruginosa. Appl. Environ. Microbiol. 53: 987-995. 40. Seo, J.-H., Bailey, J. E. 1985. Effects of recombinant plasmid content on growth properties and cloned gene product forma- tion in Escherichia coli. Biotechnol. Bioeng. 27: 1668-1674. 41. Stewart, G. J., Carlson, C.A., Ingraham, J. L. 1983. Evidence for an active role of donor cells in natural transformation of Pseudomonas sfufzeri. J. Bacteriol. 156: 30-35. 42. Stewart, G. J., Carlson, C.A. 1986. The biology of natural transformation. Ann. Rev. Microbiol. 40: 211-235. 43. Summers, D. K. 1991. The kinetics of plasmid loss. Trends Bio- technol. 9: 273-278. 44. Qagi, R. D., Ghose, T. K. 1982. Studies on immobilized Sac- charomyces cerevisiae. I. Analysis of continuous rapid ethanol fermentation in immobilized cell reactor. Biotechnol. Bioeng. 45. Weinberg, S. R., Stosky, G. 1972. Conjugation and genetic re- combination of Escherichia coli in soil. Soil Biol. Biochem. 4: 46. Wood, T. K., Kuhn, R. H., Peretti, S.W. 1990. Enhanced plas- mid stability through post-segregational killing plasmid-free cells. Biotechnol. Tech. 4: 39-44. 47. Wood, T. K., Peretti, S.W. 1991. Effects of chemically-induced, cloned-gene expression on protein synthesis in E. coli. Biotech- nol. Bioeng. 38: 397-412. 48. Wyke, A. 1988. The state of biotechnology. Chem. Eng. Prog. 49. Zeph, L.R., Onaga, M.A., Stozky, G. 1988. Transduction of Escherichia coli by bacteriophage P1 in soil. Appl. Environ. Microbiol. 54: 1731-1737. 801-808. 24: 781-796. 171-180. 84: 16-27. 220 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 41, NO. 2, JANUARY 20, 1993