1. Introduction

Escherichia coli has been extensively used for recombinant protein production ^[1,2,3] due to its fast growth at high cell densities, minimal nutrient requirements and well-characterized genetics, and due to the availability of a large number of cloning vectors and host strains ^[4]. This bacterium has the ability to accumulate recombinant proteins to at least 20% of the total cell protein ^[5] and has a good track record in drug regulatory authorities for the manufacturing of recombinant therapeutic proteins ^[3]. Recombinant protein expression in E. coli is usually achieved by inserting the gene coding for the protein of interest into a multicopy plasmid under the transcriptional control of either a constitutive or inducible promoter ^[1].

The proximity of bacterial cells in biofilms provides an excellent environment for the exchange of genetic material carried by the plasmid ^[6]. Thus, the effect of E. coli plasmids (especially conjugative plasmids) on biofilm formation by their hosts has been described in numerous studies and most of these plasmids have been found to enhance biofilm formation ^{[7,8,9,10,11,12,13,14]}. Ghigo ^[7] was the first author to show that natural conjugative plasmids induced the biofilm formation of different E. coli K-12 strains. His work suggests that the conjugative pili responsible for the horizontal transfer of the plasmid may act as adhesins, connecting the cells and stabilizing the biofilm structure in hydrodynamic systems ^[7]. Reisner’s results ^[8] also support and expand the idea that conjugative plasmids encode an important pathway for E. coli biofilm development, showing that the absence of biofilm growth was related with conjugative plasmids repressed for pili expression. Yang et al. ^[11] observed that mature biofilms harbouring an F-like conjugative plasmid (R1drd19) were thicker than biofilms not bearing that plasmid. Interestingly, the presence of R1drd19 plasmid increased the expression of numerous chromosomal genes related to envelope stress, motility and other genes known to be involved in biofilm formation ^[11]. May and Okabe ^[15] also demonstrated that the F pilus caused increased colonic acid and curli production during biofilm development, promoting cell-surface adherence ^[16]. Other conjugative plasmids of E. coli (pOLA52 and pMAS2027) have been shown to enhance biofilm formation through type 3 fimbriae ^[6,14].

Although non-conjugative plasmids are commonly used for recombinant protein production in E. coli, few studies have addressed their effect on biofilm formation. Gallant et al. ^[17] noted that plasmids encoding the common β-lactamase marker TEM-1 reduced cell adhesion and biofilm formation. Conversely, E. coli O157:H7 cells carrying a 92kb virulent and non-conjugative plasmid (pO157) ^[18,19] influenced biofilm formation and architecture ^[20]. Under smooth flow conditions, the pO157 plasmid promoted biofilm development through increased EPS production and generation of hyperadherent variants ^[20]. Concerning the impact of expression plasmids, Huang et al. ^[21,22,23] studied the production of b-galactosidase in E. coli DH5α cells carrying a non-conjugative plasmid containing the tac promoter. These authors found that plasmid-bearing cells formed biofilms with a higher cell density when compared to non-transformed cells ^[21]. Furthermore, the recombinant protein was successfully produced in biofilm cells, but at a lower level than in planktonic cells ^[21,23]. Later, O’Connell et al. ^[24] have described the first system for high level heterologous protein production in E. coli biofilm cells using a pUC-based vector for the expression of enhanced green fluorescent protein (eGFP). These authors showed that the biofilm environment enhanced the eGFP production when compared to planktonic cells. Despite the enormous potential of biofilms as an expressing system, this subject remains largely unexplored.

In a previous study, our research group has shown that the presence of the non-conjugative plasmid pET28A in E. coli JM109 (DE3) cells increased biofilm formation under turbulent flow conditions when compared to non-transformed cells ^[25]. The eGFP gene was cloned into this plasmid (resulting in plasmid pFM23) ^[26] and it was shown that heterologous protein expression increased the spatial heterogeneity of the biofilms ^[27]. Since the expression of the eGFP gene increases the metabolic drain on the host cells, it is important to assess if it also affects the process of biofilm formation and resistance. The aim of this work was to assess the effects of heterologous protein production by comparing the biofilm formation of E. coli cells harboring plasmid pET28A (devoid of the expression gene) or plasmid pFM23 (for the intracellular expression of eGFP), as well as the biofilm susceptibility to the biocide benzyldimethyldodecylammonium chloride (BDMDAC). This quaternary ammonium compound has already demonstrated strong antibacterial activity ^[28] and is often used in industrial applications ^[29].

---

2. Materials and Methods

---

2.1. Bacterial strain and plasmids

The E. coli strain JM109 (DE3) from Promega (USA) was used in this work because it is recommended for protein expression with the pET system ^[30], besides it has shown good biofilm-forming ability in turbulent flow conditions ^[25].

Competent E. coli cells were transformed by heat shock ^[31] with the control plasmid pET28A (Novagen, USA; Figure 1A) or with the expression plasmid pFM23 (Figure 1B), which was obtained by cloning the eGFP gene into the pET28A vector as previously described by Mergulhão et al. ^[26]. Transformants were selected on lysogeny broth (LB-Miller, Sigma, USA) agar supplemented with kanamycin (Eurobio, France).

---

2.2. Biofilm formation system and sampling

In order to assess the effect of heterologous protein expression on the biofilm-forming and resistance capacity of E. coli cells, the experimental apparatus presented in Figure 2 was used. It includes a recirculating tank where planktonic cells grow and a vertical flow cell for biofilm formation. Peristaltic and centrifuge pumps allow the circulation of the bacterial suspension in the system. The flow cell is composed by a semicircular Perspex duct (3.0 cm diameter and 1.2 m length, corresponding to a hydraulic equivalent diameter of 1.8 cm) with 20 apertures on its flat wall to fit removable rectangular pieces of Perspex (coupons). Polyvinyl chloride (PVC) slides (2 × 1 cm) were glued onto the coupons and were in contact with the bacterial suspension circulating in the flow cell reactor. This system allows individual sampling of each coupon without disturbing the biofilm formed on the others ^[32].

E. coli cells containing the pET28A or pFM23 plasmid were grown by recirculating the bacterial suspension at 30 ºC during 12 days under a turbulent flow with a Reynolds number of 4600 (flow rate of 255 l h^-1) ^[25]. The recirculating tank of 1 l was continuously fed at a flow rate of 0.025 l h^-1 with the nutrient medium consisting of 0.55 g l^-1 glucose, 0.25 g l^-1 peptone, 0.125 g l^-1 yeast extract and phosphate buffer (0.188 g l^-1 KH₂PO₄ and 0.26 g l^-1 Na₂HPO₄), pH 7.0 ^[25]. For maintenance of selective pressure, the growth and feeding media were supplemented with 20 µg ml^-1 kanamycin.

For biofilm sampling, the system was stopped every day to allow coupon removal and carefully started again maintaining the same flow conditions. Day 1 corresponds to the inoculation of the reactor system and the sampling was initiated on day 2 of the experiment.

---

2.3. Biofilm and planktonic cell analysis

Biofilm thickness was determined using a digital micrometer ^[32]. The needle connected to the digital micrometer was applied on the top of the biofilm and penetrated the sample to the bottom of the biofilm. The thickness was obtained by the difference between the readings made at the top and bottom of the sample. Ten measurements were made at random points on each coupon and the average value was determined. Then, the biofilm was resuspended and homogenized in 25 ml of 8.5 g l^-1 NaCl solution to assess the total cell number by staining with 4’-6-diamidino-2-phenylindole (DAPI) ^[33] during the steady-state period (between days 7 and 12).

Regarding planktonic cells, optical density (OD) at 610 nm ^[32] and total cell number were assayed. Planktonic and biofilm culturability were calculated by dividing the number of colony forming units (CFU) by the total cell number obtained by DAPI counts. For assessing CFU, planktonic and biofilm samples were serial-diluted(into 9 ml of 8.5 g l^-1 NaCl solution) and spread on top of solid growth medium (PCA; Merck, Portugal) supplemented with kanamycin (20 μg ml^-1). Enumeration of CFU was carried out after 24 h incubation at 30 ºC.

Specific eGFP production was assessed by the fluorometric method described in Mergulhão and Monteiro ^[34] during the steady-state period. A sample volume corresponding to an equivalent OD_{610 nm}= 1 was used to harvest the cells by centrifugation (3202 g for 10 min). The pellet was resuspended in 100 µL of Buffer I (50 mM Na₂HPO₄, 300 mM NaCl, pH 8) and added to a 96-well microtiter plate (Orange Scientific, USA) already containing 100 µl of Buffer I. Fluorescence was measured using a microtiter plate reader (SpectraMax M2E, Molecular Devices, Inc., UK) with the excitation filter of 488 nm and the emission filter of 507 nm.

---

2.4. Biofilm susceptibility

The quaternary ammonium compound BDMDAC (Sigma, USA) was used to assess the biocide susceptibility of biofilms formed by both E. coli strains. For these experiments, biofilms were developed for 3 days and the biocide was applied as previously described by Teodósio et al. ^[25]. The biofilm cells were exposed to the biocide for 9 h and a coupon was sampled every hour to determine the biofilm thickness (using a digital micrometer), the total number of cells (obtained by DAPI staining) and the number of culturable cells (corresponding to the CFU counts).

Time zero corresponds to the moment when the biocide was introduced on the recirculating tank and the flow was re-established using the same conditions as for biofilm formation.

To evaluate the contribution of EPS to the biofilm resistance, the biofilm matrix was extracted using a Dowex resin (50 × 8, Na⁺ form, 20-50 mesh; Fluka Chemika, Switzerland) and quantified in terms of protein and polysaccharide content ^[33]. Protein and polysaccharide assays were performed using biofilm suspensions before EPS extraction (total constituents) and with EPS (matrix constituents) after extraction.

---

2.5. Statistical analysis

The results presented in Figures 3 and 4 originated from averages of triplicate data sets obtained in independent experiments. Average standard deviations (SDs) on the triplicate sets were calculated for all analysed parameters. For biofilm formation (Figure 3), the following averages were obtained: SD < 13% for OD and SD < 27% for biofilm thickness. Concerning biofilm susceptibility (Figure 4), the following averages SDs were obtained: SD < 22% for biofilm thickness, SD < 5% for biofilm total cells, and SD < 14% for biofilm culturable cells.

Paired t-test analysis was performed based on a confidence level of 90% (differences reported as significant for P values < 0.1 and marked with *) and 95% (differences reported as significant for P values < 0.05 and marked with ).

---

3. Results

---

3.1. Biofilm formation

The effect of heterologous protein expression on the biofilm-forming capacity of E. coli JM109 (DE3) cells harbouring the pFM23 plasmid was assessed by comparison with a strain bearing the plasmid backbone (pET28A) but devoid of the eGFP gene (Figure 3).

Although higher optical density values were registered for the eGFP-expressing cells (on average 20%) in most experimental points (P < 0.05 between days 5 and 12), the growth profiles of both E. coli strains were similar (Figure 3A). Planktonic cells grew up to day 5 and then the culture cell density decreased (about 40%) until the end of the experiment. The steady state concentration of planktonic cells was 1.6 × 10⁸ cell ml^-1 for the pET28 strain, whereas this concentration was 71% higher for the pFM23 strain.

The biofilm results (Figure 3B) revealed that the eGFP-expressing strain formed thicker biofilms than the pET28A strain. The thickness of the biofilms formed by the two strains was similar at the beginning of the experiment (about 150 µm), but a 74% increase was detected for the pFM23 strain between days 3 and 4, whereas the biofilm of the pET28A strain maintained its thickness (P > 0.05). Afterwards, biofilm thickness increased for both strains with the eGFP-producing strain forming 52% thicker biofilms. Regarding biofilm cells, the total cell number was 3.5 × 10⁸ cell cm^-2 for the pET28A strain in the steady-state period and this number was 92% higher for the producing strain.

Considering planktonic and biofilm culturability, values were lower (on average 55% in steady state) for eGFP-expressing cells when compared to the non-producing strain.

Fluorescence analysis has shown that eGFP was produced in both planktonic and sessile states, but that a higher production was obtained in biofilm cells (5.8 fg cell^-1 versus 0.18 fg cell^-1 for planktonic cells in steady state).

---

3.2. Biofilm susceptibility

Biofilms formed by each of the strains were subjected to a 9 h-biocide treatment and the biofilm susceptibility is shown in Figure 4.

Figures 4A and B show that the biofilm formed by the eGFP-producing strain was reduced significantly in the first two hours of biocide action (a thickness reduction of about 46% was observed accompanied by a reduction of 53% in the total cell number), but thereafter the thickness and cell number remained approximately constant. For the pET28A strain, the cell number was relatively constant during the BDMDAC treatment (around 7.4 log cell cm^-2), but the thickness of the biofilm was reduced by 47% in the first 3 h. Therefore, the biofilm was compressed during this period and apparently had a higher cohesion since cell removal was lower than for the eGFP-expressing strain. Both biofilms reached approximately the same thickness at the end of the treatment (P > 0.05 at 9 h). The number of biofilm culturable cells (Figure 4C) decreased for both strains, although a faster inactivation of pET28A-carrying cells was observed at the beginning of the biocide treatment. A 3 log reduction on cell culturability was observed for both types of biofilms after 9 h of exposure to the biocide, but they were neither completely inactivated (Figure 4C) nor removed (Figure 4A). The most important period for the antibiofilm action of BDMDAC was the first two hours of treatment.

In order to evaluate the role of EPS on the biofilm susceptibility, the matrix of the biofilms was characterized in terms of protein and polysaccharide contents. Although biofilms formed by the eGFP-producing strain had higher total protein content (78% above than the non-producing strain), the mass percentage of proteins contained within the matrix was similar for both strains (28% for the eGFP-producing strain and 25% for the pET28A-bearing strain). The eGFP-expressing strain also produced biofilms with more polysaccharides (76% increase), however these biofilms only have 18% of the total polysaccharides in the matrix when compared to 53% in the pET28A strain.

---

4. Discussion

---

4.1. Biofilm formation

Biofilm production was enhanced in E. coli cells bearing the pFM23 plasmid for heterologous protein expression when compared to the strain harbouring the plasmid without the expression cassette (pET28A vector). Several studies suggest that both conjugative and non-conjugative plasmids can promote biofilm formation ^{[7,10,20,21,25]}. In a previous work ^[25], it was shown that E. coli JM109 (DE3) cells transformed with plasmids pET28A and pUC8 formed thicker biofilms with a higher cell concentration than the non-transformed cells. In the present study, it is shown that heterologous protein production, which is known to increase the metabolic drain on the host cell ^[35,36], intensifies this effect. Sørensen and Mortensen ^[36] reported that the stress response of maintenance of a plasmid in the host cell is increased when a target protein is highly expressed. Environmental stress situations like starvation, amino acid depletion and heat shock resemble the stress response caused by a plasmid. Although the plasmid copy number influences the host stress response, the major cause for this metabolic burden is commonly the expression of plasmid-encoded genes such as antibiotic resistance or recombinant protein genes ^[36,37,38]. It is well known that heterologous protein expression induces the production of stress proteins, elevated respiration rates and high-energy requirements ^[36,39]. Since stress conditions can favour biofilm formation ^[40], the increased metabolic burden associated with heterologous protein expression may have stimulated biofilm formation by the eGFP-producing strain.

The metabolic stress imposed by the production of eGFP in both planktonic and biofilm cells may have a detrimental effect on cell viability, as observed by Kurland and Dong for bacterial suspensions ^[41] and by Gomes et al. ^[27] for biofilms grown under the same hydrodynamic conditions used in this work. Recombinant protein production significantly affects cell metabolism by channelling resources towards producing foreign proteins ^{[36,42,43,44]}. In particular, an increased protease activity and decreased growth rate and cell viability are some stress signals that can be induced during recombinant protein synthesis ^{[35,41,45,46]}. In this work, planktonic cell growth is not decreased in eGFP-producing cells probably because their production levels are much lower than in biofilm cells.

Our results show that the biofilm state enhance recombinant protein production. This corroborates the results obtained by O’Connell et al. ^[24] using E. coli ATCC 33456 containing the plasmid pEGFP (a pUC family vector) grown in a chemostat and in a parallel plate flow cell. Furthermore, it is shown that continuous biofilm cultures for recombinant protein production are beneficial for plasmid maintenance when compared to chemostats ^[24] since sessile cells tend to grow more slowly than planktonic cells ^[47], leading to fewer divisions and correspondingly less plasmid segregation. Since the results indicate that heterologous protein production is up to 30-fold higher in biofilm cells than in planktonic cells, it is possible that a biofilm reactor system may be devised for industrial protein production. In such a reactor, the biofilm would have to be retrieved from an appropriate surface to initiate downstream processing. Besides the reactor itself, there are other surfaces (like tubing, pumping equipment, etc.) that need to be cleaned after each production cycle using standard cleaning in place (CIP) procedures. These CIP protocols often include biocides and disinfectants like BDMDAC and therefore it is useful to understand if the biofilm susceptibility to this type of agent changed upon heterologous protein production.

---

4.2. Biofilm susceptibility

The effects of biocide treatment on eGFP-producing and non-producing biofilms were assessed regarding thickness, total cell number and culturable cell number. Results show that the treatment was not sufficient for complete biofilm removal nor inactivation. In a previous study ^[48], the same concentration of benzalkonium chloride (BAC)-a quaternary ammonium compound that includes BDMDAC in its composition-was also ineffective in removal of Pseudomonas biofilms. A more recent report ^[49] showed that the percentages of killing and removal of P. fluorescens biofilms by BAC were only 15.5% and 13.8%, respectively. Simões et al. ^[48] demonstrated that the presence of BAC increases biofilm mechanical stability. On the other hand, it is known that the antimicrobial mode of action of certain cationic surfactants (like QACs) has been attributed to their positive charge, which promotes an electrostatic interaction with negatively charged sites on cell membrane ^[50]. Therefore, it is conceivable that the electrostatic interactions can increase cell-to-cell cross-linking, preventing cell removal and resulting in more compact biofilms.

During biocide treatment, the biofilm formed by the pET28A-bearing strain was more compacted than the one obtained with the eGFP-producing strain and this may be related to a higher mass percentage of polysaccharides localized at the matrix ^[51]. However, the same extent of cell killing was achieved for both strains, which indicates that mass transfer within the biofilm was not controlling biofilm susceptibility. Araújo et al. ^[49] tested the penetration ability of biocides BAC and BDMDAC in Pseudomonas biofilms and concluded that the ability of an antimicrobial agent to penetrate a biofilm is not correlated with its killing and removal efficiencies. This reinforces the fact that antimicrobial susceptibility in biofilms is multifactorial and that mass transport limitations are part of the problem but should not be considered alone.

---

5. Conclusion

This study shows that heterologousprotein expression can have a positive effect on E. coli biofilm development. It was also demonstrated that biofilm cells can produce a model recombinant protein at much higher levels than planktonic cells. Apart from affecting recombinant protein production, the physiological changes associated with the transition from planktonic to biofilm growth can impact the biofilm susceptibility to biocides. A better understanding of these processes is important for the manipulation of biofilms for biotechnological applications such as the high-level production of heterologous proteins.

---

Acknowledgment

This work was financially supported by: Project POCI-01-0145-FEDER-006939 (Laboratory for Process Engineering, Environment, Biotechnology and Energy-LEPABE funded by FEDER funds through COMPETE2020-Programa Operacional Competitividade e Internacionalização (POCI)- and by national funds through FCT-Fundacção para a Cieência e a Tecnologia. Luciana Gomes acknowledges the receipt of a Ph.D. grant from FCT (SFRH/BD/80400/2011).

---

Conflict of Interest

All authors declare no conflicts of interest in this paper.

---