A Selection of Platforms to Evaluate Surface Adhesion and Biofilm Formation in Controlled Hydrodynamic Conditions

The early colonization of surfaces and subsequent biofilm development have severe impacts in environmental, industrial, and biomedical settings since they entail high costs and health risks. To develop more effective biofilm control strategies, there is a need to obtain laboratory biofilms that resemble those found in natural or man-made settings. Since microbial adhesion and biofilm formation are strongly affected by hydrodynamics, the knowledge of flow characteristics in different marine, food processing, and medical device locations is essential. Once the hydrodynamic conditions are known, platforms for cell adhesion and biofilm formation should be selected and operated, in order to obtain reproducible biofilms that mimic those found in target scenarios. This review focuses on the most widely used platforms that enable the study of initial microbial adhesion and biofilm formation under controlled hydrodynamic conditions—modified Robbins devices, flow chambers, rotating biofilm devices, microplates, and microfluidic devices—and where numerical simulations have been used to define relevant flow characteristics, namely the shear stress and shear rate.


Introduction
Biofilms are surface-attached communities of microorganisms, establishing threedimensional structures composed of bacteria surrounded by a self-made matrix [1]. This matrix consists of polysaccharides, proteins, and extracellular DNA and influences biofilm structure and morphology [2]. It is estimated that more than 90% of the bacterial cells in natural environments reside in a biofilm [3], since it gives protection against hostile conditions (pH changes, lack of nutrients, hydrodynamics, and antimicrobial compounds), encourages gene transfer, and facilitates the colonization of niches [4].
The established model for biofilm development includes five steps, starting with the (i) reversible attachment of cells to a preconditioned surface, (ii) production of extracellular polymeric substances (EPS) causing irreversible cell attachment, (iii) early development of biofilm architecture, (iv) biofilm maturation, and (v) cell dispersion from the biofilm into the surrounding environment [5,6]. An immersed substratum is rapidly covered by molecules from the liquid, forming a conditioning film that may change the properties of that surface, making it more or less suitable for bacterial adhesion [7,8]. Then, cell adsorption at the surface occurs, followed by release or reversible adhesion. The physical forces associated with conditioning film formation and reversible adhesion are electrostatic and van der Waals forces, as well as hydrophobic interactions [9]. The next step starts when the cells become irreversibly attached to the surface due to the presence of stronger attractive forces, such as covalent and hydrogen bonds, and cellular surface structures, such as fimbriae and flagella [9]. After maturation, biofilm growth and detachment/sloughing balance each other so that the biomass amount is approximately constant in time, i.e., the steady-state is attained. gianni and Missirlis [26], there is an optimum flow rate for bacterial adhesion, reflecting the balance between the rate of cell delivery and the force acting on adhered bacteria. Furthermore, the bacteria-substratum interaction determines the shear forces that adhered bacteria will be able to withstand [26].
Besides the relevant role of hydrodynamics on the microbial adhesion step, it is also one of the most important factors in biofilm formation and structure. The fluid surrounding a biofilm is the source for nutrients and vehicle for cell by-product removal [27]. An increase in flow velocity promotes the flux of molecules (nutrients, cells, biocides, antibiotics, metabolites, etc.) by changing their concentrations in the biofilm-fluid interface. Hydrodynamics also regulates the physiological properties of the biofilm by changing the mechanical shear stresses at the interface [25]. Higher shear forces often lead to the formation of thinner, denser, and stronger biofilms [28]. Although higher flow velocities enhance molecular transport by convection, the higher density of biofilms reduces the diffusivity of the molecules inside them [29,30]. Additionally, stronger shear forces can be responsible for higher biofilm sloughing or detachment [28].
Given the importance of shear forces on initial adhesion and biofilm development, it is essential to characterize them. The vast majority of biofilm studies under flow conditions only report the tested flow rate. Nevertheless, the flow rate by itself provides little information about shear forces since it does not take into consideration the geometry of the flow system. Two main parameters should be considered to characterize shear effects: the shear rate and the shear stress. Mathematically, the shear rate is the derivative of the velocity in the perpendicular direction from the wall system [31] and quantifies the frequency at which cells contact the surface. The shear stress in Newtonian fluids is proportional to the shear rate, where fluid viscosity is the constant of proportionality [31], translating the friction from the fluid acting on the adhered cells or the biofilm. Therefore, shear stress is commonly used as a descriptor of the shear forces acting on the biofilm during maturation or detachment.
Computational fluid dynamics (CFD) are commonly used to model biofilm reactors because they enable the estimation of the fluid flow parameters of these systems, such as the shear stress and the shear rate, at relatively low cost and faster, in comparison to experimental techniques [32,33]. CFD requires that the geometry to be analyzed is divided into a finite set of volumes, called cells, forming a computational grid, called mesh. Fluid flows are described by differential equations for the conservation of mass, momentum, and energy; CFD replaces these equations with algebraic equations, which can be numerically solved for each cell, resulting in a flow field [34]. These equations describe how the single operating parameters are related. Although CFD is very useful for understanding biofilm behavior, one must bear in mind that most simulations are performed for clean surfaces. When biofilms are formed, the cross-sectional flow area is reduced, increasing the bulk flow velocity and wall shear stress. Thus, these simulations are particularly recommended for the study of initial adhesion, early stages of biofilm development (such as those usually investigated in biomedical settings), and surfaces that are frequently cleaned (as is the case with food processing equipment). In these situations, the thickness of the formed biofilms is unlikely to have a significant impact on flow dynamics and shear forces distribution [35].

Biofilm Platforms
In this context, biofilm reactors are platforms for the study of biofilms in laboratory conditions. One of the major obstacles to study in vitro biofilms is the choice of a suitable platform, where key variables such as flow rate and shear stress can be manipulated in order to mimic the conditions found in real scenarios. Although completely reproducible biofilms are nearly impossible to obtain, the development of in vitro platforms for biofilm studies is a foremost step towards the standardization of procedures and for better control of the environmental conditions that affect biofilm development [36,37].
Here, we describe the most commonly used platforms for microbial adhesion and biofilm formation in controlled hydrodynamic conditions, particularly those where CFD has been used to determine relevant flow characteristics. These platforms have advantages and limitations, which are summarized in Table 1.

Flow Cells: Robbins Device and Modifications, and Flow Chambers
Flow cells can be generally divided into two types: those that are based on the design of the Robbins device and those that are built for the direct inspection of biofilm development, here called flow chambers. In both types of flow cells, it is possible to test different surface materials simultaneously in similar nutritional and hydrodynamic conditions. Nevertheless, it is worth mentioning that modified Robbin devices have higher throughput and hydrodynamic range than flow chambers. The Robbins device and its modifications present a higher number of sampling ports available for analysis, allowing for multiple biofilm samples to be taken simultaneously, as well as for sampling more than a single time point during biofilm development [39]. Although both types of flow cells are useful tools for studying biofilm under controlled conditions, they need a specialized apparatus, are technically challenging, and are not suitable for rapid high throughput assays. Another weakness of these systems is that only a single microbial strain can be analysed per experiment.
The most straightforward configuration of a flow cell system is that of a bioreactor containing a batch culture of the desired microorganism so that the content of the reactor is pumped through the flow cell and the effluent drained to waste. This configuration may be interesting for adhesion studies, particularly if the flow rates to be tested are low, since the duration of the assay is limited by the cell suspension volume. Another configuration is to place the flow cell in a recycle loop so that the culture volume is no longer a limitation and assays can last longer and perform at higher flow rates [40]. However, it has the disadvantage that the composition of the batch culture is always changing. A third alternative is to have a chemostat feeding a recirculation loop so that the feed flow to the chemostat equals the drain flow from the loop. In this case, it is possible to feed the flow cell with a constant concentration of cells and nutrients, while decoupling the flow rate going through the flow cell from the dilution rate [41,42]. With this flow cell configuration, it is possible to work at very high flow rates and attain high shear stresses that are comparable to those found in the environment and industry [43].

Robbins Device and Modifications
The Robbins device was initially developed by Jim Robbins and Bill McCoy to monitor biofilm formation in industrial water systems [44]. Several modifications were later introduced to this design, including the use of a square-channel pipe where coupons are aligned with the inner surface without disturbing flow characteristics [45]. They are convenient for studies where a large biofilm mass amount is wanted. With the modified Robbins devices, the flow can be momentarily stopped to allow direct access to the coupon, so that time-course experiments are possible. This stop of the flow system for coupon removal involves some risk because, even if the operator is very careful that the shutdown and restart of the system are smooth, there may be some loosening of the biofilm already formed in the remaining coupons of the flow cell. For quantitative analysis of the biofilm to be carried out, destructive sampling techniques are usually required. Conventional techniques, such as total and viable cell counts, as well as protein and carbohydrate content analysis, comprise the disruption of the biofilm [42,46].
Other flow cell designs include a half-pipe geometry that more closely resembles the geometry of piping systems [43,47]. These flow cells can be operated either in laminar or turbulent regimes, but it is important to guarantee that the flow cell has an entry section that is long enough to allow for flow development before the sampling zone (thus avoiding entry effects) and that the effect of the sudden contraction on the exit zone is negligible. This will ensure that all coupons are subjected to the same hydrodynamic conditions and that biofilm samples can be directly compared [48].
In our group, a custom-made, semi-circular flow cell (identical to that shown in Figure 1) was designed to evaluate the performance of different surface coatings in preventing biofouling in the marine environment [22], food industry [24,41], and medical devices [49,50]. The hydrodynamics of this flow cell system was fully characterized by CFD [48], which allows not only the guarantee that all sampling coupons are exposed to the same shear forces but also provides knowledge of the flow rate and Reynolds number, which is necessary in order to operate this platform and simulate the shear stress and/or shear strain described for different real scenarios.

Flow Chambers
In spite of the many advantages of modified Robbins devices, they are neither adequate for monitoring the initial cell adhesion to a surface nor for the direct analysis of biofilm development. For these purposes, several models of flow chambers that can be mounted on a microscope stage and used with video capture systems have been developed, enabling real-time observation of microbial adhesion, particularly when used with transparent surfaces. The employment of fluorescent probes coupled with confocal laser scanning microscopy (CLSM) makes flow chambers especially appreciated for in situ gene expression studies [51].
The most well-known flow system to study cell adhesion is the parallel-plate flow chamber (PPFC) developed by Bos et al. [52]. Adhesion can be studied in the PPFC system under controlled hydrodynamics that mimics, for instance, physiologically relevant conditions [40,53] using a wide range of microorganisms and surfaces with different properties. This system requires low volumes and, consequently, has a reduced cost when compared to modified Robbins devices; additionally, it presents one or more glass viewing ports that permit non-destructive, real-time adhesion (single-cell visualization) and biofilm observation. Despite their versatility, one must bear in mind that PPFCs have a much lower throughput than microplates and larger flow cells based on the Robbins device. Additionally, when real-time monitoring of adhesion is performed, a decrease in the initial adhesion rates is often observed along the experimental time, which is related to a phenomenon called hydrodynamic blocking [54,55]. Hydrodynamic blocking can reduce the adhesion of cells since the area behind each adhered cell is screened from incoming cells. Adhesion rates obtained in such conditions are not truly representative of the interaction between a single cell and the surface. Thus, initial adhesion assays in these setups should be conducted so that low surface coverage is attained, and the absence of blocking should be confirmed so that consistent results can be obtained [54].
Flow chamber systems have been designed to analyse cell adhesion [23,56,57] and biofilm formation [58,59], including a PPFC coupled to a jacketed tank and connected to centrifugal pumps and a valve via a silicone tubing system ( Figure 2). The valve allows the bacterial suspension to circulate through the system at a controlled flow rate [40], and the recirculating water bath is connected to the tank jacket to enable temperature control. The PPFC is coupled to a glass tank connected to four centrifugal pumps and a tubing system to conduct adhesion or biofilm formation assays.

Rotating Biofilm Devices
Two types of rotating biofilm reactors are commonly used in the assessment of material and fluid flow effects on biofilm development: the rotating disk reactor and the rotating cylinder reactor. These reactors have different designs. The rotating disk reactor consists of a 1-L vessel with a magnetically driven rotor at the bottom, which holds removable coupons for biofilm formation ( Figure 3) [60]. The hydrodynamic conditions under which the biofilm is formed are controlled by adjusting the disk rotation speed [60], and the shear stress on the coupons' surface can be estimated from the Navier-Stokes equations. The rotating cylinder reactor is often composed of four cylindrical sections that can be rotated at variable speeds within four concentric chambers [61]. Unlike the rotating disk reactor, this platform can be used to test different cell suspensions, since each chamber of the cylinder reactor has independent feeding and sampling ports [61].

Microfluidic Devices
Microfluidic platforms have demonstrated high potential and versatility for the study of microbial adhesion and biofilm formation under different growth conditions. Compared with traditional flow cell systems, microfluidics enables greater control over flow conditions, can be used to explore a much wider range of shear rates with high flexibility in designing different flow geometries, and facilitate the parallelization of experiments [62,63]. Although microfluidic devices can be fabricated by different techniques and from a diversity of materials, the flexible elastomer polydimethylsiloxane (PDMS) has been the material of choice for the construction of these devices. Several other surfaces can be studied using xerographic construction techniques that enable different polymers to be incorporated into microfluidic flow cells [53]. Concerning the analysis methods, although off-chip detection with conventional methods is feasible, on-chip detection by optical and/or fluorescence microscopy is preferred, in order to visualize in situ and real-time effects (Figure 4) [36].  Although there is a tendency to develop biofilm models in miniaturized devices, microfluidic-based devices also have their limitations: the small liquid volumes used in microfluidics may further impede molecular analysis, and the spatial confinement may generate different biofilms from those formed in more open systems [64]. Additionally, this platform requires specialized technical abilities for device fabrication and experimental setup, and system clogging can occur due to the small dimensions [36]. Air bubbles are another recurring issue in microfluidics [65]. Because of the micrometric dimensions of the tubes and channels, air bubbles can be very difficult to remove, leading to fluid flow instability and most likely to the detachment of adhered cells or biofilm portions.

Microplates
Microplates are currently the most widely used platform for biofilm development studies. They consist of plates with multiple wells arranged in a rectangular array with a 2:3 aspect ratio, resulting in 6, 12, 24, 48, 96, and 384 wells. The volume of each well can range from tens of microliters to few milliliters, depending on the number of wells [66]. Although most researchers use microplates in static conditions, they can be placed in orbital incubators and used for dynamic biofilm studies under controlled fluid conditions [67,68]. These devices are easy to handle, which allows for studying the adhesion of different microbial strains and consequent biofilm formation in rapid and inexpensive assays, due to their reduced volume [69]. Depending on the format used, they enable high throughput at an affordable cost and sometimes non-invasive imaging through optical coherence tomography (OCT) [70,71] and confocal laser scanning microscopy (CLSM) [72]. Particularly for larger well dimensions, it is possible to place coupons at the bottom of the wells so that different surface materials can be tested [70,73,74]. The main limitations of microplates are that loosely attached cells may not be measured correctly due to detachment during washing and that biofilms formed in this platform are affected by sedimentation.

96-Well Microplates
This is the most intensively used microplate format, mainly for screening purposes. Biofilm formation in this platform is severely affected by sedimentation, and the direct inspection of the biofilm is possible but technically difficult [75,76]. They are particularly suited for short-term experiments, as they operate in batch mode with the intrinsic exhaustion of nutrients and accumulation of toxic metabolites. Results obtained in this platform often lack reproducibility, possibly due to the washing steps that are researcher-dependent and the existence of several protocol versions for biofilm analysis [36]. These plates are generally not compatible with the use of coupons, as the bottom surface is relatively small; so, only a limited number of surfaces can be assayed (limited to the construction materials of these plates).

12-and 6-Well Microplates
These microplates are very attractive formats. Although theoretically their throughput is lower than the 96-well plates, the results obtained with these platforms are more reproducible due to the higher liquid volume, decreasing the need for a large number of replicate wells. These two types of plates also sustain microbial growth for longer periods, but medium replacement can be necessary. Large coupons can be used for biofilm formation (square surfaces of up to 1.5 cm can be placed on the bottom of the 12-well plates), and uniform shear forces can be obtained. Even though the shear stress in the coupon varies with the radial distance to the center, each coupon has identical average shear stress values [71].
The hydrodynamics inside the wells of 12-well microplates have been simulated to assess the effect of orbital shaking frequency on shear stress. Numerical simulations were performed at 25 • C, with an orbital diameter of 25 mm, a liquid volume of 3 mL, and shaking frequencies of 40 and 180 rpm ( Figure 5). As expected, higher shear stresses at the bottom of the wells can be attained at higher shaking frequencies; values up to 0.07 Pa and shear rates of 42 s −1 were achieved. These values are much higher than those obtained with 96-well microplates [8,77].

Adhesion and Biofilm Studies Performed under Controlled Hydrodynamics
In this section, illustrative examples of the application of the described in vitro platforms are given, when appropriate, for the investigation of initial microbial adhesion, biofilm formation and its treatment under controlled shear conditions in different fields: environment, industry, and medicine. Table 2 presents typical shear values that can be found in the environmental field. In a natural environment, a shear rate range between 4 and 125,000 s −1 can be obtained.

Environmental Applications
Most of the research in this area has been devoted to the impact of shear and surface characteristics on biofilm formation, giving less relevance to microbial cell adhesion (Table 3). It was also observed that flow systems, namely modified Robbins devices and rotating biofilm devices, are the main choice to emulate the turbulent flows and high wall shear stresses found in water systems [79][80][81]. However, in the last few years, efforts have been made to predict flow conditions in easy-to-handle biofilm platforms like microplates [68,71]. A detailed hydrodynamic analysis of the 12-well microplates [71] allows us to define the operational conditions that should be used in the laboratory bench to further assess the biofilm formation capacity of marine bacteria [70,71] and the antibiofilm activity of novel surface coatings [22,82] under hydrodynamic conditions prevailing in natural aquatic environments.   Helicobacter pylori High shear stresses negatively influenced the adhesion to the substrata. However, the temperature and inoculation concentration appeared to not affect adhesion. [88]

Industrial Applications
Similar to what was observed in environmental systems (Table 3), in the industrial field, the modified Robbins devices and rotating devices were the most reported reactors for biofilm formation and treatment studies. Different groups have used these flow systems in shear stress intervals of great amplitude [94][95][96], covering a huge range of shear values that can be found in the industry (Table 4). Our research group, in particular, has operated a semi-circular flow cell system (Figure 1) in different conditions and was able to attain shear stress values up to 0.6 Pa during biofilm formation [42,48], confirming the versatility of this platform and its capacity to mimic the hydrodynamic conditions that can be found, for instance, in the food industry (Table 4).
When the aim was to study microbial adhesion in an industrial environment, biofilm researchers preferred to use flow chambers [97,98] or microplates [24,41], since they are faster to operate and may allow for direct inspection by microscopic techniques (Table 5). Table 4. Examples of industrial processes and their associated shear stress ranges.

Biomedical Applications
Several studies were found in the literature where biofilm assays were performed under characterized hydrodynamic conditions similar to those of medical settings. Depending on the biomedical scenario, the shear stress range can vary between 0.02 and 88.3 Pa, and the shear strain between 0.1 and 80,000 s −1 ( Table 6). Flow chambers have particularly been used in the medical field to evaluate the antiadhesive activity of novel surface materials for biomedical devices, including urinary tract and implanted devices (Table 7), since they are adequate for low fluid shear stresses and laminar flow applications, as well as for real-time insight into the dynamic process of microbial cell adhesion [21,40,57]. Furthermore, the dimensions of the flow cell or the flow rate can be adjusted to attain the required shear stress/shear rate, in order to resemble in vivo flow conditions.
Microfluidic platforms have also demonstrated high potential and flexibility for the study of microbial adhesion [113,114] and biofilm formation [115,116] under different hydrodynamic conditions.

Staphylococcus epidermidis
The increase in the ionic strength enhanced adhesion to the different surfaces, in accordance with the Derjaguin-Landau-Verwey-Overbeek (DLVO) theory, under low shear rates. The increase in the shear rate restricted the predictability of the theory. Escherichia coli Similar adhesion rates were obtained on glass and polydimethylsiloxane. The highest adhesion rates were obtained on glass and polydimethylsiloxane, and the lowest on poly(L-lactic acid). Escherichia coli Adhesion reductions of 40-50% were attained at a shear rate of 15 s −1 on the peptide-coated surfaces compared with glass. The performance of the peptide-based antifouling coating was superior to poly(L-lactic acid). [57] Effect of shear stress on bacterial adhesion to antifouling polymer brushes

Escherichia coli
The introduction of carbon nanotubes composites in the polydimethylsiloxane matrix yielded less bacterial adhesion than the polydimethylsiloxane alone. Less adhesion was obtained on the composites with pristine rather than functionalized carbon nanotubes. Incorporation of higher amounts of carbon nanotubes in polymer composites can affect bacterial adhesion by more than 40%. Composites enabling a 60% reduction in cell adhesion were obtained by carbon nanotube treatment by ball-milling. The poly(ethylene oxide) brush yielded more than 98% reduction in bacterial adhesion, although for the more hydrophobic P. aeruginosa a smaller reduction was observed. For yeast species, adhesion suppression was less effective than for the bacteria. [133] Evaluation of the role of surface free energy on bacterial adhesion to plasma-modified films 50

Pseudomonas aeruginosa
The rate of adherent cell accumulation was zero for the polyetherurethane with a poly(butyl methyacrylate) barrier membrane releasing ciprofloxacin. Tryptic soy broth decreased adhesion to polymers, when compared to phosphate-buffered saline.
The estimated values of the free energy of adhesion correlated with the amount of adherent P. aeruginosa.

Current Challenges and Future Directions of Biofilm Platforms Research
Although biofilms are a recognized problem for the environment, industry, and medicine, and act as a possible reservoir of pathogens, there is a lack of reliable standard procedures to evaluate the efficacy of methods for biofilm prevention and removal. Consequently, it is very difficult to compare data obtained in different laboratories. As discussed before, laboratory reactors are available for growing biofilms that are more representative of a clinical situation [37,146] and industrial environment [147]. Although commercially available reactors with standardized protocols exist (e.g., ASTM Method E2871-13 and 2562-12 for the CDC biofilm reactor [148]), they are usually expensive and, thus, not accessible to all biofilm researchers, besides that the operation of these reactors has specific limitations. For instance, some of them cannot be used to test different surface materials, have reduced sampling areas, require specialized labor for operation, and the fluid dynamics are rarely well-characterized. While factors such as the temperature, microbial composition, and carbon source may be similar across different protocols and biofilm platforms, the fluid dynamics, namely the shear stress and shear rate, are a defining feature of a particular reactor operation. Whether the researchers are using a commercial or custom-made biofilm setup, computational simulations of hydrodynamics are extremely valuable, as they enable a more informed decision about whether the flow behavior in that specific biofilm reactor is suitable for their research.
Nevertheless, not all interactions between early adhered cells or established biofilms and fluid flow phases (gas and/or liquid) are considered when using the CFD technique. Almost all the flows in the described biofilm reactors deal with multiphase (gas-liquid, solid-liquid, and gas-liquid-solid), but some simplifications are introduced to reduce the model complexity [78,149]. For example, the aeration of flow cell systems is often not taken into account in the CFD study [43]. Furthermore, one must bear in mind that numerical simulations are mostly performed for clean and perfectly smooth surfaces. However, as biofilms grow or different coupon materials are used, the surface properties (such as roughness and hydrophobicity) should be considered for their impact on the wall shear stress. Therefore, there is still a great challenge in the integration of physical and biological processes in biofilm reactors.
Small flow chambers and microfluidic platforms are promising for screening new possible antibiofilm approaches. They need smaller volumes of media and reagents to run continuous biofilm experiments, when compared to the Robbins device and rotating biofilm reactors, enabling high-throughput assays. Additionally, the Bioflux [150] and other microfluidic devices [151,152] are dynamic systems with significant potential for monitoring heterogeneity in the biofilm microenvironment [153]. This can be achieved with specific stains and examination by confocal microscopy. However, direct biofilm observation might not be feasible, specific stains/probes may not be available (for nutrients or metabolites), or the time scale may be too slow. Introducing sensing techniques, such as microsensors or electrochemical probes in microfluidic chips, is an important development for online biofilm detection and microenvironment analysis [153].

Conclusions
Studying microbial adhesion and biofilm growth is crucial for understanding the physiology of sessile organisms and forming the basis for the development of novel antimicrobial materials. Fluid hydrodynamics is one of the most important factors affecting cell adhesion, as well as biofilm structure and behavior. Therefore, to simulate the relevant biofilms of different fields (environment, industry, and medicine) in the laboratory, it is of utmost importance to select an adequate biofilm platform and be able to operate it at hydrodynamic conditions that are as close as possible to those encountered in a real scenario.