Substrates’ properties can affect bacterial behaviors during biofilm formation. Poly(methyl methacylate) (PMMA) surface with sub-cellular nanopillar topography not only can inhibit P. aeruginosa attachment but also prevent upstream movement and reduce the proliferation of bacteria cells (Figure 3A; Rosenzweig et al., 2019). On a micrometer-scale surface with a crystalline hemispherical pattern, bacteria behave differently depending on the diameter of the raised features. A diameter of 2 μm and larger hemispherical patterns could hinder P. aeruginosa motility, while a 1 μm pattern showed no significant difference compared with a flat surface. In addition, their travel directions also differed by diameters. On 2 μm surfaces, bacteria were more likely to move in a reverse direction. On 4 μm surfaces, they prefer to travel in approximately straight lines in the groove along the crystal axis. In comparison, they were more likely to follow an approximately hexagonal lattice of grooves on 8 μm surfaces since this feature was complex for bacteria to move across the crowns (Figure 3B; Yow-Ren Chang and Ducker, 2018). On a stepped topography, P. aeruginosa drastically reduced the probability of crossing the step compared with crossing a point on a flat surface and significantly reduced the speed perpendicular to the step when they were very close to the step. When step height was similar to the bacterial length, a time penalty to cross the step was found, while no time penalty was resolved to cross tall steps (5–9 μm) (Chang et al., 2019). The critical length-scale for topography affecting P. aeruginosa movements seems to be similar to the dimensions of the bacterium and the length of pili. They proposed that if the topography in a particular direction is not favorable for pili attachment, then that direction of motion may also be disfavored (Chang et al., 2019). Another study also reported this threshold of topographical barriers of ∼1 μm using a different pattern. It was found that a 1 μm depth of micro-fabricated furrows can inhibit the expansion of P. aeruginosa biofilm more effectively than a 0.5 μm depth one (Gloag et al., 2016). For E. coli, a 10 μm tall hexagon-shaped topographic pattern with a side length of 15 μm and inter pattern distance of 2 μm showed the interrupting biofilm formation effects. In contrast to the smooth surfaces, biofilm formation and conjugation were promoted for larger square-shaped patterns with side lengths larger than 20 μm and inter pattern distances no less than 10 μm. Besides, compared to the top and grooves of the pattern, the vertical sidewall was favored for bacterial conjugation (Figure 3C; Gu et al., 2017).

In addition to topography, substrate stiffness also affects the biofilm formation process. It has been found that the average velocity of E. coli movement was higher on rigid PDMS surfaces compared to the soft ones (Song et al., 2017). By establishing an idealized bacteria twitching mathematical model, Sabass’ group analyzed how substrate rigidity influenced bacterial migration and revealed that migration depends on the force-sensitivity of the adhesion bond. The results showed that bacteria migration speed depends non-linearly on substrate rigidity. Bacteria move faster on a rigid surface if their rear adhesion bond is more force-sensitive. However, the deformation of the soft substrate surface may block the pull force produced during migration, which negatively affects migration speed (Simsek et al., 2019).

Furthermore, substrate stiffness can also affect the pattern of growing biofilms. It has been reported that the Vibrio cholerae biofilm spread varied on the surface of different agar concentration hydrogels. The morphology of growing V. cholerae biofilms is fold-shaped, extending from the edge to the center on the soft hydrogel surface of low agar concentration. In contrast, in the high agar concentration hydrogel surface, the folds extend in the opposite direction from the center to out edge. The sliding friction between agar surface and biofilm may contribute to the morphology difference during biofilm expansion (Figure 3D; Fei et al., 2020).

During maturation, bacteria often use quorum sensing to coordinate each other and the community. A cell–cell communication system involves the production, release, accumulation, and detection of autoinducers depending on the cell density, species composition of the microbial community, and the surrounding environment. As the population density of bacteria communities increases, the amount of autoinducer accumulates in the surrounding environment, which further affects the global gene expression patterns and cell-to-cell interactions (Papenfort and Bassler, 2016; Abisado et al., 2018). The key to quorum sensing is that bacteria cells receive, recognize, and respond to these signal molecules. Many studies have demonstrated that using molecular biology techniques to alter the bacterial quorum-sensing system will affect bacterial metabolic activities and biofilm formations. Detailed information has been well-reviewed in other articles, which is not our focus here (Mukherjee and Bassler, 2019; Li and Zhao, 2020; Wang N. et al., 2021).

From a material point of view, autoinducer peptide -I (AIP-I) is an essential inducer of the S. aureus Agr quorum-sensing system and can be detected by a cognate transmembrane bound receptor. This molecule could be immobilized onto a glass substrate via a flexible linker to influence the biofilm formation process (Kim et al., 2017). Similarly, a synthetic quorum sensing inhibitor 5-methylene-1-(prop-2-enoyl)-4-(2-fluorophenyl)-dihydropyrrol-2-one can be covalently incorporated into the surface coating to hindering bacterial communication (Ozcelik et al., 2017).

The substrates’ properties can still have an impact on the matured biofilms. For example, it has been found that micropatterned pillar surfaces could modulate competition dynamics and signaling pathways in the co-culture environment of E. coli and P. aeruginosa. With the increased height of the pillar, the biofilm featured increasing volume fractions of E. coli cells. They explained that this phenomenon resulted from accumulating a signaling molecule indole, which can silence E. coli’s biofilm dispersal and inhibit P. aeruginosa’s formation. Besides, the authors also found that micropatterned pillar surfaces could affect their antibiotic susceptibilities in mono-species culture. Increased susceptibility was found in patterns with the higher pillar (Figure 4A; Bhattacharjee et al., 2017). In addition to topography, the surface charge can influence biofilm structure. For example, P. aeruginosa biofilms form mushroom shapes on negatively charged surfaces while flat on positively charged surfaces. This phenomenon is due to the negatively charged surface inhibiting motility, preventing bacteria from moving over the surface to create a flat biofilm (Rzhepishevska et al., 2013). Besides, substrates’ stiffness also affects biofilm’s susceptibility; biofilms growing on stiff substrates were less susceptible to antibiotics than those on the soft substrates (Song and Ren, 2014).

In addition to influencing naturally formed biofilms, materials could also assist bacteria in forming biofilms for better performance (Hubenova et al., 2019; Reinhardt et al., 2020). For example, alginate could be functionalized with thiazolyl blue formazan/phenazine methosulfate and showed good biocompatibility of immobilization of bacteria Pseudomonas putida 1046 for constructing artificial biofilms. This artificial biofilm exhibited good electrochemical activities due to the incorporated redox-active cross-linking network in the surrounding polymer matrix around the microbial cells. They were considered good candidates for bioanodes in microbial fuel cells for wastewater treatment (Hubenova et al., 2019). The layer-by-layer technique is another method to manipulate biofilms artificially. Rijavec et al. constructed a protective biofilm coating consisting of two strains on steel surfaces. Gramicidin S producing cells Brevibacillus brevis are encapsulated with artificial polyelectrolytes deposited on steel surfaces and then enclosed within a rubber elastomer layer. On the rubber elastomer layer surfaces, Bacillus pumilus cells were then deposited with the assistance of polyelectrolytes. They demonstrated that when applied, this artificial biofilm could affect the phylogenetic structure of the developing natural biofilm and show a possible way to reduce phylogenetic diversity and exclude undesired bacteria (Rijavec et al., 2019).

Biofilm dispersion is an active process that is usually initiated by native and environmental cues. By leaving bacteria to convert to their planktonic growth mode and leaving behind eroded biofilms, biofilm is believed to be more vulnerable at this stage. Therefore, it has been proposed that promoting biofilm dispersion can be one of the antimicrobial methods. However, more caution needs to be taken by using this method since it has been found the dispersed cells display different phenotypes compared to their planktonic and biofilm counterparts. Dispersing cells without efficient killing might cause a significant problem (Rumbaugh and Sauer, 2020). In addition to natural dispersion, materials can also assist this process. Iron oxide nanoparticles are one of the most studied candidates for assisting in removing established biofilms by either catalyzing free radical generation (Gao et al., 2016; Liu Y. et al., 2018; Hwang et al., 2019) or mechanical force like digging channels or mechanically erasing biofilm structure (Hwang et al., 2019; Quan et al., 2019). In addition to adding external particles, dynamic substrates showed promising results in effectively eliminating matured biofilms (Geilich et al., 2017; Wang W. et al., 2018; Gu et al., 2020). It has been studied on a catheter model. By applying a relatively high strain rate on silicon elastomer-based substrates, biofilms can be readily detached as large pieces once the applied strain reaches a critical value. Interestingly, when the applied strain rate is relatively low, biofilms can still be bound to the substrates even if the substrates are under high strains (Levering et al., 2014). Besides, by incorporating magnetic particles on top of the pillar, micron-sized pillars can be engineered to beat at a programmable frequency and force level when applying an electromagnetic field. These active surfaces effectively removed established biofilms of uropathogenic E. coli, P. aeruginosa, and S. aureus (Figure 4B; Gu et al., 2020). Shape memory polymers are another candidate for fabricating dynamic substrates. Gu et al. (2016b) used shape memory polymer to fabricate recessive hexagonal patterns and found a dynamic change in surface topography could remove the biofilm. However, this is a one-way shape memory polymer, which means that the material can undergo shape change only once. To overcome this drawback, Lee et al. used a reversible shape memory polymer with a shape transition temperature close to body temperature, proving that P. aeruginosa biofilm could be effectively removed. It was also shown that the detached biofilm cell had increased antibiotic susceptibility compared to the static control (Figure 4C; Lee et al., 2021).

The biofilm secretes EPSs containing mainly polysaccharides, eDNA, and proteins. Polysaccharides constitute most biofilm mass composing glucan, fructan homopolymers, glucose/mannose/rhamnose heteropolymers, cellulose, alginate, colonic acid, and N-acetylglucosamine, and are highly heterogeneous across bacterial species (Nguyen, 2017). eDNA primarily relies on cellular lysis, which is an energetically inefficient system for large-scale production (Nguyen, 2017). Functional amyloids are one of the important proteins in the biofilm matrix. They comprise stacks of β-sheets aligned perpendicular to the fibril axis (Cao and Mezzenga, 2019; Gallardo et al., 2020). Unlike amyloid fibers formed upon aberrant misfolding of proteins, which are always associated with several incurable degenerative human diseases, bacterial amyloids benefit the life cycle of the biofilm. They are the key structural components of the biofilm matrix and can mediate toxicity and cell-to-cell interactions within the biofilm. Besides, amyloids are responsible for the spatial structure, morphological differentiation, surface properties, and viral protection of the biofilm (Biesecker et al., 2018; Deshmukh et al., 2018; Erskine et al., 2018; Van Gerven et al., 2018).

Curli are one of the important amyloids that have been used as building blocks for specific applications in nanotechnology due to their high stability and physical robustness (Taglialegna et al., 2016). They are non-covalent heteropolymeric filaments of CsgA and CsgB subunits where CsgA is secreted into the extracellular milieu and self-assembled into nanofibers by seeding onto a membrane-anchored nucleation protein CsgB (Van Gerven et al., 2015; Nguyen, 2017). Therefore, by fusing functional peptides or proteins into the amyloid monomers, a large-scale nanomaterial with programmable functionality can be secreted by cells (Elizabeth et al., 2016; Knowles and Mezzenga, 2016; Li et al., 2018). Nguyen et al. presented a biofilm-integrated nanofiber (BIND) system for precise genetic programming of bacterial matrix by fusing peptide domains onto the amyloid protein CsgA. Three peptides with different functions were fused to CsgA through C-terminal for extracellular self-assembly into functionalized curli nanofibers. They demonstrated the ability to engineer biofilm to template silver nanoparticles, increase adhesion to 304L stainless steel, and covalent immobilize proteins (green fluorescent protein as an example) (Nguyen et al., 2014). Similarly, mussel food protein of Mytilus galloprovincialis can also be fused with CsgA protein to produce adhesives with better properties like strong wet bonding strength, robustness, stability, and intrinsic fluorescence (Zhong et al., 2014). Besides, the BIND platform can also be used to immobilize enzymes for biocatalytic surface application. It has been reported that upon fusing to the SpyCatcher attachment domain, a recombinant α-amylase could be immobilized onto E. coli curli fibers displaying complementary SpyTag capture domains. The obtained enzymes’ immobilized biofilms were shown to be active after exposure to various adverse conditions (Botyanszki et al., 2015). Patterns are always relatively challenging to fabricate and usually require multiple fabrication steps (Shi et al., 2020). Using inducible genetic circuits and cellular communication circuits to regulate E. coli curli amyloid production can render amyloid fibril forming patterns either autonomously or via external control (Chen et al., 2014; Wang X. et al., 2018). It has been demonstrated that amyloid fibrils assembled could co-organize and synthesize inorganic nano-objects like fluorescent quantum dots (QDs) and gold nanowires, nanorods, and nanoparticles (Chen et al., 2014). Wang et al. further pushed these bio-abiotic hybrid materials to a new complexity level by engineering E. coli to harbor a blue-light inducible gene circuit to control the expression of CsgA through programmable light regulation; it is capable of spatiotemporally controlling nano-objects assembly. They demonstrated that discrete nano-objects and complex heterogeneous structures could be assembled hierarchically on different substrates or 3D materials of complex shapes. A minimum patterning resolution of 100 μm was achieved using light-sensing cells (Wang X. et al., 2018). Resettable pressure sensors could be fabricated using amyloid fibrils as the scaffold. With the help of inkjet printing to initiate single colony localization, engineered self-patterned bacteria could grow and subsequently facilitate the assembly of nanoparticles into a 3D dome structure. Their geometry would determine the dome structure’s response to pressure (Cao et al., 2017).

In addition to E. coli cells, Bacillus subtilis is another strain capable of this purpose. B. subtilis’ biofilm amyloid fibers were composites of TasA and TapA as major and minor protein components, respectively. The tapA-sipW-tasA gene operon regulates the production of amyloid fibers. Huang et al. fused the extracellular amyloid-like protein Tas A with various other proteins or protein domains to endow the biofilm with new functionalities. They demonstrated this platform could make programmable materials with various functions, including intrinsic fluorescent, intact enzyme activity, and the capability of templating inorganic nanoparticles. Besides, they were also able to adjust biofilm’s viscoelastic properties to manipulate them into diverse shapes and microstructures independently or using 3D printing techniques (Huang et al., 2019). Similarly, Zhang et al. (2019) engineered amyloid protein with a mussel foot protein and hydrophobin-like protein to control adhesion.

Mineralization is a process in which biology regulates the formation of hierarchical architectural mineralized materials and is vital for nature and human activities. It is a prevalent process in which species capable of forming minerals could be found in six taxonomic kingdoms. Examples of natural mineral products include teeth, coral, protozoan shells, etc. (Yao et al., 2017; Ehrlich et al., 2021). The biomineralization process generally involves an organic matrix to form a framework, and then nucleation sites are constructed, followed by controlling crystal orientation, growth, and termination crystal growth (Sharma et al., 2021). It is generally classified into two classes—biologically induced and biologically controlled mineralization, depending on the degree of biological control exerted. The former refers to the passive interaction of environments that drive the precipitation. The latter means the organism directly controls the precipitation process, producing minerals of a specific size, morphology structure, and orientation. Some literature has already documented the detailed mechanism, which will not be covered here (Qin et al., 2020; Hoffmann et al., 2021). Depending on the strains, bacterium mediates different mineralization, including calcification, silicification, iron, Mn mineralization, and so on (Couasnon et al., 2020; Qin et al., 2020). The application of biomineralization covers many fields such as environmental science for pollutant removal (Long et al., 2021) or water treatment (Arias et al., 2019), civil engineering for artwork conservation or as construction material (Achal et al., 2015; Marvasi et al., 2020), biomedical engineering for cancer therapy, or tissue engineering (Chen Y. et al., 2019; Qin et al., 2020).

In addition, to use bacteria individually for nanoparticle fabrication, mineral generation, or pollutant degradation, there are studies treating biofilm as a type of material. Zeng et al. discovered two Pesudoalteromonas lipolytic variants are capable of forming wrinkled and translucent biofilms. The wrinkled biofilm was due to a nonsense mutation in AToo_o8765, causing overproduction of extracellular polysaccharides, while the translucent ones were due to a point mutation in AToo_17125. Biofilm formed by either of these two strains was proved to have antifouling activities toward larval settlement and metamorphosis of the mussel Mytilus corucus (Zeng et al., 2015). Liu et al. further applied the strain capable of overexpression cellulose for steel protection. This strain can effectively bind Ca²⁺ and form biomineralization organic-calcite hybrid film on the steel surface to provide protection. The film formed by living bacteria could also provide in situ self-healing activity (Liu T. et al., 2018). In addition to selecting natural strains, Yang et al. genetically engineered E. coli bacteria to express DDDEEK peptide, having a strong ability to absorb mineral ions and induce the formation of a biomineral. This engineered biofilm was shown to be capable of tolerance shear force and stayed on virtually any type of material surface. Interestingly, this biofilm-based coating not only exhibited good mineralization performance and better stability compared to the hydroxyapatite spray sample but also showed no extra immunogenicity and better osteogenicity and osseointegration after 12 weeks of implantation, demonstrating their potential applications in the biomedical field (Yang et al., 2018). Wang et al. genetically engineered E. coli to achieve more control over the mineralization process. CasA and mussel foot protein Mfp were fused to render higher hydroxyapatite mineralization, capacity, and interfacial binding strength. To enable E. coli biofilms to respond to environmental stimuli, such as light, they encode the expression of CsgA-Mfp3S-pep in the Dawn plasmid, which renders the biomass density and spatial pattern controllable. Interestingly, bacteria cells inside the composite remained alive and capable of responding to environmental stimuli after mineralization (Wang Y. et al., 2021).

Bacterial metabolic activities are important when they are applied for fermentation or remediation. Bacteria in their biofilm form have been reported to increase the production of value-added products and secondary metabolites. Biofilm reactors, where biofilms were immobilized onto surfaces, have been widely accepted as a promising technology in wastewater treatment, pollutant removal, and production of alcohol, organic acid, enzymes, syngas, and many other value-added products (Mahdinia et al., 2019; Abdelfattah et al., 2020; Gunes, 2021). Biofilm’s properties such as adhesion strength, biomass, and population are closely related to its efficiency. Synthetic biology techniques can be applied to genetically modify strains to significantly improve their performance. For example, the poor adhesion capacity of stain B. Subtilis 168 was the main issue that limited their application in biofilm reactors. After insertion of the sfp and epsC genes together with the deletion of the sepF gene, the biofilm’s form by this mutant strain showed increased adhesion and surfactin production (Brück et al., 2019). Pseudomonas putida is a widely used strain in bioremediation due to its remarkable ability to degrade pollutants; however, most of the laboratory-adapted P. putida strains suffer from weak biofilm formation abilities. Benedetti et al. genetically programmed their intracellular c-di-GMP levels using an inducible and tightly controlled genetic device for heterologous expression of diguanylate cyclase and phosphodiesterase genes. They then proved this biofilm formed by engineered strain displayed high dehalogenase activity (Benedetti et al., 2016). In addition to monostrain biofilms, microbial consortia behaviors could also be controlled. By manipulating communication networks, regulating gene expressions, and engineering syntrophic interactions, it is possible to control individual species’ population, distribution, and spatial organization in a microbial consortium (McCarty and Ledesma-Amaro, 2019).