The prerequisite for the formation of a biofilm is that planktonic bacteria should approach and contact a surface. During this initial step of attachment, motilities play a critical role. Brownian motion, during which particles randomly collide with each other, has been suggested to play a crucial role in the movement of non-appendaged bacteria, such as Staphylococcus sp. and Klebsiella sp. (Floyd et al., 2017). Other biological factors, such as flagella and type IV pili (TFP), have an impact on bacterial motility through fluids. Additional combinations with adhesive molecules, such as adhesive pili, proteins, and surface-bounded EPS, facilitate bacterial adhesion to a surface (Kostakioti et al., 2013; Belas, 2014; Guo et al., 2021). Despite the effective adhesins, the attachment is fragile and usually transient during this stage, because of hydrodynamics and repulsive forces at the boundary layer and the inadequate production and function of adhesive molecules (Petrova and Sauer, 2012; Kreve and Reis, 2021). Long-term attachment to the surface results in contact-dependent gene expression, which ultimately leads to the alteration from reversible to irreversible attachment.

Irreversible attachment means a sessile and sedentary adhesion to the substratum. Contact-dependent signaling pathways induce the production of important mediators for stable adhesion (Petrova and Sauer, 2012). After rounds of detachment and attachment, the second messenger cAMP creates an intermediate surface sentient state between motile and sessile in the model biofilm bacteria Pseudomonas aeruginosa. The Pil-Chp surface-sensing system contributes to an increasing level of cAMP, which oscillates and ultimately downregulates the activity of type IV pili. Multigenerational cAMP-TFP oscillations facilitate progression to the irreversible attachment stage (Luo et al., 2015; Lee et al., 2018). Another important mediator is bis-(3′-5′)-cyclic dimeric guanosine monophosphate (c-di-GMP), an essential factor that determines the change from free-moving to the attached state for several Gram-negative bacteria (Römling et al., 2013; Liu et al., 2020). C-di-GMP promotes the production of EPS. A study into P. aeruginosa suggests that exopolysaccharide Psl induces intracellular synthesis of c-di-GMP, which decreases flagella and elevates the level of Psl and other EPS molecules (Irie et al., 2012). Moreover, in P. aeruginosa, c-di-GMP can repress the synthesis of flagella by binding with transcription factor FleQ (Baraquet and Harwood, 2013). During the irreversible attachment stage, microbial cells accumulate and aggregate in layers and produce massive EPS.

As an increasing number of bacteria attach irreversibly and the attached bacteria multiply, biofilms gradually form with complex structures and diverse bacterial species. Then, microcolonies, the biofilm precursor, are built and further develop into matured macrocolony biofilms. On the formation of microcolonies, encapsulated bacteria produce large amounts of EPS, which supports inner microbes and shields them against antibacterial molecules. EPS components vary between different bacterial species or even strains of the same species but commonly include polysaccharides, proteins, and DNAs (Floyd et al., 2017). During cariogenic biofilm formation, the binding interactions between Streptococcus mutans and Candida albicans are selectively increased when C. albicans cell wall is coated with extracellular glucans. Of interest, co-existing C. albicans promotes S. mutans biofilm maturation and maintains an acidic environment, which contributes to the pathogenesis of dental caries (Kim et al., 2020). Escherichia coli mainly produces cellulose as the exopolysaccharide constituent and P. aeruginosa secretes polysaccharides that involve Psl, Pel, and alginate. However, not all P. aeruginosa strains produce these three types of compounds. Mucoid strains of P. aeruginosa are the causative agent of cystic fibrosis and excrete all three polysaccharides, of which alginate is dominant. Alginate is associated with persistence and immune evasion and promotes P. aeruginosa coinfection with Staphylococcus aureus in cystic fibrosis disease. Non-mucoid P. aeruginosa strains, PAO1 and PA14, only produce Psl and Pel (Wozniak et al., 2003; Irie et al., 2012; Wei and Ma, 2013; Limoli et al., 2017).

During this phase, bacteria continuously grow and proliferate with new inner cells that attach to the surface and the whole biofilm spatially expands upward. Matured biofilm is characterized by a mushroom-shaped three-dimensional formation (Muhammad et al., 2020). A recent study into biofilm-forming Vibrio cholerae suggested that intercellular mechanical potentials resulted in cellular order and overall construction by the production of specific EPS ingredients. In addition, shear flow, one of the external factors, plays a key role in shaping biofilm architecture by competing against cell–cell attraction forces (Hartmann et al., 2019). Mature biofilms encompass internal microbes and protect them from being killed by chemotherapies and host defenses and continue to exist for a prolonged period (Breidenstein et al., 2011).

As a biofilm matures, some microorganisms detach from the biomass and disperse. These discrete bacteria could reattach into a surface and form a secondary biofilm or float freely in the intermediate environment (Kaplan, 2010). Two main mechanisms are involved in the dispersal of biofilm bacteria (Rumbaugh and Sauer, 2020; Wille and Coenye, 2020). Passive dispersion is decided by mechanical or shear stress. For example, the weight of a 67 h-old S. mutans biofilm biomass was reduced to only 15% of primary mass after treatment by 10 times increased shear stresses (Hwang et al., 2014). Based on this knowledge, high-pressure pulsating water is now used to clear dental plaque that causes dental decay when deposited on tooth enamel. The water jet is much more efficient than toothbrushes to eradicate dental plaque from hidden and narrow areas (Sharma et al., 2008). In contrast to passive dispersion, active biofilm dispersal is initiated by bacterial responses to triggers, such as nutrient abundance and atmospheric triggers (hypoxia or low nitric oxide), and results from the activation of several signaling factors (James et al., 1995; Jiang et al., 2020). One of the main mechanisms is the reduction of intracellular c-di-GMP. For example, accumulated nitric oxide is sensed by NbdA and activates phosphodiesterase, which degrades c-di-GMP in P. aeruginosa. Decreased levels of c-di-GMP are associated with elevated motility and downregulated EPS synthesis, leading to lower tackiness and ultimate dissemination (Li et al., 2013; Cutruzzolà and Frankenberg-Dinkel, 2016). Another important mechanism is the involvement of extracellular enzymes that target structural EPS ingredients. S. aureus secretes four types of proteases, serine protease (SspA), cysteine protease (SspB), staphopain (ScpA), and metalloprotease (Aur), which lead to the degradation of extracellular proteins (Fleming and Rumbaugh, 2017). Dispersin B, produced by several bacteria including S. aureus, Klebsiella pneumoniae, and E. coli, is a glycoside hydrolase that degrades the exopolysaccharide and displays biofilm disorganizing functions (Fleming and Rumbaugh, 2017; Guilhen et al., 2017; Wille and Coenye, 2020).

Of note, different types of substrates have great impact on the formation and development of biofilms. On artificial substrates (carbon fiber and polyvinyl chloride), bacterial network patterns are more complex than those formed on natural surfaces (pebble and wood). It is also reported that bacteria colonized on artificial substrates are more powerful in metabolizing nitrogen, carbon, and arsenic sources (Miao et al., 2021). Therefore, the effects of different substrate types on the dynamics of biofilm community could attract more attention in the field of biofilm research. In addition, modeling biofilm formation in vitro allows the biofilm to be studied more closely and helps us to explore potential treatments for the targeted biofilms. For instance, flow chambers are promising approaches that could be used to assess bacterial adhesion to biomaterial surfaces, which is evaluated by microscopy (Gomes and Mergulhão, 2021). Microfluidic platforms contain multiple channels to study bacterial adhesion, biofilm formation, or antimicrobial treatments. A newly designed microfluid chip, BiofilmChip, recreates polymicrobial biofilm formation under clinical conditions and monitors the heterogeneous biofilm microenvironment (Blanco-Cabra et al., 2021). To gain knowledge on biofilm formation and to test antimicrobial agents directly, investigations are required to validate existing in vitro platforms further and to develop more effective techniques.

The prerequisite to forming mixed-species biofilms is the coexistence of distinctive microbes (Li and Tian, 2016). A recent study into 29 bacteria and 1 fungus isolated from various environments demonstrated that 77% of these 30 strains coaggregated with at least another strain and 70% coaggregated with other archetypic strains. This coaggregation was mostly mediated by the interaction between lectin and saccharide (Stevens et al., 2015). To date, the most studied cohesion and coaggregation model are the oral biofilm–dental plaque. Streptococcus gordonii, a first colonizer on dental plaque, provides specific cues to colonize with Porphyromonas gingivalis into a polymicrobial biofilm. Apart from the interaction of SspB (a surface polypeptide on S. gordonii) with short fimbriae on P. gingivalis, some components identified in S. gordonii are associated with this interconnection, including adhesive proteins and extracellular capsules (Park et al., 2005; Kuboniwa et al., 2006). Another initial colonizer, Streptococcus oralis, secretes glyceraldehyde-3-phosphate dehydrogenase, which acts as a coadhesin by binding to major fimbriae on P. gingivalis (Maeda et al., 2004). In addition to the motility organelles involved in cohesion, to unravel the interaction between two initiators of dental plaque, S. gordonii and Veillonella parvula, dual-RNA sequences on each strain indicated that oxidative stress-associated processes dominated the V. parvula coaggregation responses and S. gordonii mainly focused on carbohydrate metabolism during this process (Mutha et al., 2019).

In summary, microbes initiate cohesion and coaggregation by producing several adhesive components and induce intercellular interactions to promote multispecies coexistence within a biofilm.

After interbacterial coaggregation, which results in efficient cell-to-cell communication, the metabolic interactions emerged as another important cooperative strategy.

Different species cooperate to break down a common substrate into usable nutrients and achieve nutritional interdependence. For example, inside a multispecies biofilm consortium with Acinetobacter johnsonii strain C6 (Kaas et al., 2017), Pseudomonas putida evolves into rough colony variants by natural selection (Hansen et al., 2007). Specifically, when both bacteria originally used aromatic benzyl alcohol as their only carbon source, the P. putida variant was dependent on the metabolite, benzoate, from the A. johnsonii and could survive oxygen starvation in the mixed-species biofilm. The variant of wapH mutation increased the production of a cellulose-like polymer, therefore, enhancing adherence with A. johnsonii (Hansen et al., 2007).

However, when different strains have access to distinctive nutrient substrates, they have the potential to exchange metabolites and catalyze them further in another pathway. This cross-feeding or syntrophy is a common mechanism during metabolic interactions (Adamowicz et al., 2018; Evans et al., 2020). Using oral biofilms as an example, Veillonella sp. rely on lactate as a nutrient source. Veillonella sp. are frequently in symbiosis with lactic acid-producing bacteria, such as S. oralis and S. mutans. This type of cross-feeding on lactate has been observed between S. gordonii and Aggregatibacter actinomycetemcomitans (Guo et al., 2014; Bowen et al., 2018). In addition, S. gordonii expresses putative arginine-ornithine antiporter ArcD, which plays an important role in arginine uptake and maintenance of nutrient supply. Arginine is commonly used by S. gordonii and transformed into ATP and NH₃, which are beneficial for bacterial growth and homeostasis. Ornithine is exported by ArcD from S. gordonii to Fusobacterium nucleatum, in which orthenine is broken down by ornithine decarboxylase into putrescine. Simultaneously, a dual-species biofilm of F. nucleatum and S. gordonii is attenuated by arcD depletion, whereas restoring ornithine reverses inhibited biomass of this mixed-species biofilm, which suggests the need for ornithine cross-feeding to maintain biofilm development (Jakubovics et al., 2008; Sakanaka et al., 2015). Despite the well-studied cross-feeding between different species, unidirectional cross-feeding in single-species biofilm has been identified. Investigations into the sputum from cystic fibrosis patients found oxic and hypoxic zones. Under anaerobic conditions, P. aeruginosa reduces pyruvate to lactate through lactate dehydrogenase, which is further used as a nutritional supplement. Lactate induces the active expression of IldE, the enzyme that catalyzes D-lactate oxidation during aerobic growth. Due to the spatial distribution of oxygen, lactate produced by P. aeruginosa in an anoxic zone further feeds the bacteria in the oxic area and activates LldE expression (Lin et al., 2018; Evans et al., 2020). In addition, P. aeruginosa could survive on propionate and acetate converted from mucin by commensal anaerobic bacteria, and genes involved in propionate utilization are activated during in vivo cystic fibrosis. P. aeruginosa uses mucin as another carbon source by cross-feeding from other species (Flynn et al., 2016). Cross-feeding is beneficial because it gives single or multispecies biofilm systems higher metabolic efficiency by maximally utilizing waste products and makes nutrient sources available for the microorganisms to support their growths.

Through quorum sensing (QS), in which small diffusible signal molecules (autoinducers) are released and are detected to induce gene expression in a coordinated manner, mono or polymicrobial communities achieve cell–cell communication in response to microbial density (Abisado et al., 2018). In general, Gram-positive bacteria produce autoinducing peptides for QS and Gram-negative bacteria use acyl-homoserine lactones as signal factors whose synthesis depends on the LuxI-like protein. Another type of QS, LuxS-encoded autoinducer 2 (AI-2), is found in Gram-negative and Gram-positive bacteria, which mediates interspecies interactions (Li and Tian, 2012; Mukherjee and Bassler, 2019). For example, AI-2 produced by S. oralis promotes the development of a mixed biofilm with Actinomyces naeslundii, which was demonstrated by the inability of S. oralis that had mutated LuxS to form a biofilm with A. naeslundii and the recovered ability to form this dual-species biofilm after restoring AI-2 synthesis (Rickard et al., 2006). In addition, F. nucleatum secretes AI-2 that significantly induces coaggregation with each species in a red complex (Porphyromonas gingivalis, Treponema denticola, and Tannerella forsythia) and this process is inhibited by QS inhibitors. Specifically, AI-2 triggers the production of bacterial surface adhesins and mediates interspecies interactions between period into pathogens (Jang et al., 2013). As in gastrointestinal niches, Gram-positive bacterium Enterococcus faecalis produces AI-2 that attracts E. coli and maintains their expression of lsr operon even at low cell densities, which results in enhanced aggregation of microcolonies (Laganenka and Sourjik, 2018). Apart from QS between bacterial species, interkingdom interactions via QS have been demonstrated. S. gordonii produces AI-2 that is sensed by the fungi C. albicans and induces elevated hyphae synthesis and modulates the expression of mitogen-activated protein kinases (Bamford et al., 2009). In addition, the concentration of signal molecules has a role. A QS molecule from C. albicans, farnesol, enhances growth and microcolony formation of S. mutans at relatively low levels, whereas a higher concentration of this molecule inhibits S. mutans proliferation (Kim et al., 2017).

In summary, QS plays an indispensable role in biofilm formation and interspecies or cross-kingdom communication by inducing associated gene expressions.

Close cell–cell contact in biofilms creates a favorable condition for HGT. To date, it is well known that HGT occurs with increased competence in biofilms. Several classic mechanisms are involved in HGT, which include conjugation, transformation, and transduction (Molin and Tolker-Nielsen, 2003). In addition, recent studies have revealed other mechanisms that include membrane vesicles (MVs) and gene transfer agents (GTAs; Figure 2).

Conjugation is a direct cell-to-cell transfer by plasmids or transposons. For instance, in oral biofilm structures, live donor bacterium Veillonella dispar can transfer the conjugative transposon Tn916 to four different Streptococcus spp. that are recipients in a multispecies oral consortium that grows as a biofilm (Hannan et al., 2010). In addition, one study into soil bacteria indicated plasmid permissiveness during conjugation in a recipient-dependent manner. E. coli transfers the IncP1 plasmid, pKJK10, to a mixed collection of 15 bacterial strains; however, only the transfer of this plasmid from P. putida to Stenotrophomonas rhizophila succeeded, and transfer to the mixed community was not observed (de la Cruz-Perera et al., 2013).

Transformation is the simple use of eDNA in the surrounding environments. In marine biofilms on chitinous surfaces, QS in V. cholerae that was mediated by AI-2 and CAI-I enhanced the competence of eDNA uptake by inducing competent gene comEA expression when cocultured in a mixed-species biofilm (Antonova and Hammer, 2011). eDNA is often released after cell autolysis. In an experimental biofilm, the transformation of purified V. dispar genomic DNA to Streptococcus mitis gave S. mitis tetracycline resistance (Hannan et al., 2010).

Transduction refers to the transfer of host genomic DNA in which bacteriophages are the carriers. Bacteriophages that carry Shiga toxin (Stx)-encoding genes promoted the emergence of a new Stx-producing E. coli (STEC) within the biofilm at 37°C or 20°C, which suggested that a biofilm environment was suitable for transduction and promoted interactions between strains of the same species (Solheim et al., 2013).

MVs are released by bacteria and carry genetic materials that fuse with the recipient bacteria for HGT (Toyofuku et al., 2019). GTAs are phage-like particles that are produced by bacteria, which carry random DNA fragments from the original bacteria and transfer this into the recipient bacteria (Lang et al., 2012; Bárdy et al., 2020). MVs and GTAs are important factors for HGT in a marine planktonic ecological environment (Paul, 2008; Lang et al., 2012; Lossouarn et al., 2015; Biller et al., 2017). Recent research found that MVs promoted the formation of single-strain bacterial biofilms (Seike et al., 2020). The MVs from S. mutans augmented C. albicans biofilm development but had no significant effect on C. albicans growth under planktonic conditions (Wu et al., 2020). However, the role of MVs and GTAs in multistrain bacterial biofilms remain unclear. Further investigations into MVs and GTAs are required to expand the knowledge on HGT in biofilms.

Of interest, previous studies suggested that the density of cell populations in biofilms promoted plasmid dispersal via conjugation, which then induced biofilm maturation (Molin and Tolker-Nielsen, 2003; Madsen et al., 2012). In addition, released DNA enhanced the stability of biofilms by synthesizing EPS (Montanaro et al., 2011). In summary, a biofilm provides a suitable environment for HGT, which then produces populations with antibiotic resistance or higher pathogenicity, which promotes the homeostasis of the biofilm (Madsen et al., 2012).

Microorganisms in polymicrobial biofilms benefit each other by secreting certain molecules. Within polymicrobial biofilms of otitis media, Moraxella catarrhalis secretes β-lactamase and protects Streptococcus pneumoniae from death by β-lactam, and a quorum signal (AI-2) generated by S. pneumoniae further increases M. catarrhalis colonization (Perez et al., 2014). This type of positive feedback loop makes either microorganism more tolerant to clearance (Perez et al., 2014). In dental plaque biofilms, for example, interkingdom synergism has been characterized. Glucosyltransferase B, an exoenzyme derived from S. mutans, binds to the mannan layer of C. albicans, which augments EPS formation and the development of mixed-species biofilm (Ellepola et al., 2017; Hwang et al., 2017). In turn, C. albicans produces polysaccharides to support colonization and adhesion of S. mutans, which promotes mixed biofilm formation and induces dental caries (Khoury et al., 2020). Apart from mutually supportive interactions, biofilm bacteria can protect other microorganisms. Streptococcus sanguinis in oral cavities produces hydrogen peroxide (H₂O₂), which suppresses the survival of P. gingivalis. The presence of A. actinomycetemcomitans reverses this inhibition through the secretion of KatA-encoded catalase, which reduces H₂O₂ concentrations, and therefore, supports the survival of P. gingivalis in this tri-species biofilm (Zhu et al., 2019). Thus, when the diversity of microorganisms in a biofilm increases, mechanisms underlying the interactions between different species become more complex. To gain more knowledge about the panorama of biofilm community, we need to understand the competitive and antagonistic interactions in addition to cooperation and synergism.

Due to the requirements for limited nutrients, different bacterial species compete for nutrients to survive. For example, obligate aerobe Brevibacillus sp. M1-5 primarily dominates at the air–water interface, the emergence of the facultative aerobe Pseudoxanthomonas sp. M1-3 competes for oxygen, which decreases the number of viable Brevibacillus sp. (Yamamoto et al., 2010). Another example is iron chelation that uses siderophore. In the dual-species biofilm composed of Pseudomonas fluorescens and P. putida, chelator production provides an absolute growth advantage over non-chelator populations (Eberl and Collinson, 2009). Uropathogenic strains of K. pneumoniae and E. coli compete for iron, in which K. pneumoniae grows better than E. coli by the production of siderophores. The competition for nutrients, in particular iron, ultimately modulates composition (K. pneumoniae: E. coli ratio, 55:1; Juarez and Galván, 2018). Therefore, the ability to effectively use nutrients promotes bacterial growth to the detriment of its competitors, which leads to changes in biofilm composition.

In a biofilm, several bacteria could produce antibacterial or bacteriostatic proteins, for example, bacteriocins. The main difference between bacteriocins and antibiotics is that bacteriocins show unique activity to strains that belong to the (same) producing species (García-Bayona and Comstock, 2018). For example, Streptococcus salivarius in the oral cavity contributes to the production of large amounts of bacteriocins. Santagati et al. identified 13 from 81 α-hemolytic streptococci strains, which were isolated from healthy children, produced bacteriocins, and inhibited different Gram-positive pathogens, such as S. pneumoniae (Santagati et al., 2012). Another study demonstrated the probiotic ability of S. salivarius 24SMB and S. oralis 89a, which disturbed the biofilm development or even dissolved preexisting biofilms from typical upper respiratory tract pathogens that involved Streptococcus pyogenes, Streptococcus pneumoniae, Moraxella catarrhalis, S. aureus, Staphylococcus epidermidis, and Propionibacterium acnes (Bidossi et al., 2018).

Studies have found that about 30–50% of E. coli strains are efficient producers of colicin (Riley and Gordon, 1992; Gordon et al., 1998; Smarda and Smajs, 1998; Gillor et al., 2004). Colicin is possibly encoded by the kil gene-containing plasmids and released into the extracellular environment by cell lysis (Rendueles and Ghigo, 2015). Colicins are a particular group of bacteriocins that are produced by and toxic to some strains of E. coli and other enteric bacteria and colicin-producing strains are resistant to the colicin they produce (Hibbing et al., 2010). By binding to surface receptors and then entering colicin-sensitive cells, colicin mediates cell death by producing pores on the membrane or degrading DNA and RNA. In response to SOS stress, colicin R is induced in E. coli ROAR029 that resides in a matured mixed biofilm rather than in a free-swimming status, which exerts highly competitive functions against enteroaggregative E. coli LF82 by pore formation (Rendueles et al., 2014).

Various bacteria produce toxic molecules that kill neighboring bacteria. A well-studied low molecular weight compound is H₂O_2. Studies into oral species S. gordonii, S. sanguinis, and caries-inducing S. mutans demonstrated the importance of H₂O_2. S. gordonii and S. sanguinis produce H₂O₂ via SpxB-encoded pyruvate oxidase to inhibit pathogenic S. mutans. This antagonism is modulated by glucose and oxygen availability. In addition, S. mutans produce mutacins, which have cytolytic functions on other Streptococcus species (Kreth et al., 2008). A. actinomycetemcomitans have developed responses against H₂O₂ by secreting catalase (KatA) and Dispersin B (DspB) that detoxifies H₂O₂ and assists bacteria to escape from damage, respectively, which suggests an adaptive mechanism to counteract disruptions by toxic compounds (Stacy et al., 2014).