The extracellular matrix (ECM), defined as the three-dimensional network of high-molecular-weight extracellular molecules that surround the surface of the cells that produce them, is one of the few structures conserved during evolution and shared through the three biological domains of life: Archaea, Bacteria, and Eukarya (Stavolone and Lionetti, 2017). Although its composition varies according to the type of organism, i.e., prokaryote vs. eukaryote organisms, or unicellular vs. multicellular organisms, the ECM remains required for conserved functions, such as cell adhesion, intercellular communication, and differentiation (Ozbek et al., 2010; Limoli et al., 2015).

In bacteria, the ECM consists of three major polymer components: (i) polysaccharides, (ii) proteins, and (iii) extracellular DNA (eDNA). Bacterial ECM may serve several functions, such as protection of free-living bacteria (planktonic life), promotion of migration or genetic exchanges, and ion reservoirs. In addition, the ECM is an essential building material for biofilm formation (Dragoš and Kovács, 2017). Bacterial biofilms (sessile life) represent multicellular assemblies, in contact with a surface, in which bacteria stick to each other and are shielded by the ECM. They represent the most ubiquitous bacterial lifestyle in natural and artificial environments.

The eukaryotic ECM, whether in tissues or on circulating cells, is a network of extracellular macromolecules made up of core proteins (collagens, glycoproteins, and proteoglycans) and of ECM-associated components, including adhesion proteins and cellular receptors (Mouw et al., 2014). The ECM serves as a physical scaffold to the cell but it is also a dynamic structure interacting with other cells to regulate diverse cellular functions, including survival, growth, migration, and differentiation, as well as response to diseases (Bellincampi et al., 2014; Bonnans et al., 2014; Humphrey et al., 2014). Interestingly, the ECM can be strongly remodeled early upon viral infection. Viruses can use the ECM for attachment to target cells and subsequent interactions with cell-surface receptors allowing their entry. Besides its role to favor initial infection, modulation of the ECM from infected cells can impact viral disease manifestation directly or indirectly. For example, modification of the ECM by viruses may modulate inflammation. This is observed upon respiratory syncytial virus (RSV) infection of fibroblasts via hyaluronan-dependent mechanisms, which enhances mast cell binding as well as mast cell protease expression via direct interactions with the ECM (Reeves et al., 2020). It can also influence the process of carcinogenesis. For instance, Hepatitis B virus (HBV) encodes a viral onco-protein, transactivator protein X, involved in the ECM remodeling through the HIF-1α/LOX pathway, which is shown to promote hepatocellular carcinoma metastasis (Tse et al., 2018). After productive infection, modulation of the ECM can also enhance viral transmission through cell-to-cell contact. Indeed, infected cells can establish contacts between cells that are not normally in physical contact using for example, increased migration of infected cells toward non-infected cells. Alternatively, infected cells can exploit existing cell-cell interactions, such as immunological synapses, to increase adhesiveness and viral transfer through the immune synapse. For that purpose, some viruses such as human T-cell leukemia virus type 1 (HTLV-1) can upregulate the expression of endogenous cell adhesion molecules (CAM), such as Intercellular adhesion molecule-1 (ICAM-1), and of other components of the ECM (Nakachi et al., 2011; Gross and Thoma-Kress, 2016), while others can produce their own adhesion proteins. For example, the glycoprotein Env from Murine Leukemia Virus (MLV) can act as a viral adhesion molecule (VAM), mimicking the behavior of a CAM (Sherer et al., 2007; Mothes et al., 2010). Moreover, these virally induced areas of contact can be highly specialized and fully dedicated to viral transmission, as exemplified by the ability of herpes simplex virus (HSV), human immunodeficiency virus (HIV) and HTLV-1 to promote the formation of virological synapses (Vasiliver-Shamis et al., 2008; Nejmeddine and Bangham, 2010; Abaitua et al., 2013).

Among several viruses, HTLV-1 masters the art of remodeling the ECM, by forming a viral biofilm at the surface of infected cells (Pais-Correia et al., 2010). In this structure, numerous virions are embedded in the ECM of infected cells. Interestingly, spreading of virions in multiple collective entities have been described for many other viruses (Li et al., 2014; Sanjuán and Thoulouze, 2019), notably through transfer of previously accumulated virions in membrane invaginations or as aggregates, demonstrated for HIV. This collective spreading is hypothesized to provide a selective advantage by increasing virions stability, the number of virions delivered to target cells and thus global infectivity, and protection from immune response. However, virions accumulation in a viral biofilm has been clearly and convincingly demonstrated for HTLV-1 only (Li et al., 2014; Sanjuán and Thoulouze, 2019). Interestingly, the HTLV-1 biofilm appears to be very similar to bacterial biofilms (Thoulouze and Alcover, 2011), both consist in a microbial community embedded in an adhesive ECM. However, the similarities and differences between these bacterial and viral types of biofilms have been little discussed so far. In this review, we compare the molecular composition of bacterial and viral biofilms, and their pathophysiological impact on the host and on therapeutic strategies (summarized in Table 1). This comparative analysis of bacterial and viral biofilms highlights aspects of viral biofilms that are poorly understood, and shows how the understanding of bacterial biofilms may inspire future work on viral biofilms.

Bacterial biofilms can be formed on abiotic and biotic surfaces. Indeed, a wide range of abiotic surfaces may be colonized by bacterial biofilms, such as implanted medical devices, surgery equipment, or river stones. Biotic surfaces may be host tissues, skin, or medical devices that have been rapidly coated by host ECM proteins such as fibrinogen or fibronectin in the blood plasma (Ratner, 2015; Khatoon et al., 2018). Surfaces have different chemical (charge and hydrophobicity) and mechanical properties (topography, roughness, and stiffness), which influence the speed of bacterial biofilm formation and strength of attachment. Typically, 5–35% of the biofilm volume is constituted by the microorganisms while the remaining volume is ECM. The ECM composition can vary greatly between biofilms, depending on the strains/species present, the shear forces experienced, the temperature and the availability of nutrients (Barbosa et al., 2015; Oliveira et al., 2015; Mizan et al., 2016). However, all bacterial biofilms share general attributes. They contain polymeric molecules constituting the ECM involved in intercellular adhesion. Matrix polymers include exopolysaccharides, proteins, teichoic acids, and molecules released by eukaryotic and prokaryotic dead cells, which are mainly negatively charged DNA, called eDNA after its release (López et al., 2010; Bjarnsholt et al., 2018).

Exopolysaccharides are the ECM components most frequently detected in biofilms. Most are long molecules, linear or branched, which can be homopolysaccharides or heteropolysaccharides consisting, for the latter, of a mixture of neutral and charged sugar residues. Pseudomonas aeruginosa, an important Gram-negative pathogen and a strong biofilm inducer, produces at least three distinct exopolysaccharides that contribute to biofilm development and architecture: Alginate, Pellicle (Pel), and Polysaccharide Synthesis Locus (Psl). Alginate is overproduced by mucoid strains isolated from lungs of cystic fibrosis patients, in which it is involved in the initiation of biofilm formation and in the mechanical stability of mature biofilms. Pel and Psl are involved in the establishment of biofilms in non-mucoid wild-type strains, which do not produce alginate. Pel is a glucose-rich polymer, essential for the formation of biofilms at air–liquid interfaces (called pellicles) and of biofilms that are attached to a surface, while Psl, a mannose-rich polymer, is involved in the adherence to abiotic and biotic surfaces and in the maintenance of biofilm architecture (Branda et al., 2005; Ghafoor et al., 2011). For the Gram-positive pathogen Staphylococcus aureus, polysaccharide intercellular adhesin (PIA), also known as poly-N-acetyl glucosamine, is a major exopolysaccharide found in biofilms. However, the presence of PIA is not universal and is for example, dispensable for the formation of staphylococcal biofilms (Mack et al., 1996; Brooks and Jefferson, 2014).

Unlike bacterial biofilm, the viral biofilm components that have been identified so far are not encoded by the virus itself nor synthesized by viral enzymes. Besides exploitation of existing cell pathways and functions for their own replication, viruses also alter physiological processes important for their dissemination. HTLV-1 can remodel several components of the eukaryotic ECM, enhancing their expression and modifying their spatial organization, thus leading to the formation of the viral biofilm (Thoulouze and Alcover, 2011). Extracellular components in which viruses are embedded in biofilm at the surface of the infected cells are made of complex polysaccharide assemblies, enriched in heparan sulfate, mannose and galactose exposing mainly the Thomsen-Friedenreich (TF) antigen, a desialylated Galβ1-3GalNAc glycosylated structure, and the tetrasaccharide sialyl-LewisX (sLex; Hiraiwa et al., 1997, 2003; Pais-Correia et al., 2010; Assil et al., 2019; Nakamura et al., 2019). Interestingly, Galectin-3 and Tetherin, have been found in viral biofilm and could be involved in the retaining molecules early steps (Pais-Correia et al., 2010; Nakamura et al., 2019). The viral biofilm is also composed of ECM components, like Agrin, HSPG, and Collagen IV (COL4A1 and COL4A2; Pais-Correia et al., 2010; Gross and Thoma-Kress, 2016; Millen et al., 2019). Moreover, Tarasevich et al. showed that the cell surface proteins CD4, CD150, CD70, CD80, and CD25 are also present in viral biofilm structures (Tarasevich et al., 2015). Although O-glycosylated surface receptors CD43 and CD45 do not seem to co-localize with viral particles, they were suggested to play an important role in the accumulation of viral biofilm-like structures on one pole of the HTLV-1-infected cell, facilitating contact with the target cell and efficient transmission of viral material (Mazurov et al., 2012).

Thus, polysaccharides and protein components are shared between bacterial and viral biofilm, while in contrast eDNA is exclusively found in bacterial biofilm. In addition, while direct interactions among bacteria have been documented in the bacterial biofilm, data on the interaction between viral particles in the viral biofilm are missing.

Generally, bacterial biofilm formation is described as a cyclic process involving four main steps: adhesion, micro-colonies, maturation, and detachment (Figure 1). In the first step, bacteria have to adhere to a surface serving as a support (Armitano et al., 2013; Gagné-Thivierge et al., 2018). Mobile bacteria, such as P. aeruginosa and Escherichia coli, have a great advantage in this initial attachment, by utilizing flagella to overcome hydrodynamic and repulsive forces (O’Toole and Kolter, 1998; Wood et al., 2006; Liaqat et al., 2019). On abiotic surfaces, attachment depends on the surface properties of materials and bacteria, and involves hydrophobic interactions or Van der Waals forces (Renner and Weibel, 2011; Carniello et al., 2018). Wall components of several bacteria also serve as specific determinants for attachment. As examples, wall teichoic acids, lipoteichoic acids, and cell wall anchored proteins of staphylococci promote adhesion to abiotic surfaces (Vergara-Irigaray et al., 2008; Heilmann, 2011), while flagella, pili, or fimbria of P. aeruginosa play a role in surface attachment (Déziel et al., 2001; Vallet et al., 2001). When occurring on tissues or medical devices in the human body, adhesion is mainly governed by interaction of surface proteins from bacteria with human matrix proteins. In staphylococci, surface proteins belonging to the family of microbial surface components recognizing adhesive matrix molecules (MSCRAMMs) interact with fibronectin, fibrinogen, vitronectin, collagen, and other matrix molecules to allow adhesion of bacteria on host cell surfaces (Foster et al., 2014; Otto, 2018).

Once adhered to surfaces, bacteria form irreversible micro-colonies that are maintained through intercellular adhesion, leading to three-dimensional mature biofilm. This intercellular adhesion is allowed by the secretion of polymer molecules, which may be of carbohydrate, protein, or nucleic acid nature. In particular, the polyanionic charge of DNA is an important intercellular adhesion factor used to link molecules together. Of note, DNA release in biofilm ECM can be independent of bacterial lysis (Rajendran et al., 2015). Biofilm grows until it becomes a macroscopic “mushroom”-like structure. Such biofilms are not uniform and contain channels that are believed to be essential for providing nutrients to bacteria embedded deeper in biofilm layers. Indeed, bacteria are present in different physiological states (anaerobic, dormancy,…) depending on their location in the biofilm, and on the availability of substrates, oxygenation, pH, or exposure to metabolites present in the biofilm.

After biofilm maturation, isolated or aggregated bacteria can dissociate and end up in the bloodstream. This detachment stage allows colonization of other sites, leading to the appearance of “metastasis” of the infection. Bacterial release from the biofilm requires degradation of biofilm polymers by enzymes and surfactant molecules that reduce surface-bacterial interactions (Boles et al., 2005; Harmsen et al., 2010; Boles and Horswill, 2011; Kiedrowski et al., 2011; Otto, 2013). Several studies have demonstrated that this detachment process is subjected to quorum sensing, a regulatory mechanism that coordinates genes expression in the population (Boles and Horswill, 2008; Solano et al., 2014).

Although the processes leading to the synthesis, stability, and regulation of bacterial biofilms have been widely studied, the kinetics and dynamics of ECM remodeling mechanisms leading to the formation of viral biofilm have not been investigated yet, but can be drawn from scarce literature data (Figure 2). Interestingly, the viral oncoprotein Tax induces the expression of genes involved in the synthesis of some HTLV-1 biofilm components, suggesting that HTLV-1 might directly drive the formation of this biofilm. It is for example, the case of Fucosyltransferase Fuc-T VII, which synthetizes sialyl Lewis X, but also of collagen IV (COL4A1, COL4A2; Hiraiwa et al., 1997, 2003; Pais-Correia et al., 2010) and of ICAM-1 (Hiraiwa et al., 2003; Fazio et al., 2019). Using transcriptome analysis, Millen et al. showed recently that col4A1/2 genes are upregulated both in HTLV-1 infected cells and in Tax-induced cells, upon Tax-dependent transactivation of their promoter (Millen et al., 2019). Overexpressed COL4 proteins co-localized with the viral Gag protein in viral biofilm and were involved in viral transmission. In addition, Tax also induces expression of cell adhesion molecule-1 (CADM-1; Masuda et al., 2010), a protein involved in adhesion and cell infiltration, that could participate in the adhesiveness of the viral biofilm to target cells. Despite these studies, little is known about the role of Tax in formation of the viral biofilm. Thus, it would be interesting to study the formation of viral biofilm in HTLV-1 infected cells, in which viral expression is repressed but inducible such as MT-1 or ED-40515 cells (Mahgoub et al., 2018; Hong et al., 2020).

On the other hand, other components of the viral biofilm might not be induced directly by viral proteins, but rather indirectly, as a cellular response to infection. As an example, it is well established that Tetherin expression is induced in cells infected by various viruses (Evans et al., 2010), as a type I interferon (IFN-I)-induced innate immune response that restricts viral spread (Venkatesh and Bieniasz, 2013). Indeed, Tetherin expression dramatically inhibits the release of cell-free HTLV-1 particles by tethering them to the infected cell and to each other, without affecting viral protein expression (Jouvenet et al., 2009; Kuhl et al., 2011). Interestingly, while Tetherin restriction is counteracted by a variety of viral proteins (Evans et al., 2010), including Vpu from HIV-1 or Env from HIV-2 (Jolly et al., 2010), such a viral antagonism has not been described in the case of HTLV-1 infection, suggesting that Tetherin might not impact HTLV-1 transmission. Indeed, Tetherin knockdown has been shown to severely decrease HTLV-1 cell-free transmission but not cell-to-cell infection (Ilinskaya et al., 2013). Strikingly, cell-to-cell transmission analyzed in this study did not imply biofilm because viral transmission was measured using pseudovirus-expressing cells, thus devoid of HTLV-1 biofilm. Thus, the role of Tetherin in biofilm-based viral transmission still remains to be determined.

Knowledge is also missing on how viral biofilm is detached from the infected cell surface, and whether these released structures retain infectivity. However, this hypothesis is supported by the observation that removal of viral biofilm from the infected cells using heparin washes reduced HTLV-1 cell-to-cell transmission by 80% (Pais-Correia et al., 2010), and that purified HTLV-1 biofilm is the infectious entity responsible of viral transmission to target cells (Pais-Correia et al., 2010; Alais et al., 2015), probably more efficiently than transmission through the formation of a viral synapse (Igakura et al., 2003; Majorovits et al., 2008) or through cell-free viral particles, that are known to be poorly infectious in vitro and also rarely found in vivo (Li et al., 2008). Indeed, a ratio of around 10⁶ HTLV-1 virions is needed to infect one cell, while a ratio of 75 HIV virions is sufficient to infect one target cell (Derse et al., 2001). This lack of infectivity of cell-free virions has been linked to a reduced ability of Env protein to induce viral membrane fusion following isomerization of intersubunit disulfide-bound interactions observed in cell supernatant (Wallin et al., 2004; Li et al., 2008; Shinagawa et al., 2012). Thus, accumulation of virions inside the HTLV-1 biofilm might protect Env proteins from destabilization and could further maintain virus infectivity.

Alternatively, viral biofilm might favor cell-cell contacts through increased adhesiveness properties. This translates into close interactions among adhesion proteins (ICAM-1 and Lymphocyte function-associated antigen-1; LFA-1) and between viral and cellular proteins (Env and CD82/NRP-1). Those interactions were shown to trigger the polarization of the microtubule-organizing center (MTOC) in the infected cell, and as a result, the accumulation of viral proteins and viral RNA at the contact point, thus leading to the formation of a viral synapse and allowing viral particles’ budding at the synaptic cleft (Igakura et al., 2003; Majorovits et al., 2008). MTOC polarization in the infected cell upon cell-cell contact could also favor the delivery of extracellular vesicles (EV) to the target cell, a mechanism recently shown to increase the permissiveness of target cells (Pinto et al., 2019). In addition, polarization of the infected cells involves modification of actin polymerization, a process controlled by Tax through its induction of Gem, Fascin, and Collapsin response mediator protein-2 (CRMP-2; Masuda et al., 2010; Varrin-Doyer et al., 2012; Chevalier et al., 2014; Gross et al., 2016) and leading to increased cell-cell contacts and intercellular conduits (Chevalier et al., 2014; Gross et al., 2016). The latter are involved in the transfer of the p8 viral protein to target cells (Van Prooyen et al., 2010), which induces LFA-1 up-regulation. This in turn reinforces cell-cell adhesion through LFA-1 interaction with ICAM-1 (Van Prooyen et al., 2010; Malbec et al., 2011; Pinto et al., 2019).

It is unanimously accepted that Glucose Transporter 1 (GLUT-1), Neuropilin-1 (NRP-1), and Heparan Sulfate Proteoglycans (HSPG) are the three cell surface receptors involved in binding and entry of HTLV-1 cell-free virions in CD4+ T cells (Jones et al., 2011; Hoshino, 2012). Despite these advances, several questions remain unanswered, including their specificity and affinity for binding HTLV-1 present in biofilm and the precise role of each of these molecules during HTLV-1 entry in other types of cells. For example, HTLV-1 entry in dendritic cells (DC) seems to rely on endocytosis, as inhibition of micropinocytosis or clathrin-dependent endocytosis completely abolished HTLV-1 capture by immature and mature DC, as well as subsequent productive infection of immature DC (Rizkallah et al., 2017). Besides, the role of Glut-1 in HTLV-1 viral biofilm entry in plasmacytoid dendritic cell (pDC) was confirmed, but not that of NRP-1 (Assil et al., 2019). In addition, while the same study also confirmed the role of Glut-1 in productive infection of T-cells after contact with HTLV-1 viral biofilm, it failed to demonstrate the need of NRP-1, suggesting that the proposed interaction of HTLV-1 envelop with NRP-1 as prerequisite before interaction with Glut-1 (Lambert et al., 2009) might not be mandatory when virions are embedded in biofilm and not cell-free. However, whether viral biofilm is captured as a collective entity by the target cell, or whether individual virions are extracted from the biofilm before endocytosis is still unknown. Alternatively, one could hypothesize that because virions are associated with ECM in the viral biofilm, viral entry could be induced by interactions of ECM proteins such as integrins with their ligand, thus using regular routes of receptor recycling through endocytosis to enter in target cells. Then, interaction with Glut-1 could occur inside vesicles after the relief of ECM components. Indeed, this mode of entry is used by several bacterial species such as S. aureus and Streptococcus pyogenes (Rohde and Cleary, 2016; Josse et al., 2017), that expressed adhesions proteins able to bind fibronectin from ECM. Then, fibronectin interacts with the α₅β₁ integrin thus allowing the endocytosis of the complexed-bacteria in intracellular compartment.

Altogether, these results suggest that viral biofilm formation (i) might protect extracellular virions from Env protein instability, (ii) might increase cell-cell adhesiveness to favor cell-cell transmission, and (iii) might overcome the low infectivity of core particles, potentially through close interactions permitting transfer of viral proteins or cellular EV that might control target cell permissiveness.

HTLV-1 is almost exclusively transmitted during cell-cell contacts. The concentration of viruses at the viral synapse or in the viral biofilm allows the transfer of virions in a directional and extremely efficient way. This strategy limits the interactions of the virus with the environment, which could allow it to escape therapeutics strategies and immune responses. In comparison with bacterial biofilms, our understanding of the impact of the viral biofilm on the viral cycle is still very limited. In particular, whether the presence of virions in a biofilm modifies viral entry into the target cells is not completely understood. Is there viral diffusion inside the biofilm to reach the target plasma membrane and allow subsequent membrane fusion, or are viral particles and ECM endocytosed, then leading to fusion in endosomes? What are the molecular mechanisms that regulate these pathways? Moreover, is viral biofilm present on different HTLV-1 infected cell lines similar in terms of composition, structure, and topography, given that these infected cells have different infectivity and ability to escape to innate sensing (Alais et al., 2015; Assil et al., 2019)? Viral biofilm has only been reported in in vitro models using HTLV-1-infected cell lines (Pais-Correia et al., 2010; Nakamura et al., 2019). Further studies are needed to demonstrate the presence and the role of biofilm structures in in vivo models. One of the major difficulties is that the sites of viral transmission in vivo are not identified, probably because expression of HTLV-1 transcripts occurs in bursts or intermittently in infected cells, which considerably reduces the windows of observation. Furthermore, it is still very challenging to discriminate viral transmission using viral synapse or biofilm transfer given that these two mechanisms share several identical molecular players. Major questions also remain open about the role of viral biofilm during inter-individual viral transmission (Okochi and Sato, 1984; Kaplan et al., 1996; Hino, 2011) or its presence in biological fluids, such as maternal milk, blood, and saliva. Finally, analyzing whether other viruses transmitted by cell-cell contacts are also able to form a viral biofilm is of uttermost importance to build a generalized concept of viral biofilm.

All authors contributed to manuscript revisions, read and approved the submitted version. YM wrote the first draft of the manuscript.