**4. An example of complex biofilm: biofilm formation at the surface of nanofiltration membranes used for drinking water production.**

We and others have previously studied very complex biofilms formed on nanofiltration (NF) membranes during surface water filtration in drinking water production processes [63, 64]. After several years of filtration, the foulant consists in a brown viscous layer covering the entire surface of the membrane [65] (Figure 11).

**Figure 11.** Visual examination of a fouled NF membrane

Dry weight of the foulant is about 2 g/m2. The NF biofilms harbours mainly exopolysaccharides and proteins, as shown by characteristic ATR-FTIR signals near 1650 cm-1 (amide I), 1550 cm-1 (amide II), 1450 cm-1 (due in part to C-H deformation), 1400 cm-1 (due in part to symetric stretch for the carboxylate ion), 1250 cm-1 (P=O and C-O-C stretching and/or amide III), and in a broad complex region from 1250 to 900 cm-1 (due in part to C-O-

C, C-O, ring-stretching vibrations of polysaccharides and the P=O stretch of phosphodiesters) (Figure 12).

**Figure 12.** ATR-FTIR spectra of a virgin membrane (plain line) and of a fouled membrane (dotted line)

Fluorescence microscopy observations after nucleic acid staining with DAPI and polysaccharides staining with lectins labelled with fluorescein isothiocyanate or tetramethylrhodamine isothiocyanate indicate a high spatial heterogeneity inside the foulant matter with a mean thickness of 32.5 ± 17.7 µm [66] (Figure 13). Examples of lectins that can be used for such polysaccharides staining experiments are peanut agglutinin (PNA) targeting β-gal(1->3)galNAc residues, wheat germ agglutinin (WGA) targeting (glcNAc)2 and NeuNAc residues, *Bandeiraea simplicifolia* (BS-1) agglutinin targeting α-gal and α galNAc residues and Concanavalin A (ConA) targeting α-man and α-glc residues.

**Figure 13.** CLSM visualization of the heterogeneity of a NF biofilm after staining with DAPI, TRITC and TITC-labelled lectins. Magnification x630

The microbial cells, mainly composed of bacteria, are localized in the superficial layer of the fouling material and are organized as microcolonies interspersed at the membrane surface. Some algae are also present, as shown by autofluorescence properties. The presence of a dense and wide polysaccharide matrix harbouring few microbial cells at the NF membrane surface has been associated with differences in the efficiency of cleaning procedures against different foulants categories [65, 67]. Polysaccharide residues are found in areas where microcolonies are present and in areas devoid of microbial cells. This polysaccharide organization has been previously observed with environmental biofilms grown in vitro with river water as the sole source of carbon and nutrients [68]. High staining with PNA and BS-1, respectively reveals high occurrence of galactosides residues in the polysaccharide components of the foulants. The BS-1 lectin staining pattern indicates a high degree of spatial organisation with the observation of long and entangled fibers. WGA staining shows short fibers and cloud stained areas. PNA and ConA lectin staining are more interspersed. The polysaccharide composition of the fouling layer changes quantitatively and qualitatively during spring and summer [64]. Lectin staining increases from March to September for all the lectins used. Staining with BS-1 increases constantly in March, June and September. A high increase of binding with PNA, and ConA is observed between March and June, but the binding of these two lectins does not change between June and September. Staining with the WGA is weak in March and June and is higher in September. The lectin-binding changes with time may be linked to an increase of the biomass attached at the membrane surface and to changes among the populations of attached cells. Nutrients, oxygen level and the concentration of metals can influence the exopolymer abundance of environmental model biofilms grown in vitro with river water as the sole source of carbon and nutrients [69]. The modification of these parameters leads to a shift in the glycoconjugate makeup of the biofilms.

384 The Complex World of Polysaccharides

phosphodiesters) (Figure 12).

C, C-O, ring-stretching vibrations of polysaccharides and the P=O stretch of

**Figure 12.** ATR-FTIR spectra of a virgin membrane (plain line) and of a fouled membrane (dotted line)

Fluorescence microscopy observations after nucleic acid staining with DAPI and polysaccharides staining with lectins labelled with fluorescein isothiocyanate or tetramethylrhodamine isothiocyanate indicate a high spatial heterogeneity inside the foulant matter with a mean thickness of 32.5 ± 17.7 µm [66] (Figure 13). Examples of lectins that can be used for such polysaccharides staining experiments are peanut agglutinin (PNA) targeting β-gal(1->3)galNAc residues, wheat germ agglutinin (WGA) targeting (glcNAc)2 and NeuNAc residues, *Bandeiraea simplicifolia* (BS-1) agglutinin targeting α-gal and α -

galNAc residues and Concanavalin A (ConA) targeting α-man and α-glc residues.

**Figure 13.** CLSM visualization of the heterogeneity of a NF biofilm after staining with DAPI, TRITC

and TITC-labelled lectins. Magnification x630

Biofilms may be considered to be highly porous polymer gels [70] and diffusion studies demonstrate gel-like characteristics [71]. Previous work has suggested that laboratorygrown and some natural biofilms are viscoelastic in nature [3, 8, 72]. During rotation analysis, a rheofluidification behaviour is observed for NF biofilms [66]. Different mechanisms can explain shear thinning of a biofilm. Break down of links between polymers in the biofilm matrix or deflocculation of particles corresponding to an irreversible modification of the biofilm structure can occur. Such irreversible modifications are unlikely in the experimental conditions published because of the reversibility of viscosity changes with shear rate [66]. Shear thinning of NF biofilms may be related to the polymeric composition of the biofilm matrix. With shear acceleration, polymers may follow the direction of the flow leading to viscosity decrease. This has been previously observed with purified polysaccharides like cellulose [73]. Moreover, bending of biofilm structures in the shear direction during the application of shear stress has been mentioned to explain the viscoelastic response of a mixed culture biofilm [72]. NF biofilms have been submitted to oscillation analysis with a cone-plate rheometer [66]. In such experiments, a sinusoidal oscillation of defined maximum strain and oscillatory frequency is applied to a sample and the storage (*G'*) and lost (*G"*) modulus are measured. The storage modulus characterizes the ability of the material to store energy, whereas the loss modulus characterizes energy dissipation in the material under dynamic excitations. If the material is perfectly elastic then the resultant stress wave is exactly in phase with the strain wave. By contrast, when the rate of change of the sinusoidal oscillation is a maximum and the strain is zero, for purely viscous systems, the resultant stress wave will be exactly 90° out-of phase with the imposed deformation. For NF biofilms, during oscillation analysis, values of storage modulus (*G'*) stay higher than values of loss modulus (*G"*) over the entire range of frequencies covered, indicating that the NF biofilm behave like a highly elastic physical gel [74]. Polysaccharides alone, like alginates, are known to realize a sol-gel transition under adequate physicochemical conditions [75]. The physicochemical microenvironment inside NF biofilms may be permissive to exopolysaccharides sol-gel transition. The gel state is resistant enough and presents a micro porosity favourable for resistance to flow forces, microcolonies development and cell nutrition inside the biofilm structure. This model of sol-gel transition of polysaccharides inside biofilms is consistent with rheological properties previously demonstrated for other biofilms: *Streptococcus mutans* biofilms have elasticity and viscous behaviour analogous to NF biofilms for a range of frequencies between 0.1 to 20 Hz [76]. The rheofluidification behaviour and gel-type rheological properties shared by different type of biofilms and purified polysaccharides suggest that the critical components of the biofilm matrix determining the biofilm texture are polysaccharides.

The time-dependent strain response observed in the creep curves clearly indicated that NF biofilms exhibited viscoelastic behaviour. Viscoelasticity is thought to be a general mechanical property of biofilms. A very wide range of elasticity and viscosity values has been previously observed for a wide sample of biofilms formed artificially in laboratory experiments or coming from natural aquatic environments [4, 72, 76]. Thus, it wasn't surprising to observe that the rheological properties of NF biofilms are different from the ones of natural biofilms from different aquatic environments like Nymph Creek (Yellowstone National Park) and Chico Hot springs (Montana) algal biofilms [4]. These differences in viscosity and elasticity between biofilms can be related to different exopolysaccharide contents and to different shearing strains. Bacterial and algal alginates are known to have different monomeric composition leading to a stronger binding of cations for bacteria, a property involved in the formation of a stable gel in the presence of ambient Ca2+ cations [77].

The specificity of NF biofilms may be the necessity to resist shear forces applied to the membrane during the filtration process. In the Méry-sur-Oise plant, NF membranes are operated at feed pressure of approximately 10 bars [78]. The high membrane feed pressures may influence the rheological properties of NF biofilms by increasing cohesive forces in the biofilm bulk, increasing forces, which keep the exopolymers to the membrane surface, and thus strengthening the mechanical stability of the biofilm. This may explain at least in part the NF biofilms resistance to industrial cleaning [65].

Shaw et al. have previously shown that the elastic relaxation time varied much less between biofilms of different origins. was estimated to be the time required for viscous creep length to equal elastic deformation length (so that memory of initial conditions is lost), i.e., ≈ *η* /*G*.

The elastic relaxation time of about 30 minutes lies within the range previously determined for various biofilms [4]. The universality of the viscoelastic transition of biofilms has been suspected to have critical survival impact [4]. The ability of biofilms to deform in response to mechanical stress may be a conserved strategy of defence to enable persistence on surfaces in different flow conditions.
