Bacterial attachment on optical fibre surfaces

Optical fibres have received considerable attention as high-density sensor arrays suitable for both in vitro and in vivo measurements of biomolecules and biological processes in living organisms and/or nano-environments. The fibre surface was chemically modified by exposure to a selective etchant that preferentially erodes the fibre cores relative to the surrounding cladding material, thus producing a regular pattern of cylindrical wells of approximately 2.5 μm in diameter and 2.5 μm deep. The surface hydrophobicity of the etched and non-etched optical fibres was analysed using the sessile pico-drop method. The surface topography was characterised by atomic force microscopy (AFM), while the surface chemistry was probed by time-of-flight secondary ion mass spectrometry (ToF-SIMS). Six taxonomically different bacterial strains showed a consistent preference for attachment to the nano-scale smoother (R q = 273 nm), non-etched fibre surfaces (water contact angle, θ = 106° ± 4°). In comparison, the surfaces of the etched optical fibres (water contact angle, θ = 96° ± 10°) were not found to be amenable to bacterial attachment. Bacterial attachment on the non-etched optical fibre substrata varied among different strains.


Introduction
In recent years optical fibre bundles have been widely investigated as a base for developing sensors suitable for both in vitro and in vivo measurements of biological processes in living cells or nano-environments (Bernhard et al. 2001;Epstein et al. 2003;Stoddart and Brack 2007). Optical fibre arrays in a 96 element matrix have been commercialised by Illumina Inc. and are used to study gene expression and perform wholegenome SNP genotyping (Fan et al. 2004;Kuhn et al. 2004). Fibre bundles can also serve as a platform for constructing a highly sensitive bio-sensing system that can operate at the single cell level (Vo-Dinh and Kasili 2005;Fritzsche et al. 2009) or as live cell array biosensors for the detection of various chemical compounds (Biran et al. 2003). One of the most recent applications of such fibres is the design of whole-cell biosensors for environmental monitoring (Biran et al. 2003;Kuang et al. 2004). The latter application relies on random assembly of the cells by sedimentation, after which bacterial cells can remain viable, sustained by the reservoir of nutrients above the surface (Kuang et al. 2004). The array platform allows the physiological and genetic variabilities between cells and their response to various stimuli to be continuously monitored. The main concerns in designing a whole-cell biosensor is the immobilisation of live cells onto the fibre surface and maintaining their viability (Udd 1995;Gadelmawla et al. 2002;Lee 2003;Lee et al. 2005Lee et al. , 2009Dhawan et al. 2009). This is mainly because most of the steps involved in cell immobilization frequently result in cell death or damage and subsequent impaired sensitivity. Moreover, in order to gain the full benefits from the optical fibre array sensor platform, it is important to ensure that collections of separated cells respond in the same way as a collection of unconstrained cells. Thus, a detailed understanding of the cell-fibre surface interactions is believed to be the key issue in the design of functional, long lasting and effective biosensing devices based on optical fibre substrata (Polwart et al. 2000;D'Souza 2001;Vo-Dinh and Stokes 2002;Biran et al. 2003;Vo-Dinh and Kasili 2005;White and Stoddart 2005;White et al. 2007).
Despite being the subject of intensive research for various applications in clinical diagnosis, biomedical and environmental monitoring, or industrial and food technology, very little is known about the interaction between the optical fibre surfaces and the biological materials (including bacteria) that they come into contact with. The present study aimed to investigate the effect that the surface characteristics of optical imaging fibre have on the retention pattern of medically and environmentally significant microorganisms of different taxonomic affiliations. Two types of optical fibre substrata were used in all experiments: standard, from now on referred to as 'plain' and chemically modified, referred to as 'etched'. Bacterial and fibre surface characteristics were analysed in order to better understand the characteristics that influence the extent of bacterial adhesion to the fibre substrata and the associated bacterial metabolic activity. Particular attention was given to the role the surface microand nano-scale morphology play in the bacterial adhesion process. Existing work has indicated that the nano-scale changes in surface roughness of various materials may have a significant effect on bacterial adhesion Mitik-Dineva et al. 2008b, 2009a, 2009b.
The optical imaging fibres used in this study were made from germanium doped silica glass cores surrounded by fluorine-doped silica cladding and consisted of *70,000 individual single-mode fibres (known as pixels) fused together into a bundle that was coated with silicone resin. The total outer diameter of the final fibre bundle was about 1.3 mm ( Figure 1d). Workable (5 mm long) fibre substrata were obtained by cleaving the fibre bundle received from the manufacturer, which yielded relatively smooth (R max up to 200 nm) flat surfaces on the fibre-end faces, as shown in the high-resolution SEM images presented in Figure 1a,b). In order to fabricate micro-structured surfaces, samples of the standard optical imaging fibre were further modified by chemical etching. This was achieved by immersing one half of the 5 mm long samples into buffered hydrofluoric acid (BHF) etching solution for 20 min (White and Stoddart 2005). Because the glass cores etch at a faster rate than the surrounding cladding, treating the fibre samples with BHF resulted in creating a general hexagonal pattern of cylindrical wells, approximately 2.5 mm in diameter and 2.5 mm deep, on the fibre surfaces ( Figure 1).
Upon treatment all samples were rinsed with nanopure H 2 O (18.2 MOcm 71 Barnstead/Thermolyne NANOpure Infinity water purification system), sterilised and stored under sterile conditions prior to inoculation. The rinse solution was tested for acidity with phenolphthalein and bromothymol blue indicator dyes to confirm that all of the BHF had been removed from the fibre surfaces.
The surface hydrophobicity of the substrata was determined via contact angle measurements using the sessile pico-drop method as described elsewhere (Taylor et al. 2007;Urquhart et al. 2008). Measurements were performed using nanopure water on a Kru¨ss DSA 100 apparatus fitted with a piezo-doser head. The piezo-doser allowed small nanopure water droplets (100 pl) to be deposited onto the fibre facets (Taylor et al. 2007). Sample positioning and data acquisition took place automatically, with droplet side profiles being recorded (a dual camera system was used, one to record a side profile of a spot and the other to record a plan view to ensure that the water droplet was deposited at the center of each spot) for data analysis. Water contact angle (WCA) calculations were performed using a circle segment function as required for small water droplets. Measurements were taken over six areas of two samples of each fibre; in total 12 droplets on both, plain and etched fibre samples, were analysed. Results were averaged and standard deviation (SD) values were calculated (Table 1, Figure 2). The contact angles of formamide (Sigma) and diiodomethane (Sigma) were also measured using an FTA1000c (First Ten Å ngstroms Inc) equipped with a nano-dispenser. An average of at least five measurements was taken for each solvent and titanium surface. Each measurement of a particular contact angle was recorded in 50 images in 2 s with a Prosilica Model Navitar 444037 camera, and the contact angle was determined as a result of images analysed using the FTA Windows Mode 32 software. The average contact angle for each of the three solvents on each surface was used to calculate the surface free energy and its components, based on the Lewis acid/base method (Ö ner and McCarthy 2000).
The topographical features of both the etched and plain fibre substrata were analysed using scanning electron microscopy (SEM) and AFM as described elsewhere , 2009a. Highresolution images of the samples containing adsorbed bacterial cells were taken using a FESEM (ZEISS SUPRA 40VP) at 3 kV with 10006, 50006 and 20,0006 magnifications. Images with 10006 and 50006 magnification were used to estimate the number of bacteria adhering to the fibre surfaces.
A scanning probe microscope (SPM) (Solver P7LS, NT-MDT) was used to image the fibre surface topography whilst also providing a quantitative analysis of the surface roughness (Table 1, Figure 3). The analysis was performed in the semi-contact mode, which reduces the interaction between the tip and the sample and thus allowed the destructive action of lateral forces that exist in contact mode to be avoided. The height of the surface features was measured with a resolution of a fraction of a nanometer and the surface roughness of the areas investigated could be statistically analysed using the standard instrument software (LS7-SPM v.8.58). Five samples of both etched and plain fibre surfaces were investigated. Each sample was briefly scanned to evaluate the overall homogeneity of the surface and then one typical topographical profile was studied in detail. Statistical data processing was performed using SPSS 15.0 (SPSS Inc, Chicago, Illinois, USA). Single independent group T-tests were performed to evaluate the consistency of surface roughness parameters.

Reconstruction of interactive three-dimensional images
Interactive three-dimensional (3D) visualization of the fibre surfaces was undertaken using a custom C-code and the S2PLOT graphics library (Barnes et al. 2006;Ivanova et al. 2009). The input data files were in NT-MDT format and were read into the custom viewing tool (mdt-view) using a modification of the nt-mdt module of Gwyddion by Necas and Klapetek (http:// gwyddion.net/, Version 2.12). NT-MDT files were then converted into a three-dimensional surface, coloured according to height, and displayed with the S2PLOT s2surpa function. Visualizations were exported from mdt-view to an intermediate VRML format, with textures for axis labels in TGA format. Textures were converted to PNG format and the VRML model was then imported into Adobe Acrobat 3D Version 8 to create an interactive figure, using the approach described by Barnes and Fluke (2008). Simple Java Script commands were used to provide additional functionality. The two interactive panels in Figure 3 can be viewed in the pdf format of this paper by clicking on either panel (C) or (F) with the mouse pointer, provided Adobe Reader Version 8.0 or higher is used to view this article. This opens a window where the surface can be examined interactively using the mouse to control the camera orientation and zoom level.

Time-of-Flight Secondary Ion Mass Spectrometry analysis
A Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS) was used to investigate the surface chemistry of the two fibre surfaces and provide an insight into compositional differences that might have influenced the extent of bacterial adsorption. All measurements were performed using a ToF-SIMS IV instrument (ION-ToF GmbH, Munster, Germany) with a reflection analyser and a pulsed electron flood source for charge neutralization. Both positive and negative spectra were acquired from a 100 mm 6 100 mm area. Samples were exposed to the atmosphere for less than 5 min during mounting in the TOF-SIMS instrument. All experiments were performed using a cycle time of 100 ms. A monoisotopic 69 Ga þ primary ion source was operated at 25 keV in the ''burst alignment'' mode, which afforded very high  spatial resolution at the expense of mass resolution and positive and negative spectra were acquired with a mass resolution typically larger than 6000 at m/z ¼ 27, sufficient to identify most of the fragments (supplemental material). The resulting spectra acquired were analysed using the standard instrument software (Supplemental online material [Supplementary material is available via a multimedia link on the online article webpage]).

Microorganisms, culture conditions and sample preparation
Escherichia coli K12, Staphylococcus aureus CIP 68.5, Pseudomonas aeruginosa ATCC 9027; and three marine bacteria: Cobetia marina DSM 4741 T , Pseudoalteromonas issachenkonii KMM 3549 T , and Sulfitobacter guttiformis DSM 11458 T were used in this study. The bacterial strains were obtained from American Type Culture Collection (ATCC, USA), Culture Collection of the Institute Pasteur (CIP, France) and German Culture Collections (DSM, Germany). Selected marine bacteria derived from our previous studies (Mitik-Dineva et al. 2009a). All bacteria were routinely cultured in a nutrient (Merck) or marine (Oxoid) agar and stored at 7808C as described elsewhere (Ivanova et al. 2002). A fresh bacterial suspension was prepared for each of the strains grown overnight in 100 ml of nutrient or marine broth (in 0.5 l Erlenmeyer flasks) at 378C with shaking (120 rpm) prior to each experiment (Mitik-Dineva et al. 2009a). Bacterial cells were collected during the logarithmic phase of growth as confirmed by growth curves (data not shown). However, because cell densities varied after incubation for 12 h, the cell density of each strain was adjusted to OD 600 ¼ 0.3 in order to achieve an approximately similar number of cells in each sample (OD 600 ¼ 1 corresponds to 8 6 10 8 cells ml 71 for E. coli, however, this estimate may vary for different bacteria). Therefore, the bacterial cell suspensions were further subjected to direct counting using a haemocytometer to confirm the number of bacterial cells for each strain used in experiments (Mather and Roberts 1998). On the day of the experiment, a 2 ml aliquot of log-phase bacterial suspension was adjusted to OD 600 ¼ 0.3 in nutrient/ marine broth and kept in centrifuge tubes. Duplicate samples of both the as-received and etched fibres were placed into each of the tubes and were incubated for 12 h at room temperature (ca 228C). After incubation, all samples were rinsed three times with sterilized nanopure H 2 O, attached to glass supports and stored under sterile condition until needed. In SEM experiments, samples with adsorbed bacteria were initially coated with 20 nm gold thin films using a Dynavac CS300 according to the procedure developed previously (Mitik-Dineva et al. 2009b;Truong et al. 2009). The lower detection limit was estimated as 1.1 6 10 3 cells mm 72 according to Morono et al. (2009) using the following formula: where n is the number of cells required giving a probability p (p ¼ 0.95, 95% chance to find one bacterial cell) of detecting a cell, T fov is total area of fields of view, C fov is the number of fields of view, and the total fibre area is 1.3267 mm 2 . Bacterial cell surface hydrophobicity was determined from a series of static contact angle measurements using the sessile drop method as described elsewhere ). The bacterial surface charge (Table 2) was determined via measurement of the electrophoretic mobility (EPM) of the cells. The EPM was converted to zeta potential using the Smoluchowski's approximation (Korenevsky and Beveridge 2007;Soni et al. 2007;Mitik-Dineva et al. 2008a). The zeta potentials of all six strains were measured as previously described (de Kerchove and Elimelech 2005;van Merode et al. 2007).

Visualization and quantification of the bacterial adsorption and metabolic activity
Bacterial cell density quantification was performed using SEM images; bacterial morphological changes are not described in the results. Cell numbers from at least ten representative images/areas were transformed into a ''number of bacteria per unit area'' to allow the quantity of bacteria attaching to the substratum surface to be determined. The average densities have estimated errors of approximately 10-15% due to the local variability in the surface coverage.
Viable bacterial cells were visualised using SYTO 1 17 Red (Molecular Probes TM , Invitrogen) (data not shown) and bacterial production of extracellular polymeric substances (EPS) on the fibre surfaces using Concanavalin A Alexa Fluor 488 Conjugate (Molecular Probes Inc.). This dye selectively binds to a-mannopyranosyl and a-glucopyranosyl residues in EPS (Goldstein et al. 1964). The EPS was visualised and analysed using confocal laser scanning microscopy (CLSM, Olympus Fluoview FV1000 Spectroscopic Confocal System). The fluorescence intensities were scanned at 543 for SYTO 1 17 Red and 488 for Alexa Fluor 488/Ex/nm.

Fibre surface characterisation
As indicated by the high resolution SEM (Figure 1) and AFM (Figure 3) images, the surface topography of the optical fibres had significantly changed as a result of the chemical etching. The differences in the surface topography are summarised in Table 1 and confirmed by a statistical analysis of the surface roughness parameters. The results indicated that the etched, micro-structured, honeycomb patterned fibre surface was significantly rougher than that of the non-etched fibre surface. All roughness parameters, including the average surface roughness (R a ) which represents the average/absolute deviation of the surface irregularity from the mean line over one sampling length, the root mean square roughness (R q ) defined as the SD of the distribution of surface height, and the peak-to-peak roughness (R max ) which represents the vertical distance between the highest peak and the lowest valley along the assessment length of the profile, appeared to be considerably higher on the etched fibre substrata than those of the non-etched samples as confirmed by statistical analysis (p 5 0.05). The most significant increase, more than tenfold, was observed for the (R q ) which is believed to be more sensitive than the (R a ) parameter when considerable deviations from the mean line occur (as is the case in this instance) (Gadelmawla et al. 2002). Analysis of the surface chemistry of the plain and etched fibres using ToF-SIMS revealed appreciable differences between the surface chemical characteristics of both substrata. In the case of the plain fibre, ToF SIMS analysis revealed the presence of silicone contamination deriving from the coating material with characteristic fingerprint of poly(dimethyl siloxane) Figure S1). Positive and negative ToF-SIMS spectra, images and detailed analysis of both fibre substrata are presented in supplementary online materials.
The etching process resulted in a slight decrease in the fibre surface hydrophobicity, with the WCA decreasing from y ¼ 106 + 48 on the as-received, to y ¼ 96.0 + 10.18on the etched fibre surface. Markedly, surface tension values remained very similar for both types of surfaces; 9.18 mJ m 72 and 9.66 mJ m 72 for asreceived and etched fibres, respectively. The large deviation in the WCA measured on the etched surface is likely to be due to the variable position of the small drops relative to the topography for repeated measurements. Removal of silicone contamination after etching might also play a role in slight decrease of WCA on the etched fibre surface. The composite contact angle measured on the etched fibre surface (which is comprised of solid together with air filled pores) can be used to calculate the contact angle of the solid part of the etched surface using the Cassie-Baxter theory for non-wetting surfaces. However, it is recognised that there has been some debate as to the applicability of this equation for solid surfaces where the size of the surface roughness is small compared to that of the liquid-vapour interface, which is comparable to the size of the droplet used to measure the contract angle (Ö ner and McCarthy 2000;Gao and McCarthy 2007;Nosonovsky 2007). According to this theory, it can be assumed that the liquid droplet in contact with the etched fibre surface is in contact with both the solid surface and the air trapped in the pores formed by the etching process (Abdelsalam et al. 2005). The apparent contact angle (y*) of the composite surface allows the contact angle of the solid itself (y) (ie in the absence of air filled voids) to be calculated according to the following relationship: where f 1 and f 2 are the area fractions of the solid and air filled voids, respectively, with f 1 þ f 2 ¼ 1. The area fraction of the air filled voids was calculated to be 0.20 (f 2 ), and since y* was measured as 96 + 108, the contact angle on the solid itself (y) was calculated to be 83 + 128.

Surface characteristics of the bacterial cells
The surface hydrophobicity of the bacterial cells and surface charge (Table 2) varied amongst the species, probably reflecting the different chemical composition of the surface-associated EPS. WCA (y) values of 608 for bacterial cell surfaces were considered to be the borderline between hydrophilic and hydrophobic behaviour (Vogler 1998), which allows the observation that the surface of four of the studied strains, E. coli, S. guttiformis, P. aeruginosa and P. issachenkonii, exhibited slightly hydrophilic characteristics, whereas the cell surfaces of C. marina and S. aureus were found to exhibit hydrophobic characteristics.
The surface charge of each of the bacterial cells is presented in Table 2. It can be seen from the data presented that the least negatively charged bacterium was P. aeruginosa (z ¼ 714.4 + 0.7 mV), whereas the most negatively charged bacterium was S. guttiformis (z ¼ 743.2 + 0.2 mV). If these results are considered in the context of the suggested inverse correlation between cell surface charge and bacterial adhesion (Li and Logan 2004) and the electrostatic repulsion between negatively charged bacteria and negatively charged surfaces under commonly encountered pH conditions (Jucker et al. 1996), S. guttiformis would be expected to exhibit the weakest and P. aeruginosa the strongest attachment preferences to the negatively charged fibre substrata used in this study.

Bacterial attachment patterns on the non-etched and etched fibre surfaces
The number of retained bacteria on both fibre surfaces after 12 h incubation was determined and statistically analysed (Table 2, Figure 4, Supplementary online material Figure S2). It is clear that all six strains maintained their presence on the smoother, plain fibre substrata, but not on the etched surfaces. C. marina and P. issachenkonii were the most prominent colonizers with 55,000 and 53,000 attached cells mm 72 , respectively. The attachment patterns of the six bacterial strains can be seen on the high resolution SEM images (Figure 4). Granular deposits of variable size were also detected for E. coli, P. aeruginosa, P. issachenkonii and S. guttiformis on both types of surfaces. These deposits are presumed to be EPS secreted by adhering cells and it is likely that they serve as primers that facilitate bacterial adhesion. The secretion of EPS by these strains during colonization of other surfaces has previously been reported (Goldstein et al. 1964;Ivanova et al. 2008aIvanova et al. , 2008bIvanova et al. , 2009Mitik-Dineva et al. 2008b, 2009b, 2009a. Interestingly, C. marina and S. aureus cells, while also being successful colonizers of the non-etched fibre surface, did not produce EPS to the same extent as E. coli, P. aeruginosa, P. issachenkonii and S. guttiformis. A remarkably different bacterial response was observed on the etched fibre substrata. Although none of the tested strains were able to remain attached to the etched fibre substrata, varying quantities of EPS were still detected around and inside the wells. EPS aggregations produced by E. coli, P. aeruginosa and P. issachenkonii were mostly located on and around the fibre cores, whilst granular EPS of different sizes produced by S. guttiformis (Figure 4f) was randomly deposited over the fibre cores and the surrounding cladding. Neither the SEM nor the CLSM analysis showed any C. marina or S. aureus cells remaining on either of the fibre substrata. In addition, the EPS produced by both strains was considerably less than that detected for the other strains.

Discussion
Even though the relationship between bacterial attachment patterns and cell surface characteristics (such as hydrophobicity and surface charge) has been studied intensively over the past few decades, the somewhat contradictory results have not allowed the formulation of a reliable correlation (van Loosdrecht et al. 1990;Li and Logan 2004). This is most likely due to the fact that bacterial survival strategies include a dynamic attachment process that is dependent on the presence, chemical composition and structure of extra-cellular surface components (Auerbach et al. 2000;Danese et al. 2000;Jain et al. 2007).
In the present study, the two strains with highly hydrophobic cell surfaces, C. marina and S. aureus, displayed rather different attachment tendencies. C. marina cells were successful colonizers of the asreceived fibre substrata, while S. aureus cells did not attach onto these to any significant extent (Figure 4, Supplemental online material Figure 2S). Considering the electrostatic properties of bacterial cell surfaces, S. guttiformis cells, having the highest negative charge, would be expected to have the weakest propensity for attachment to both fibre substrata and P. aeruginosa the strongest propensity for attachment. However S. guttiformis cells were seen to attach in greater numbers than P. aeruginosa (Figure 4, Table 2). The most successful colonizers overall were C. marina and P. issachenkonii (e) and S. guttiformis (f) on the non-etched (left and middle columns) and etched (right columns) fibre surfaces. The EPS produced by the cells was visualized by labelling with the fluorescent dye Concanavalin A. Images a, b, e and f confirm the presence of EPS produced by E. coli, P. aeruginosa, P. issachenkonii and S. guttiformis after 12 h incubation on the non-etched fibre surfaces. Neither the SEM nor the CLSM analysis showed any S. aureus (c) or C. marina (d) cells remaining on either of the fibre substrata. In addition, the EPS produced by both strains was considerably less than that detected for the other strains.
isachenkonii, both of which exhibited a significantly higher surface charge than P. aeruginosa. The observation reported here that the bacteria responded differently on the same surfaces, is supported by data from earlier studies, where it was reported that bacterial cells can exhibit variable surface characteristics due to the presence of EPS on the outer surface, their constant dynamic motion and the presence or absence of molecular areas of variable polarity or charge (Li and Logan 2004;Vadillo-Rodriguez et al. 2004). The EPS material usually encapsulates the cell and/or is released into the liquid phase as planktonic EPS. These substances are believed to serve as promoting factors for bacterial attachment and biofilm formation (Beech et al. 1991;Favre-Bonte et al. 1999;Evans 2000). Nevertheless, an increasing amount of information suggests that the function of the EPS depends on its composition (Sutherland 2001). For example, capsular polysaccharides of Vibrio vulnificus are mainly composed of uronic acids, and contribute to an increase in negative charge and hydrophilicity (Wright et al. 1990) and hence inhibit the attachment of the bacterium. Other chemically different polysaccharides, eg those of P. aeruginosa composed of neutral sugars, may promote adhesion to the same surfaces (Yildiz and Schoolnik 1999;Wozniak et al. 2003).
In previous studies the authors have tested the attachment behaviour of a number of medically and environmentally challenging bacteria to various substrata, such as glass, polymer and titanium Mitik-Dineva et al. 2008b, 2009a, 2009b. Results obtained so far indicate that neither the cell surface hydrophobicity nor the charge correlates with the level of bacterial attachment. Hence neither of the cell surface characteristics studied here could provide a reasonable explanation for the different bacterial responses towards the two fibre substrata.
Substratum physico-chemical characteristics such as hydrophobicity, charge, chemistry and topography have also been intensively studied in an attempt to predict trends in bacterial attachment behaviour (Busscher and van der Mei 1997;Bos et al. 1999Bos et al. , 2000Teixeira and Oliveira 1999;Pereira et al. 2000). The results of the present study also indicate that the only property of the imaging fibre substrata that is significantly affected by the chemical modification is the surface roughness. All of the other surface parameters remained nearly constant. Therefore the observed difference in the bacterial response to the two types of fibre substrata appears to be principally associated with the change in the surface structure. Importantly, irrespective of their taxonomic affiliations and species-specific characteristics, all of the studied strains attached to the smoother, plain fibre substrata, while no bacterial cells were retained on the etched fibre surfaces.
The current understanding of the effects of surface topography on bacterial adhesion, including the ''attachment point'' theory, suggests that bacteria prefer microscopic surface irregularities as the starting point for their attachment, as these provide shelter from unfavorable environmental influences (Howell and Behrends 2006;Riedewald 2006;Scardino et al. 2006;Whitehead and Verran 2006). However, recent work in this field, suggests that this might not always be the case Mitik-Dineva et al. 2008b, 2009b. In particular, nano-scale surface roughness may have a significant effect on bacterial adhesion, with differences in the surface roughness of just a few nanometers appearing to exert a strong influence on the cellular response to certain surfaces (Howell and Behrends 2006;Riedewald 2006;Whitehead and Verran 2006;Ivanova et al. 2008bIvanova et al. , 2009Mitik-Dineva et al. 2008b. The results presented here show that a variable number of bacterial cells were able to colonize the nano-scale rough, non-etched fibre substrata, whereas less smooth, micro-scale rough, etched fibre substrata might sustain cellular attachment below the detection limit. It is notable that the same adhesion tendency was observed for all of the tested strains on the etched fibres, regardless of their taxonomic affiliation and their cell surface characteristics. In summary, an improved understanding of cellsurface interactions may facilitate the design and manufacture of optical fibre sensing surfaces with cyto-attractive or cyto-repellent characteristics, depending on the particular application requirements (Bos et al. 2000;Polwart et al. 2000;Bernhard et al. 2001;D'Souza 2001;Biran et al. 2003;Kuang et al. 2004;Howell and Behrends 2006;Riedewald et al. 2006). The results of this study suggest that plain FIGH-70-1300N optical fibres would make suitable substrata for the development of whole cell biosensors, due to the fact that they possess surface characteristics that allow bacterial attachment. On the other hand, the modified fibre surfaces were found to be amenable to bacterial attachment and might have cyto-repellent potential. They may be more suitable for the development of Surface Enhanced Raman Spectroscopy (SERS) substrata or optic probes where a bacteriafree environment is desirable. The results of this study will be of use for the construction of chemical sensors, whole-cell biosensors, SERS probes or other optical fibre instrumentation.