w a t e r r e s e a r c h 6 4 ( 2 0 1 4 ) 1 2 3 e1 3 3

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/watres

Combination of synchrotron radiation-based Fourier transforms infrared microspectroscopy and confocal laser scanning microscopy to understand spatial heterogeneity in aquatic multispecies biofilms Sheela Reuben a,1, Krzysztof Banas b,1, Agnieszka Banas b, Sanjay Swarup c,d,e,* a

Singapore Delft Water Alliance (SDWA), National University of Singapore, 2 Engineering Drive 2, Engineering Workshop 1, #02-05 Singapore 117577, Singapore b Singapore Synchrotron Light Source (SSLS), National University of Singapore, 5 Research Link, Singapore 117603, Singapore c Metabolites Biology Laboratory, Department of Biological Sciences, National University of Singapore, Singapore 117543, Singapore d NUS Environmental Research Institute (NERI), T-Lab Building, 5A Engineering Drive 1, Singapore 117411, Singapore e Singapore Center for Environmental Life Sciences Engineering (SCELSE), Nanyang Technological University, 60 Nanyang Drive, SBS-01N-27, Singapore 637551, Singapore

article info

abstract

Article history:

Understanding the spatial heterogeneity within environmental biofilms can provide an

Received 17 January 2014

insight into compartmentalization of different functions in biofilm communities. We used

Received in revised form

a non-destructive and label-free method by combining Synchrotron Radiation-based

18 June 2014

Fourier Transform Infrared Microspectroscopy (SR-FTIR) with Confocal Laser Scanning

Accepted 30 June 2014

Microscopy (CLSM) to distinguish the spatial chemical changes within multispecies bio-

Available online 9 July 2014

films grown from natural storm waters in flow cells. Among the different surfaces tested for biofilm growth and optimal imaging, mylar membranes were most suited and it

Keywords:

enabled successful spatial infrared imaging of natural biofilms for obtaining reliable and

Biofilm heterogeneity

interpretable FTIR spectra. Time series analysis of biofilm growth showed that influx of

Aquatic biofilms

water during biofilm growth, results in significant changes in biofilm formation. Early

Water

biofilms showed active nutrient acquisition and desiccation tolerance mechanisms cor-

Synchrotron

responding with accumulation of secreted proteins. Statistical approach used for the evaluation of chemical spectra allowed for clustering and classification of various regions of the biofilm. Microheterogeneity was observed in the polymeric components of the biofilm matrix, including cellulose, glycocalyx and dextran-like molecules. Fructan and glycan-rich regions were distinguishable and glycocalyx was abundant in the strongly adhering peripheral regions of biofilms. Inner core showed coexistence of oxygen dimers

* Corresponding author. Metabolites Biology Laboratory, Department of Biological Sciences, National University of Singapore, Singapore 117543, Singapore. Tel.: þ65 6516 7933. E-mail address: [email protected] (S. Swarup). 1 Contributed equally to the paper. http://dx.doi.org/10.1016/j.watres.2014.06.039 0043-1354/© 2014 Elsevier Ltd. All rights reserved.

124

w a t e r r e s e a r c h 6 4 ( 2 0 1 4 ) 1 2 3 e1 3 3

and ferrihydrite that will likely support growth of Fe (II)-oxidising bacteria. The combined SR-FTIR microspectroscopy and CSLM approach for complex natural biofilms described here will be useful both in understanding heterogeneity of matrix components and in correlating functions of juxtaposed microbial species in complex natural biofilms with physicochemical microenvironment to which they are exposed. © 2014 Elsevier Ltd. All rights reserved.

1.

Introduction

Biofilms are aggregations of microorganisms embedded in extrapolymeric substances (EPS). Biofilms can have populations of single species or a community derived from multiple microbial species, which can form on a vast array of abiotic and biotic surfaces. Naturally occurring biofilms consist of multiple species, which can be as varied as algae, fungi and bacteria. Natural biofilms are key mediators in the degradation of environmental pollutants and recycling of elements (Davey and O'Toole, 2000). The availability of nutrients, physical conditions such as water flow, turbulence, environmental parameters such as pH, temperature, ecological conditions combined with the stage of biofilm growth determine the size, shape and function of the biofilms (Costerton et al., 1995). Microbes within the biofilms are not randomly located but form defined spatial arrangements that are specialized in a variety of functions such as shielding from environmental stressors, degradation of chemicals and access to nutrients. The function of a group of organisms may differ from others in a community and hence the communities are structured such that it is conducive for the group to perform its function. Biofilm heterogeneity arises due to differential and combinatorial phenotypic expression in various subpopulations localized to different regions in a biofilm (Spormann, 2008). Visualization of microorganisms and understanding their metabolic profiles is critical in gaining an insight into their function in the biofilm community and microbial interrelationships within the biofilms. Higher resolution corresponding to a single cell is useful because it can relate spatial design of microbial communities to the ecophysiology of its members (Wagner et al., 2006). Though researchers have attempted to understand biofilm compositional heterogeneity in cultured cells (Ngo Thi and Naumann, 2007; Choo-Smith et al., 2001), insight into spatial heterogeneity of natural biofilms is limited. Studies using isotope labelling or fluorescently tagging approaches have been conducted recently to understand spatial heterogeneity (Lee et al., 1999; Huang et al., 2007). However, these techniques require use of labels that can alter the physiology of the microorganisms. Analytical methods such as fluorescence microscopy (Yu and McFeters, 1994; GarciaBetancur et al., 2012), magnetic resonance microscopy (MRM) (Seymour et al., 2004; Gjersing et al., 2005), and Raman microscopy (Virdis et al., 2012; Beier et al., 2010) have been used for biofilms characterisation and monitoring. These techniques have their own constraints, when applied to the studies of living organisms. In fluorescence microscopy, cells

are required to have either an addition of fluorescent labels or the use of genetically modified strains that produce fluorescent proteins. MRM for carbon resonance also requires labelled substrates. Hence, label-requiring technologies may affect biofilm physiology. Vibrational spectroscopic methods are procedures in which there is no need to add dyes or labels for spectral measurements. These non-destructive techniques are based on the absorption (FTIR) or scattering (Raman) of light directed onto a sample. FTIR microspectroscopy can be considered as a non-destructive chemical mapping, where each pixel within an image corresponds to a complete infrared spectrum that reflects the chemical composition of the analysed spot. Many common biomolecules, such as nucleic acids, carbohydrates, proteins and lipids have characteristic and well-defined IRactive vibrational modes. Raman and IR spectroscopy are complementary techniques, which can provide more complete biochemical information within a sample. Analysis of hydrated samples is quite difficult using FTIR spectroscopy since water absorbs so strongly that its signal masks other interesting peaks in the spectrum. On the other hand, water is less problematic in Raman spectroscopy, enabling measurements of hydrated samples. However, the signal-to-noise ratio of the resulting Raman spectrum is overall poorer compared to FTIR spectra measured within the same time range. Moreover, Raman microscopy suffers from low signal yielded relative to the incident power (Ivleva et al., 2010). Raman confocal microscopy study revealed that even medium laser powers, in the order of tens of mW (visible and near-IR); can affect cellular activity by generation of heat and photoproducts, which makes this technique rather destructive for biological samples (Kang et al., 2008). For these reasons, FTIR microspectroscopy as a label-free and high sensitivity approach seems to be reasonable choice for evaluation of biochemical characteristics of biofilms. SR-FTIR microscopy has been used as a label-free approach to track biogeochemical changes with high sensitivity and micrometre spatial resolution in real time (Holman et al., 2009). The mean power of a synchrotron mid-IR beam is around 1 mW at the focused sample position, which is below the threshold for cytotoxic effects (Holman et al., 2002). This enables capturing the chemical composition without affecting the living cells in the biofilm. Several reports have established FTIR spectroscopy as a method of rapid identification and differentiation of microorganisms at the genus, species and strain level (Preisner et al., 2010; Lamprell et al., 2006; Kuhm et al., 2009; Naumann, 2001). FTIR microscopy has also been used to characterize growth heterogeneity within microbial micro-

w a t e r r e s e a r c h 6 4 ( 2 0 1 4 ) 1 2 3 e1 3 3

colonies. Application of this method for studying biofilms in their natural environment is a major challenge because of the strong absorption of the water in the mid-IR region. One approach used to suppress this effect is attenuated total reflectance (ATR) sampling (Kazarian and Chan, 2013). In this case, however, penetration depth of the evanescent wave is below 1 mm and captures only information from the molecules near the surface. An alternative approach is to use a closedchannel microfluidic flow cell with thin spacers to reduce the thickness of the aqueous layer to less than 15 mm. This approach suffers from the significant interference fringes that are not easily removable from the resulting spectrum. There is also an option to use open-channel microfluidic approach to minimize water absorption and the interference fringe problem while maintaining the functionality of biofilms (Holman et al., 2009). However, ability to use this microfluidic approach to study biofilms analysis is challenging due to its complicated construction. Confocal Laser Scanning Microscopy (CLSM) is a powerful technique that not only gives images with microscopic resolution but also reveals a three-dimensional structure (Palmer Jr. and Sternberg., 1999). This technique, however, does not give information on the chemical composition of the biofilm. Combining CLSM with SR-FTIR imaging helps to better understand the spatial chemical composition of biofilms. Here, we have used these two imaging techniques to study the spatial heterogeneity based on chemical composition from biofilms grown on mylar foil using water from a storm water canal. We present the results of search for optimal support material in terms of biocompatibility and usability in FTIR experiments. Subsequently, we scrutinize a single biofilm to better understand the changes in chemical composition across the biofilm. We show here the application of this technique in studying the biological and chemical microheterogeneity in complex environmental biofilms, identifying the chemical basis of this heterogeneity and mapping the dynamics of chemical profile in the extracellular domains at various stages of the biofilm growth.

sheets were placed in small sterile Petri dishes in sterile water and taken for SR-FTIR imaging. The sheets were rinsed once with distilled water to remove unattached or loosely attached cells and then imaged. Six different materials were tested for their propensity to form biofilms as well as give good signals for the spectral data: polished and unpolished steel, transparent nano porous alumina membrane (Smart Membranes, Germany), standard microscopic glass, mirrIR microscopic slide (Kevley Technologies, USA) and mylar foil. For time series experiment of biofilm growth, 4 sets of mylar sheets were placed in the flow cell as mentioned above. Each set had two replicate mylar sheets. Canal water was used as source for biofilm growth on mylar sheets. The first set of samples was collected at the end of the first and second week, respectively. After two weeks, fresh water from the canal was introduced to the system to understand the effect of new source of water on the chemical composition of biofilm. The third and fourth set of samples was collected after three and four weeks, respectively.

2.2.

Material and methods

2.1.

Biofilm growth conditions

Samples of water were collected from a storm water canal in Singapore, transported to the laboratory immediately and used for the experiment on the same day. Experimental set-up was built to allow water circulation in a closed system consisting of a 10 L water reservoir, bubble trap, peristaltic pump (Cole Parmer, USA) and flow cell with dual chamber (FC 270) (Biosurface Technologies, USA) connected via silicone tubing of diameter 1.5875 mm (1/16 inch) ID (Cole Parmer, USA). The flow cell had indentations for placing four disc coupons of 0.5 inch diameter and 0.15 inch thickness (Reuben et al., 2012). Mylar films (Chemplex Industries, USA) were placed with metal supports in these recesses instead of coupons. Canal water was circulated in the flow cell system for 12 h at a rate of 8 ml/min at room temperature for the biofilm growth to occur. Slow flow rate was chosen to replicate conditions in the canal when there is no rainfall. At the end of the experiment, mylar

Confocal scanning laser microscopy (CSLM)

Confocal scanning microscope used for this study was LSM META510, Carl Zeiss, Germany. Carl Zeiss LSM software (ver. 4.0 sp2 with Physiology) was used to capture the images. The CSLM was equipped with an inverted microscope and Argon multi-line gas laser (458, 477, 488, 514 nm, 30.0 mW). Imaging was performed using two channels for green and red light 497e529 nm and 561e679 nm, respectively. For CSLM, the biofilms were stained using the Live-dead BacLight Bacterial Viability Kit (Molecular Probes Inc, USA) by incubating them at room temperature in the dark for 15 min, according to the manufacturer's manual. The biofilms were scanned using objective lens LD Plan-Neofluar 20x/0.4 Corr. Z-series was generated that dissect through the specimen to get 3D imaging at default intervals (~3 mm) depending on the thickness of the biofilms.

2.3.

2.

125

FTIR settings

IR microscope Hyperion 2000 (Bruker Optik, Germany) equipped with 15x infrared Schwarzschild objective combined with spectrometer IFS 66v/s was used during FTIR experiments. Both sources (classical Globar and synchrotron radiation e SR) were tested during the measurements performed at ISMI beamline (Bahou et al., 2007) of Singapore Synchrotron Light Source. At ISMI SR is extracted from the edge region of dipole of the compact superconducting electron storage ring. The nominal source point is located at half the maximum field, i.e. at 2.25 T. With the system of mirrors the light is collected, transferred and focused at the entrance port of the Bruker IFS 66v/S spectrometer. A high photon flux and brilliance of synchrotron radiation enable IR experiments with high spatial resolution difficult to reach with conventional black body radiation (Globar). Advantage of using synchrotron radiation as the IR source over Globar source is due to its high brilliance. In synchrotron related research brilliance is measured in photons/s/mrad2/ mm2/0.1% band width. Brilliance advantage manifests itself when measured area is set to small value (less than 15 microns

126

w a t e r r e s e a r c h 6 4 ( 2 0 1 4 ) 1 2 3 e1 3 3

by 15 microns in our case). Due to relatively small size of the beam (as compared to isotropic radiation of classical sources) photon flux is in this case much bigger for synchrotron based radiation. However it is worth to mention that total deposited power is the same or even smaller when compared with classical source and thus there is no risk of any changes in biological material with the incoming IR. Colour LCD monitor allowed for easy and real time control of the inspected sample area, and video camera enabled to take a picture of analysed sample. Motor driven xey stage for precise sample positioning allowed for performing 2D experiments in an automatic way. FTIR absorption spectra were recorded from the biofilms grown on different support materials. Depending on the type of support material reflection or transmission mode was used.

Regions of biofilm samples measured by FTIR chemical mapping were carefully chosen by using visual overview in FTIR microscope. Characteristic shapes identified later served as the landmarks for identification when the samples were transferred for CLSM measurement. The region of interest was chosen by the system of horizontal and vertical slits set typically to 15 by 15 mm2. Usually 600 scans at each point were collected at 4 cm1 energy resolution within the wavenumber range of 4000 to 400 cm1. During all measurements MCT (Mercury Cadmium Telluride) detector cooled to liquid nitrogen temperature (77 K) was used. Fourier transformation was done using a BlackmannHarris three-term apodization function and a zero-filling factor of 4. Spatial heterogeneity of the microcolonies was

Fig. 1 e SR-FTIR spectra of 2-day old biofilms on different surfaces. A. Reflection or transmission spectra of biofilms on different surfaces. Depending on the type of material, reflection (unpolished and polished steel, mirrIR slide) or transmission (transparent alumna membrane and microscopic glass) mode was used. B: SR-FTIR spectrum of 2-day old biofilms on mylar foil measured in transmission mode.

127

w a t e r r e s e a r c h 6 4 ( 2 0 1 4 ) 1 2 3 e1 3 3

examined by mapping the sample in 15 mm steps in the x and y directions. The size of the map (165 by 150 mm2 with the pixel size 15 mm2) was determined by the dimensions of the scan area set prior to the measurements. Background spectrum was collected on pure mylar foil by using the same experimental settings. Initial pre-processing of the collected spectra was performed by means of OPUS software ver. 5.5 provided by the manufacturer of the spectrometer and microscope, then spectra were exported in text format. Further spectral processing as well as statistical analysis of the spatial data was done with help of R platform, open source software environment for statistical computing and graphics (R Development Core Team, 2014).

2.4.

Data analysis

2.4.1.

Time series data

3.

Results and discussion

3.1. Search for the substrate suitable for biofilm growth and SR-FTIR spectroscopy

Time series data were normalized and baseline corrected. Pareto scaling was performed using datPAV software (Biswas et al., 2011). Three spectra from central areas of biofilms were averaged and compared for each week. Hierarchical clustering (Everitt, 2011; Williams, 2011) was performed using Euclidean distance and group average linkage method. This is a procedure that objectively groups the input cases (i.e. the spectra in our research) based on similarities of their properties (the spectral characteristics). The result of the analysis forms a dendrogram that is the relationship between the input cases represented by the distance at which they connect on a dissimilarity scale. Since the spectral information reflects the biochemistry of the sample measured, the distance in the dendrogram can be interpreted as a measure of biochemical diversity in various spectra. Cluster analysis was performed using Primer V6 package (Clarke and Gorley, 2006) (Primer-E Ltd., Plymouth, United Kingdom).

2.4.2.

information assigned to each spectrum (hyperSpec object). Pre-processing included smoothing, baseline correction and normalization in order to remove sample thickness effects. Subsequent analysis included integration over certain spectral ranges in order to visualise spatial distribution of the groups of interest (i.e. amides, anions and carbohydrates). Hierarchical cluster analysis was applied for unsupervised classification and clustering of the spectra. All calculations (data processing and multivariate analysis) and graphical presentations were done with the package hyperSpec (Beleites and Sergo, 2014) working in R platform version 3.02.

Spatial data

IR spectra collected for points in the rectangular grid (SR-FTIR imaging) were arranged in a matrix with additional coordinate

The analysis of biofilms by means of FTIR is a major challenge, firstly because of finding proper support material to prepare the sample for the experiments and secondly because of the strong water absorption in the mid-IR region. Ideal support material for this application needs to have two important characteristics: ability to support biofilm growth (biocompatibility) and minimal background signal in mid-IR region when imaging. In order to identify such a material, complex environmental biofilms were grown on various substrates using water from a storm water canal (Sungei Ulu Pandan, Singapore). Two-day old biofilms were then directly imaged and analysed in situ. Though all tested materials allowed for good biofilm growth, they had varying capabilities for FTIR imaging by showing strong absorption in spectral regions of interest, which led to various signal to noise ratios (Fig. 1A). FTIR spectra collected for the biofilms that were grown on standard microscopic glass slides, showed high background signal. Though mirrIR slides are known to produce good FTIR spectra, in this setup, they failed to yield good quality spectra. The possible reason for this was the specific condition of the investigated system. Most of the reported successful

Table 1 e Assignment of selected IR spectral bands for the investigated biofilms. Wavenumber [cm1]

Assignment

References

C]O stretch

Water molecules, anions Water molecules, anions Water molecules, anions Amide I

1542

NeH bend and CeH stretch

Amide II

1457 1116

Bending CH2/CH3

Holman et al., 2009 Holman et al., 2009 Holman et al., 2009 Delille et al., 2007 Ngo Thi and Naumann, 2007 Choo-Smith et al., 2001 Delille et al., 2007 Ngo Thi and Naumann, 2007 Ojeda et al., 2008, 2009 Delille et al., 2007 Ngo Thi and Naumann, 2007 Delille et al., 2007 Ngo Thi and Naumann, 2007 Holman et al., 2009 Delille et al., 2007 Ngo Thi and Naumann, 2007 Delille et al., 2007 Ngo Thi and Naumann, 2007

3697 3653 3620 1648

1031

1007 915

Bond

Polyglucose

128

w a t e r r e s e a r c h 6 4 ( 2 0 1 4 ) 1 2 3 e1 3 3

applications of mirrIR slides are for frozen tissue samples sectioned with the use of microtome (Dobson et al., 2008). Transparent alumina gave some peaks in the spectrum of biofilms, but the noise level was very high. Both polished and unpolished steel produced better spectra. Among all the tested surfaces, mylar membranes provided the best spectra, which were confirmed by both visual inspection and comparison of signal-to-noise ratios for respective spectra. As an example Fig. 1B presents SR-FTIR spectrum of 2-day old biofilms on mylar foil measured in transmission mode. Normally, reflectance spectra are of poorer quality (with exception for thin coatings on metallic substrates) than transmittance spectra due to the fact that typically only less that 5% of the IR radiation is reflected. By using mylar membranes that have no interfering absorption in mid-IR region (4000e400 cm1) as the substrate material, one can perform experiments in transmission mode. Mean value of calculated SNR (signal divided by root mean square) for mylar was 420, while for other materials measured in the same experimental conditions was less than 100. SNR was calculated by using standard (for mid-IR) spectral range 2100e1900 cm1 after 100% transmission line (for mylar, transparent alumina and microscopic glass) or 100% reflection line experiment (for polished and unpolished steel and mirrIR slides) under the IR microscope with slits set to 15 by 15 mm2. Time of the single measurement was

set to 60 s. Mylar membranes were used for all further experiments.

3.2.

Bands assignment and selection

Original spectra showed well-resolved vibrational bands from both small molecules and key macromolecules superimposed on a broad baseline feature, due to different background scatterings, biofilm thickness effects and continuum water absorption. Fig. 1B shows median spectrum with 18th and 86th percentiles for representative regions with highest variability. Spectral analyses revealed many interesting peaks. We selected for biofilm study the absorption peaks at 3697, 3653 and 3620 cm1 (anions), 1648 cm1 and 1542 cm1 (amide I and amide II, respectively), 1457 cm1 (bending of CH2/CH3) and four of most prominent absorption bands from carbohydrates region: 1116, 1031, 1007 and 915 cm1. All dominant bands observed for biofilms under investigation with their assignment are included in Table 1.

3.3.

Biofilm growth studies

The chemical composition of biofilms grown from one to four weeks were analysed to investigate the differences in their chemical composition. The biofilms were randomly scanned at different points on the mylar sheet for each week.

Fig. 2 e SR-FTIR spectra of biofilms on mylar membrane A. Spectra for various time-points (from 1 to 4 weeks). B. Hierarchical clustering of weekly absorption spectra.

w a t e r r e s e a r c h 6 4 ( 2 0 1 4 ) 1 2 3 e1 3 3

Hierarchical cluster analysis showed two distinct groups. Biofilms from week 3 stage had the most distinct spectra and significantly differed from the others (ANOVA test p ¼ 3.41E10). It was observed that after fresh canal water was introduced into the system, there were significant changes from first, second and fourth week (Fig. 2A & B). Hierarchical clustering showed that spectra obtained for week 1 and 2 did not group separately. Spectra of four- week- old biofilms were closer to those of week 1 and 2 than week 3. However, some peak differences were observed with maximal at 1641 cm1, which correspond to amide I (stretching of the carbonyl coupled to the CeN) and in the range of 900e1200 cm1, which is assigned to polysaccharides. It can, hence be concluded that biofilms of week 1 and 2 stage had higher polysaccharides but lesser amide I and vice versa for week 4. Spectra of 3 week old biofilms showed higher amide II bands. Despite low levels of noise in spectra for amide II bands, their signals were consistently observed in several spectra from the early stage biofilms. In terms of biovolume, typically, cellular material of biofilms is reported to be just 15%, the remainder being matrix (Sutherland, 2001). Biofilm matrix is composed of peptidoglycan and other macromolecules, such as polysaccharides, proteins, DNA, RNA and other metabolic products and hence, referred to as extrapolymeric substances (EPS). In river biofilms, EPS act as buffer against nutrient changes and accumulate in environments, actively participating in biogeochemical reactions (Neu and Lawrence, 2009). Hence, EPS has a very dynamic composition. The amide changes likely correspond to secreted proteins in the biofilm. Such secreted proteins have been shown to be mostly proteases that are involved in nutrient acquisition (Rice et al., 1999). The major changes occur during week 3 due to the

129

influx of nutrients as well as new bacterial species from the canal water, which resulted in drastic changes in the composition of proteins. These could be due to changes in metabolism, defence, utilization of available new sources and responses to accommodate new species. As biofilm age progresses there is a change in the amount of polysaccharides and proteins in week 4 compared to the first two weeks. However, the spectra are similar to week 1 and 2 indicating that a more steady state of biofilm composition is achieved without major perturbation.

3.4. Chemical mapping and spatial heterogeneity within a biofilm In order to study the chemical heterogeneity within biofilm samples 2-dimensional scans were performed over the region of interest. Fig. 3A shows the microscopic overview of the biofilm area with region of interest marked with red rectangle. Size of the scanned area was 165 by 150 mm2 with the pixel size 15 mm2. The same region of the sample was also evaluated by confocal scanning microscopy. Stained biofilm shows a central dark region with peripheral lighter regions. Image shows the live cells as green and dead cells as red. The regions that overlapped are seen as yellow colour (Fig. 3B). Traditional univariate approach to FTIR micro-spectroscopy is the integration over the certain bands in order to get their spatial distributions is shown in the Fig. 3CeE (spatial distribution of amide I and II range 1480e1680 cm1 (C), carbohydrates range 820e1200 cm1 (D) and anions range 3550e3750 cm1 (E). In order to obtain a consolidated view of the differences in spectral distribution and classify parts of the biofilms into various categories, a multivariate statistical approach was used. Hierarchical cluster analysis enables unsupervised

Fig. 3 e 2D scanning of biofilms grown on mylar membranes at week 4 stage. A Overview of the region of interest (ROI), B CSLM imaging of the ROI, C Integration over 1542 cm¡1 bands e spatial distribution of amide II, D Integration over 1031 cm¡1 band e spatial distribution of polysaccharide, E Integration over 3697, 3653 and 3620 cm¡1 bands e spatial distribution of anions, F Results of HCA statistical analysis e location of three clusters.

130

w a t e r r e s e a r c h 6 4 ( 2 0 1 4 ) 1 2 3 e1 3 3

classification of the cases, which we applied in our case for spectra obtained for every point of the map. Fig. 3F shows visualisation of the spatial distribution of three clusters, corresponding to biofilm inner part, outer part and support material region, respectively, in the measured area. Further, to perform a detailed analysis of biofilms, an overlap between confocal images and the chemical imaging was carried out (Fig. 4C). Confocal microscope images of biofilms showed distinct light and dark regions. In order to understand the differences between these regions, four areas of nearly uniform sizes were chosen for each group (bright and dark regions). Confocal image was overlaid with FTIR image in order to obtain the locations of the chemical maps. IR mean spectra from these regions were then compared for the two groups (Fig. 4B). Although these spectra were highly similar, any differences that exist were identified at a finer resolution, by performing deconvolution of the spectra (Jung, 2000; Petibois et al., 2006; Belbachir et al., 2009). Due to the number of free parameters careful inspection of the results of fitting is always required. Various initial conditions (including the number of components used in fitting) should be tested. In our case ten peaks were finally chosen because of the most stable solution and reasonable fit quality. Fig. 4D and E present fitting results of ten peaks in the spectral range 850e1200 cm1 for bright and dark area mean spectrum, respectively. A number of differences were identified between dark and bright areas in several spectral regions. These include increased intensity observed for dark areas for the bands 1068e69 cm1, 1090 cm1, 1011e1012 cm1 and 937 cm1 and increased intensity observed for bands 985 cm1, 999e1002 cm1, 1116e17 cm1 and 914 cm1 for bright areas.

Bands observed for 1031e32 cm1 and 1050 cm1 wavenumbers do not differ between dark and bright areas. The major difference between the dark and bright regions tested pertained to EPS. For the wavenumbers 1068e69 cm1 assigned to fructose vibration (Vodnar et al., 2012) and regions 999e1002 cm1, 1116e1117 cm1 and 914 cm1 assigned to fingerprint region of cellulose vibration (Gorassini et al., 2008), glycocalyx (Holman et al., 2009) and rotational isomerism about C1O1,bond of dextran molecules (bond 1,6) (Carmona et al., 1997) respectively, they all showed differences between dark and bright regions. EPS are regarded as a major factor that influence biofilm formation and are responsible to strengthen the interactions between microorganisms and determines the formation of cell aggregates formation process on solid surface (Czaczyk and Myszka, 2007). The changes observed indicate that not only the amount of exopolysaccharides, but even the type varies spatially within the biofilm. These differences can be attributed to the type of organisms associated specifically to that region in the biofilm. Regions associated with bright areas most likely would be dominated by bacteria with fructans in their EPS, whereas those in the dark regions most probably would be dominated by species which have a-D-glucans in their EPS. a-D-glucans mostly contain a(1 / 6) linked D-glucosyl units (peak at 914 cm1). Glycocalyx is bacterial carbohydrate that facilitates strong adhesion of bacteria to surfaces and between cells and other key biomolecules during biofilm development in microscopic space and is a prerequisite for formation of bacterial biofilms (Kang et al., 2008). In this study, glycocalyx carbohydrates showed an increase in the light regions which are located towards the periphery of the biofilm. These

Fig. 4 e Spectral differences in different regions of mature biofilms. A Comparison of mean spectra of bright and dark areas in the polysaccharides region (1200e850 cm¡1) of biofilms at week 4 stage. B Comparison of mean spectra of bright and dark areas in the amide region (1700e1400 cm¡1). C CSLM image of biofilm on mylar membrane with overlaid FTIR measurements grid, bright (letter B) and dark (letter D) areas are marked. D Deconvolution of the mean spectrum in 850e1200 cm¡1 range for bright area. E Deconvolution of the mean spectrum in 850e1200 cm¡1 range for dark area.

w a t e r r e s e a r c h 6 4 ( 2 0 1 4 ) 1 2 3 e1 3 3

regions would require stronger adhesion to the surface for sustenance of the biofilm. Cellulose, on the other hand is a polysaccharide that is not involved in biofilm formation, but its production increase tolerance to desiccation, suggesting that the function of cellulose might be related to resistance to environmental stresses (Gualdi et al., 2008). The presence of cellulose in the brighter regions could be a defence mechanism for the biofilm as it is exposed constantly to environmental pressures compared to the darker inner regions. The other difference observed was the increased intensity of the bands in the 1011e1012 cm1 and 937 cm1 spectral regions for the darker regions of biofilm, which can be assigned to oxygen dimer (Markevich et al., 1996) and d (OH) ferrihydrite, respectively (Thieme et al., 2010). This implies an abundance of ferrihydrite deposits in the dark regions, where possibly Fe (II)-oxidizing bacteria subpopulations thrive. As was mentioned by Gault (Gault et al., 2011): “for Fe (II)oxidizing bacteria to effectively compete with abiotic Fe (II) oxidation in circumneutral waters they must occupy an environment, where the dissolved oxygen concentration slows the chemical oxidation of Fe (II) sufficiently”. The ferric iron thus produced by Fe (II)-oxidizing bacteria can be recycled by Fe (III)-reducing bacteria that may inhabit anoxic microenvironments adjacent to the areas occupied by microaerophilic Fe (II)-oxidizers. It can be surmised that in natural biofilms both aerobic and anaerobic subpopulations coexist depending on the function and microenvironment presented to them. Biofilms that grow on rocks or near iron oxide rich sediments have a tendency to absorb metal ions to their EPS (Yamamoto et al., 2011) or biofilms would contain precipitates of hydrous iron oxides described as ferrihydrite as seen in metal interactions studies with varied pH (Smith and Ferris, 2003). These findings could be confirmed by the elemental analysis of the trace elements by using micro X-ray fluorescence (m-XRF) spectroscopy and especially its synchrotron radiation based equivalent e SRIXE (Synchrotron Radiation Induced X-ray Emission). When amide and anion wavelength regions were compared for the dark and light regions, no major changes were observed. Intensity of the amide bands was, however, slightly higher in the darker regions (Fig. 4C). Since the cell density in the dark regions are lesser as seen in confocal images, the amides may not be a part of the cell structure but may belong to excreted/secreted proteins by the bacterial cells. It is well known that proteins form a part of the extrapolymeric substances in several species of bacteria contributing to functions such as surface adhesion and maintain fitness through nutrient acquisition by proteases in different ecological niches (Absalon et al., 2011; Sikora, 2013). This study hence shows the advantages of combining synchrotron e FTIR with confocal microscopy to understand the structure and properties of biofilms. The technique may be used to study and quantify other features of the biofilms such as the EPS diversity changes under different environmental conditions especially hydrodynamic forces and environmental stresses for aquatic multispecies biofilms. The microspectroscopy can be combined with fluorescent in situ hybridization techniques to visualize gene expression patterns in relation to physiochemical microenvironment.

4.

131

Conclusions

Our preliminary results revealed that the analysis of natural multispecies biofilms by means of FTIR is possible. Major technical challenge remains due to the strong water absorption in the mid-infrared region. Mylar foil is a reasonably good substratum for SR-FTIR analysis with a SNR of 340. Spatiotemporal changes in chemical composition in biofilms especially in the biofilm matrix can be obtained using SR-FTIR. Correlating SRFTIR with CSLM helps to distinguish areas with high cell density and void regions that can be correlated with chemical composition to obtain better understanding of biofilm structure and function at micron scale dimensions in non-destructive and label-free manner. This method provides further insights into linking heterogeneity of structure and function within natural multispecies biofilms in a dynamic manner.

Acknowledgements The authors gratefully acknowledge the support and contributions of Singapore-Delft Water Alliance (SDWA) R-264-001002-272. The research presented in this work was carried out as part of the SDWA's ASC Pandan research programme. The authors also thank Singapore Synchrotron Light Source and funds from NUS Core Support C-380-003-003-001, A*STAR/ MOE RP 3979908M and A*STAR 12 105 0038 grants for imaging studies. Authors thank Staffan Kjelleberg and Yehuda Cohen of the Singapore Centre on Environmental Life Sciences Engineering (SCELSE) for critical comments during manuscript preparation.

references

Absalon, C., Van Dellen, K., Watnick, P.I., 2011. A communal bacterial adhesin anchors biofilm and bystander cells to surfaces. PLoS Pathog. 7 (8) http://dx.doi.org/10.1371/ journal.ppat.1002210. Bahou, M., Wen, L., Ding, X., Casse, B.D.F., Heussler, S.P., Gu, P., Mathis, Y.-L.Y.-L., 2007. Infrared spectro/microscopy at SSLS e edge effect source in a compact superconducting storage ring. AIP Conf. Proc. 879 (1), 603e606. http://dx.doi.org/10.1063/ 1.2436133. Beier, B.D., Quivey, R.G., Berger, A.J., 2010. Identification of different bacterial species in biofilms using confocal Raman microscopy. J. Biomed. Opt. 15 (6), 066001e1-5. Belbachir, K., Noreen, R., Gouspillou, G., Petibois, C., 2009. Collagen types analysis and differentiation by FTIR spectroscopy. Anal. Bioanal. Chem. 395 (3), 829e837. Beleites, C., Valter Sergo, V., 2014. HyperSpec: a package to handle hyperspectral data sets in R, Rpackage version 0.98-20140523. URL http://hyperspec.r-forge.r-project.org. Biswas, A., Rao, R., Umashankar, S., Mynampati, K.C., Reuben, S., Parab, G., Swarup, S., 2011. datPAVean online processing, analysis and visualization tool for exploratory investigation of experimental data. Bioinformatics 27 (11), 1585e1586. http:// dx.doi.org/10.1093/bioinformatics/btr179. Carmona, P., Navarro, R., Hernanz, A. (Eds.), 1997. Spectroscopy of Biological Molecules: Modern Trends. Kluwer Academic Publishers, ISBN 978-94-011-5622-6.

132

w a t e r r e s e a r c h 6 4 ( 2 0 1 4 ) 1 2 3 e1 3 3

Choo-Smith, L.P., Maquelin, K., van Vreeswijk, T., Bruining, H.A., Puppels, G.J., Ngo Thi, N.A., Kirschner, C., Naumann, D., Ami, D., Villa, A.M., Orsini, F., Doglia, S.M., Lamfarraj, H., Sockalingum, G.D., Manfait, M., Allouch, P., Endtz, H.P., 2001. Investigating microbial (micro)colony heterogeneity by vibrational spectroscopy. Appl. Environ. Microbiol. 67 (4), 1461e1469. http://dx.doi.org/10.1128/AEM.67.4.1461-1469.2001. Clarke, K.R., Gorley, R.N., 2006. PRIMER V6:User Manual/Tutorial. PRIMER-E, Plymouth, United Kingdom. Costerton, J.W., Lewandowski, Z., Caldwell, D.E., Korber, D.R., Lappin-Scott, H.M., 1995. Microbial biofilms. Annu. Rev. Microbiol. 49, 711e745. http://dx.doi.org/10.1146/ annurev.mi.49.100195.003431. Czaczyk, K., Myszka, K., 2007. Biosynthesis of extracellular polymeric substances (EPS) and its role in microbial biofilm formation. Pol. J. Environ. Stud. 16 (6), 799e806. Davey, M.E., O'Toole, G.,A., 2000. Microbial biofilms: from ecology to molecular genetics. Microbiol. Mol. Biol. Rev. 64 (4), 847e867. Delille, A., Quiles, F., Humbert, F., 2007. In situ monitoring of the nascent Pseudomonas fluorescens biofilm response to variations in the dissolved organic carbon level in lownutrient water by attenuated total reflectance-Fourier transform infrared spectroscopy. Appl. Environ. Microbiol. 73 (18), 5782e5788. http://dx.doi.org/10.1128/AEM.00838-07. Dobson, R.L., Motlagh, S., Quijano, M., Cambron, R.T., Baker, T.R., Pullen, A.M., Regg, B.T., Bigalow-Kern, A.S., Vennard, T., Fix, A., Reimschuessel, R., Overmann, G., Shan, Y., Daston, G.P., 2008. Identification and characterization of toxicity of contaminants in pet food leading to an outbreak of renal toxicity in cats and dogs. Toxicol. Sci. 106 (1), 251e262. http://dx.doi.org/10.1093/toxsci/kfn160. Everitt, B., 2011. Hierarchical Clustering in Cluster Analysis. Wiley, Hoboken, pp. 71e110. Garcia-Betancur, J.C., Yepes, A., Schneider, J., Lopez, D., 2012. Single-cell analysis of Bacillus subtilis biofilms using fluorescence microscopy and flow cytometry. J. Vis. Exp. 15 (60), 3796. Gault, A.G., Ibrahim, A., Langley, S., Renaud, R., Takahashi, Y., Boothman, C., Lloyd, J.R., Clark, I.D., Ferris, F.G., Fortin, D., 2011. Microbial and geochemical features suggest iron redox cycling within bacteriogenic iron oxide-rich sediments. Chem. Geol. 281 (1e2), 41e51. http://dx.doi.org/10.1016/ j.chemgeo.2010.11.027. Gjersing, E.L., Codd, S.L., Seymour, J.D., Stewart, P.S., 2005. Magnetic resonance microscopy analysis of advective transport in a biofilm reactor. Biotechnol. Bioeng. 89 (7), 822e834. Gorassini, A., Calvini, P., Baldin, A., 2008. Fourier Transform Infrared Spectroscopy (FTIR) Analysis of Historic Paper Documents as a Preliminary Step for Chemometrical Analysis, Multivariate Analysis and Chemometry Applied to Environment and Cultural Heritage, second ed. Ventotene Island, Italy. Gualdi, L., Tagliabue, L., Bertagnoli, S., Ierano, T., De Castro, C., Landini, P., 2008. Cellulose modulates biofilm formation by counteracting curil-mediated colonization of solid surfaces in Escherichia coli. Microbiology-Sgm 154, 2017e2024. http:// dx.doi.org/10.1099/mic.0.2008/018093-0. Holman, H.Y., Bjornstad, K.A., McNamara, M.P., Martin, M.C., McKinney, W.R., Blakely, E.A., 2002. Synchrotron infrared spectromicroscopy as a novel bioanalytical microprobe for individual living cells: cytotoxicity considerations. J. Biomed. Opt. 7 (3), 417e424. http://dx.doi.org/10.1117/1.1485299. Holman, H.Y.N., Wozei, E., Lin, Z., Comolli, L.R., Ball, D.A., Borglin, S., Fields, M.W., Hazen, T.C., Downing, K.H., 2009. Real-time molecular monitoring of chemical environment in obligate anaerobes during oxygen adaptive response. Proc.

Natl. Acad. Sci. U S A 106 (31), 12599e12604. http://dx.doi.org/ 10.1073/pnas.0902070106. Huang, W.E., Stoecker, K., Griffiths, R., Newbold, L., Daims, H., Whiteley, A.S., Wagner, M., 2007. Raman-FISH: combining stable-isotope Raman spectroscopy and fluorescence in situ hybridization for the single cell analysis of identity and function. Environ. Microbiol. 9 (8), 1878e1889. http:// dx.doi.org/10.1111/j.1462-2920.2007.01352.x. Ivleva, N.P., Wagner, M., Horn, H., Niessner, R., Haisch, C., 2010. Raman microscopy and surface-enhanced Raman scattering (SERS) for in situ analysis of biofilms. J. Biophot. 3, 548e556. Jung, C., 2000. Insight into protein structure and proteineligand recognition by Fourier transform infrared spectroscopy. J. Mol. Recognit. 13, 325e351. Kang, L.L., Huang, Y.X., Luo, M., 2008. Technique of confocal Raman microscopy on erythrocytes. Spectrosc. Spectr. Anal. 28 (10), 2343e2347. http://dx.doi.org/10.3964/j.issn.1000059310-2343-05. Kazarian, S.G., Chan, K.L.A., 2013. ATR-FTIR spectroscopic imaging: recent advances and applications to biological systems. Analyst 138, 1940e1951. Kuhm, A.E., Suter, D., Felleisen, R., Rau, J., 2009. Identification of Yersinia enterocolitica at the species and subspecies levels by Fourier transform infrared spectroscopy. Appl. Environ. Microbiol. 75 (18), 5809e5813. AEM.00206-09 [pii]. Lamprell, H., Mazerolles, G., Kodjo, A., Chamba, J.F., Noel, Y., Beuvier, E., 2006. Discrimination of Staphylococcus aureus strains from different species of Staphylococcus using Fourier transform infrared (FTIR) spectroscopy. Int. J. Food Microbiol. 108 (1), 125e129. http://dx.doi.org/10.1016/ j.ijfoodmicro.2005.10.015. Lee, N., Nielsen, P.H., Andreasen, K.H., Juretschko, S., Nielsen, J.L., Schleifer, K.H., Wagner, M., 1999. Combination of fluorescent in situ hybridization and microautoradiography-a new tool for structure-function analyses in microbial ecology. Appl. Environ. Microbiol. 65 (3), 1289e1297. Markevich, V.P., Medvedeva, I.F., Murin, L.I., 1996. Effect of hydrogen on oxygen-related defect reactions in silicon at elevated temperatures. Early Stages Oxyg. Precip. Silicon 17, 103e122. Naumann, D., 2001. FT-infrared and FT-Raman spectroscopy in biomedical research. In: Gremlich, H.-U., Yang, B. (Eds.), Infrared and Raman Spectroscopy of Biological Material. Marcel Dekker, Inc., New York, NY, pp. 323e377. Neu, T.R., Lawrence, J.R., 2009. In: Moran, A., Brennan, P., Holst, O., von Itzstein, M. (Eds.), Microbial Glycobiology: Structures, Relevance and Applications. Elsevier, San Diego, pp. 735e758. Ngo Thi, N.A., Naumann, D., 2007. Investigating the heterogeneity of cell growth in microbial colonies by FTIR microspectroscopy. Anal. Bioanal. Chem. 387 (5), 1769e1777. http://dx.doi.org/10.1007/s00216-006-0829-z. Ojeda, J.J., Romero-Gonzalez, M.E., Bachmann, R.T., Edyvean, R.G.J., Banwart, S.A., 2008. Characterization of the cell surface and cell wall chemistry of drinking water bacteria by combining XPS, FTIR spectroscopy, modeling, and potentiometric titrations. Langmuir 24 (8), 4032e4040. http:// dx.doi.org/10.1021/la702284b. Ojeda, J.J., Romero-Gonzalez, M.E., Banwart, S.A., 2009. Analysis of bacteria on steel surfaces using reflectance micro-Fourier transform infrared spectroscopy. Anal. Chem. 81 (15), 6467e6473. http://dx.doi.org/10.1021/ac900841c. Palmer Jr., R.J., Sternberg, C., 1999. Modern microscopy in biofilm research: confocal microscopy and other approaches. Curr. Opin. Biotechnol. 10 (3), 263e268. http://dx.doi.org/10.1016/ S0958-1669(99)80046-9. Petibois, C., Gionnet, K., Gonc¸alves, M., Perromat, A., Moenner, M.,  le ris, G., 2006. Analytical performances of FT-IR De

w a t e r r e s e a r c h 6 4 ( 2 0 1 4 ) 1 2 3 e1 3 3

spectrometry and imaging for concentration measurements within biological fluids, cells, and tissues. Analyst 131 (5), 640. Preisner, O., Guiomar, R., Machado, J., Menezes, J.C., Lopes, J.A., 2010. Application of Fourier transform infrared spectroscopy and chemometrics for differentiation of Salmonella enterica serovar Enteritidis phage types. Appl. Environ. Microbiol. 76 (11), 3538e3544. DOI: AEM.01589-09 [pii]. R Core Team, 2014. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. Austria. URL: http://www.R-project.org/. Reuben, S., Chua, C.L.N., Fam, K.D., Thian, Z.Y.A., Kang, M.K., Swarup, S., 2012. Bacterial diversity on different surfaces in urban freshwater. Water Sci. Technol. 65 (10), 1869e1874. http://dx.doi.org/10.2166/Wst.2012.952. Rice, S.A., Givskov, M., Steinberg, P., Kjelleberg, S., 1999. Bacterial signals and antagonists: The interaction between bacteria and higher organisms. J. Mol. Microbiol. Biotechnol. 1 (1), 23e31. Seymour, J.D., Codd, S.L., Gjersing, E.L., Stewart, P.S., 2004. Magnetic resonance microscopy of biofilm structure and impact on transport in a capillary bioreactor. J. Magn. Reson. 167 (2), 322e327. Sikora, A.E., 2013. Proteins secreted via the type II secretion system: smart strategies of Vibrio cholerae to maintain fitness in different ecological niches. PLoS Pathog. 9 (2). Smith, D.S., Ferris, F.G., 2003. Specific surface chemical interactions between hydrous ferric oxide and iron-reducing bacteria determined using pK(a) spectra. J. Colloid Interface Sci. 266 (1), 60e67. http://dx.doi.org/10.1016/S0021-9797(03)00667-2. Spormann, A.M., 2008. Physiology of microbes in biofilms. Curr. Top. Microbiol. Immunol. 322, 17e36.

133

Sutherland, I.W., 2001. The biofilm matrixean immobilized but dynamic microbial environment. Trends Microbiol. 9 (5), 222e227. Thieme, J., Gleber, S.-C., Selmer, J., 2010. In: Frimmel, F.H., Niessner, R. (Eds.), Nanoparticles in the Water Cycle Properties, Analysis and Environmental Relevance. Virdis, B., Harnisch, F., Batstone, D.J., Rabaeyac, K., Donosea, B.C., 2012. Non-invasive characterization of electrochemically active microbial biofilms using confocal Raman microscopy. Energy Environ. Sci. 5, 7017e7024. Vodnar, D.C., Pop, O.L., Socaciu, C., 2012. Monitoring Lactic acid fermentation in media containing dandelion (Taraxacum officinale) by FTIR spectroscopy. Not. Bot. Horti Agrobot. ClujNapoca 40 (1), 65e68. Wagner, M., Nielsen, P.H., Loy, A., Nielsen, J.L., Daims, H., 2006. Linking microbial community structure with function: fluorescence in situ hybridization-microautoradiography and isotope arrays. Curr. Opin. Biotechnol. 17 (1), 83e91. S0958e1669(05)00198-9 [pii]. Williams, G., 2011. Cluster Analysis in Data Mining with Rattle and R. New York, pp. 179e192. Yamamoto, K., Arai, H., Ishii, M., Igarashi, Y., 2011. Trade-off between oxygen and iron acquisition in bacterial cells at the air-liquid interface. FEMS Microbiol. Ecol. 77 (1), 83e94. http:// dx.doi.org/10.1111/j.1574-6941.2011.01087.x. Yu, F.P., McFeters, G.A., 1994. Physiological responses of bacteria in biofilms to disinfection. Appl. Environ. Microbiol. 60, 2462e2466.

Combination of synchrotron radiation-based Fourier transforms infrared microspectroscopy and confocal laser scanning microscopy to understand spatial heterogeneity in aquatic multispecies biofilms.

Understanding the spatial heterogeneity within environmental biofilms can provide an insight into compartmentalization of different functions in biofi...
1MB Sizes 0 Downloads 3 Views