II-1. Bacterial strain and growth condition
Rhodococcus sp. SD-74 was kindly provided by Professor Nobuhiko Nomura
(University of Tsukuba, Japan) [17, 23]. This strain was pre-cultured on a PMY agar plate at 30 °C for 1.5 days. After that, it was maintained at 4 °C. PMY agar medium contained 1.0%
(v/v) glycerol, 0.5% (w/v) polypepton (Laboratorios CONDA), 0.3% yeast extract (Acumedia), 0.3% malt extract (Bacto), 0.1% KH2PO4, 0.1% K2HPO4, and 1.5% agarose (YEASTERN BIOTECH CO., LTD.). Besides, the pH of the medium was adjusted to 7.0 using KOH.
II-2. Preparation of biofilm samples
A single colony of Rhodococcus sp. SD-74 was picked up from the PMY agar plate and cultured in a tilted glass bottom dish (MatTek; P35G-1.5-20-C) containing 2 mL of TSB medium (Bacto) (Fig. II-1). Rhodococcus sp. SD-74 grew and formed biofilms at the air–medium interface in an incubator in which the temperature was set to 30 °C and high humidity was maintained. Biofilms were allowed to develop for different periods (1, 3, 5, 7 and 9 days) and under different illumination conditions (dark and bright). In the bright condition, a compact fluorescent lamp (Philips; 929689958698) was used to illuminate the biofilm sample with white light (~30 W/m2). The emission spectrum of the lamp is shown in Fig. II-2. Before Raman spectral measurements, excess culture medium was gently removed from the edge of the dish by a pipette, leaving biofilms on the glass bottom dish. This dish
was directly transferred to the microscope stage for subsequent Raman measurements. After measurements, the sample was discarded.
II-3. Confocal Raman microspectroscopy and imaging
Raman spectral acquisition and imaging experiments were performed with a laboratory-built confocal Raman microspectrometer (Fig. II-3) [24]. The 632.8 nm output of a He–Ne laser (Thorlabs) was used as the Raman excitation light. The laser beam was magnified by a factor of ~2.7 in order to cover the exit pupil of the objective used and to better use a high NA of the objective. The expanded beam was introduced to a custom-made inverted microscope (Nikon) by a pair of an edge filter (Semrock) and a hot mirror (KJ). This microscope was modified from a TE2000-U microscope in collaboration with Nikon engineers. The beam was focused onto the sample by an oil-immersion objective (CFI Plan Fluor; 100×, NA = 1.3). Back-scattered light was collected by the same objective and guided along the opposite direction to the incoming path. Rayleigh scattering and anti-Stokes Raman scattering were eliminated by the edge filter with high blocking of OD > 7, and only Stokes Raman scattering was transmitted.
Stokes Raman scattering light was first focused onto a 100 μm pinhole by a 150 mm lens and then collimated by another 150 mm lens. This confocal setup enables the elimination of light coming from off-focal positions and achieved an axial (Z) resolution of 2.4 μm (see below for details about how we obtained this number). Stokes Raman scattering light was
dispersed by an imaging spectrometer (HORIBA Scientific; iHR320) and detected by a back-illuminated, deep-depletion, liquid-N2 cooled CCD detector (Princeton Instruments;
Spec-10:100) with 100 × 1340 pixels operating at −120 °C. A 600 grooves/mm grating was used to cover a wide spectral range (> 2000 cm−1) with an effective spectral resolution of 7 cm−1. For bright-field observation, the sample was illuminated by a halogen lamp and optical images were acquired by a digital camera (Nikon; DS-Ri1) equipped with the microscope.
In Raman imaging experiments, a high-precision piezoelectric nanopositioning stage (PI;
P-563.3CD) equipped on the microscope stage was used to translate the sample horizontally.
In the present study, the sample was translated by 16 μm with a 0.4-μm step in both X and Y directions (41 × 41 pixels). Spectral acquisition was synchronized with sample scanning by the computer program LabVIEW (National Instruments).
For the purpose of avoiding possible photobleaching and heating effects induced by laser irradiation, experimental data were acquired under conditions of low laser power (2.6 mW at the sample point) throughout the present study and sufficiently short exposure time (1 s per pixel) in Raman imaging experiments.
Lateral and axial resolutions were estimated by using intensity profiles of selected Raman bands. Calibration results are shown in Fig. II-3. By scanning the laser spot continuously to cross an interface that is assumed to be infinitely sharp, we can obtain a plot of the intensity of a given Raman band versus the distance. This distance profile of Raman
intensity is fit to a model function f(x). f(x) is equal to a Heaviside function convoluted
result was used to determine the spatial resolution based on the following relation:
σ
In lateral direction, a sharp edge of silicon wafer was chosen as an interface with respect to air for resolution determination. The intensity of a second-order phonon band of silicon at 520 cm−1 was used to plot a calibration curve (Fig. II-4a). In contrast, axial resolution was determined at a glass–indene interface. The intensity of a Raman band of indene at 1018.3 cm−1 was used to plot a calibration curve (Fig. II-4b).II-4. Singular value decomposition
In a Raman imaging experiment, a large number of Raman spectra needs to be acquired within a short period of time (1681 spectra within ~30 min in the present case). To reduce noise in the spectra acquired with a short exposure time, singular value decomposition (SVD) analysis was performed [26, 27]. SVD is a purely mathematical procedure that decomposes a real m × n matrix A according to the following equations (II-3 to II-7): [28]
UWVT
)
The diagonal elements of W are non-negative and called singular values. In this work, the first 500 pixels of Raman spectra, corresponding to 2800–2000 cm−1, were removed to increase computational accuracy and to save time. Raman spectra in the remaining 840 pixels (2000–290 cm−1), which covers the fingerprint region and is of our interest, were arranged according to mapped positions to construct an 840 × 1681 matrix A. The matrices U and VT represent the spectral and positional matrices, respectively. The singular value of each component was used to decide whether the contribution of the component to the original matrix is significant or not. Components of U and VT associated with small singular values were neglected when doing reconstruction of the matrix A. The number of singular values used for reconstruction depends on images, but it was less than 10 in all cases. The SVD was computed in Igor Pro (WaveMetrics) using LAPACK routines.
Biofilm developed in a tilted glass bottom dish
Figure II-1. Scheme of the biofilm sample preparation.
Remove excess medium
Objective
Figure II-2. Emission spectrum of the light source used to illuminate the biofilm sample (see Chapter III-3).
He-Ne laser (632.8 nm)
Laser line filter
Beam expander
Hot mirror
Edge filter Spectrometer
Pinhole (Φ 100 μm) CCD detector
To eyepieces or digital camera
Figure II-3. Laboratory-built confocal Raman microspectrometer.
Sample
Objective
(100× oil, NA=1.3)
(a) (b)
Figure II-4. Calibration curves and fitted results for the laboratory-built confocal Raman microspectrometer in lateral (XY) direction (a) and axial (Z) direction (b). Red lines, observed intensity change; blue lines, best fit to the model function f(x).
Chapter III
Results and Discussion
III-1. Raman spectra of Rhodococcus sp. SD-74: planktonic versus biofilm forms
Under the static culture conditions, Rhodococcus sp. SD-74 shows two types of the living form: planktonic form and biofilms (Fig. III-1). In the planktonic form (Fig. III-1a), the bacterium can move freely in the medium and usually occurs as a single cell unless it undergoes cell division. In contrast, bacterial cells in the biofilm (Fig. III-1b) adhere to the glass surface of the culture dish and often form a colony. To examine chemical compositions of R. sp. SD-74 cells in its planktonic and biofilm states, we measured their Raman spectra (Fig. III-2). The assignments of the Raman bands observed in Fig. III-2 are summarized in Table III-1 [29-31]. The Raman spectrum of the planktonic R. sp. SD-74 cell (Fig. III-2a), although noisy, shows prominent Raman bands at 1649, 1440, 1001, and 782 cm−1. The Raman band at around 1650 cm−1 is assigned to a combination of the cis-C=C stretching mode of lipids and the amide I mode of proteins [30]. The Raman band at ~1440 cm−1 is attributable to the CH2 scissoring (1439 cm−1) and CH3 degenerate deformation (1456 cm−1) modes of both proteins and lipids. The 1001 cm−1 band is due to the ring breathing vibration of phenylalanine residues in proteins. The band at 782 cm−1 is assigned to the O–P–O symmetric stretch plus contributions of cytosine and thymine vibrational modes [31], characteristic of DNA.
The Raman spectrum of the biofilm (Fig. III-2b) shows a clearly distinct pattern from that of the planktonic cell. In addition to the same Raman bands as observed in the planktonic
spectrum, much more intense bands appear at 1004, 1157, and 1516 cm−1. All these features are well-known Raman bands of carotenoids: C=C stretch (1516 cm−1), C–C stretch (1157 cm−1), and in-plane CH3 rocking (1004 cm−1). It is interesting to note that carotenoids are detected only when R. sp. SD-74 forms aggregates.
III-2. Chemical composition and distribution changes during biofilm development
Biofilm development is intrinsically a dynamic process in which not only chemical compositions but also the distributions of the biofilm constituents within the matrix vary drastically with time. To monitor such spatiotemporal evolution of the biofilm constituents during biofilm development in situ, we first performed space-resolved Raman spectral measurements with a sufficiently long exposure time to achieve a high signal-to-noise ratio (S/N). We focus on the behavior of the Raman bands of carotenoids at 1516, 1157, and 1004 cm−1. We then performed multimode Raman imaging to study the spatial distribution of carotenoids as well as other biofilm constituents, and its temporal evolution.
III-2-1. Space-resolved Raman spectra
Raman spectra recorded with a long acquisition time have a high S/N and show clear spectral features. Figure III-3 shows Raman spectra of Rhodococcus sp. SD-74 biofilms grown for different periods (1, 3, 5, 7, and 9 days). These spectra were measured at around the center of the biofilm matrix. Note that the five biofilms were all different. For various reasons, it was difficult for us to trace the development of one particular biofilm for such a long time
as 9 days. We make a comparison between these five spectra based on the spectral pattern and relative Raman band intensity, which represent differences in chemical composition and differences in relative concentration, respectively. First, let us consider the chemical composition change in the biofilm during its development. The five spectra show nearly the same spectral pattern as in Fig. III-2a (see Table III-1 for band assignments) and do not change up to 9 days, indicating that there were no significant changes in both chemical components and their environments in the biofilm.
Second, consider the relative concentration changes of each component in the biofilm.
Remarkably, the intensities of the carotenoid Raman bands increase with the biofilm growth time. This observation indicates that the carotenoid concentration in the biofilm increased with biofilm development. Besides, the band at 782 cm−1, which arises from DNA, shows a similar tendency to those of carotenoids. Namely, their intensity increased slightly as R. sp.
SD-74 biofilm developed. In contrast, bands at 1452 and 1658 cm−1, which are assigned to both proteins and lipids, do not show clear time dependence.
III-2-2. Multimode Raman images
Although the space-resolved Raman spectra shown in Fig. III-3 allow us to glimpse the dynamic changes of the biofilm constituents, they cannot provide information on their spatial distributions. To determine in a quantitative manner how the spatial distributions change with biofilm development, we performed Raman imaging experiments on Rhodococcus sp. SD-74
biofilms at 1, 3, 5, 7, and 9 days. We use band area intensities of six Raman bands at 782, 1004, 1157, 1452, 1516, and 1658 cm−1 to construct a series of biofilm Raman images at 1, 3, 5, 7, and 9 days. The Raman images thus obtained are shown in Fig. III-4, together with the optical images of the biofilms. Because each band exhibits a dramatic variation in area intensity, we adopt different color scales for different Raman images. Red color in Raman images indicates the highest intensity and purple color indicates the lowest intensity. In this way, we are able to compare the concentration change of each component during biofilm development.
As in the previous discussion, we classify six Raman bands into three groups:
carotenoids, DNA, and admixtures of proteins and lipids. Unlike the Raman images of proteins and lipids, which display an almost uniform pattern of concentration distribution within the biofilm, the concentration distributions of carotenoids and DNA are found to be more inhomogeneous. To best see this, compare the Raman images at 7 and 9 days in Fig.
III-4. No appreciable difference is seen between the optical images of the biofilms at 7 and 9 days. However, the corresponding Raman images of carotenoids and DNA do show a complicated, highly inhomogeneous pattern inside the biofilms.
Figure III-5 shows the dependence of the averaged Raman intensities of the three components (carotenoids, DNA, and proteins and lipids) on biofilm development time.
Because the size of biofilm differs from time to time as shown in the optical images in Fig.
III-4, we select the first 500 pixels for each Raman image which give large intensities of the 1516 cm−1 band and average the intensities at those 500 pixels to plot the time dependence.
Error bars shown in Fig. III-5 were estimated from standard deviations of the 500 values. The same averaging protocol was employed for all development times. As shown in Fig. III-5, the concentration changes derived from the Raman images show the same tendency as space-resolved Raman spectra: the concentrations of carotenoids and DNA increase with the development time, whereas that of proteins and lipids does not show a noticeable increase.
At this point, we asked why the concentration of carotenoids increases as the Rhodococcus sp. SD-74 biofilm develops. Carotenoids are common pigments that are widely
found in nature and play important roles in many species [32]. Functions of carotenoids include light harvesting in photosynthesis, antioxidation, and structural factor in purple bacteria [33, 34]. In the course of our investigation, we observed that the color of the culture medium containing R. sp. SD-74 in a tube under the stir conditions became darker with culture time (Fig. III-6). In both stir and static culture conditions, we did not supply fresh medium. This observation leads us to hypothesize that carotenoids observed in Rhodococcus sp. SD-74 biofilm act as an antioxidant and that the increase in their concentration may be a
consequence of some oxidative stress induced by nutrition deficiency in the biofilm during its development.
III-3. Raman imaging of Rhodococcus sp. SD-74 biofilm under distinct light illumination
conditions
To test our hypothesis, we performed Raman imaging experiments on Rhodococcus sp.
SD-74 biofilms that were allowed to develop under different oxidative stress conditions, namely, different light illumination conditions. Intensive light illumination has been shown to generate high levels of reactive oxygen species (ROSs), such as O2−, H2O2, and •OH, in bacteria [35, 36]. These oxygen species are unstable due to the lack of electrons, so they are prone to react readily with most of bio-molecules and lead to oxidation. Once the concentration of ROSs becomes higher than the bacterial defense capacity level, they induce oxidative stress and in some cases cause serious damages to bacterial cells [37-39].
In the experiments, we continuously illuminated biofilm samples with the emission from a compact fluorescent lamp. Because we did not intend to kill the bacteria, light intensity applied was adjusted to be not very high (~30 W/m2) [40]. Control experiments were also performed without light illumination (dark condition).
III-3-1. Multimode Raman images
Figure III-6 displays multimode Raman images of R. sp. SD-74 biofilms developed for 1, 3, and 5 days under dark (control) and bright conditions. As shown in Fig. III-7, there is no appreciable difference in molecular distributions between the two conditions. In contrast, temporal concentration changes are quite different between the two conditions. Using the same method as described earlier, we calculate averaged Raman intensity of each band at each
time. The time profiles of the averaged Raman intensities of the six bands under the dark and bright conditions are compared in Fig. III-8. From Fig. III-8, we find the following:
(1) The concentration of carotenoids increased with a very similar rate in both dark and bright conditions. More importantly, the concentration of carotenoids was always higher under the bright conditions than under the dark conditions.
(2) The Raman bands attributable to proteins and lipids (1452 and 1658 cm−1) appear to be independent of illumination conditions.
(3) The DNA band at 782 cm−1 increased slightly in the control experiment but did not increase when the samples were continuously illuminated by the lamp.
III-3-2. Possible origin of the carotenoid accumulation
In this section, we discuss plausible accounts for the carotenoid accumulation observed in Rhodococcus sp. SD-74 biofilms. According to our hypothesis, carotenoids in the present case are acting as an antioxidant. Because oxidative stress induced in the biofilm is considered to be greater under the bright conditions than under the dark conditions, a larger amount of ROSs might be generated in the former case, which requires a larger number of antioxidants, namely, carotenoids, to protect essential bio-molecules from highly reactive ROSs. Therefore, we can anticipate that the concentration of carotenoids under the bright conditions is higher than under the dark conditions. This is exactly what we see in the top panel of Fig. III-8, and the experimental results support our hypothesis. We conclude that the concentration of
carotenoids and the extent of oxidative stress in Rhodococcus sp. SD-74 biofilms are positively correlated. The increase of carotenoids with biofilm development implies that oxidative stress increases with time under the present culture conditions and that the bacterium utilizes carotenoids to counter this effect.
Table III-1. Band assignments for Raman spectra of Rhodococcus sp. SD-74 in its planktonic
and biofilm forms.
Wavenumber (cm−1) Assignment
782 DNA (O–P–O symmetric stretch plus cytosine and thymine vibrations)
1001 Ring breathing mode of phenylalanine residues 1004 Carotenoids (in-plane CH3 rocking)
1157 Carotenoids (in-phase C–C stretching)
1440 CH2 scissoring and CH3 degenerate deformation 1452 CH2 scissoring and CH3 degenerate deformation 1516 Carotenoids (in-phase C=C stretching)
1649 cis-C=C stretching of unsaturated lipid chains and amide I mode 1658 cis-C=C stretching of unsaturated lipid chains and amide I mode
(a) Planktonic form (b) Biofilm
Figure III-1. Bright-field optical images of planktonic form (a) and biofilm (b) of Rhodococcus sp. SD-74. Scale bar measures 2 μm in a and 10 μm in b.
Figure III-2. Raman spectra of Rhodococcus sp. SD-74 in planktonic form (a) and in biofilm (b).
(b)
(a)
Figure III-3. Space-resolved Raman spectra of Rhodococcus sp. SD-74 biofilms developed for 1 (a), 3 (b), 5 (c), 7 (d), and 9 days (e).
(e)
(d)
(c)
(b)
(a)
782
1004 1157 1516 1452 1658
1 day
3 day
5 day
7 day
9 day
Carotenoids DNA Admixture of
proteins and lipids
Figure III-4. Multimode Raman images of Rhodococcus sp. SD-74 biofilms at the
Raman shift of 782, 1004, 1157, 1452, 1516, and 1658 cm−1 at the development time of 1, 3, 5, 7, and 9 days. The first column shows bright-field optical images of the biofilms studied. All the images presented here have the same size, i.e., 41 × 41 pixels (16 × 16 μm2), for consistency.
Band area intensity
High Low
1516 cm−1 1157 cm−1
1452 cm−1 1658 cm−1
782 cm−1 1004 cm−1
Figure III-5. Time dependence of averaged Raman band area intensities at 782, 1004, 1157, 1452, 1516, and 1658 cm−1 shown in Fig. III-4.
Carotenoids
Proteins plus lipids
DNA
1 day 2 day 3 day Cultured time
Figure III-6. Picture of the culture medium containing Rhodococcus sp. SD-74 under the stir conditions, which shows that the color of the medium becomes darker with culture time.
1 day
3 day
5 day
Carotenoids DNA Admixture of
proteins and lipids
Figure III-7. Multimode Raman images of Rhodococcus sp. SD-74 biofilms at the
Raman shift of 782, 1004, 1157, 1452, 1516, and 1658 cm−1 with the development periods of 1, 3, and 5 days under the dark (a) and bright (b) conditions. The first column shows bright-field optical images of the biofilms studied. All the images presented here have the same size, i.e., 41 × 41 pixels (16 × 16 μm2), for consistency.
(a) Dark
(b) Bright
1516 cm−1 1157 cm−1
1452 cm−1
1658 cm−1
782 cm−1 1004 cm−1
Figure III-8. Time dependences of averaged Raman band area intensities at 782, 1004,
1157, 1452, 1516, and 1658 cm−1 shown in Fig. III-6, under the dark (blue, closed circle) and bright (red, open circle) conditions.
Carotenoids
Proteins plus lipids
DNA 1516 cm−1
1157 cm−1
Chapter IV
Summary
In the present study, we have investigated in vivo the biofilm development process of Rhodococcus sp. SD-74 by using Raman microspectroscopy and imaging. First, we explored
Raman spectral differences between planktonic form and biofilm of this species. We then examined changes with biofilm development time in the concentration of each biofilm component. We found that the concentration of carotenoids within the biofilm drastically
Raman spectral differences between planktonic form and biofilm of this species. We then examined changes with biofilm development time in the concentration of each biofilm component. We found that the concentration of carotenoids within the biofilm drastically