Raman spectral analysis of microbial pigment compositions in vegetative cells and heterocysts of multicellular cyanobacterium

The one-dimensional multicellular cyanobacterium, Anabaena sp. PCC 7120, exhibits a simple topology consisting of two types of cells under the nitrogen-depleted conditions. Although the differentiated (heterocyst) and undifferentiated cells (vegetative cells) were distinguished by their cellular shapes, we found that their internal states, that is, microbial pigment compositions, were distinguished by using a Raman microscope. Almost of Raman bands of the cellular components were assigned to vibrations of the pigments; chlorophyll a, β-carotene, phycocyanin, and allophycocyanin. We found that the Raman spectral measurement can detect the decomposition of both phycocyanin and allophycocyanin, which are components of the light-harvesting phycobilisome complex in the photosystem II. We observed that the Raman bands of phycocyanin and allophycocyanin exhibited more remarkable decrease in the heterocysts when compared to those of chlorophyll a and β-carotene. This result indicated the prior decomposition of phycobilisome in the heterocysts. We show that the Raman measurement is useful to detect the change of pigment composition in the cell differentiation.


Introduction
Cyanobacterium is a gram-negative prokaryote that carries out oxygenic photosynthesis [1]. Among of them, Anabaena sp. PCC 7120 (hereafter named Anabaena) is a filamentous and multicellular cyanobacterium [2][3][4] (Fig. 1A). The filament is composed of a lot of cells connected in a one-dimensional manner.
In nitrogen-depleted culture medium, Anabaena initiates cellular differentiation [2][3][4][5][6][7][8]. The differentiation occurs at an interval of approximately ten cells along the filament. Differentiated and undifferentiated cells are named as heterocysts and vegetative cells, respectively. A heterocyst is easily distinguished by using a standard microscope because it is larger in size and rounder in shape. Once a cell is fully differentiated, it never reverts vegetative cell. As the number of vegetative cells increases by the cell division, a new heterocyst is differentiated approximately midway between two older heterocysts. A heterocyst fixes dissolved nitrogen to the reduced nitrogen species, e.g., ammonium ion [6,7]. In addition, the light-harvesting phycobilisome complex is chemically decomposed or inactivated, leading to suppress the oxygen-evolving photosystem II (PSII) [6][7][8][9][10]. This is because a nitrogen-fixing enzyme, nitrogenase, is sensitive to oxygen. On the other hands, vegetative cells are dedicated for the photosynthesis and contain both PSI and PSII, while heterocysts mainly contain PSI. As the photosynthesis and nitrogen fixation are incompatible in the same cell, the heterocysts and vegetative cells exchange metabolites produced by the nitrogen fixation or photosynthesis with the neighboring cells [4][5][6][7] (Fig. 1B).
A fluorescence microscope has been used to detect the differences of the photosynthetic systems between vegetative cells and heterocysts [9][10][11][12]. However, the fluorescence microscope is difficult to detect pigments such as carotenoids with no or weak fluorescence. Alternatively, a Raman microscope has been used to observe the microbial pigments [13]. The Raman microscope can be applied to detect and differentiate no or weak fluorescent molecules with characteristic fingerprints of molecular vibrational spectra [14][15][16]. This is because the biomolecules, such as atoms combined with electrons, vibrate with different wavenumbers. These oscillators interact with the light and are detected as Raman scattering [14,16]. Especially, the shorter wavelength excitation is useful for resonance effects of the Raman scattering and selectivity of target molecules [14,17]. One disadvantage of the resonance effects is excessive excitation that may cause disturbance of visible light-induced physiological phenomena and/or damages in viability of cells [18,19]. However, it becomes possible to measure Raman spectra of a single cyanobacterial cell by controlling the excessive excitation [20]. Since the weak fluorescent carotenoids for light-harvesting are abundant in cyanobacteria [21], the Raman microscope has been applied to the studies of multicellular cyanobacteria [20,22].
In this study, we analyzed the distribution of biomolecules including small molecules in living vegetative cells and heterocysts in a noninvasive and non-labeling manner using Raman signals. We found that vegetative cells and heterocysts were classified by 16 Raman bands in the Raman spectra by excitation at 785 nm laser. These Raman bands were assigned to vibrational modes of resonance Raman bands of four pigments known as light-harvesting pigments, chlorophyll a, β-carotene, phycocyanin, and allophycocyanin. We also found that the components in the phycobilisome in PSII, that is, phycocyanin and allophycocyanin, were detected by analyzing the high resolution of Raman bands with sharp band widths. We calculated the correlations of band intensities among four pigments in vegetative cells and heterocysts. As a result, the intensities of Raman bands of phycocyanin and allophycocyanin in the heterocysts were remarkably decreased when compared to those of chlorophyll a and β-carotene. Our result shows good correspondence with the previous studies that phycobilisome in PSII is decomposed during the differentiation process [9,10]. Raman spectroscopic data as a function of cell differentiation state help us to analyze the change of the PSII components before and after the differentiation.

Bacterial strains and culture
Anabaena sp. PCC 7120 (wild type) were grown in 25 ml of BG-11 0 (lacking sodium nitrate) liquid medium in 50 ml flasks at 30 • C under illumination with white fluorescent lamps (FL30SW-B, Hitachi co.) at 45 μM photons M − 2 s − 1 . The culture was being shook at 120 rpm until an optimal density at 730 nm (OD 730 ) reached about 0.4-0.5. The liquid culture was washed three times with BG11 0 , diluted to an OD730 of ~0.2, and underlain beneath a fresh BG-11 0 solid medium plate containing 1.5% agar solution (Becton, Dickinson and company, USA) with a bottom dish glass. The sample was placed in a Raman microscope (as mentioned below) kept at 30 • C under illumination with white fluorescent lamps at 45 μM photons m − 2 s − 1 .

Reference pigments
Raman spectra of pigments were obtained by using chlorophyll a (Sigma-Aldrich, C6144-1 MG), β-carotene (Wako Pure Chemical Industries, 035-05531), phycocyanin (Sigma-Aldrich, P2172), and allophycocyanin (Sigma-Aldrich, A7472). To measure the spectra, these pigments were dissolved in BG11 0 liquid medium and underlain beneath a fresh BG-11 0 solid medium plate containing 1.5% agar solution with a bottom dish glass. The condition of measurement was the same as that mentioned above.

Raman microscope and spectral pre-treatments
In Via confocal Raman spectrometer equipped with a CCD camera (inVia Reflex, Renishaw co.) was used to measure the Raman spectrum. The excitation wavelength was at 785 nm. We measured Raman spectra of individual cells by selecting the central points of the cells. A typical Raman spectrum of a small confocal volume in the cytoplasm (horizontal diameter, ~1 μm) of a single living vegetative cell (~3 μm diameter) yields a sufficient signal-to-noise ratio for analysis (~1 s per pixel, with a 785 nm laser at ~20 mW directed at the confocal volume). In this study, the baselines of Raman spectra were corrected. The baseline-corrected Raman spectrum y' (ν) was calculated as y' (ν) = y (ν) − y poly (ν), in which y poly (ν) is a fitted polynomial curve constructed with the following procedures. (i) For a spectrum truncated between the minimum Raman shift position ν min and the maximum position ν max , the degree of the function d was selected to fit the baseline using a polynomial function (this time d = 3). (ii) Using the least squares method, the polynomial function y poly was first fitted to the Raman spectrum y. (iii) The Raman spectrum y was divided into upper and lower parts, relative to the fitted baseline y poly . (iv) The number of data points on the upper side of y was designated N A , and the number on the lower side of y was designated N B . If N A < N B , the upper part of y was removed from the whole of y, and the Raman spectrum y was replaced with the lower part of the spectrum. Then, procedure (ii) was repeated. When N A ≥ N B , the baseline was considered the best fit and optimal.

Raman spectral measurements of vegetative cells, heterocysts, and four pigments
The average Raman spectra of vegetative cells and heterocysts are shown in Fig. 2A and B. The procedure to obtain the Raman spectrum is explained in Method section. Here briefly, fifteen vegetative cells (or fifteen heterocysts) were selected from three Anabaena filaments, and the fifteen Raman spectra were measured for every single cell. The intensity values of a Raman spectrum in the region from 990 to 1770 cm − 1 were normalized to unity. The fifteen normalized spectra of the vegetative cells (or heterocysts) were averaged shown in Fig. 2A and B.
The number of Raman band peaks in vegetative cells was almost identical to that in heterocysts (it was 16 bands), and the band positions in vegetative cells were nearly the same as those in heterocysts (the difference was within 2 cm − 1 ) ( Fig. 2A and B). We found that the intensities of several Raman bands in vegetative cells were different from those in heterocysts. To show the difference of spectral features more clearly, we conducted that the Raman spectrum in heterocysts was (caption on next column) Fig. 2. Raman spectra of living cells and photosynthetic pigments obtained with excitation at 785 nm (A, B) The normalized intensities of Raman spectra of the vegetative cells and heterocysts. To calculate the Raman spectra, fifteen vegetative cells and fifteen heterocysts were selected from three Anabaena filaments, respectively. The Raman bands labeled by arrows in (A) were representative bands assigned to vibrations of chlorophyll a, β− carotene, phycocyanin, and allophycocyanin. These bands were also selected in Fig. 3
subtracted from that in vegetative cells. As a result, the intensities of several peaks were different, i.e., 1240, 1249, 1365, 1371, 1580, 1592, 1629, and 1639 cm − 1 (Fig. 2C). This indicates that these peaks are potential differentiation markers for Anabaena. That is, heterocysts can be distinguished from vegetative cells solely using a Raman microscope.

Comparison of the normalized band intensities between vegetative cells and heterocysts
To elucidate the difference of composition of four pigments between vegetative cells and heterocysts, we compared the normalized intensities of the Raman bands of vegetative cells with those of heterocysts (Fig. 3). The band positions in the Raman spectra of vegetative cells and heterocysts were almost identical to those in the Raman spectra of four pigments. Some Raman bands in the Raman spectra of the living cells were assigned to more than two vibrational modes of resonance Raman bands of pigments. For example, the Raman band at 1188 cm − 1 in the Raman spectrum of chlorophyll a was very closed to that at 1190 cm − 1 in the spectrum of β-carotene. Moreover, the Raman band at 1285 cm − 1 in the Raman spectrum of chlorophyll a was also very closed to that at 1284 cm − 1 in the spectrum of allophycocyanin. Consequently, the sixteen major bands (four times four) were selected for chlorophyll a, β-carotene, phycocyanin, and allophycocyanin.
All the band intensities of chlorophyll a of hetrocysts were decreased comparing with those of the vegetative cells (p < 0.01, Welch's t-test) (Fig. 3A). Similarly, all the band intensities of phycocyanin and allophycocyanin of hetrocysts were decreased comparing with those of vegetative cells (p < 0.005) ( Fig. 3C and D). However, the Raman band intensities of β-carotene were almost identical between the vegetative cells and heterocysts (Fig. 3B), suggesting that β-carotene is chemically stable. It is known that while the amount of β-carotene is constant through the differentiation [9,10,22], the chlorophyll a, phycocyanin, and allophycocyanin are chemically decomposed, leading to impaired PSII. Therefore, we considered that the differences of the Raman bands of the chlorophyll a, phycocyanin, and allophycocyanin between the vegetive cells and heterocysts were caused by the chemical decompositions (a chemical decomposition reduces the amount of the target molecule and the corresponding Raman intensity).
Here, we found that the Raman band intensities of phycocyanin and allophycocyanin were largely decreased after the differentiation when compared to those of chlorophyll a (Fig. 3A, C, and D). We considered that this is because chlorophyll a is a component of both the PSI and PSII, but phycocyanin and allophycocyanin, which form the phycobilisome as a protein complex unit, exist mainly in the PSII. Therefore, the selective decomposition of the PSII in the heterocyst resulted in the remarkable decrease of Raman bands of phycocyanin and allophycocyanin. However, the Raman band intensities of phycocyanin and allophycocyanin were far from zero in the heterocysts, suggesting that phycobilisome existed in the heterocysts. The gap of the values was explained in the Discussion section by referring latest studies. Fig. 2A and B The sixteen major bands (four times four) were selected for the chlorophyll a, β-carotene, phycocyanin, and allopycocyanin, respectively. Gray and black graphs show the normalized band intensities of the vegetative cells and heterocysts at each Raman shift (cm − 1 ). Error bar indicates the standard deviations (n = 15). Statistical difference is indicated by the following symbols. *: p < 0.01, **: p < 0.005.

Pigment compositions and correlations in the same cell
Correlations among intracellular pigment compositions were analyzed by selecting a representative Raman band among the four major bands (Fig. 3) in each pigment. For example, in the case of chlorophyll a, the normalized band intensity at 1328 cm − 1 was highly correlated to the other three normalized intensities at 1046, 1142 and 1535 cm − 1 among 15 vegetative cells (r = 0.71-0.72, Supplemental  Fig. 1). These results suggest that the quantitative variation of chlorophyll a was reflected as the normalized band intensities in the living cells. Hereafter, the normalized band intensity at 1328 cm − 1 was selectively used as the signal of chlorophyll a. In the case of β-carotene, phycocyanin, and allopycocyanin, the normalized band intensities at 1523, 1629, and 1639 cm − 1 were also highly correlated to the other three normalized band intensities in each pigment (r = 0.71-0.74, Supplemental Fig. 1). Thus, the normalized band intensities at 1523, 1629, and 1639 Cm − 1 were used as the representative signals of β-carotene, phycocyanin, and allophycocyanin, respectively.
In vegetative cells, almost no correlations (r = − 0.19-0.03) were found between β-carotene and other three pigments (Fig. 4A, D, and E). That is, the amount of β-carotene was unrelated to that of other pigments in the PSI and PSII. On the other hands, moderate correlations (r = 0.42 and 0.63) were found between chlorophyll a and the other two pigments in the phycobilisome (phycocyanin and allophycocyanin), respectively ( Fig. 4B and C). We consider that this correlation was observed because chlorophyll a, phycocyanin, and allophycocyanin are the components of PSII. Especially, high correlation (r = 0.70) was found between phycocyanin and allophycocyanin (Fig. 4F). This is probably related to the fact that six pieces of phycocyanin aggregates locate outside, surrounding a single piece of allophycocyanin aggregate for a single unit of the phycobilisome (Fig. 1C). That is, the quantitative ratio of phycocyanin to allophycocyanin was tightly regulated in the phycobilisome.
In heterocysts, we observed insignificant correlations (r = − 0.09 and − 0.13) between chlorophyll a and other two pigments in the phycobilisome (phycocyanin and allophycocyanin), respectively ( Fig. 4B and C). Moreover, the correlation between phycocyanin and allophycocyanin was moderate (r = 0.45) when compared to the correlation value in the vegetative cells (Fig. 4F). We considered that the PSII is decomposed in the heterocyst, and therefore, the correlations observed in the vegetative cells were disappeared or much weaker.

Discussion
We measured the Raman spectra of the vegetative cells and heterocysts in Anabaena filaments. The Raman bands in the spectra were assigned to vibrations of chlorophyll a, β-carotene, phycocyanin, and allophycocyanin. In the heterocysts, the Raman band intensities of chlorophyll a, phycocyanin, and allophycocyanin were significantly decreased (Fig. 3A, C, and D). We considered that this is because chlorophyll a, phycocyanin, and allophycocyanin are chemically decomposed with the PSII unfunctionalized through the differentiation. Especially, in the heterocysts, the band intensities of phycocyanin and allophycocyanin were remarkably decreased when compared to those of chlorophyll a. That is, we can distinguish the vegetative cells and heterocysts by analyzing the band intensities of phycocyanin and allophycocyanin. In the heterocysts, it is known that phycobilisome in PSII was decomposed [9,10]. Therefore, we considered that the decrease of the band intensities of phycocyanin and allophycocyanin was due to the decomposition of phycobilisome in the heterocysts.
In heterocysts, efficient energy transfer was proposed to occur from phycocyanin to PSI to facilitate nitrogen fixation and other reactions [36], but the transfer mechanism and structural details had been unknown. Watanabe et al. reported an isolation of a unique phycobilisome-PSI supercomplex from heterocysts of Anabaena sp. PCC 7120 [37]. Biochemical and spectral analysis revealed that phycocyanin was included mainly in this type of phycobilisome, which was functionally connected to the PSI tetramer via a new connecting component, CpcL [37]. Watanabe et al. explains such new type of antenna complex may play an important role on light harvesting in PSI-driven cyclic electron transport to facilitate nitrogen-fixation and other reactions [37]. Therefore, phycocyanin was not decomposed perfectly and continued to exist in heterocysts. Our analysis corresponded to this result well because the band intensities of phycocyanin and allophycocyanin were not zero in the heterocysts, suggesting that phycobilisome was not perfectly decomposed.
Moreover, Watanabe et al. reported that allophycocyanin was decomposed in preference to phycocyanin [37]. Here, we focused on how the normalized Raman band intensities of phycocyanin and allophycocyanin were different in the vegetative cells and heterocysts. After the differentiation, the four band intensities of phycocyanin were decreased by 37 In the calculated percentages were large, all the mean percentages for allophycocyanin were larger than those for phycocyanin. Our analysis suggests that the Raman bands of allophycocyanin were more decreased after the differentiation when compared to those of phycocyanin. That is, Watanabe et al. and our study suggest that phycocyanin and allophycocyanin were decomposed or restored to PSI by different mechanisms, respectively. As a future work, it should be addressed how phycocyanin and allophycocyanin are liberated from PSII, and how they are decomposed or restored in the heterocysts.
As mentioned in the Introduction section, the Raman microscope can detect no or weak fluorescent molecules in living cells differently from the fluorescence microscope. From the Raman spectral analysis, we revealed that the Raman band intensities of β-carotene were not significantly different after the differentiation. This result suggests that β-carotene is little decomposed in heterocysts. Moreover, the Raman spectral analysis enabled us to analyze the correlations among four kinds of pigment compositions in vegetative cells and heteroysts. Sugiura et al. also reported the quantitative correlations among microbial pigments in vegetative cells and heterocysts of the other multicellular cyanobacterium, Nostoc sp., by using a confocal laser microscope [38]. However, the confocal laser microscope cannot detect the carotenoid molecules, and therefore, they measured three kinds of pigments (chlorophyll a, phycocyanin, and allophycocyanin). The advantage of using the Raman microscopic analysis is to measure not only the strong fluorescent molecules but also the weak fluorescent molecules comprehensively.
In this study, we distinguished the differentiated cells (heterocysts) from the undifferentiated cells (vegetative cells) by analyzing the pigment composition from the Raman spectra. The measurement of the Raman spectrum shown in this study is a useful technique to analyze the intracellular chemical composition without external probes. This methodology could have the potential to apply to other purpose, for example, the selection of focal bacterial cells from environments.

Author contributions
JI conceived the project, performed the experiments, and analyzed the data. JI and HT wrote the manuscript and approved the submitted version.

Declaration of competing interest
The authors declare that they have no conflicts of interest.

Data availability
Data will be made available on request.