Detection of biogenic amorphous calcium carbonate (ACC) formed by bacteria using FTIR spectroscopy

: 19 While the formation of intracellular amorphous calcium carbonate (ACC) by living organisms is 20 widespread, its detection in prokaryotes remains difficult owing to its susceptibility to transform 21 or dissolve upon sample preparation. Because of these challenges, a large number of ACC-22 forming prokaryotes may have been undetected and their abundance in the natural environment 23 is possibly underestimated . This study identifies diagnostic spectral markers of ACC-forming 24 prokaryotes that facilitate their detection in the environment. Accordingly, ACC formed by 25 cyanobacteria was characterized using Fourier transform infrared (FTIR) spectroscopy in near-IR, 26 mid-IR, and far-IR spectral regions. Two characteristic FTIR vibrations of ACC, at ~860 cm −1 and 27 ~306 cm −1 , were identified as reliable spectral probes to rapidly detect prokaryotic ACC. Using these 28 spectral probes, several Microcystis strains whose ACC-forming capability was unknown, were 29 tested. Four out of eight Microcystis strains were identified as possessing ACC-forming capability 30 and these findings were confirmed by scanning electron microscopy (SEM) observations. Overall, 31 our findings provide a systematic characterization of prokaryotic ACC that facilitate rapid detection 32 of ACC forming prokaryotes in the environment, a prerequisite to shed light on the role of ACC-33 forming prokaryotes in the geochemical cycle of Ca in the environment.


Introduction
Compared to the crystalline polymorphs of calcium carbonate (CaCO 3 ) such as calcite, vaterite and aragonite, amorphous calcium carbonate (ACC) is more soluble and therefore fairly unstable except in several biotic cases 1 .Biogenic ACC was first documented in eukaryotic organisms such as mollusks, sea urchins, sponges, ascidians, and crustaceans, where ACC acts as a precursor phase for the formation of calcite and aragonite minerals in skeletal architectures [1][2][3][4][5][6] .
Several studies have shown that diverse prokaryotic organisms are also capable of forming ACC.
For example, the large uncultured sulfur bacterium Achromatium oxaliferum, widely distributed at the oxic-anoxic boundary in sediments of freshwater, brackish, and marine environments, was shown to form numerous intracellular ACC inclusions filling most of the cell volume 7 .Recently, a magnetotactic bacterium affiliated to the Alphaproteobacteria was shown to form intracellular ACC 8 .The repertoire of prokaryotes forming ACC also includes numerous phylogenetically diverse cyanobacteria 9,10 .Detection of biogenic ACC in cyanobacteria was particularly interesting because they are the only known ACC-forming prokaryotes that are culturable in the laboratory.These strains have the highest Ca demand among all cyanobacteria, hypothesized to be driven by their need to form intracellular ACC, which suggests a biological function of ACC.This role possibly consists in an intracellular pH buffer or a storage form of inorganic carbon and/or Ca 11,12 .The formation of ACC by bacteria has significant geochemical implications.First, intracellular ACC inclusions formed by some strains were shown to be associated with their capability to strongly accumulate radioactive alkaline earth elements such as 90 Sr and 226 Ra, offering some potential for remediating pollutions by these elements 13 .Moreover, some strains of ACC-forming cyanobacteria are shown to sequester high concentrations of alkaline earth elements such as Ba and Sr.This raises questions about the current understanding of the biological role of Ba/Sr in cyanobacteria and, in reverse, the role of these cyanobacteria in the geochemical cycling of Ba and Sr, especially in environments where they might be abundant 11,14- 16 .The biogeochemical implications of bacterial ACC are not only limited to cyanobacteria.
Owing to the exceptional size of A.oxaliferum, often greater than 17 µm in length and 10 µm in width, ACC within A. oxaliferum cells can represent most of the solid calcium present in some sediments 17 .
Despite the increasing number of reports about ACC-forming prokaryotes and their biogeochemical significance, detection of ACC in prokaryotes remains a difficult task.Pure abiogenic ACC is highly unstable, transforming within minutes to crystalline calcium carbonate polymorphs such as calcite, or aragonite 18 .In contrast, ACC found in prokaryotes remains relatively stable intracellularly with no obvious spontaneous transformation to crystalline phases such as calcite.The mechanisms underlying such an unprecedented stability of biogenic ACC remain unknown.Various organic or inorganic additives (Mg, P) and/or confinement of ACC within small volumes, have been shown to stabilize ACC [19][20][21][22] .This stabilization process is interrupted upon cell lysis, or cell death, which therefore induces ACC loss.Techniques such as scanning electron microscope (SEM) with energy dispersive X-ray spectrometry (EDXS), transmission electron microscope (TEM) with EDXS, and Raman microspectroscopy (but not for ACC in autofluorescent cyanobacteria) have been successfully used to detect ACC in bacteria 23,24 (Table 1).While these techniques are interesting as they provide information at the single-cell scale, their use is time-consuming, which can become prohibitive when several samples (e.g., a screening of numerous bacterial strains) require a preliminary confirmation of the presence or absence of ACC, and when a more global average view of the ACC content of a sample is needed.Moreover, it has been shown that bacterial ACC can be lost and/or transformed upon some sample preparation protocols (Table 1) 8,23 .. Because of these challenges, a large number of ACC-forming prokaryotes may have remained undetected and their abundance in the natural environment possibly underestimated.Overall, there is thus a need for the use of techniques allowing efficient detection of ACC in these complex samples with minimal sample preparation.
6][27][28] ), capable of rapidly detecting CaCO 3 polymorphs in situ, non-destructively, and without complex sample preparation, making it an ideal approach for detection of ACC in prokaryotes at a bulk scale.FTIR spectroscopy has been a popular choice to study the kinetics of ACC crystallization in eukaryotic biominerals (e.g. 2,291][32] ).In the mid-IR region, crystalline carbonate phases such as calcite have distinct bands at ~714 cm −1 (Ʋ 4 ), ~866 cm −1 (Ʋ 2 ), ~1084 cm −1 (Ʋ 1 ) and 1420-1470 cm −1 (Ʋ 3 ).ACC shares Ʋ 1 -Ʋ 3 vibrations with crystalline CaCO 3 but lacks the distinct vibrational band at ~714 cm −1 and has an additional shoulder on the Ʋ 3 band 33 .The lack of Ʋ 4 band in ACC along with the characteristic Ʋ 3 shoulder have been used extensively as a diagnostic feature to detect   Summary of techniques used for characterization of ACC in bacteria and their limitations biogenic ACC present in various eukaryotic carbonates such as spicules of ascidian 6 , spicules of sea urchin 29 , fish-gut carbonates 34 , and earthworm secreted carbonate granules 35 .In eukaryotic carbonates, where crystalline CaCO 3 coexists with ACC, the Ʋ 3 / Ʋ 4 and Ʋ 2 / Ʋ 4 peak intensity ratios have been shown to provide semi-quantitative information on the fraction of amorphous and crystalline phases 18 35 .Unlike the mid-IR spectral range, the far-infrared (FIR) spectral region (80-650 cm -1 ) offers valuable insights about lower energy lattice vibrations that correspond to vibrations among different carbonate units 36 .Brusentsova et al. compiled a comprehensive dataset of FIR spectra measured on several reference crystalline carbonate polymorphs, but no dataset exists for synthetic or biogenic ACC 37 .While FTIR spectroscopy has been widely used as a tool to detect ACC in eukaryotic carbonates, the application of FTIR spectroscopy to the detection of prokaryotic ACC is limited and not straightforward 38 .Unlike eukaryotic carbonates that are typically composed of a large fraction of crystalline carbonate with ACC and low organic content, prokaryotic ACC consists of a low ACC content mixed with high organic content.Indeed, they are commonly found at low abundance in the environment and surrounded by a complex assemblage of cellular biomolecules 38,39 .The high organic content could mask the characteristic asymmetric vibration (Ʋ 3 ) band in the region 1500-1400 cm −1 and symmetric vibration band (Ʋ 4 ) at 714 cm −1 due to overlapping bands from proteins (amide I and II peaks) and rocking vibration of methylene group of long-chain alkanes 40 .Such interferences may prevent the direct use of diagnostic indicators of ACC (e.g., absence of Ʋ 4 band), previously established based on eukaryotes [39][40][41] .Mechanically separating the organics from ACC in prokaryotes does not appear as a viable approach as this may result in the rapid transformation of ACC outside the cell, as mentioned previously.Owing to these complications, the question remains as to what IR spectral features of ACC are suitable for its detection in systems dominated by prokaryotes?Within this context, we conduct a comprehensive study of the key spectral features of ACC across the near-IR, mid-IR and far-IR, spectral range in ACC-forming cyanobacteria and test these features in a set of cyanobacterial strains whose ACC-forming capabilities remain unknown.

Synthesis of abiotic ACC
An abiotic amorphous calcium/magnesium carbonate (referred as ACC hereafter) was prepared in-house following the protocol described in 42 .The sample was precipitated by mixing two 100 ml solutions: (i) an aqueous carbonate solution containing 0.05 M of K 2 CO 3 (pH = 11.4); and (ii) an aqueous calcium and magnesium solution containing 0.045 M of CaCl 2 .2H 2 O (Sigma-Aldrich) and 0.053 M of MgCl 2 .6H 2 O (adjusted to pH = 8.6).The calcium-and magnesiumcontaining solution was poured directly into the carbonate solution.The resulting precipitate was isolated by centrifugation, washed twice with anhydrous ethanol, and dried at 37°C for 7 days.
Bulk mineralogy of the dried powder was analyzed using a Panalytical powder X-ray diffractometer, equipped with a Co-anode X-ray source.Patterns were recorded between 2theta=3-70 degrees, with 0.001 degree steps and 2 milliseconds/step.The characterization of synthesized ACC using X-ray diffraction (XRD) showed only broad peaks, confirming that the synthesis resulted in an amorphous phase (Fig. S1).Synthesized ACC was stored at room temperature in a desiccator.The synthesized ACC remained amorphous for at least 1 year, as confirmed by XRD (data not shown).Thermogravimetric analysis-differential scanning calorimetry (TGA-DSC) analyses were performed using a SDT Q 600 TA.About 15 mg of powder were placed in an aluminum crucible and the sample was heated up to 900 o C under a 100 ml-per-minute nitrogen flow.The temperature scan rate was set at 5 o C per minute.Mass spectrometry analyses were coupled with the TGA-DSC analyses using a ThermoStare GSD 301 T3 (Pfeiffer Vacuum).The signal corresponding to 44 amu (CO 2 ) and 18 amu (H 2 O) were recorded.

Culture conditions for cyanobacteria strains
Here, we will refer to strains forming intracellular ACC as ACC+ strains and strains not forming intracellular ACC as ACC-strains.Two reference strains, Synechocystis PCC 6803 (ACC-) and Cyanothece PCC 7425 (ACC+) were cultured in the BG-11 medium, a standard growth medium used for freshwater cyanobacteria and the chemical composition of which is provided in Stanier et al. 43 .Cultures were grown with agitation and continuous light (5-10 µmol m -2 s -1 ).When cell suspensions reached an optical density of ~1, cells were harvested by centrifugation, rinsed, and dried at 40 o C for 48 hours.In addition, Microcystis spp.strains from the Paris Museum Collection (PMC) were included in this study as strains whose capacity to form intracellular ACC was initially unknown.Eight different Microcystis strains were cultured to reflect some phenotypic diversity.These strains were grown in the standard BG-11 medium, without agitation, at 23 °C, with a 12:12 photoperiod and a light intensity of ~ 5 µmol m -2 s -1 .The cells were harvested via centrifugation and dried at 40 o C for 48 hours.The total Ca concentration of the dried biomass was measured by acid digestion of ~5 mg of dry biomass and subsequent analysis of the digestate using inductively coupled plasma -optical emission spectrometry (ICP-OES).

SEM analysis
Microcystis cell suspensions were filtered on 0.22 µm polyethersulfone (PES) filters, then rinsed three times with 10 mL MilliQ® water.The filters were dried at ambient temperature and mounted on aluminum stubs using double-sided carbon tape.The samples were carbon coated prior to SEM analyses.Scanning electron microscopy (SEM) observations were performed using a Zeiss Ultra55 SEM at IMPMC.The electron accelerating voltage used was 10 or 15 kV with a working distance of 7.5 mm and a 60 μm aperture.Images were collected with a backscattered electron detector (AsB detector) and the maps of the different elements (Ca, P, K, Mg, Si, Fe) were retrieved from energy dispersive X-ray spectrometry (EDXS) hyperspectral data (HyperMap).

FTIR spectroscopy analyses
The abiotic ACC and dried biomass of the different strains were finely ground in an agate mortar.Attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectra were acquired over the mid-IR region (4000-600 cm −1 ) using a NICOLET 6700 FTIR spectrometer equipped with a diamond internal reflection element (IRE), a KBr beam splitter and a mercury cadmium telluride (MCT) detector.The sample was mounted on the diamond IRE.Spectra were accumulated by collecting 64-100 scans with a resolution of 1 cm -1 .Spectra processing including baseline correction, normalization, and peak fit was carried out using the Origin software™.Baseline corrected mid-IR spectra were fitted using a peak analyzer (based on the Levenberg-Marquardt algorithm).All biomass spectra were fitted in the 945±5 to 845±5 cm −1 frequency range.The initial peak position was determined by the local maximum method, which finds the frequency corresponding to the local maximum within the 20-point range.The spectra were fitted with linear combinations of Voigt lineshapes, the centroids of which were initially set to values determined from the local maxima peak analysis and remained unfixed during the fit unless noted otherwise.The baseline was fixed for the fit (cf.section 3.3).The area under the spectrum and above the background was used for spectral normalization (cf.section 3.4).
For the near-infrared analyses, bi-directional reflectance spectra were obtained using the SHADOWS instrument 44 in the 500-4000 nm (20000-2500 cm −1 ) spectral range.Samples were deposited in a cylindrical sample holder (13 mm in diameter, 5 mm in depth) and measured under a standard observation geometry (incidence=0°, emergence=30°, phase=30°).Spectra were calibrated using the spectral reflectance spectralon TM and infragold TM and corrected for the nonlambertian behavior of spectralon.Last, for far infrared analysis, approximately 6 mg of dried sample were mixed with 80 mg of polyethylene (PE).Pellets were formed in a vacuum press under ten tons of pressure for ~5 min, and spectra were acquired in transmission mode using the Bruker IFS 66v/s FTIR spectrometer in the 600-50 cm −1 spectral range, at a resolution of 4 cm -1 and using a deuterated triglycine sulfate (DTGS) detector and a multilayer mylar beam splitter.
No baseline correction was performed because it was difficult to define and subtract a baseline accurately.S1.The X-axis is in wavelength, where 500 nm corresponds to 2x 10 4 cm −1 and 4000 nm corresponds to 2.5 x 10 3 cm −1 .

Spectral features of synthetic ACC across near-IR, mid-IR and far-IR
The characteristic peaks of synthetic ACC across the spectral range of 20,000-50 cm −1 are shown in Fig. 1 and summarized in Table S1.The band assignments of the IR peaks observed in ACC are listed in Table S1.In the mid-IR spectral region, ACC is specifically indicated by the presence of a shoulder at 1474 cm −1 and the absence of a peak at 712 cm −1 (Fig. 1, Inset 1-2) 32 .This spectral range also showed for ACC a broad band between 2950 and 3700 cm −1 (O-H stretching) and at 1650 cm −1 (O-H bending), both corresponding to structural water within ACC 32 .Consistently, the water content in ACC was estimated to 17 wt% (corresponding to 0.94 mole of water per carbonate) based on TGA-DSC-MS analyses (Fig. S2).The out-of-plane bending absorption (Ʋ 2 ) at ~860 cm −1 (FWHM=25 cm −1 ) was observed.This band is sensitive to differences in chemical environment changes from amorphous to crystalline forms 30 .Indeed, with the transformation of an amorphous phase to a crystalline phase, the Ʋ 2 band position shifts from 860 cm −1 (ACC) to 870 cm −1 (calcite) (Fig. 1A).The Ʋ 2 band position is also sensitive to the Mg content in ACC: the band shifts towards lower wavenumbers with increasing Mg content in ACC and can vary between 860-864 cm −1 20,27 .The remaining bands in the mid-IR spectral range were characteristic vibrations of the carbonate ion and are common in both ACC and crystalline polymorph 32 .
In the FIR region, ACC shows a broad peak around ~306 cm -1 and a weak shoulder on this broad peak at ~212 cm −1 (Fig. 1B).For comparison, calcite FIR spectra show several sharp peaks at 310 (with a shoulder at 361), 280, 220, 107 (shoulder at 96) cm −1 , consistent with previous studies (e.g. 36,37).These sharp bands are due to IR-active collective lattice vibrational modes that involve relative translations of the calcium and carbonate ions.In contrast, ACC is rotationally disordered which results in a broad envelope across the available frequencies 45 .Lastly, in the nm (Fig. 1C).The peaks <2500 nm are attributed to O-H stretching overtones originating from structural and/or weakly bonded water 46 .Above 2500 nm the reflectance is close to zero, because of the saturation of the fundamental OH stretching mode.Moreover, the typical carbonate vibrations in region >2500 nm (see calcite NIR spectra in Fig. 1C) are masked by strong waterrelated absorption bands in ACC 47,48 49 .Overall, due to the presence of water in ACC, the NIR region offers no diagnostic spectral features of ACC.
Several key insights can be drawn from the above analyses.All bands of ACC in the NIR region are masked by structural water.Almost all biological occurrences of ACC are known to be hydrous, and thus, NIR is not a particularly useful IR fingerprinting region for ACC.In contrast, mid-IR and far-IR regions offer promising diagnostic spectral features that we test below to detect ACC in cyanobacteria.Numbers on the spectra correspond to the band# in Table 2.

FTIR spectra of ACC+ and ACC--cyanobacteria in the mid-IR and far-IR ranges
Several absorption bands were identified in the mid-IR spectra of Cyanothece PCC 7425 (ACC+) (Fig. 2).The spectral band assignment for the ACC+ and ACC-strains was achieved based on a comparison and cross-referencing against existing FTIR literature about photorophic microorganisms ( 41,(50)(51)(52)(53) ) and is summarized in Table 2. Bands [1-10] were present in both Cyanothece PCC 7425 and Synechocystis PCC 6803 and were attributed to non-ACC components.These components included O-H vibrations derived from water (band 1), methyl and methylene group vibrations from lipids and fatty acids (band 2), amide I and amide II from proteins (bands 3,4), asymmetric bending of methyl groups of lipids, and symmetric stretching from C-O group in carboxylic acids (bands 5,6), a combination of asymmetric stretching from P=O group of nucleic acids and/or polyphosphates and C-O stretching vibrations from polysaccharides (bands 7, 8, 9, 10) 50,51,54 .The bands between 2500-2000 cm −1 are due to ambient CO 2 and insufficient purging of the sample chamber.Cyanothece PCC 7425 had an additional absorption band present at ~860 cm −1 (band 11) which was absent in Synechocystis PCC 6803.The peak position of band 11 coincides with the ACC peak at 860 cm −1 (Fig. 2).
Moreover, the FWHM of the fitted peak at 860 cm −1 in Cyanothece PCC 7425 was 25.1 cm −1 , matching that of the 860 cm −1 peak of ACC.No absorption band was found around ~700 cm −1 region in Cyanothece PCC 7425 suggesting the absence of crystalline carbonate polymorphs.This is consistent with the under-saturation of the BG-11 growth medium with respect to carbonate phases, which does not allow the precipitation of extracellular crystalline carbonates in the growth solution (Fig. S3).Spectral features of ACC apart from the860 cm −1 , namely 3600-2700 cm −1 (O-H stretching), 1650 cm −1 (O-H bending), 1470-1420 cm −1 (Ʋ 3 ) and 1084 cm −1 (Ʋ 1 ) had strong interference with the lipids and water bands (1, 2), protein bands (3, 4), carboxylic Table 2: Wavenumbers and band assignment for characteristics vibrations found in the mid IR spectra of Cyanothece PCC 7425 and Synechocystis PCC 6803.The band assignment was primarily based on the 50,51,54 .
In the NIR region, the spectra of Cyanothece PCC 7425 and Synechocystis PCC 6803 looked almost identical, which is not surprising based on the lack of ACC features in the NIR region due to superimposition with structural water (Fig. S4).The different peaks found in the NIR spectra of cyanobacteria were primarily attributed to lipids, proteins, and water (Fig. S5) in accordance to Vaidyanathan et al. 55 .
In the FIR region, broad features around 150 cm −1 were observed in both ACC+ and ACCcyanobacteria strains, possibly corresponding to torsional frequencies of primary amines 56,57 .
Figure 3: FIR spectra of the ACC+ and ACC-cyanobacteria.The peak labeled marked as "a" denotes the location of the characteristic features of ACC in FIR at ~306 cm −1 .This peak was at ~320 cm −1 in Cyanothece PCC 7425 (ACC+).
Interestingly, FIR spectra of Cyanothece PCC 7425 showed a broad but strong peak around ~320 cm −1 , which was absent in Synechocystis PCC 6803 (Fig. 3).The asymmetric line shape is characteristic for samples prepared with PE as filling material, and it is not found in the spectra of samples prepared without PE 58 .The position of this peak coincides with the broad peak observed in FIR spectra of synthetic ACC (Fig. 1), suggesting that the peak might be due to ACC in Cyanothece PCC 7425.
Comparing the mid-IR and far-IR spectra of ACC+ and ACC-cyanobacteria strains, only the Ʋ 2 band at 860 cm −1 and the broad peak at ~306 cm −1 (characteristic vibrations of ACC) are free from overlaps with spectral features attributed to organics.Consistently, they are present in ACC+ cyanobacteria, while absent in ACC-cyanobacteria.Indeed, ACC features in spectral region >900 cm −1 are masked by vibrations from organics, suggesting that spectral features of ACC in the spectral region > 900 cm −1 are not reliable diagnostic indicator of presence of ACC in a sample dominated by organics.Such complex combination between ACC bands and organic bands is shown to be absent in organic-poor eukaryotic ACC, where the intense vibrational bands from ACC mask the bands from the matrix biomolecules, making the ACC mid-IR bands in regions > 900 cm −1 detectable 59 .Based on these results, we argue that the use of spectral features of ACC in region > 900 cm −1 to confirm ACC in an organic-rich matrix could result in some misleading results.For instance, Enyedi et al. used the presence of bands at 1494 cm −1 and at 1066 cm −1 in an organic-rich bacterial extra-polymer-substance (EPS) to infer the presence of ACC in bacterial EPS 38 .Fortunately, this interpretation was further supported by other complementary techniques (electron diffraction and Raman spectroscopy).However, we note that the interpretation of the FTIR data by Enyedi et al. is not accurate, since the characteristic 20 peaks of ACC at 1494 cm −1 and at 1066 cm −1 could be attributed to the vibrations from organic components present in the samples, such as proteins, nucleic acid, and phosphorylated polysaccharides.
Whether these spectral features apply to a broader diversity of prokaryotes remains an open question.The characteristic peaks of major biomolecules (e.g., proteins, polysaccharides) can be expected to remain the same across different prokaryotic taxa, but some minor differences have been reported due to solution chemistry, growth conditions and taxonomy 41,[60][61][62] .In the next section, we assess the effect of physiological differences on the above-identified ACC spectral features by surveying different cyanobacterial strains affiliated to the same Microcystis genus.

1
Figure 5: Peak fit of Microcystsis strains in the frequency region 940-840 cm -1 .For strains PMC 569, PMC 810 and PMC 828, best fit was achieved by using 2 band components (~918 cm −1 , 882 cm −1 ); for PMC 824, best fit was achieved by using 3 band components (918 cm −1 , 882 cm −1 and 855 cm −1 ); for strains PMC 814, 820 and 831, 3 band components (918 cm −1 , 882 cm −1 and 860 cm −1 ) yielded best fit.The peak positions of the fitted bands are noted in table 3 and fit quality summarized in the table S2 While ACC formation has been evidenced in a broad diversity of cyanobacteria 10 , Microcystis has so far been poorly studied in this perspective.Here, we acquired FTIR spectra of several different Microcystis strains in the mid-IR region to investigate if some of them may form ACC (Fig. 4).Characteristic bands of proteins, lipids, polysaccharides, and phosphate groups (bands 1-10) were observed in the 4000-900 cm −1 spectral region in all Microcystis strains, similarly to what was observed in Cyanothece PCC 7425 and Synechocystis PCC 6803 (Fig. 4A).The band location and assignment were in good agreement with the previously reported bands in Microcystis in the 4000-1000 cm −1 frequency range 41 .Visual inspection of Microcystis strains in the 950-800 cm −1 region reveals that all Microcystis spectra had a clearly defined band at ~918 cm −1 , labeled as band 10 in Fig. 4C.Additionally, a broad peak at ~880 cm −1 was clearly observed in Microcystis PMC 569, PMC 810, PMC 824, and PMC 828, while this band appeared visually as an asymmetric broad band for the other Microcystis strains (PMC 814, PMC 820, PMC 827 and PMC 831) (Fig. 4C).A sharp peak at ~855 cm −1 was observed in Microcystis PMC 824 only.In Microcystis PMC 820, PMC 827 and PMC 831, a band close to 860 cm −1 was observed.Unlike what was observed for Cyanothece PCC 7425, this band, possibly affiliated to the 860 cm −1 peak of ACC, was not well-defined and overlapped with other neighboring peaks in the 950-800 cm −1 spectral range (Fig. 4B).
To overcome the challenge of overlapping bands, we fitted the experimental spectra of all Microcystis strains in the 950-800 cm −1 frequency range with a combination of four (Voigt) peak components: fit peak 1 at ~918 cm −1 and fit peak 2 at ~880 cm −1 , fit peak 3 at ~860 cm −1 , and fit peak 4 at ~855 cm −1 .Fig. 5 shows the best fits for different Microcystis strains, and the peak properties are listed in table 3. The fit quality parameters are listed in Table S2.Based on the curve fit, we find that fit peak 1 at ~918 cm −1 and fit peak 2 at ~880 cm −1 were present in all Microcystis strains (Table 3).The fit peak 1 at ~918 cm −1 was also present in Synechocystis PCC 6803 and Cyanothece PCC 7425 (peak labeled 10 in Fig. 2 and Fig. 4) and is attributed to a combination of asymmetric stretching from the P=O group of nucleic acids and/or polyphosphates and C-O stretching vibrations from polysaccharides.Intra-strain variations in the intensity of fit peak 1 at ~918 cm −1 could potentially be due to variations in the polyphosphate and polysaccharide content of the cells.Polyphosphates and polysaccharides serve as carbon, phosphorus and energy reserves, and their content is known to vary during the growth of the cyanobacteria 41,52 .The fit peak 2 at ~880 cm −1 could match with vibrations of crystalline carbonates.However, no peak at ~714 cm −1 was found suggesting the absence of such phases.
Alternatively, fit peak 2 may be assigned to bicarbonate 67 .Cyanobacteria are known to accumulate high amount of intracellular bicarbonate owing to carbon concentrating mechanisms 68 .The bicarbonate band is generally weak and sensitive to the degree of hydration of bicarbonate ions, i.e. the band is red shifted and becomes broader with increasing hydration 69 .Unfortunately, it is not possible to further confirm this band assignment based on other bands characteristic of bicarbonate since they overlap with carbonate vibrations and/or are masked by vibrations of biomolecules.A third possible affiliation of fit peak 2 can be made as strong =C-H out-of-the plane bending vibrations of alkenes 54 .The other peaks characteristics of alkenes are stretching =C-H vibrations at 3100-3000 cm −1 and C=C stretching vibrations at 1660-1500 cm −1 , both of which overlap with band #1 (attributed to water) and band #3, #4 (attributed to proteins) in all Microcystis strains 70 .Alkenes typically form part of the structure of fatty acids, and/or carotenoid pigments (commonly present in photosynthetic cells and one of the main components of the phytoaccumulation complex 70,71 ).Last, fit peak 2 at ~880 cm −1 could be due to the P=O group vibrations of polyphosphates as proposed by Wang et al. 72 .This assignment is also consistent with the presence of other characteristic bands (bands 7, 8, 9, 10) of polyphosphates observed in all the Microcystis strains (Table 2).As a summary, fit peak 2 may result from vibrations in bicarbonates, alkenes and/or polyphosphates.Interestingly, fit peak 4 at 855 cm −1 was only found in Microcystis PMC 824 and had a relatively narrow FWHM (8 cm −1 ) in comparison to the 860 cm −1 (Ʋ 2 ) band of ACC (FWHM=25 cm −1 ) suggesting that the band at 855 cm −1 is not linked to the 860 cm −1 band of ACC.Nitrates and chloride salts, contained by the BG-11 growth medium are known to give rise to a sharp peak at ~855 cm −173 .Thus, the band at ~855 cm −1 could be assigned to nitrate salts from the growth medium and may result from insufficient rinsing of the Microcystis PMC 824 cell suspension.
Last, peak fit of the Microcystis spectra revealed that fit peak 3 at 860 cm −1 was only present in strains PMC 814, PMC 820, PMC 827, and PMC 831 (Fig. 5 and Table 3).The overall quality of the fit for these strains was poorer when fit peak 3 was not included in the fit components (Fig. S5 and Table S2).The FWHM of fit peak 3 at 860 cm −1 in Microcystis strains was slightly narrower (~ 21 cm −1 ) but still comparable with the 860 cm −1 band of synthetic ACC (25 cm −1 ) (Table 3).By contrast, poor fit quality was obtained upon fitting a band at 860 cm -1 in the experimental spectra of the PMC 824, PMC 569, PMC 810, and PMC 828 strains, suggesting that the band at 860 cm −1 was not present in these spectra (Fig. S5 and Table S2).Thus, based on the presence/absence of the 860 cm −1 band in Microcystis strains, we were able to screen four (PMC 814, PMC 820, PMC 827 and PMC 831) out of eight strains as promising candidates to possess ACC-forming capabilities.These findings were further reinforced by analyzing the FIR region spectra of Microcystis strains.We compared the FIR spectra of two Microcystis strains: Microcystis PMC 827 and Microcystis PMC 824.We found broad features around 150 cm −1 in both PMC 827 and PMC 824 strains, similar to that observed in ACC+ and ACC-cyanobacterial strains and possibly corresponding to torsional frequencies of primary amines 56,57 .FIR spectra of Microcystis PMC 827 showed a broad but strong peak around ~320 cm −1 , which was absent in Microcystis PMC 824 (Fig. 6).The position of this peak coincides with the broad peak observed in FIR spectra of synthetic ACC as well as in ACC+ forming cyanobacteria strain (Fig. 3), confirming the presence of ACC in Microcystis PMC 827 and its absence in PMC 824.This is consistent with observations on the ACC-forming potential of these strains derived from the mid-IR spectra of these strains.
To further confirm our findings on the capability of Microcystis strains to form ACC, we analyzed all the Microcystis strains by SEM-EDXS (Fig. 6 and Fig. S6).For each strain, several cell aggregates were randomly selected, and EDXS maps were obtained.Each cell had a variable number of inclusions, ranging from zero to a hundred.Based on EDXS elemental mapping, all Microcystis strains contained inclusions enriched in P, containing Mg, K, and Ca were interpreted as polyphosphate (polyP) granules.Several functions of polyP inclusions have been 30 inferred in bacteria, including the provision of an alternative source of energy under some specific environmental conditions [10].Observation of polyP inclusions in Microcystis is consistent with the presence of polyP IR bands [8-10] in all Microcystis strains [Table 2, Fig. 4].
Qualitatively, the number of polyP inclusions vary among different Microcystis strains as observed in SEM micrographs, which suggests that the amount of polyP content varies across Microcysits strains (Fig. 6 and Fig. S6).The variation in polyP content across Microcystis strains is also reflected in the differences observed in peak height of the polyP bands [8-10] present in Microcystis strains (Fig. 4).In contrast to polyP inclusions, inclusions enriched in Ca with little or no P were interpreted as ACC.They were only found in strains PMC 814, 820, 827, and 831, which was perfectly consistent with FTIR results.Altogether, based on the presence/absence of the mid-IR Ʋ 2 absorption band at 860 cm -1 and a broad band at ~300 cm −1 in FIR region, both characteristics of ACC, and using SEM-EDXS as a supporting tool, we identify four out of eight Microcystis strains that are able to form ACC, whereas the rest of the Microcystis strains do not form ACC, at least not in sufficient amount to be detected by FTIR spectroscopy and/or under the growth conditions that were used in the present study.Microcystis strains.As shown in Table 4, the "calculated ACC" content in ACC+ Microcystis strains varied between 0.5-3% wt ACC (Table3).This calculated ACC content was compared with the ACC content estimated from the measured total Ca concentration in the dried biomass, referred to as the "derived ACC content" in Table 4.Although the estimated absolute value of derived and calculated ACC content in strains were different, the trends were similar, that is Microcystis PMC 820 had the highest ACC content, followed by Microcystis PMC 827.Similar variations in the ACC content of diverse strains have been observed before.For the known ACC+ cyanobacteria strains, the maximum amount of ACC content can vary from 2-8 wt% 11 .

Semi-quantitative estimation of ACC in Microcystis
Reasons for such variations in the ACC content are not clear and will need to be further assessed by future studies.It could be linked with the varying physiological state of the cells, or a different function of ACC in these strains 10,11 .

Conclusions
This study probed intact cells of cyanobacteria forming intracellular ACC inclusions using FTIR spectroscopy, to identify diagnostic IR spectral features of ACC that are suitable for its detection in systems dominated by prokaryotes.ACC has several IR active bands that have been widely used to discriminate ACC from other CaCO 3 polymorphs.While several of these bands were found to overlap with characteristic bands of proteins, lipids, polysaccharides, and phosphate groups, two ACC characteristic bands at 860 cm −1 and a broad peak at ~306 cm −1 were free from any overlap from biomolecules.Moreover, the bands at 860 cm −1 and a broad peak at ~306 cm −1 were systematically present only in ACC+ cyanobacteria and absent in ACC-cyanobacteria, suggesting that both bands could serve as a spectral marker to screen strains for their capability of forming ACC.The reliability of these diagnostic features of ACC was tested on several Microcystis strains whose ACC-forming capabilities were unknown.Using mid-IR and far-IR ACC spectral features, we identified four out of eight Microcystis strains as ACC forming strains.These findings were confirmed by SEM-EDXS observations.Lastly, we quantified the ACC content of Microcystis strains using the variations in 860 cm -1 band intensity.While the semi-quantification approach used here is an approximation, it provides some insights into assessing the Ca accumulation potential of the ACC-forming Microcystis strains and other prokaryotes and their relevance as a geochemical reservoir of Ca in the environment.Overall, FTIR spectroscopy in the mid-IR and far-IR range appear as a very valuable and reliable tool to diagnose the presence of ACC in bacteria and screen efficiently numerous samples before using more time-consuming techniques such as electron microscopy.Whether these spectral features are effective in detecting ACC−forming prokaryotes in a complex environmental sample composed of different microbial populations as well as extracellular minerals remains an interesting topic for future research.

Figure 1 :
Figure 1: (A) ATR-FTIR spectra of synthetic ACC and calcite in the mid-IR spectral range.The arrows correspond to characteristic mid-IR active frequencies of ACC.The dashed rectangle envelops the zoomed regions of mid-IR ACC spectra shown in the inset 1 (1500-1200 cm −1 ) and inset 2 (750-680 cm −1 ).The bands between 2500-2000 cm -1 in ACC spectra are due to ambient CO 2 and insufficient purging of the sample chamber.(B) Far-IR spectra of synthetic ACC, and calcite.Arrows denote the characteristic features of ACC and calcite.(C) Near-IR spectra of synthetic ACC, and calcite.Details on the assignment of the vibrations of ACC across the entire region are listed in TableS1.The X-axis is in wavelength, where 500 nm corresponds to 2x 10 4 cm −1 and 4000 nm corresponds to 2.5 x 10 3 cm −1 .

Figure 2 :
Figure 2: (A) ATR-FTIR spectra of the Cyanothece PCC 7425 (ACC+), Synechocystis PCC 6803(ACC-), and synthetic ACC.The bands between 2500-2000 cm −1 in the spectra are due to ambient CO 2 and H 2 O resulting from insufficient purging of the sample chamber.The dashed rectangle envelops the zoomed regions of mid-IR ACC spectra shown in the inset 1 (1800-900 cm −1 ) and inset 2 (900-820 cm −1 ).Numbers on the spectra correspond to the band# in Table2.

Figure 4 :
Figure 4: (A) ATR-FTIR spectra of PMC Microcystis sp.strains in the mid IR range.The dashed rectangle envelops the zoomed regions of mid-IR ACC spectra shown in the inset 1 (1800-900 cm −1 ) and inset 2 (950-820 cm −1 ).Numbers on the spectra correspond to the band numbers and their assignment are provided in Table1.The vertical gray dash lines in Inset 1 are guidelines to facilitate identification of the position of bands (1-10) in the different PMC Microcystis sp.strains.The solid black line in Inset 2 denotes the location of characteristic Ʋ 2 band of ACC (~860 cm −1 ).

Figure 6 :
Figure 6: FIR spectra of Microcystis PMC 827 and Microcystis PMC 824.The peak labeled marked as "*" denotes the characteristic features of ACC in FIR at ~300-320 cm −1 .

Figure 7 :
Figure 7: (A) ATR-FTIR spectra of Synechocystis PCC 6803-ACC mixtures.(B) Zoomed plot of the boxed region in (A).The peak intensity of Ʋ 2 band of ACC was normalized by the frequency region marked by dashed rectangle in (A).(C) The FTIR calibration curve of normalized peak area of Ʋ 2 band for different Synechocystis PCC 6803-ACC mixtures.

Table 3 :
Peak spectral properties of the fitted peaks in Microcystis strains corresponding to the best fit for each strain.

Table 4 :
The calculated ACC (wt %) content in 4 Microcystis strains, using the calibration curve equations determined on artificial Synechocystis PCC 6803 -ACC mixtures.