Calcium carbonates : Induced biomineralization with controlled macromorphology

Biomineralization of (magnesium) calcite and vaterite by bacterial isolates has been known for quite some time. However, the extracellular precipitation hardly has been linked to different mophologies of the minerals that are observed. Here, isolates from limestone associated groundwater, rock and soil were shown to form calcite; magnesium calcite or vaterite was also present. More than 92 % of isolates indeed could form carbonates, while abiotic controls failed in mineral 15 formation. The crystal morphologies varied, including rhombohedra, prisms and pyramid-like macro-morphologies. Using different conditions like varying temperature, pH or media components, but also co-cultivation to test for collaborative effects of sympatriotic bacteria, were used to differentiate mechanisms of calcium carbonate formation. Single crystallites were cemented with bacterial cells; these may have served as nucleation sites by providing a basic pH at short distance from the cells. A calculation of potential calcite formation of up to 2 g per liter of solution could link the microbial activity to 20 geological processes.


Introduction
The processes of carbonate biomineralization by bacteria are usually linked to an alkaline microenvironment enhancing the potential for carbonate precipitation at small spatial scales (Hammes and Verstraete, 2002). By uptake of CO 2 , precipitation of calcium carbonates is favored, especially in presence of Ca 2+ . Thus, the process clearly may be defined as microbially 25 induced biomineralization, in contrast to microbially controlled processes best seen with intracellular, compartment-bound formation of magnetite by magnetotactic bacteria. The cell surface can act as a nucleation site for mineralization (Schultze-Lam et al., 1996). Since properties are different for Gram positives and Gram negatives, both binding divalent cations such as Mg 2+ or Ca 2+ , a determination of the microbial clades is necessary (Chahal et al., 2011;Douglas and Beveridge, 1998;Zhu et al., 2017). Depending on different cell surface structures, different crystal morphologies have been reported (Cao et al., biominerals were described (Tisato et al., 2015). Thus, a more thorough investigation of microbially induced biomineralization seems warranted.
The occurrence of carbonate forming bacteria has been investigated with respect to different environments (Andrei et al., 2017;Gray and Engels, 2013;Horath and Bachofen, 2009;Kang and Roh, 2016;Rusznyak et al., 2012). The process generally was linked to changes induced in the direct microenvironment of the growing bacteria through metabolic features, 5 or to providing nucleation sites for crystallization in a supersaturated environment (Roberts et al., 2013;Seifan et al., 2017;Yan et al., 2017). Either microbiomes of a specific habitat, or physiological and biomineralization properties of a single isolate have been investigated. To bridge that gap, an investigation of more isolates from one environment seems indicated.
Basic scientific questions addressed include the nature and rates of biotic versus abiotic nature of calcite formation. For one species, 33 g/L calcite precipitation was reported (Seifan et al., 2016). The same assumption has been tested for the formation of rhodoliths in marine systems, where carbonate rocks evolve (Cavalcanti et al., 2014). The formation of CO 2 15 through respiration clearly is a factor that would imply aerobic microbes (Keiner et al., 2013). Hence, a marine sediment derived limestone habitat and isolation of aerobic bacteria from that environment seemed indicated (compare Meier et al., 2017).
We thus were interested to identify different bacteria from a carbonate rock-water-soil system to address the question of how bacteria may contribute to different morphologies, even at a distance to the cells, and how much calcium carbonates may be 20 formed in vitro under conditions more likely to occur in nature than previously used (Castanier et al., 1999). To obtain a variety of calcium carbonate forming bacterial isolates, we focused on marine Triassic limestones exposed in central Germany. We obtained Gram-positive and Gram-negative bacteria from the limestones, groundwater and rendzina soil developed on the limestones near the limestone quarry Bad Kösen (Thuringia, Germany). The isolated bacteria represented major taxa present in the microbiomes (Meier et al., 2017). They were incubated in single culture or in co-cultivation under 25 different conditions to address biomineralization relevant to carbonate formation. With our results, we can advance the background on microbially induced biomineralization with respect to different mechanisms.

Sampling
The Muschelkalk quarry in Bad Kösen is characterized by Lower Muschelkalk (Jena Formation, Lower Wellenkalk core 30 sampled) and Middle Muschelkalk (Karstadt Formation, upper Schaumkalkbank sampled). Rock samples were taken from cores under sterile conditions. The groundwater wells near Bad Kösen, in Stöben (Hy Camburg 13/198; 4478864N, 5660183E; Lower Muschelkalk sampled at 34 m depth) and Wichmar (Hy Camburg 121/1988; 4478030N, 5655906E; 120 m, Middle Muschelkalk sampled at 17 m) were sampled with the help of an electric pump after reaching constant pH and temperature. All samples were stored at 6 °C prior to analysis.
Soil was sampled at 40 cm depth from 15 sampling points at the Bad Kösen quarry for rendzina on the respective limestone 5 bed in a radius of about 1 km and homogenized.

Isolation and characterization of bacteria
For the isolation of bacteria from the limestones, Std I (Carl Roth; supplied with NaCl at 3, 5, 7 % if indicated by sample chemistry), minimal AM (Amoroso et al., 2002) and oligotrophic R 2 A (Reasoner and Geldreich, 1985) media were applied.
Calcification promoting B-4 medium without pH adjustment (Banks et al., 2010) was used for isolating limestone associated 10 bacteria.
For rock samples, 5 g of powdered rock sample were added to 45 ml sterile 0.9 % NaCl, followed by vortexing for 20 min.
Subsequently, sonication was applied for 15 min and filtered supernatant was plated.
A total of 2.5 L of groundwater samples were filtered before culturing using 0.2 μm membrane filters (Omnipore membrane filters, Millipore) in a filtration system (GP Millipore Express plus membrane, Millipore). The filters were placed on agar 15 and incubated at 10 °C for three days. Additionally, dilution series in sterile 0.9 % NaCl solution were plated. Surface sterilized rock samples were powdered and dilution series were prepared, cultured in liquid medium or placed as small particles on nutrient agar.
Soil extracts from 100 g mixed soil samples, dried at 40 °C overnight were prepared using 500 ml 10 mM 3-(Nmorpholino)propansulfonic acid (MOPS, Serva). The suspension was incubated at 28 °C for 1 hour prior to filtration with a 20 vacuum pump system. The extract was supplemented with 100 mg/L glucose, 1 mg/L casein hydrolysate, 1 mg/L yeast extract and 18 mg/L agar and sterilized before adding to cultures obtained from rock and soil samples. Incubation was performed at 28 °C and 10 °C for 3 to 5 days in duplicates, with constant shaking at 120 rpm for liquid cultures.

Biomineralization assays
To study effects on carbonate mineralization, B-4 agar plates or liquid cultures were incubated at 28 °C or 10 °C for one to three weeks. CaCO 3 (B-4CO) or Ca 3 (PO 4 ) 2 (B-4CP) as source of calcium were used replacing the calcium acetate of the medium. For differentiation between biological and chemical mechanisms, non-inoculated plates and plates streaked with dead biomass were used as negative controls. Bromthymol blue (Merck, Darmstadt) was used as pH indicator where 5 appropriate.
Selected experiments were designed to investigate competition or induction of activities of the bacterial strains streaked towards each other on B-4 agar. Two-or four-strain interactions were tested using compatriot isolates from the same habitat.

Mineralogical investigations
Solid products were visualized with a stereomicroscope (Zeiss) and sampled in 0.2 ml reaction tubes. Powder X-ray 10 diffractometer (Bruker D8 Advance) with Cu Kα radiation (λ=1.54058 Å) and LynxEye detector was used on powdered crystals transferred to a zero-background silicon sample holder and measured with following parameters: 18-70 °2θ, a step size of 0.02 º2Θ and dwell of 0.5 sec.
Scanning electron micrographs (SEM; Quanta 3D FEG; FEI) were taken from carbon or gold sputtered samples imaged in secondary (SE) or back-scattered electron mode (BSE) at an acceleration voltage of typically 10 kV. Semi-quantitative 15 chemical analysis was conducted using an energy-dispersive X-ray (EDX) spectrometer (EDAX, Mahwah).

Quantification of precipitate formation
Liquid medium inoculated with Agrococcus jejuensis sMM51, Bacillus muralis rLMd or Bacillus sp. rMM9 was incubated at 28 °C for at least three weeks. Precipitates were harvested using a sterile cell strainer (easystrainer, 40µm for 50 ml tubes, GBO) and rinsed with pure water to separate cells from the solids. The crystals were dried and weighed. Bacterial growth 20 was determined with cells diluted 1:1000 with ISOTON Diluent (Beckman Coulter, Brea) in a Coulter cell counter in triplicates.
For calcultation of yearly biomineral formation, we used the amount produced in our cultures (approx. 100 mg) during the time of incubation (3 weeks) to calculate how much this would make in 52 weeks, a full year. 1700 mg would be the result of such an approximate calculation for new mineral formation during a year. This is an underestimation, since the nucleation 25 time is needed only once.

Biomineralization activities of bacterial isolates
This study is focused on bacterial isolates from two different limestones, Lower Muschelkalk (LM) and Middle Muschelkalk (MM), and three compartments for each lithotype representing rock (r), groundwater (gw) and soil (s). Of 140 isolates, only 10 formed no crystals under the conditions tested (Figs. 1-6). This shows that in a carboniferous environment, strains 5 forming carbonates are highly enriched.
As to the means of inducing mineralization, the influence of pH could be tested. While 89 strains did not change medium pH, 43 produced an alkaline environment supportive of carbonate formation. However, two of those alkalinity producing strains did not produce biominerals, all other non-producers did not change pH. In addition, crystal formation was observed with all strains which actually lowered pH. The temperature clearly had an effect, although there was no clear correlation. 10 While on one medium, higher temperature might have induced formation of crystals not observed at low temperature, the opposite effect was visible on the next medium. Any combination of traits was observed. This indicates that rather differences among strains than a mere general influence on the micro-environmental conditions leading to abiotic carbonate formation was observed.
The highest incidence of mineral production was observed with soil isolates obtained from soil developed on Middle 15 Muschelkalk. Biomineralization potential was not associated to phylogeny. Eight isolates of the genus Bacillus isolated from Lower Muschelkalk rock samples induced precipitation, whereas six strains of the same genus were negative in biomineralization.

Identification of calcium carbonate biominerals
From each habitat, the four most prevalent isolates were selected for further study (Tab. 1). Of these 24 strains, 16 were 20

Variation in crystal morphologies
Color and morphology of crystal aggregates were highly variable, ranging from colorless to brownish or purple, and morphologies from individual rhombohedral crystals, round or acicular aggregates or rosettes to laminated crusts. 5 Aggregates located within the agar mostly formed small rhombohedra or spheres. Biominerals were on top or below the biomass; for some strains the crystals formed always in a certain distance from the colonies, which might indicate a zone of change in pH around the culture (Fig. 7). Indicator plates with bromthymol blue indicated mainly change to basic pH > 7.6, followed by crystal precipitation. A few isolates of each environment revealed acidification, most pronounced for Lower Muschelkalk Bacillus strains and Middle Muschelkalk Streptomyces with pH < 10 6.0.
A recurring morphological feature was subparallel intergrowth of µm to sub-µm sized crystallites. In simple cases, the resulting aggregates resembled the morphology of a single crystal, in other cases, more complex aggregates were found ( Fig.   8; for further morphologies of other isolates, see supplemental Fig. S1). Crystal aggregates produced by L. amnigena S_H3 exhibited platy crystals arranged into a rosette. 15

Impact of cultivation conditions
Phenotypic differences in crystal morphology were readily observed, dependent also on temperature (Fig. 9). On plates containing calcium carbonate or calcium phosphate, more bacterial growth, but less biomineralization was observed. On the medium with CaCO 3 , the groundwater isolates showed temperature-dependent biomineralization. Crystals were formed at 10 °C, whereas the temperature of 28 °C induced mineral dissolution indicated by halo formation. 20

Direct impact of bacteria
The bacterial influence was visible also by direct associations. R. erythropolis WSd5 crystals were coated by bacterial cells (see Fig. 8). Some areas free of cells contained oval, clearly delimited holes of the same size and shape as the surrounding bacterial cells. Occasionally, these holes were characterized by an extended trace (a tail) on one side.
Two or four strains of one community were co-cultivated to determine their interactions. There was either no interaction, 25 inhibition leading to lack of growth, or parasitism visible by overgrowing the competing strain(s). Concerning biomineralization, the co-cultivation had either no effect, or it enhanced precipitation, producing larger crystals in the contact zone (Tab. 2). The biotic interactions showed an impact on biomineralization, e.g. with L. amnigena S_H3 with more crystals developing at the contact zones, Advenella sp. W_Sd3 from groundwater of Middle Muschelkalk was overgrown by all interaction partners without a change in crystal distribution, and Streptomyces sp. sLM17 from the Middle Muschelkalk 30 rock sample inhibited A. cerinus sLM10 without crystal formation.

Quantification of precipitates
The quantification of crystals formed resulted in similar cell counts ranging from 2.06 * 10 7 to 8.65 * 10 7 . Since the bacterial biomass within the centrifugation pellet thus was similar, the dry mass was compared to see different rates of biomineral formation. The dry weight of precipitates varied, with highest amounts and thus the highest ability for crystal formation seen with Bacillus sp. rMM9 (0.104 g/L), followed by Agrococcus jejuensis sMM 51 (0.096 g/L). Bacillus muralis rLMd 5 demonstrated the lowest amount of crystals wit 0.064 g/L (Tab. 3).

Discussion
Microbially induced calcite precipitation has been known as a general phenomenon since the 1970's and found application in different fields such as cementation of cracks in historical memorials, buildings or sculptures made of limestone by cementing cracks, mostly through ureolytic bacteria (van Tittelboom et al., 2010;Wong, 2015;Zhu and Dittrich, 2016). The 10 general availability of bacterial isolates from different environments associated with limestones, however, was not accessed to the full (Gonzales-Martinez et al., 2017;Seifan et al., 2017). Here, two lithotypes of Lower and Middle Muschelkalk were assessed for the prevalence of carbonate precipitating bacteria.
Calcite, magnesium calcite and vaterite could be formed by the bacteria growing on standard laboratory media. Since vaterite occurs in aqueous, supersaturated solutions, high water content in the precipitates might be explained with water 15 available from the agar. Magnesium calcite was likely formed when the bacterial cell surfaces accumulated the element (see also Cui et al., 2015;Rusznyak et al., 2012). Few trace elements, such as sulfur, chloride and phosphorus probably originating from cell components or salts included in the medium have been detected (see also Rivadeneyra et al., 2000).
Different crystal aggregates with macro-morphologies such as rhombohedra, rosettes and spheres were detected, occurring either in the medium at a distance to the inoculated bacteria, or below or on top of the cultures. Morphological variations of 20 crystal shapes revealed a microbial impact on mineral precipitation (compare Branson et al., 2016). Crystal colors by impurities incorporated in the lattice such as salts or secreted pigments were clearly observed, as calcite can incorporate metal ions in its crystal structure (Kang et al., 2014).
With respect to alkaline pH favorable for calcium carbonate precipitation, a probable process might be the ammonification of amino acids, deriving from yeast extract added to the medium. By degrading amino acids, ammonia develops which 25 increases the pH to alkaline conditions. So far, mostly urease activity has been implied for pH increase in biogenic calcite formation (Bachmeier et al., 2002;Okyay et al. 2016;Wei-Soon et al., 2012). However, our results with strains acidifying the medium and still precipitating calcium carbonate clearly shows that other mechanisms are involved as well.
Biotic effects were investigated by co-cultivation. Growth inhibition of a compatriote strain could be a result of secreted antimicrobial compounds or competition (Hibbing et al., 2009). Growth promotion, like with Flavobacterium limicola S_5.3, 30 may be due to growth promoting compounds for interspecies communication of xenosiderophore use. Increased crystallization at contact zones was noticed in specific bacterial combinations, e.g. with Rhizobium sp. S_4.1a and Pseudomonas sp. S_H4.
Biogenic calcite precipitation may contribute to limestone sedimentation (compare Garcia et al., 2016). With 0.104 g/L dry weight for Bacillus sp.rMM9, a high yield of carbonate precipitation was found. Bacterial cells, especially spores with their high surface to volume ratio and specific cell wall structure, can serve as nucleation sites. Spores encapsulated by calcite 5 might survive for long periods of time, and can be reactivated, followed by germination and growth of new bacterial generations (Murai and Yoshida, 2013). Hence, ample sporulation may explain differences in the amounts recorded between different strains of the genus Bacillus (Yasuda-Yasaki et al., 1978).
Bacteria are known to contribute to the growth of carbonate stalactites that grow by a few mm per year (Genty et al., 2011).
Comparing these observations to our experiment, bacterial isolates may exert a meaningful impact on limestone deposition. 10 From the 104 mg/L after three weeks, up to yearly 2 g/L can be calculated. This compares well to a model experiment which resulted in ~ 2 g abiotic CaCO 3 /yr (Short et al., 2005). As we could show increase in productivity in mutual interactions, microbial communities might well reach even higher rates (Castanier et al., 1999).
As a main result of our investigation of 140 isolates of two lithotypes of limestone in Germany, we can conclude that (magnesium) calcite and vaterite production can be induced through medium alkalinity, through direct surface interation for 15 nucleation visible in close associations, but also in acidified media and a distance apart from the growing bacteria. This indicates, that within the microbially induced calcium carbonate precipitation, mechanistically different routes of biomineralization are possible. The control of morphologies at a distance to the colony seems specifically interesting. We propose that molecules secreted by the bacteria, e.g. specific proteins, might lead to preferential crystal growth at different mineral surfaces due to coating. This clearly warrants further, more molecular studies. 20