Article pubs.acs.org/est
Submicron-Scale Heterogeneities in Nickel Sorption of Various Cell− Mineral Aggregates Formed by Fe(II)-Oxidizing Bacteria Gregor Schmid,† Fabian Zeitvogel,† Likai Hao,† Pablo Ingino,† Irini Adaktylou,† Merle Eickhoff,† and Martin Obst*,† †
Environmental Analytical Microscopy, Center for Applied Geoscience, University of Tübingen, Hölderlinstrasse 12, 72074 Tübingen, Germany S Supporting Information *
ABSTRACT: Fe(II)-oxidizing bacteria form biogenic cell−mineral aggregates (CMAs) composed of microbial cells, extracellular organic compounds, and ferric iron minerals. CMAs are capable of immobilizing large quantities of heavy metals, such as nickel, via sorption processes. CMAs play an important role for the fate of heavy metals in the environment, particularly in systems characterized by elevated concentrations of dissolved metals, such as mine drainage or contaminated sediments. We applied scanning transmission (soft) X-ray microscopy (STXM) spectrotomography for detailed 3D chemical mapping of nickel sorbed to CMAs on the submicron scale. We analyzed different CMAs produced by phototrophic or nitrate-reducing microbial Fe(II) oxidation and, in addition, a twisted stalk structure obtained from an environmental biofilm. Nickel showed a heterogeneous distribution and was found to be preferentially sorbed to biogenically precipitated iron minerals such as Fe(III)-(oxyhydr)oxides and, to a minor extent, associated with organic compounds. Some distinct nickel accumulations were identified on the surfaces of CMAs. Additional information obtained from scatter plots and angular distance maps, showing variations in the nickel−iron and nickel−organic carbon ratios, also revealed a general correlation between nickel and iron. Although a high correlation between nickel and iron was observed in 2D maps, 3D maps revealed this to be partly due to projection artifacts. In summary, by combining different approaches for data analysis, we unambiguously showed the heterogeneous sorption behavior of nickel to CMAs.
■
oxidizing bacteria.15 The sorption behavior of nickel was found to be different for abiogenically formed Fe(III)-(oxyhydr)oxides as compared to biogenically formed iron minerals.4,15 The sorption of heavy metals can also be influenced by microbially excreted EPS forming organic coatings on the surface of iron minerals.16−18 Due to the heterogeneous composition of such (biogenic) minerals and CMAs, which varies on submicron spatial scales, the sorption behavior of nickel and the individual contributions of the organic and inorganic constituents in such complex systems still remain poorly understood. To gain a better understanding of these processes, we analyzed the distribution of nickel on the submicron scale in CMAs formed by three physiologically different types of Fe(II)oxidizing bacteria, namely nitrate-reducing, phototrophic, and microaerophilic Fe(II)-oxidizing bacteria. For the first two types, we were able to use well-studied laboratory model strains that form heterogeneous aggregates of cells, EPS, and biogenic Fe(III) minerals. For the latter, we used an environmental biofilm sample containing so-called twisted stalks, which are
INTRODUCTION The anthropogenic input of nickel into the environment has been increasing over the last few decades by diverse sources such as combustion of coal, fuel oil, waste, or sewage and the use of fertilizers or by industrial or mining activities.1,2 Nickel is a heavy metal,3 and it can be immobilized by sorption to iron minerals4−6 to ferromanganese precipitates7,8 or to functional groups associated with organic carbon.2,9 Cell−mineral aggregates (CMA) formed by Fe(II)-oxidizing bacteria are heterogeneous structures composed of mineral particles, microbial cells, and extracellular polymeric substances (EPS). In environments with elevated metal concentrations such as mine drainage, CMAs can contribute to the immobilization of dissolved nickel.10−12 Fe(II)-oxidizing bacteria are known to cause the precipitation of ferric iron minerals and at the same time excrete considerable amounts of extracellular polymeric substances (EPS).13 Different microbially mediated mechanisms of Fe(II) oxidation are known, such as nitrate-reducing, phototrophic, or microaerophilic metabolisms;13,14 thus, the biogenic precipitation of iron minerals can occur in many different habitats. It has been shown that biogenic iron minerals in sediments or in microbial biofilms can affect the distribution and bioavailability of nickel in the environment.10 Heterogeneities in nickel sorption on the submicron scale have been investigated for biofilms11,12 and for phototrophic, Fe(II)© XXXX American Chemical Society
Received: June 18, 2015 Revised: October 29, 2015 Accepted: November 20, 2015
A
DOI: 10.1021/acs.est.5b02955 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
cultivated in 58 mL serum bottles with 25 mL of low (15−20 μM) phosphate medium (N2/CO2 headspace, v/v of 90/10), pH of 6.8−7.0, Fe(II) concentrations of 2−4 mM, and 5% inoculum (pregrown on H2), as described elsewhere.15 Anoxically prepared 200 mg/L NiCl2 stock solution was prepared by dissolving NiCl2 (Merck, Darmstadt, Germany) in Milli-Q water and added to the serum bottles prior to inoculation with bacteria to a final concentration of 10 mg/L (∼170 μM) to study sorption (i.e., adsorption and coprecipitation during microbial Fe(II)-oxidation). Prior to the inoculation with bacteria, no precipitation of any mineral phase was observed. BoFeN1 cultures were incubated at 28 °C in the dark, whereas SW2 was incubated at 20 °C under permanent illumination (>600 l× tungsten light bulbs). Fe(II) concentrations were measured during microbial Fe(II) oxidation by the ferrozine assay. All Fe(II) was oxidized after 7−8 days and 12−16 days, respectively. To extend our study to microaerophilic Fe(II)-oxidizing bacteria, we collected organic carbon fibers mineralized in iron minerals, the so-called twisted stalk, from a microbial biofilm that was sampled from the wall of an abandoned silver mine (Segen Gottes Mine, Haslach, Germany).21 The mat is mainly composed of Fe(III)-(oxyhydr)oxides encrusted twisted stalks formed by microaerophilic bacteria, such as Gallionella sp.19,20 The biofilm sample (denoted SG) was suspended in sterilefiltered water collected from the sampling site (0.22 μm mixed cellulose esters, Fisherbrand, Loughborough, England). To study the sorption of nickel to these structures, we added NiCl2 stock solution to a final concentration of ∼30 mg/L (∼510 μM) and incubated it for 3 days. Metal Analytics. Sample aliquots of BoFeN1 and SW2 to determine Fe(II) and nickel concentration were taken at several time points during microbial Fe(II) oxidation under N2 atmosphere in a glovebox chamber (GS Glovebox Systemtechnik, Malsch, Germany). Fe(II) concentrations of BoFeN1 cultures were quantified spectrophotometrically using a FlashScan 550 plate reader (Jena Analytics, Jena, Germany) by the ferrozine assay,38 modified after Klueglein and Kappler,39 which prevents abiotic Fe(II) oxidation by nitrite (being a possible denitrification product of BoFeN1 during microbial Fe(II) oxidation). For the determination of Fe(II) concentrations of SW2 cultures, the ferrozine assay was modified after Hegler et al.40 Nickel concentrations were quantified by ICPOES with a PerkinElmer Optima 5300 DV (PerkinElmer, Waltham, MA). BoFeN1 sample aliquots were centrifuged, and the supernatants were removed and acidified to 1% with 69% concentrated HNO3 (Fluka Analytical, Steinheim, Germany). SW2 sample aliquots for ICP-OES were filtered using 0.22 μm nylon filters (Costar, Corning, NY) and acidified to 1% with 69% concentrated HNO3. The geochemistry of the biofilm from Segen Gottes sampling site is explained in detail by Picard et al.21 Sample Preparation for STXM. BoFeN1 was sampled after complete Fe(II) oxidation (8 days) for STXM spectrotomography. SW2 was sampled after 13 days, when at least ∼75% of Fe(II) was oxidized. Aliquots of both samples were taken and prepared for STXM spectrotomography in a glove box chamber. Samples were centrifuged and resuspended twice in sterile deionized water (DI water). A total of 2−5 μL of the sample were transferred onto Formvar-coated Cu TEM grids (200 or 300 mesh, Plano, Wetzlar, Germany) and dried under N2 atmosphere. The SG sample was taken 3 days after the addition of NiCl2 under oxic conditions but without washing
organic polymer structures that are encrusted in Fe(III)(oxyhydr)oxides, formed by microaerophilic Fe(II)-oxidizing bacteria.19−21 This three-model system represents Fe(II)oxidizing metabolisms that occur under different environmental conditions. The nitrate-reducing, anaerobic Fe(II)-oxidizing Acidovorax sp. strain BoFeN122 is a model organism for nitratereducing Fe(II) oxidation, which is shown by many nitratereducing bacteria.23 CMAs similar to the ones produced by BoFeN1 have been observed for a number of other nitratereducing bacterial strains.24 Nitrate-reducing Fe(II) oxidation is a relevant Fe(II)-oxidizing mechanism, particularly in sediments.23 CMAs formed by phototrophic Fe(II) oxidizers, as well as by the photoferrotrophic, anaerobic Rhodobacter ferrooxidans strain SW2,25 are common, especially in aquatic environments,23,26 and they are thought to have played an important role in the formation of banded iron formations.27 Finally, microaerophilic Fe(II) oxidziers can be found in environments containing both Fe(II) and low concentrations of oxygen, e.g., creeks,28 mines,6 and water distribution pipelines,29 making them of high environmental importance. We applied synchrotron-based scanning transmission (soft) X-ray microscopy (STXM) in combination with angle-scan tomography for spectromicroscopic 3D chemical mapping. Using a soft X-ray spectrotomography technique, it is possible to resolve structural features and information about compositional distributions in three dimensions, in contrast to 2D images, which represent information integrated along the beam path.30 By acquiring near-edge X-ray fine structure (NEXAFS) image stacks, we obtained the chemical distribution of organic carbon, iron, and nickel in 3D and 2D. Previous synchrotron studies with BoFeN1 revealed different patterns of cellular encrustation with iron minerals.31 The formed CMAs showed the capability for heavy metal sorption.32,33 Heavy metal sorption onto CMAs has also been observed via CMAs formed by SW2,15,34 where microbial Fe(II)-oxidation occurs enzymatically via an Fe(II) oxidoreductase.35 Previous studies suggested that, to prevent cell encrustation, SW2 cells excrete organic polymers such as lipids and polysaccharides to generate an iron redox gradient close to the cell with precipitation of nanogoethite as the final Fe(III)-(oxyhydr)oxide phase.26 Both strains (BoFeN1 and SW2) were used as lab cultures in this study and grown under controlled conditions. In addition to CMAs formed by the lab cultures, we analyzed nickel sorption to twisted stalks sampled from an environmental biofilm formed by microaerophilic Fe(II)-oxidizing bacteria. Microaerophilic Fe(II)-oxidizing bacteria are prelevant in natural systems21,28,36,37 and complete our study to reveal nickel sorption properties different from those of nitratereducing and phototrophic Fe(II)-oxidizing strains.
■
MATERIALS AND METHODS Cell Cultivation and Biofilm Sampling. All glassware used for cell cultivation and sorption experiments was prewashed with 1 M HCl and subsequently washed three times with Milli-Q water to remove residual traces of nickel and iron. The chemo-organotrophic, nitrate-reducing ß-proteobacterium Acidovorax sp. strain BoFeN122 was cultivated anoxically in 58 mL serum bottles with 25 mL of low (15−20 μM) phosphate medium (N2/CO2 headspace, v/v of 90/10) and a pH of 6.8−7.0, 10 mM Na nitrate, 5 mM acetate, 8−9 mM Fe(II), and 5% inoculum (pregrown on acetate−nitrate), as described elsewhere in detail.31 The photoferrotrophic, αproteobacterium Rhodobacter ferrooxidans strain SW225 was B
DOI: 10.1021/acs.est.5b02955 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 1. 3D chemical maps of BoFeN1 with segmented cell and shell structures, Fe/OC, Ni/OC, and Ni/Fe scatter plots with respective Pearson’s r correlation coefficients (r). Angular distance maps showing the 3D variability of Fe/OC, Ni/OC, and Ni/Fe ratios in orthoslice representation. For a detailed description, see the text. (Scale bar: 1000 nm).
(typically) ±72°.30,42 Images were acquired at specific X-ray absorption energies (denoted as stack maps, consisting of images in pre-edge and at a characteristic X-ray absorption edge) at the C 1s, Fe 2p, and Ni 2p absorption edges. X-ray energies at the C 1s absorption edge were calibrated by measuring the 3p Rydberg peak of gaseous CO2 for Fe 2p by the dominant L3-edge peak of Fe(III) at 708.7 eV44 and for Ni 2p by using the resonance peak of pure NiCl2 at 852.7 eV.45 Selected energies for stack maps at the C 1s absorption edge were 288.2 eV, dominated by the C 1s → π*C=0 transition of proteins46,47 with some minor contributions of polysaccharides and lipids (therefore used as an approximation for organic
steps to prevent the desorption of sorbed Ni2+. To prevent artificial deposition of NiCl2 on the stalks during drying, we blotted the grids carefully using filter paper. For STXM spectrotomography measurements, 300 mesh TEM grids were sliced, and single strips were mounted on a brass rods, as described in detail elsewhere.41,42 STXM Data Acquisition. STXM spectrotomography measurements were performed at the SM 10ID-1 beamline of the Canadian Light Source (CLS, Saskatoon, Canada).43 Sample strips were mounted onto the tomography stage,30,41 allowing for sample rotation to acquire a series of images at different tilt angles. The sample strips were tilted in steps of 4° in a range of C
DOI: 10.1021/acs.est.5b02955 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 2. 3D chemical maps of SW2 with segmented cell and extracellular structure, Fe/OC, Ni/OC, and Ni/Fe scatter plots with respective Pearson’s r correlation coefficient (r). Angular distance maps showing the 3D variability of Fe/OC, Ni/OC, and Ni/Fe ratios in orthoslice representation. For a detailed description, see the text. (Scale bar: 1000 nm).
STXM Data Processing and Evaluation. NEXAFS image stacks and stack maps were converted from transmission scale into linear absorbance scale (optical density, OD = −ln(I/I0), where I is the measured intensity of a pixel, and I0 is the incident flux intensity in an empty area adjacent to the sample) using aXis2000.48 Quantitative species maps (denoted as chemical maps) for each rotation step were obtained by subtracting the image acquired in the pre-edge from the image at the characteristic X-ray absorption energy (ODedge − ODpre‑edge). 3D chemical maps were reconstructed using IMOD49,50 using the serial iterative reconstruction (SIRT) algorithm51 with 400 iterations (radial filter cutoff 0.4, falloff 0.5). Chimera52 was used for the visualization of the reconstructed 3D maps. Pixel spacings of 50 × 50 nm2 were
carbon) and 280 eV for an energy in the pre-edge with no Cspecific X-ray absorption. At the Fe 2p edge, energies were chosen at 723.5 eV for the characteristic X-ray absorption peak of iron minerals44 and at 705 eV for the pre-edge. At the Ni 2p edge, 852.7 eV was selected for the characteristic Ni2+ absorption peak45 and 850 eV for the pre-edge. In addition, 2D stack maps of various sample spots of all sample types were acquired to statistically verify the chemical tomography results. Furthermore, image series across an X-ray absorption edge, socalled stacks, were acquired at the Fe 2p and the C 1s edges for BoFeN1 and SG and at the C 1s edge for SW2 to obtain information on the iron mineralogy and organic carbon composition. D
DOI: 10.1021/acs.est.5b02955 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology selected for BoFeN1 and SG and 25 × 25 nm2 for SW2. The spatial resolution in the reconstructed 3D data sets is explained in detail within the Supporting Information. A simple background correction was applied after SIRT reconstruction by setting the center of the Gaussian-like background noise peak in the histogram to zero. Additionally, a Gaussian blur filter (σ = 0.5 pixels) was applied to the 3D maps in Fiji53 to reduce noise. To remove streaking artifacts introduced by SIRT due to missing-wedge effects,54 we segmented 3D maps of BoFeN1 and SW2 using the program ilastik55 to create binary masks representing the individual structures, such as bacterial cells and mineral shells, or extracellular structures. These masks were then multiplied with the reconstructed 3D maps. Because there were no such individual structures in the SG samples that were composed of twisted stalks only, no segmentation was necessary for these. The Fiji plugin ScatterJ56 was used on the resulting 3D maps for the analysis of correlations and trends, the back-mapping of regions of interest (ROI), the calculation of Pearson’s r coefficients, and for the calculation of angular distance maps for organic carbon (C 1s), iron (Fe 2p), and nickel (Ni 2p). For the calculation of Pearson’s r coefficient, gray values
