Bacterial S-layer preservation and rare arsenic-antimony-sulphide bioimmobilization in siliceous sediments from Champagne Pool hot spring, Waiotapu, New Zealand

The biomineralization of microorganisms has been studied in detail at many hot-spring systems around the world, where the discharged fluids are commonly supersaturated with respect to several mineral phases (Ferris et al. 1986; Pentecost 1995; Schultze-Lam et al. 1995; Cady & Farmer 1996; Hinman & Lindstrom 1996; Jones et al. 1998; Konhauser et al. 2001; Mountain et al. 2003). A major reason for such investigations is to provide insights into the fossilization processes. By increasing knowledge of how organic structures are preserved in mineral matrices, the interpretation of the ancient microfossil record may be greatly enhanced (Schultze-Lam et al. 1995). Biomineralization and ultimately fossilization in high-temperature hot springs are passive processes, such that bacteria simply act as nucleation sites for mineral formation. Therefore, the composition of the biomineral is controlled by the chemistry of the spring water. Upon expulsion from the hot-spring vent, many geothermal fluids become supersaturated with respect to amorphous silica as a result of cooling and (or) evaporation, or supersaturated with respect to CaCOj because of rapid CO: degassing. It follows that most hot-spring biomineralization is either amorphous silica (opal-?), which may incorporate high levels of Fe (e.g. Ferris et al. 1986; Schultze-Lam et al. 1995; Cady & Farmer 1996; Hinman & Lindstrom 1996; Renaut & Jones 2000; Asada & Tazaki 2001; Konhauser et al. 2001; Mountain et al. 2001), or calcite and (or) aragonite (e.g. Chafetz & Folk 1984; Pentecost 1995).

Biomineralization begins as fine precipitates (microcrysts) on the outer surface of the microorganism, commonly while the microbe is still alive (Phoenix et al. 2000, 2001). Then, as biomineralization progresses, the microcrysts may merge and coalesce to form thicker precipitates that gradually destroy any fine structure of the cell. The cell wall and extracellular material are often only crudely preserved in this process, and the internal cytoplasm may degrade and be destroyed (Schultze-Lam et al. 1995; Konhauser& Ferris 1996).

This paper describes the exceptional preservation of ultrastructural details in biomineralized microorganisms from Champagne Pool in the Waiotapu geothermal area in New Zealand, and considers their implications for fossilization. Detailed analyses of these deposits have revealed the preservation of S-layers, a crystalline and ultra-fine mosaic of interconnected proteinaceous subunits that are present on the outer surface of cell walls. These matrices are the finest ultrastructural details yet found that have been preserved by hot-spring biomineralization. S-layers commonly vary between species and can thus be used as a 'fingerprint" for their identification. This new evidence increases our ability to identify species preserved by mineralization in the geological record. Identification is normally difficult because of the loss of morphological detail during preservation (see Jones et al. 200 Ib). Furthermore, many of the microorganisms at Champagne Pool are preserved by As-Sb-S sorption onto cell wall and extracellular polymers. Such As-Sb-S-enriched compositions have not been previously documented in fossilized bacteria at hot springs, and a new mechanism is invoked for their involvement in the preservation of ultrastructural details. The results of this study suggest that rapid adsorption of heavy metals onto cell wall and extracellular polymers is fundamental in enhancing microfossil preservation.

Study location

Champagne Pool is a large (65 m diameter, 150 m deep) hot spring in the Waiotapu geothennal area of North Island. New Zealand (Fig. Ia and b) (Lloyd 1959). The spring fills a hydrothermal eruption crater that formed about 900 years ago (Hedenquist & Henley 1985). The spring discharges water of a mildly acid chloride type, with a pH of c. 5.2 and a constant temperature of 75 °C (Table 1) (Hedenquist 1991; Jones el al. 2001a). This slight acidity is partly due to the upward flow of dissolved CO2 from the deep geothennal source (Hedenquist & Henley 1985; Brown 1986). The geothennal fluid has boiled but is largely undiluted by groundwater (Hedenquist 1991; Alfaro 1998). It has a high gas flux, evolving high quantities of gases in the following proportions: CO2 82.1%, H^sub 2^S 7.2%, CH^sub 4^ 2.9%, N^sub 2^ 6.2% (Giggenbach el al. 1994; Jones el al. 2001a). The arsenic, antimony and H^sub 2^S(aq) concentrations in the spring water are 5.3 ppm, 3.5 ppb and 7 ppm, respectively (Table 1) (Jones et al. 2001 a). The spring water is supersaturated with respect to amorphous silica (logQ/K = +0.18) and is anoxic or strongly dysoxic, even at the surface (Jones et al. 2001a).

The spring pool has a subaerial rim of grey siliceous sinter up to 2 m wide and 0.5m high (Figs Ic and 2a). A shallow (<0.5m deep), subaqueous sinter shelf up to 2 m wide lies around the inner pool margins (Fig. 2a). The shelf is composed of sinter (mainly opal-?) that is rich in arsenic and antimony sulphides, which gives it, and associated loose sediments, a distinctive orange colour. Much of the sinter is stromatolitic and is built by a low-diversity community of filamentous microbes (Fig. 2b). Loose siliceous sediments mantle the stromatolites and much of the shelf. Similar orange particles are found in constant suspension in the pool. Jones el al. (200Iu) showed that the loose sediments and suspended floes are formed of filamentous microbes that have been encrusted and (or) partly replaced by opaline silica precipitates rich in amorphous As-Sb sulphides. The sinters and loose sediments also contain high concentrations of Au (115 ppm) and Ag (370 ppm). Jones et al. (2001a) proposed that these microbes, which are morphologically similar to those building the stromatolites, are providing templates for precipitation of the amorphous silica and associated sulphides.

Methods

Sediments from the SE shore of Champagne Pool (Site A in Fig. Ib) were collected in metal-free glass vials and suspended in 2.5% glutaraldehyde, a fixative for electron microscopy. Floes of the suspended orange paniculate matter were similarly collected from spring water at the NE shore (Site B in Fig. Ib). Small samples of sinter were removed from the submerged shelf and stored in airtight polythene bags. On return to the laboratory, sinter material was collected by scraping the surface with a sterile scalpel; this material was then preserved in 2.5% glutaraldehyde. Samples were then processed for transmission electron microscopy (TEM) as described by Phoenix el al. (2000), using Embed 812 resin. Ultrathin sections (c. 70 nm thick) were prepared on a Reichert-Jung Ultracut E ultramicrotome. sections were mounted on copper grids. Some grids were stained with uranyl acetate and lead citrate to improve contrast during imaging, whereas others were left unstained to prevent misidentification of elements because of overlap of peaks during energy dispersive spectroscopy (EDS). Staining with lead citrate and uranyl acetate is required as organic material is very electron translucent, and thus contrast is poor without a heavy metal stain. EDS analysis was performed to determine solid-phase chemistry using a Philips EM 40OT transmission electron microscope equipped with an EDAX Sapphire EDS detector. EDS data were processed using the Quant code of the EDAX Phoenix analyzer software to determine wt% concentrations of individual elements. Wt% concentrations do not include oxygen, as oxygen associated with mineral phases is very difficult to quantify as it is also associated with the resin and biological material. Because the sections analysed were only c. 70 nm thick, normal ZAF corrections could not be applied to the EDS spectra. It follows that the wt% concentrations determined here should be viewed simply as good approximations. Such good approximations are sufficient for this study. Selected area electron diffraction (SAED) was performed on the Philips EM 400T to determine mineral structure. Ultrathin section images were collected using a LEO 912AB transmission electron microscope with a IK × IK ProScan CCD camera using SIS ESIvision software. Both transmission electron microscopes were operated at 100 kV.

To highlight the structure of S-layer lattices, tracings ot the S-layers in digital images obtained by TEM were constructed using CorelDRAW v. 9. Tracing lines were drawn over the areas that are either lightly mineralized or not mineralized; these were assumed to represent the boundaries between S-layer subunits, thus creating a tracing of the Slayer structure.

Biomineralization: TEM and EDS analyses

Loose sediment and floes

Loose sediments collected from the SE shore of Champagne Pool (Site A) contain abundant rod-shaped bacteria that are c. 1 µm long and coccoid forms c. 1.25 µm in diameter. Their morphology and size are similar to those reported by Jones et al. (2001a). Only cell wall material and extracellular capsular material (Fig. 3a and b) are preserved. The intracellular mineralization is rare, but where present, has an irregular structure and does not preserve any structural detail of the cytoplasm. S-layers (a crystalline mosaic of polypeptides found on the outer cell surface) are preserved, although not as extensively as in other samples (see below).

Extensive biomineralization (i.e. crusts of the order of hundreds of nanometres in thickness) is generally absent in these samples. EDS analyses, however, commonly detected significant quantities of As, Sb and S sorbed onto the organic polymers (Fig. 3 inset). Although precipitates were rare, where visible they form small microcrysts (Fig. 3a). Material sorbed to cell walls typically contains 1-25 wt% silica, c. 20% sulphur, 50-60% antimony and 15-25% arsenic (Fig. 4a and b; open circles) (oxygen is not included in wt% calculations). Other biominerals in sediments collected from the SE shore of Champagne Pool are relatively more depleted in As, S and Sb, and are dominated by Si (up to c. 100%), Fe (up to 25%), and Na (up to 20%) (Fig. 4b, open circles).

The microbes in the orange floes and sediments that were collected from the NE shore of Champagne Pool (Site B) similarly show excellent preservation of cell walls through metal sorption onto the organic matrix (Fig. 5). As at Site A, the preservation of cytoplasmic material is not seen. The microbial population is dominated by rods of 1 -3 µm × 0.5 µm (Fig. 5). Where more extensive biomineralization occurs, the details of the cell wall morphology are destroyed, leaving a thicker, amorphous biomineral frame that simply records the original shape of the cell (Fig. 6). EDS analyses show that the cell walls are preserved in material rich in Si, S, Sb and As (Fig. 4a and b; open triangles). As the cell wall biomineralization thickens, the biominerals tend to become enriched in Si (<95 wt%) with some K. (<3 wt%), Fe (c. 18 wt%), and Al (<5 wt%), but depleted in Sb, As and S (Fig. 4b; open triangles).

Siliceous sinter from the submerged shelf

Samples from subaqueous orange sinter deposits around the pool margin appear to be dominated by small coccoid cells, 500 nm to 1 µm in diameter, and rods c. 1 µm long. Cell walls are preserved by extensive mineral replacement of cell wall material. These samples commonly show fossilized S-layers (Fig. 7), which are preserved in thick (200 nm), Al-rich (10wt%) silica (Fig. 4b; open squares). SAED shows that this phase is amorphous. Other details of cell morphology, such as cytoplasmic structure, are not preserved. EDS analyses show that metal immobilization varies from the sorption of Sb-As-S-enriched material to thicker Al-rich silica precipitates that commonly dominate the biomineralization that preserves the S-layers (Fig. 4a and b; open squares).

The morphology of preserved S-layers is depicted in Figure 8 as a framework of lines, with the lines outlining each subunit. There is a consistent pattern between the analysed S-layers, in both morphology and size. Measurements indicate an average unit-to-unit spacing of 9.7 ± 1.6 nm. These values fall within the expected range of 5-32 nm (Sleytr & Messner 1983; Sara & Sleytr 1987), supporting the interpretation that these mosaics are indeed S-layers.

Discussion

Arsenic and antimony speciation

EDS analysis of metals immobilized onto cell surfaces demonstrates considerably high concentrations of As (up to 33 wt%), Sb (up to 60wt%) and S (up to 20 wt%). Considering the concentrations of As, Sb and S in the spring water (Table 1 ), this demonstrates a considerable enrichment of these elements onto the cell surface. Furthermore, where As, Sb and S are immobilized onto the bacterial surface, they are immobilized in much greater concentrations than the other elements present in the spring water. However, they do not form extensive precipitates on the bacterial surface.

Arsenic is transported as either negative or neutrally charged thio-, oxide or hydroxide complexes in anoxic geothermal fluids (e.g. Mironova et al. 1984, 1990; Webster 1990; Zotov et al. 1994; HeIz et al. 1995; Cleverley et al. 2003; Nordstrom & Archer 2003; Webster & Nordstrom 2003). This is highlighted by calculating the probable species in the spring water at Champagne Pool using the GWB code. At the oxygen fugacity of Champagne Pool, arsenic species are likely to be dominated by hydroxide complexes such as H^sub 3^AsO^sub 3(aq)^ and thio-complexes such as HAs^sub 2^S^sub 4^^sup -^, As^sub 2^S^sub 4^^sup 2-^ and As^sub 2^S^sub 3(aq)^ (Fig. 9a) (this is in agreement with speciation calculations of similar geothermal fluids of the Uzon Caldera geothermal system; Cleverley et al. 2003). The spring water at Champagne Pool is close to the boundary between the sulphide-complex dominant (in particular HAs^sub 2^S^sub 4^^sup -^) and hydroxide-complex dominant (in particular H^sub 3^AsO^sub 3(aq)^) system (Fig. 9a) (this is based upon an assumption of equilibrium between the aqueous sulphur species; although it is likely the Champagne Pool spring water exhibits some degree of redox disequilibrium, the modelling results are still useful as a guideline to spring water chemistry for the purposes of this study). Other arsenic species that may occur under the low f O^sub 2(g)^ and sulphide-enriched conditions of Champagne Pool include H^sub 2^As^sub 3^S^sub 6^ (Webster 1990; HeIz et al. 1995; Nordstrom & Archer 2003; Webster & Nordstrom 2003), which would substitute for HAs^sub 2^S^sub 4^^sup -^ (Webster 1990), H^sub 2^AsS^sub 3^^sup -^ (HeIz et al. 1995) and AsO(SH)^sub 2^^sup -^ (HeIz et al. 1995; Nordstrom & Archer 2003). These do not appear in our speciation calculations as they were not incorporated into the database, but still demonstrate that negatively charged thio-complexes are likely to play an important role in arsenic transport in the spring water.

The immobilization of As, Sb and S onto bacterial surfaces may occur through either precipitation or adsorption. As shown in Figure 9b, all As-S and Sb-S minerals in the Champagne Pool spring water are undersaturated, except for orpiment, which exhibits a log Q/K. value close to zero under these conditions. It is therefore possible that the As content of the spring water is controlled by equilibrium with orpiment, and thus As-S immobilization onto the cell wall may be the result of orpiment precipitation. The failure, however, of SAED to detect crystalline orpiment on cell walls suggests that precipitation of orpiment is not an important mechanism of As-S immobilization onto cell surfaces here, although we cannot rule out the precipitation of poorly ordered As^sub 2^S^sub 3^ phases at this stage. Stibnite (Sb^sub 2^S^sub 3^) is undersaturated and would therefore require co-precipitation with the As^sub 2^S^sub 3^ phase to induce its immobilization. Although Sb is a common contaminant of orpiment (Webster 1990), such a coprecipitation mechanism appears unlikely here as often concentrations of immobilized Sb are greater than those of As (Fig. 4a). Alternatively, immobilization of As, Sb and S onto the cell surface may occur through adsorption onto cell surface functional groups. At the pH of the spring water (pH 5.2) bacterial surfaces will have abundant deprotonated, negatively charged carboxyl groups that typically deprotonate between pH 2 and 6 (Perdue 1985; Fein et al. 1997; Cox et al. 1999). Phosphoryl groups exhibit two deprotonation constants, one at c. pK^sub a^ 3 and one at c. pK^sub a^ 7, and therefore will provide both negatively and neutrally charged functionalities at this pH (Fein et al. 1997; Cox et al. 1999). Additionally, amine groups upon the bacterial surface will be positively charged at this pH, because they do not deprotonate to neutrally charged species until pH c. 9 (Cox et al. 1999; Sokolov et al. 2001). Under the conditions at Champagne Pool, aqueous arsenic is likely to form neutral or negatively charged sulphide and hydroxide complexes (Fig. 9a). The abundance of sulphur associated with cell wall immobilized As suggests that adsorption of As-S species could be important. Considering there are no positively charged As-S species to react with negative cell functionalities, it is possible that the adsorption of negative As-S (e.g. HAs^sub 2^S^sub 4^^sup -^) species onto positive amine groups may be a favourable immobilization mechanism. Calculations using GWB suggest that Sb species in the hot spring water are similarly likely to be dominated by either negative or neutrally charged sulphur and hydroxide species. At the estimated oxygen fugacity of the hot spring, the dominant species are HSb2S^sub 4^^sup -^ and Sb(OH)^sub 3^^sup 0^^sub (aq)^, in approximately equally concentrations. The Sb speciation results appear reasonable when compared with previous studies. For example, several studies suggested that Sb(OH)^sub 3^^sup 0^^sub (aq)^ is commonly the most important antimony-bearing complex (Wood 1987; Krupp 1988; Zotov et al. 1994), and would almost certainly be the most important at higher temperatures (200 °C or higher). At lower temperatures, however, such as the conditions at this hot spring (c. 75 °C), thioantimonite species such as HSb^sub 2^S^sub 4^^sup -^ become increasingly important (Kolpakova 1982; Krupp 1988). Furthermore, all antimony mineral phases are undersaturated (Fig. 9b). Thus, we tentatively suggest that antimony is also immobilized onto the bacterial polymers through adsorption. Considering the high concentrations of S associated with Sb on bacterial surfaces, it is possible that the adsorption of negatively charged thioantimonite complexes onto positive amine groups is important. The hypothesis that Sb and As are adsorbed onto bacterial polymers through interaction with positive amine groups is speculative. Clearly, these conclusions should be considered with caution and other adsorption mechanisms and precipitation processes are not ruled out at this stage.

In addition to the sorption of As, Sb and S onto cell wall and extracellular polymers, other cells exhibit more extensive biomineralization, dominated by amorphous silica. These varying degrees of mineralization probably reflect the duration of mineralization. Considering this, it is proposed that the early stages of biomineralization are dominated by the adsorption of As, Sb and S complexes. Then, as the reactive surface of the bacterium is masked, further adsorption is prevented and the precipitation of the supersaturated amorphous-silica phase becomes dominant (logQ/K = +0.18).

S-layer preservation and micmfbssil identification

S-layers, which form an ordered mosaic of proteins on the outer surface of the cell wall, can be found in both Eubacteria and Archaea. S-layers are commonly composed of hexagonal (p6), square (p4; found especially in Archaea), or oblique (p2) lattices with centre-to-centre spacing between morphological units of 532 nm (Beveridge 1981; Sleytr & Messner 1983; Sleytr et al. 1986a, b; Sara & Sleytr 1987; Engelhardt & Peters 1998). S-layers are commonly highly reactive, and contain abundant ionizable carboxyl and amine groups. Indeed, S-layers commonly contain significant amounts of lysine [(CH^sub 2^)^sub 4^-NH^sub 3^^sup +^] (Sara & Sleytr 1987), which is characterized by a free terminal amino group.

Previous work by Ferris et al. (1988) showed that Fe enhances bacterial preservation as it degrades the autolysins that are responsible for the breakdown of bacterial polymers. It is possible that heavy metals such as As and Sb may also enhance preservation because of autolysin degradation. Furthermore, the excellent preservation may have been enhanced because the hot spring waters are anaerobic, which inhibits the rapid oxidation of organic matter.

From knowledge of the S-layer morphology and size, and the ecological conditions, it is possible to suggest the microorganism that produced these S-layers. Reviews by Sleytr & Messner (1983) and Sleytr et al. (1986a) recorded more than 100 strains of Eubacteria and about 30 strains of Archaea that exhibit Slayers. Many of these microorganisms, however, will not inhabit this hot spring environment (i.e. pH 5.2, 75 °C, anaerobic). Furthermore, any microorganisms with S-layer unit-to-unit spacings much greater than the S-layers preserved here can be discarded. This reduces the numbers to a few strong candidates. A common bacterium possessing S-layers that is well documented from hot springs is the thermophilic anaerobe Clostridiiim (Sara et al. 1988; Krivenko et al. 1990; Karnauchow et al. 1992; Sonne-Hansen et al. 1993; Caganella & Trovatelli 1995). Most species of Clostridium grow at pH >6, whereas Champagne Pool has a pH between 5 and 6. The species Clostridium thermohydrosulfuricum, however, can grow at these lower pH values (Sonne-Hansen et al. 1993), and has been shown to grow well at 75 °C. Additionally, this microorganism is rod shaped, up to 0.6 nm wide, and has an oval terminal spore when sporulating (Sonne-Hansen et al. 1993). Similar rod-shaped bacteria with terminal spores have been found preserved in silica at Champagne Pool (Jones et al. 2001a). C. thermohydrosulfitricum shows a hexagonal S-layer fabric with a unit-to-unit spacing of about 14 nm, which is similar to the mosaic spacing of c. 10 nm recorded here. These values are close when considering distortions during preservation and shrinkage during TEM preparation. Furthermore, from a study of the surface charge characteristics of C. thermohydrosulfuricum by Sara et al. (1988), it seems clear that at the pH of the spring waters at Champagne Pool the S-layer surface would be dominated by abundant positively charged amine groups, which would be able to react with any negatively charged metal complexes in the spring water. A second anaerobic thermophile isolated from hot springs that shows a suitable S-layer is Desulfotomaculum nigrifacans, which has a square lattice with unit-to-unit spacings of 11 nm (Sleytr et al. 1986ft). Bacteria with square lattice Slayers are considered, as little distortion is needed to alter the apparent lattice structure of an S-layer. Such distortions could occur either during or after preservation.

The conditions at Champagne Pool are ideal for thermophilic, mildly acidophilic archaeans (see Jones et al. 2001a) such as the Sulfobales and Thcrmoproteales, both of which commonly display S-layers. The unit-to-unit spacings for these microorganisms, however, are between 17 and 31 nm (Sleytr et al. 1986a) and are therefore considered too large to be responsible for the S-layers preserved here.

Conclusions

From this study of microbe fossilization in Champagne Pool, it is clear that detailed ultrastructure on the scale of nanometres (here exemplified by the preservation of S-layers) can be preserved through mineralization at hot springs, provided the appropriate conditions are met. Those conditions include anoxia to retard organic degradation and an abundance of reactive aqueous metal species capable of rapid adsorption onto functional groups within the bacterial polymers. The adsorption of heavy metals probably precedes amorphous silica precipitation, highlighting the morphology of the microbe. Thus, it appears that rapid adsorption of heavy metals onto organic polymers is a prerequisite in enhancing microfossil preservation. Moreover, preservation of fine ultrastructures, such as the S-layers described here, enhances our ability to identify such microfossils preserved in the geological record.

This work was supported by grants from the Natural Sciences and Engineering Research Council (NSERC) of Canada to R.W.R. (GP0000629), BJ. (A6090) and F.G.F., and an Ontario Premier's Research Excellence Award to F.G.F. We thank the Department of Conservation, New Zealand, for permission to collect samples; A. Leinhardt and the staff of the Waiotapu Geothermal Site for kindly allowing us access; and M. Rosen (USGS) and B. Mountain (IGNS) for their help with this study. We thank R. Harris for his assistance with the TEM, and J. Cleverley for assistance with GWB. We thank D. Polya, J. Macquaker and C. Rice for their helpful reviews.

© 2005 Geological Society Publishing House Provided by ProQuest LLC. All Rights Reserved.