Natural synthesis of bioactive greigite by solid-gas reactions

Abstract Greigite, a ferrimagnetic iron sulfide Fe(II)Fe(III)2S4, is thought to have played an essential role in chemical evolution leading to the origin of life. Greigite contains a [4Fe–4S] cluster-like structure and has been synthesized in the laboratory by liquid-state reactions. However, it is unclear how greigite can be synthesized in nature. Herein, we show that greigite is synthesized by the solid–gas reaction of Fe(III)-oxide-hydroxides and H2S. We discovered that the hyperthermophilic hydrogenotrophic methanogen Methanocaldococcus jannaschii reduced elemental sulfur, and the resulting sulfide generated greigite from hematite. The time course and pH dependence of the reaction respectively indicated the involvement of amorphous FeS and H2S as reaction intermediates. An abiotic solid–gas reaction of hematite and H2S (g) under strictly anaerobic conditions was developed. The solid–gas reaction fully converted hematite to greigite/pyrite at 40–120 °C within 12 h and was unaffected by the bulk gas phase. Similar abiotic reactions occurred, but relatively slowly, with aqueous H2S in acidulous liquids using hematite, magnetite, or amorphous FeO(OH) as starting materials, suggesting that greigite was extensively produced in the Hadean Eon as these Fe(III)-oxide-hydroxides were shown to be present or routinely produced during that era. Surprisingly, the obtained greigite induced methanogenesis and growth of hydrogenotrophic methanogens, suggesting that the external greigite crystals enhanced reactions that would otherwise require enzymes, such as [4Fe–4S] cluster-harboring membrane-bound hydrogenases. These data suggested that the greigite produced by the solid–gas and solid–dissolved gas reactions was bioactive.


INTRODUCTION
Iron sulfides are thought to have played an essential role in chemical evolution (Wächtershäuser, 1992;Russell and Hall, 1997;Roldan et al., 2015). Among them, greigite (Fe(II)Fe(III) 2 S 4 ) contains the cubic Fe 4 S 4 unit (Rickard and Luther, 2007), similar to the [4Fe-4S] cluster that is crucial to life (Beinert et al., 1997). Fe-S clusters are biosynthesized by cysteine desulfurase (CDS) (Bandyopadhyay et al., 2008). However, numerous phylogenetic groups in the Euryarchaeota and Crenarchaeota phyla lack CDS genes (Liu et al., 2010). Thus, the origin of Fe-S clusters in the first life forms remains unknown. Although greigite is postulated as the origin of [4Fe-4S] clusters (Russell
Greigite can be synthesized in the laboratory by various methods (Berner, 1964;Wada, 1977;Dekkers and Schoonen, 1994;Chen et al., 2005;Cao et al., 2009;Paolella et al., 2011;Feng et al., 2013;Bauer et al., 2014). Among them, the polysulfide pathway, which involves mixing acidic Fe(II) and alkaline polysulfide solutions, is thought to represent the natural synthesis at alkaline hydrothermal vents (Russell and Hall, 1997). In this pathway, the initially formed amorphous FeS is converted to mackinawite (FeS), which is then converted, via the metastable intermediate greigite, to pyrite (FeS 2 ) (Berner, 1984;Hunger and Benning, 2007). In the conversion from mackinawite to greigite, Fe(II) is oxidized to Fe(III) under strictly anaerobic conditions in which polysulfides function as oxidizing agents (Berner, 1964;Rickard and Luther, 2007). This oxidation is thought to be the critical step in greigite synthesis. Interestingly, Fe(III)-oxide-hydroxides have not been tested as starting materials for greigite synthesis; they are expected to simply provide Fe(III).
Greigite is biologically synthesized extracellularly (Lefèvre et al., 2011) or intracellularly (Bertel et al., 2012) by sulfate-reducing bacteria (SRB). The product of sulfate reduction, sulfide, is crucial to the production of greigite. Sulfide is also produced by elemental sulfur (S 0 )-reducing (hyper)thermophilic heterotrophs; however, these organisms are not thought to produce greigite (Igarashi and Kuwabara, 2014). The third known sulfide-producer is methanogens; some of these organisms have been reported to reduce S 0 and produce sulfide (Stetter and Gaag, 1983). However, since methanogenesis and sulfidogenesis are thought to compete for reducing equivalents, S 0 -reducing ability is not considered useful for methanogenesis (Stetter and Gaag, 1983). Thus, whether methanogens produce greigite has not yet been studied to our knowledge. In the present study, we found that a hyperthermophilic hydrogenotrophic methanogen is a potent greigiteproducer and developed an abiotic solid-gas reaction to synthesize bioactive greigite based on synthesis by the methanogen. Analogous solid-dissolved H 2 S reactions likely reflect the natural synthesis of greigite because of its simplicity and versatility.

Materials
Powdered S 0 (Wako, Osaka, Japan) was suspended in distilled water at 3 mol/L and sterilized by Tyndallization; namely, the suspension was autoclaved twice at 110°C for 30 min with an interval of 24 h. Hematite (purity >97% [Wako] for growth medium; or purity = 99.9%, aver-age particle size = 0.3 lm [Kojundo Chemical Laboratories, Sakado, Japan] for the solid-gas reaction), magnetite (purity = 99%, particle size 61 lm; Kojundo Chemical Laboratories), and amorphous FeO(OH), prepared as described previously (Lovley and Phillips, 1986), were washed with and resuspended in distilled water at 0.6 mol/L, sterilized by autoclaving at 121°C for 20 min, and stored under air at room temperature. Amorphous FeS, prepared as described previously (Brock and Od'ea, 1977), was washed with distilled water that had been bubbled with N 2 gas for 20 min, resuspended in the degassed distilled water at 0.6 mol/L, sterilized by autoclaving at 121°C for 20 min, and stored at room temperature in an anaerobic workstation (Igarashi and Kuwabara, 2014) in which the gas phase was N 2 :H 2 :CO 2 (80:10:10).

Cultivation
The hyperthermophilic hydrogenotrophic methanogen Methanocaldococcus jannaschii JAL-1 (Jones et al., 1983) was obtained from the Japan Collection of Microorganisms (JCM) and maintained by successive cultivation in JCM 232 medium devoid of NaHCO 3 (Mc medium, pH 6.0) under a headspace of H 2 :CO 2 (80:20) at 80°C for 24 h. M. jannaschii was inoculated at 1 Â 10 5 cells/mL into 12 mL Mc medium supplemented with 312 mmol/L S 0 and 30 mmol/L hematite (Mc + S 0 + hematite medium) in 68-mL serum bottles in the anaerobic workstation. After replacing the headspace with H 2 :CO 2 (80:20), cultivation was performed at 80°C for 32 h unless otherwise stated. The pH of the medium was lowered to 5.6 before cultivation due to the dissolution of CO 2 .

Quantification of magnet-attracted substances
Cultures of M. jannaschii in Mc + S 0 + hematite medium were individually transferred to 15-mL disposable tubes, which were then capped after sealing with a piece of Teflon tape in the anaerobic workstation. The tubes were centrifuged at 1670g for 1 min. The supernatant was discarded, and the sediment was washed twice with the N 2bubbled distilled water and once with 70% ethanol. The sediment was then vacuum-dried using a centrifugal concentrator (CC-105; TOMY, Tokyo, Japan) and dried overnight at room temperature in a desiccator under N 2 . The dried sediment was ground with a mortar and pestle and weighed on a piece of powder paper. By applying a ferrite magnet purchased from a local market from underneath the powder paper, substances attracted by the magnet were transferred onto another piece of powder paper and weighed. The magnet-attracted substances were quantified as the weight percent of the total sediment. Notably, nonmagnetic substances bound to ferrimagnetic substances would be counted as magnet-attracted substances.

Solid-gas and solid-dissolved gas reactions
The 68-mL serum bottles were prepared in the anaerobic workstation, each accommodating 1 mL 1 M Na 2 S (without adjustment of the pH) and a test tube (/ = 1 cm, L = 5 cm) containing 0.63 mmol Fe in a dry mineral, either hematite (average particle size = 0.3 or 1.0 lm) or magnetite (particle size 6 1 lm). The dry mineral was prepared by subjecting the precipitates from centrifugation of the autoclaved suspension to drying in a clean oven at 50°C for up to 2 days. After sealing the serum bottle with a butyl rubber stopper (Nichiden-Rika Glass; Codes 309018) and an aluminum seal, the gas phase was replaced with Ar (purity = 99.999%), unless otherwise indicated, at ambient atmospheric pressure (1.0 atm). Gaseous H 2 S (H 2 S (g)) was generated in the serum bottles by injecting 0.55 mL 2 M H 2 SO 4 into the 1 M Na 2 S using a disposable Myjector syringe (TERUMO, Tokyo, Japan) ( Fig. 1). The volume of H 2 SO 4 was set so that the pH of the resulting solution became 5.6 ± 0.1 (n = 3), resembling the pH of M. jannaschii cultures. The serum bottles were incubated at 80°C , unless otherwise stated, for the designated periods. This type of reaction was designated as the solid-gas reaction type. In similar solid-dissolved gas reactions, the iron sources were suspended in 1 M Na 2 S, supplemented with H 2 SO 4 as described above, and reacted with aqueous sulfides. The volume of 2 M H 2 SO 4 was slightly different for some of the solid-dissolved gas reactions (e.g., 0.6 mL for amorphous FeO(OH) and 0.57 mL for amorphous FeS) to bring the pH of the solutions to 5.6. The products from amorphous FeO(OH) were washed with 0.5 N HCl and then washed six times with N 2 -bubbled distilled water.

X-ray diffraction analysis (XRD)
Culture sediments were collected by centrifugation at 1670g for 1 min. S 0 was mostly located within the lower part of the sediment. After discarding the supernatant, the upper part of the sediment was transferred using a spatula onto a silicon holder (Silicon Zero Background Sample Plate; SanyuShoko, Tokyo, Japan); this was expected to simplify the profile by eliminating the peaks derived from S 0 . Homogenized sediments were also analyzed to determine whether any other minerals were formed. The holder was sealed with polyimide film (Nilaco Corporation, Tokyo, Japan) and vacuum grease to avoid desiccation and possible oxidation during the analysis. XRD profiles were obtained using an X-ray diffractometer (Ultima IV; Rigaku, Tokyo, Japan) for CuKa radiation scanning at a step interval of 0.02°2h and a counting time of 2 s with a 2h range from 15°to 60°, operating at an accelerating voltage of 40 kV at 30 mA. Products of solid-gas and solidÀdissolved gas reactions were analyzed as described above, except that the polyimide film and vacuum grease were not used with the products of solid-gas reactions; the polyimide film caused a broad peak at approximately 20°. All the products were analyzed on the day of preparation. Obtained profiles were compared with the Powder Diffraction File TM (ICDD, 2011) for identification of minerals.

Field-emission scanning electron microscopy (FE-SEM)
Culture sediments were placed onto SEM glass plates (diameter, 18 mm; Okenshoji, Tokyo, Japan), which had been coated with 0.1% (w/v) poly-L-lysine. After standing for 30 min at room temperature, the specimens were fixed with 2% (w/v) glutaraldehyde in 0.2 M sodium cacodylate (pH 7.2) for 2 h at room temperature and processed as described previously (Kuwabara and Igarashi, 2012). Products of solid-gas and solidÀdissolved gas reactions were attached to carbon tape, which was placed on an aluminum stub. The specimens were observed using a field-emission scanning electron microscope (JSM-6330F; JEOL, Tokyo, Japan) as described previously (Igarashi and Kuwabara, 2014).

Other analytical methods
Cell density, methane, H 2 , sulfide, and Fe(II) were determined from three independent serum bottles unless otherwise stated, and duplicate measurements were averaged for each culture. Densities of free cells in the liquid phase were determined by direct counting based on the autofluorescence from coenzyme F 420 of the methanogen using an optical microscope (Eclipse E600; Nikon, Tokyo, Japan) (Igarashi and Kuwabara, 2014) and bacteria counting chambers. Methane and H 2 in the gas phase were determined using gas chromatography (GC-8A; Shimadzu, Kyoto, Japan) with a Molecular Sieve 5A 60-80 column (2.0 m; Shinwa Chemical Industries, Ltd., Kyoto, Japan) (Igarashi and Kuwabara, 2014).
Fe(II) and sulfide in culture sediments were quantified as follows. Culture media were centrifuged to separate sediments and supernatants, as described previously (Igarashi and Kuwabara, 2014). The sediments were washed with and resuspended in 12 mL of N 2 -bubbled water. During the homogenization of precipitates by swirling, 0.1-mL aliquots were transferred to 68-mL serum bottles in the anaerobic workstation. The serum bottles were capped with butyl rubber stoppers and aluminum seals, and the gas phases were replaced with N 2 (99.9999% pure). The samples were supplemented with 0.4 mL of 7.5 N HCl and each bottle was incubated at 100°C in an oven for 20 min to dissolve the amorphous FeS and greigite (Cornwell and Morse, 1987). A 0.1-mL aliquot of each sample was taken using a Myjecter syringe (TERUMO, Tokyo, Japan), 0.02 mL of which was added to 1 ml of 0.1% 3-(2-pyridyl)-5,6bis(4-phenylsulfonic acid)-1,2,4-triazine (ferrozine) in 50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)-NaOH (pH 7.0) (Sørensen, 1982) to determine the Fe(II) content by the ferrozine method, using an e 562 of 27,900 M À1 cm À1 (Stookey, 1970). For sulfide determination, 10 mL of 1.2% Zn(O 2 CCH 3 ) 2 (H 2 O) 2 (zinc acetate dihydrate) and 1.3 mL of 12% NaOH were added to the 0.4 mL of sample remaining in each serum bottle, which was then vigorously shaken to trap both gaseous and dissolved sulfides. Then, 2 mL of 0.1% N,N,-dimethyl-pphenylenediamine in 5.5 N HCl and 1 mL of 0.046 M FeCl 3 in 1.2 N HCl were added (Chen and Mortenson, 1977) to determine sulfide by the methylene blue method using an e 664 of 95,000 M À1 cm À1 (Cenens and Schoonheydt, 1988). Fe(II) and sulfide in the culture supernatants were similarly determined by the ferrozine (Stookey, 1970;Sørensen, 1982) and the methylene blue (Chen and Mortenson, 1977;Cenens and Schoonheydt, 1988) methods, respectively.

Greigite synthesis by M. jannaschii
A previous study showed that hematite had the ability to recover methane production by fermenter (Thermosipho globiformans)-methanogen (M. jannaschii) syntrophy in the presence of S 0 (Igarashi and Kuwabara, 2014), suggesting that non-nutrient environmental minerals could enhance microbial activities. Because we are interested in the reduction of hematite mediated by S 0 reduction and the feedback from the resulting iron sulfides to the biological system, we attempted to evaluate the ability of the methanogen to reduce S 0 .
M. jannaschii did reduce S 0 during growth at 80°C, as evidenced by the production of aqueous sulfides (Fig. 2b). The reduction did not significantly affect methanogenesis or growth (Fig. 2a), contrary to the previously reported mesophilic and thermophilic S 0 -reducing methanogens (Stetter and Gaag, 1983). Similarly, hematite alone had little or no effect on methanogenesis and growth, probably because it lacks cell permeability. Surprisingly, exposure to both of these substances almost quadrupled methanogenesis without affecting growth (Fig. 2a), suggesting the uncoupling of methanogenesis from ATP synthesis (Mountfort and Asher, 1979). During cultivation, the hematite turned black, suggesting its reduction to a possible iron sulfide by the methanogen-produced sulfide. This black product was not generated by the incubation of Mc + S 0 + hematite medium without the microorganism, and the amounts of sulfides solubilized from the sediments before and after the incubation were 0.19 ± 0.1 and 0.36 ± 0.1 lmol per serum bottle. The increase by the incubation (0.17 lmol/serum bottle) was far less than that by the cultivation (Fig. 2b), indicating little reduction of hematite without the microorganism. Interestingly, the black product was attracted by a magnet (Fig. 2c) but did not attract metal Fe. This ferrimagnetic product could be involved in the enhancement of methanogenesis.
Time courses of cultivation-dependent formation of ferrimagnetic product suggested that the iron sulfide initially formed at 12 h was nonmagnetic and then became ferrimagnetic after 16 h (Fig. S1a). These observations were supported by quantification of Fe(II) and sulfide, which were solubilized in 6 N HCl in butyl rubber-stoppered serum bottles containing magnetically attracted substances (Fig. S1b-d). When the nonmagnetic, intermediate preparations produced after 12 h were quantitated for Fe(II) and sulfides, their 6 N HCl-solubilizable substances showed an Fe(II)-to-S ratio of 1.0 (mol/mol), while these preparations showed only peaks of hematite upon XRD (Fig. 3a), suggesting that the intermediate was undetectable by XRD and thus likely to contain amorphous FeS. Sediments from the 16-h to 32-h cultures showed decreasing hematite and increasing greigite peaks with time (Fig. 3b, c), whereas those from the 96-h culture exhibited peaks of greigite and pyrite but not hematite (Fig. 3d). These results suggested that hematite was first reduced to amorphous FeS, which was then converted to greigite and subsequently to pyrite, similar to the sequence in the polysulfide pathway (Hunger and Benning, 2007) except for the absence of mackinawite. Among these minerals, only greigite is ferrimagnetic; thus, we concluded that the ferrimagnetism originated from greigite.
FE-SEM of the sediments from the 32-h cultures showed nanoflakes, which are characteristic of greigite (Cao et al., 2009) (Fig. 3f), while those from the 96-h cultures showed framboidal pyrite-like botryoids (Butler and Rickard,Fig. 3. Changes in minerals during cultivation. M. jannaschii was cultured in Mc + S 0 + hematite medium for the designated periods. The sediments were analyzed by XRD (a-d) and FE-SEM (e-g). Greigite, closed triangles. Hematite, closed squares. Pyrite, closed circles. The broad peak around 2h = 20°was derived from the polyimide film. The ordinate shows arbitrary signal intensity. A dividing M. jannaschii cell (M), as inferred from the cell shape, and an accompanying bridge (arrow) made from flagella (Igarashi and Kuwabara, 2014); hematite (H); greigite-like nanoflakes (G); and framboidal pyrite-like botryoids (P). Bars: 1 lm. 2000) and layered structures with smooth surfaces (Fig. 3g), which may have contained greigite; the nanoflakes were very rare in the 96-h sediments. None of these structures were observed in the sediments at 12 h; instead, complex petal-like structures were observed (Fig. 3e). Cells of M. jannaschii attached to hematite (Igarashi and Kuwabara, 2014), possible amorphous FeS (Fig. 3e), and the nanoflakes (Fig. 3f), suggesting that the sulfide emitted from the methanogen directly reduced the Fe(III) in the sediments.
The absence of greigite formation by S 0 -reducing heterotrophs (Igarashi and Kuwabara, 2014) implied the inhibition by organics. Inclusion of nonproteinaceous organics (i.e., glucose, glycine, lactate, acetate, Casamino acids, or soluble starch) at 0.3% (w/v) in the medium did not interfere with greigite formation (Fig. S2). However, addition of tryptone or yeast extract apparently inhibited greigite formation in a concentration-dependent manner, although growth and methanogenesis were not apparently inhibited. Sufficient greigite to accelerate methanogenesis was probably formed even in the presence of proteinaceous organics, as described below. Inhibition of the formation of appreciable amounts of greigite by proteinaceous organics is consistent with the previously observed inability of (hyper) thermophilic heterotrophs to produce greigite.

Greigite synthesis by solid-gas and solid-dissolved gas reactions
The pH of the culture medium (5.6) suggested that the sulfide involved in greigite formation was H 2 S (aq) based on the pH equilibrium profile of sulfides (Snoeyink and Jenkins, 1980). In fact, the pH dependence of cultivationmediated greigite formation (Fig. S3) resembled the pH equilibrium profile. Based on this finding, we developed an abiotic solid-gas reaction of hematite and H 2 S (g) to synthesize greigite, in which dry hematite in a test tube was reacted with H 2 S (g) generated from 1 M Na 2 S by the addition of 2 M H 2 SO 4 in serum bottles (the solid-gas reaction; Fig. 1). The solid-gas reaction at 80°C completely converted hematite to greigite and pyrite within 1 h (Fig. 4a). In contrast, when hematite was suspended in 1 M Na 2 S followed by the addition of 2 M H 2 SO 4 (the solid-dissolved gas reaction), only slight conversion occurred in the 1-h reactions (Fig. 4d), suggesting that reactions involving ionic sulfide species (HSand S 2-) were not responsible for the synthesis of greigite and pyrite. Prolonged solid-dissolved gas reactions for 3 h increased the amounts of greigite and pyrite, and decreased the amount of hematite (Fig. 4g), suggesting that essentially the same mechanism operated in the solid-gas and solid-dissolved gas reactions. The concentration of aqueous sulfides after the 1-h solid-gas reaction was 7.7 ± 0.8 mM, suggesting that the H 2 S (g) concentration at the start of reaction is calculated to have been 15 mmol/L, about 2-fold that of aqueous sulfides. When hematite with an average particle size of 1 lm, instead of 0.3 lm, was used for the solid-gas reaction, the reactivity decreased significantly (Fig. S4e), suggesting that internal moieties in hematite grains did not react during the limited reaction time. Greigite and pyrite also formed from magnetite, with increased formation observed in the solid-gas reaction than in the solid-dissolved gas reaction (Fig. 4b, e). Prolonged solid-dissolved gas reactions with magnetite increased the amounts of greigite and pyrite (Fig. 4h). These characteristics are similar to those of hematite. The 1-h solid-gas reactions of hematite and magnetite produced 0.25 ± 0.04 and 0.20 ± 0.02 lmol H 2 per serum bottle, respectively. Unlike the polysulfide pathway (Hunger and Benning, 2007), mackinawite was not detected in these reactions, even after short reaction times (Fig. S4a-d).
Desiccation of amorphous FeS and amorphous FeO (OH) caused the formation of hardened large grains unsuitable for solid-gas reactions. When wet amorphous FeS was subjected to the solid-dissolved gas reactions, only mackinawite formed, even after 144 h (Fig. S4f). The absence of the expected conversion from mackinawite to greigite suggested that mackinawite was not a precursor for greigite. In contrast, the amorphous FeO(OH) turned black upon suspension in Na 2 S and produced mackinawite and greigite in the 1-h solid-dissolved gas reaction (Fig. 4c). The difference in the reactivity between amorphous FeS and amorphous FeO(OH) suggested that inner moieties of amorphous FeO(OH) particles remained unreduced in 1 M Na 2 S and thus were able to provide Fe(III) for greigite synthesis in the solid-dissolved gas reaction. The mackinawite would be produced by a bypass reaction from amorphous FeS (Fig. S4f) and was removed by washing the product with 0.5 N HCl (Lovley and Phillips, 1986) (Fig. 4f). Prolonged solid-dissolved gas reactions of amorphous FeO(OH) produced pyrite, apparently with maintenance of the relative levels of mackinawite and greigite (Fig. 4i), consistent with the conversion of greigite to pyrite and the de novo synthesis of greigite, as in the polysulfide pathway (Hunger and Benning, 2007). No products showed XRD peaks for S 0 , a possible oxidation product from H 2 S. Thus, S 0 may not be a product of the solid-gas and solidÀdissolved gas reactions.
On FE-SEM, the products synthesized by the 1-h solidgas reaction of hematite and magnetite showed flower-like structures encompassing microcavities (Fig. 4j) and spherical structures composed of variously sized grains with rugged surfaces (Fig. 4k), respectively. The products obtained by the 1-h solid-dissolved gas reaction of amorphous FeO(OH) followed by washing with 0.5 N HCl showed structures resembling an aggregate of 400 nmlong rods (Fig. 4l), which are in the range of nanobacteria (50 nm to 500 lm in length) (Kajander et al., 2003).
Although the solid-gas reactions produced H 2 , the bulk gas phase of H 2 did not affect greigite synthesis (Fig. S5a). Similarly, N 2 , CO 2 , and CH 4 exhibited essentially no inhibition (Fig. S5c, e, f). As with ordinary chemical reactions, the higher the temperature, the faster the formation of greigite; for example, greigite synthesis was almost completed by 12 h at 40°C and by 0.5 h at 120°C (Fig. S5b, d).

Promotion of methanogenesis and growth of methanogens by greigite
To test the possible enhancement of methanogenesis by greigite, the products (0.63 mmol Fe) of solid-gas reactions prepared from hematite and magnetite, and of solid-dissolved gas reaction prepared from amorphous FeO(OH) followed by washing with 0.5 N HCl were individually tested by adding to Mc medium devoid of cysteine, a substrate of CDS involved in the biosynthesis of Fe-S clusters (Bandyopadhyay et al., 2008), although M. jannaschii was reported to have no CDS gene (Liu et al., 2010). Addition of either greigite preparation dramatically elevated stationary levels of methane concentration and the cell density to approximately 10-and 2-fold those in the absence of additives, respectively (Table 1). These results suggest that greigite promoted methanogenesis and growth, and the coexisting pyrite did not hinder the promotion. The concen-tration dependence indicated that sufficient enhancement of methanogenesis and growth occurred with 5 mg of the products from hematite (Table 1), and 1 mg was enough to promote methanogenesis to the level attained in the presence of 0.3% yeast extract (Fig. S2a).
The methanogenesis and growth effects might be related to the hydrogenotrophic nature and/or the S 0 -reducing ability of M. jannaschii (Fig. 2). Therefore, other methanogens, including the hydrogenotrophic methanogen Methanocaldococcus indicus capable of reducing S 0 (L' Haridon et al., 2003), the hydrogenotrophic methanogen Methanothermobacter thermautotrophicus (Zeikus and Wolee, 1972) incapable of reducing S 0 , and a heterotrophic Fig. 4. Greigite preparations produced by solid-gas and solid-dissolved gas reactions. Hematite, magnetite, and amorphous FeO(OH) were subjected to solid-gas or solid-dissolved gas reactions for 1 h, unless otherwise indicated. Solid-dissolved gas reactions were also conducted for 3 h (g-i). The products were analyzed by XRD (a-i) and FE-SEM (j-l). Greigite preparations produced by solid-gas (a, j) and soliddissolved gas (d, g) reactions with hematite, by solid-gas (b, k) and solid-dissolved gas (e, h) reactions with magnetite, and by solid-dissolved gas reaction with amorphous FeO(OH) (c, i) and the product from amorphous FeO(OH) after washing with 0. 5 N HCl (f, l) methanogen Methanolobus tindarius (Kö nig and Stetter, 1982) capable of reducing S 0 (Stetter and Gaag, 1983), were tested for any potential functional enhancement. Promotion of methanogenesis and growth was observed for M. indicus and Mt. thermautotrophicus but not for Ml. tindarius (Fig. 5), suggesting that the hydrogenotrophic nature, but not the S 0 -reducing ability, may be related to these effects of greigite.

Comparison of solid-gas and solid-dissolved gas reactions
In this study, we developed abiotic solid-gas and soliddissolved gas reactions of Fe(III)-oxide-hydroxides and H 2 S, referring to the greigite synthesis by M. jannaschii from S 0 and hematite. The reactions appeared to depend essentially on the concentration and motility of H 2 S. The higher reactivity observed in the solid-gas reactions than in the solid-dissolved gas reactions could be explained by the frequency of collisions between H 2 S and Fe(III)-oxide-hydroxides, as both concentration and motility were higher in H 2 S (g) than in H 2 S (aq). Furthermore, the presence of charges at the structured water layer on the mineral surface in the solid-dissolved gas reaction may have interfered with the reactions involving H 2 S (aq). During greigite synthesis in the solid-gas reactions, preservation of pre-existing Fe(III) under the completely anaerobic H 2 S atmosphere must be essential, as Fe(II) oxidation under anaerobic conditions is essential for the polysulfide pathway.

Formulas of greigite/pyrite synthesis in solid-gas and solid-dissolved gas reactions
The conversion of hematite to amorphous FeS would be expressed as follows, assuming SO 2 as the oxidation product of H 2 S; the other possible oxidation product, S 0 , was never detected in the reaction.
Amorphous FeS was apparently transformed to greigite but amorphous FeS alone was not found to be a good substrate for greigite synthesis (Fig. S4f), indicating that both hematite and amorphous FeS were necessary to synthesize greigite.
The greigite synthesis from hematite and H 2 S is expressed by removing FeS from formulas (1) and (2).
In contrast, greigite synthesis from magnetite could be explained by substituting S for O as an overall reaction via an intermediate.
The difference in the reaction sequence could explain the difference in the morphologies between the hematitederived flower-like structures and the magnetite-derived spherical structures of greigite/pyrite. Greigite synthesis from amorphous FeO(OH) would involve the formation of amorphous FeS from the reduction of amorphous FeO  (OH)). The whole product or a designated amount of preparation from hematite was added to the culture. Solid-gas reactions and methanogen cultivation were performed in triplicate, and the results are shown as mean ± SD. GPH and GPM, products of solid-gas reactions of hematite ( Fig. 4a) and magnetite (Fig. 4b), respectively. G, product of the solid-dissolved gas reaction of amorphous FeO(OH) followed by washing with 0.5 N HCl (Fig. 4f). Fig. 5. Enhancement of methanogenesis and growth of methanogens by greigite. Methanogens were cultured in the absence or presence of greigite, and methane concentrations in the gas phase (green bars) and densities of free cells (plots) were measured. GPH, products of solid-gas reactions of hematite (Fig. 4a). The entire product obtained from a 1-h reaction (0.63 mmol Fe source) was added to the cultures. Mean ± SD (n = 3). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) (OH) with H 2 S and the subsequent formation of greigite from amorphous FeO(OH), amorphous FeS, and H 2 S. The total reaction would be expressed as follows.
The mechanism of pyrite formation from greigite may not differ depending on the iron source for greigite synthesis; the S(-II) of H 2 S and greigite would reduce the Fe(III) in greigite, as in the case of pyrite synthesis from amorphous FeS and H 2 S (Drobner et al., 1990).
According to the above formulas, the syntheses of greigite/pyrite must produce stoichiometric amounts of H 2 , which should be usable as the reductive force for production of organics (Heinen and Lauwers, 1996;Cody et al., 2000). In the solid-gas reactions of hematite and magnetite, however, H 2 production was over three orders of magnitude smaller than that of iron sources. Furthermore, the reaction of hematite was not inhibited by the bulk gas phase of H 2 . These apparent discrepancies appear to be similar to that of pyrite synthesis (Drobner et al., 1990) and could be due to possible H 2 -adsorbing properties of greigite (Cao et al., 2009) and/or pyrite. To test this hypothesis, liberation of H 2 from the minerals must be tested by applying voltage (Cao et al., 2009).

Stimulation of hydrogenotrophic methanogens by greigite
Greigite enhanced the methanogenesis and growth of hydrogenotrophic methanogens. Considering that these methanogens have membrane-bound hydrogenases harboring [4Fe-4S] clusters for the extraction of electrons from environmental H 2 (Thauer et al., 2010), we hypothesize that the Fe 4 S 4 unit of greigite increased the amounts of these hydrogenases. The resulting increase in the hydrogenase activities may have contributed to the enhanced methanogenesis and growth. From a more speculative viewpoint, the Fe 4 S 4 structure of greigite could be incorporated into the cells, considering that the methanogenesis and growth require proteins harboring different types of Fe-S clusters, which are known to be interconvertible even in vitro (Beinert et al., 1997). However, incorporation of a solid structure into a membrane component has not been demonstrated, and thus further research is required to clarify the mechanism of the enhancements. Liu et al. (2010) suggested that Fe-S clusters of archaea without CDS genes, such as M. jannaschii, are derived from inorganic sulfide. This study strongly suggests that greigite is one of these inorganic sulfides.

Biogenic greigite/pyrite
SRB are thought to be the only biological producer of pyrite in paleontology (Astafieva, 2005) because of their sulfide-producing nature. However, because the pH optima of most SRB are not lower than 6.5 (https://www.dsmz. de/catalogues/catalogue-microorganisms.html), their greigite/pyrite synthesis must be slower than that by M. jannaschii. Considering the pH dependence of greigite synthesis and that most Methanocaldococcus species reported to date are suggested to grow well at pH 6.0 (http://jcm.brc.riken.jp/ja/catalogue), the possible involvement of Methanocaldococcus species in biological greigite/pyrite production cannot be ignored. Nevertheless, the predominance of SRB in biological greigite/pyrite synthesis seems to stand at non-hyperthermophilic temperatures, since most non-hyperthermophilic methanogens have pH optima in the range similar to that of SRB (https://www.dsmz.de/catalogues/catalogue-microorganisms. html) and thus are not expected to be involved in greigite synthesis, even if they can reduce S 0 .

Greigite synthesis in the Hadean Eon
Even in the anaerobic Hadean Eon, hematite was present (Hazen, 2013), and magnetite and amorphous FeO (OH) were routinely produced by the serpentinization of olivine (Janecky and Seyfried, 1986) and UV oxidation of oceanic surface Fe(II) (Braterman et al., 1983), respectively. These Fe(III)-oxide-hydroxides must have reacted with H 2 S (aq) emitted from acidic hydrothermal vents in the acidulous ocean (Macleod et al., 1994;Russell and Hall, 2006) to produce greigite. Greigite synthesis may have been more ubiquitous than was expected from the polysulfide pathway, given that the polysulfide pathway occurs only at alkaline hydrothermal vents, which are quite rare compared to acidic hydrothermal vents (Kelley et al., 2001). Greigite synthesis by the solid-gas reactions may have astrobiological significance. Fe-S clusters are necessarily present in organisms on Earth, and possibly in organisms on other planets. If greigite is the origin of Fe-S clusters, as postulated by Russell and Martin (2004) and supported by data from the present study, then the presence of greigite would be a prerequisite for the development of similar life forms. Results from the present study suggested that the presence of a combination of Fe(III)-oxide-hydroxides and H 2 S is sufficient to predict the occurrence of greigite.

CONCLUSIONS
The hyperthermophilic hydrogenotrophic methanogen M. jannaschii was found to be able to reduce S 0 . The resulting H 2 S (aq) produced amorphous FeS, greigite, and pyrite from extracellular hematite. Inspired by the greigite synthesis by M. jannaschii, we established conditions for abiotic reactions of hematite with H 2 S for greigite formation. Greigite formed faster with H 2 S (g) than with H 2 S (aq), indicating that sulfide ions may not be involved in greigite formation. Magnetite and amorphous FeO(OH) were also converted to greigite by reactions with H 2 S (g) and H 2 S (aq). The Fe(III) present in oxides and hydroxides, which was preserved under H 2 S-derived completely anaerobic conditions, is likely essential for greigite synthesis. The solid-gas reaction of hematite and H 2 S (g) occurred without preference for a pre-existing anaerobic gas phase and at temperatures as low as 40°C. From a practical aspect, the solid-dissolved gas reactions appeared to represent greigite synthesis in nature. The obtained greigite enhanced methanogenesis and growth of hydrogenotrophic methanogens, i.e., it was bioactive. This finding reinforces the perspective that greigite is the origin of the [4Fe-4S] cluster.

FUNDING
This study was supported by the Japan Society for the Promotion of Science KAKENHI Grant Number 26420881 to T.K. and by a Sasakawa Scientific Research Grant from The Japan Science Society, Grant Number 25-637, to K.I. The funding sources had no role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.