bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1 Sulfur reduction coupling with anaerobic ammonium oxidation drives proto-anabolic

2 networks

3 Peng Bao1, 2, *, Guo-Xiang Li1, 2, 3, Hu Li1

4

5 1Key Lab of Urban Environment and Health, Institute of Urban Environment, Chinese

6 Academy of Sciences, Xiamen 361021, P. R. China

7 2Ningbo Urban Environment Observation and Station, Chinese Academy of Sciences,

8 Ningbo 315800, P. R. China

9 3University of Chinese Academy of Sciences, Beijing 100049, P. R. China

10

11 Address correspondence to:

12 Dr. Peng Bao

13 Institute of Urban Environment, Chinese Academy of Sciences, Xiamen 361021, P. R.

14 China

15 E-mail: [email protected]; [email protected]

16

17

18 bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

19 Abstract

20 The geochemical energy that drives the transition from non-life to life is as yet

21 unknown. Here we show that thiosulfate/sulfate reduction coupling with anaerobic

22 ammonium oxidation (Sammox), could provide the primordial redox equivalents and

23 energy for prebiotic proto-anabolic networks, the reductive acetyl-CoA pathway

24 combined with incomplete reductive tricarboxylic acid (rTCA) cycle, reductive

25 amination and pyrrole synthesis. Fe-S mineral catalysis and thiols/thioesters as energy

26 couplers enhance the efficiency of prebiotic proto-anabolic networks in

27 thiosulfate-fueled Sammox reaction systems under hydrothermal conditions. Results

28 implied that thiosulfate-fueled Sammox should be the primordial reaction driving the

29 construction of proto-anabolic networks rather than sulfate-fueled Sammox, as it

30 could be catalyzed, and also sulfate would have been severely limiting in ancient

31 oceans. To confirm our findings, we isolated and identified a mixtrophic Sammox

32 bacterium, Ralstonia thioammoniphilus GX3-BWBA, which prefer thiosulfate to

33 sulfate. Genomic analysis of R. thioammoniphilus GX3-BWBA implied that this

34 ancient metabolism in modern microbes should contain two stages according to

35 ammonium transformation, —oxidation of ammonium to nitrite and denitrification.

36 The incomplete rTCA cycle and reductive acetyl-CoA pathway were all identified in

37 R. thioammoniphilus GX3-BWBA metabolic networks, which were responsible for

38 chemolithotrophic metabolism. We inferred that Sammox drove the coupling of the

39 biochemical transformation of C, H, O, N, S, and/or Fe simultaneously in Hadean

40 alkaline hydrothermal systems, thereby permitting the emergence of life. The results bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

41 bridged the gap in the transition from geochemistry to biochemistry.

42

43 bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

44 The transition from non-life to life occurred in the context of geochemical energy

45 derived from element-coupled transformation1-3. The chemistry of life is based on

46 redox reactions, that is, successive transfers of electrons and protons from the six

1,4 47 major elements, i.e., C, H, O, N, S, and P . The H2/CO2 redox couple, which can

48 occur in submarine hydrothermal vents, has been proposed to drive the reductive

49 acetyl-CoA pathway, an ancient metabolic route3,5-7. Nevertheless, the primordial

50 energy source of the H2/CO2 redox couple suffers from the difficulty that the

8 51 exergonic reaction competes with the endergonic reaction for available H2 . The

52 subsequent surface metabolism and thioester world theories still could not answer the

53 question, how the required reduced carbon compounds have been synthesized9,10.

54 Despite this limitation, these theories have emphasized the important roles of

55 thioesters and Fe-S mineral catalysis for driving the primordial rTCA cycle, a central

56 anabolic biochemical pathway whose origins are proposed to trace back to

57 geochemistry9,11,12.

58 Native iron/metals were recently shown to promote the reductive acetyl-CoA

59 pathway and rTCA cycle and strongly support the feasibility of these two primordial

60 synthetic pathways13,14, although they were generally considered to be rare near the

61 Earth’s surface13 and cannot support the long elemental transition from geochemistry

62 to biochemistry. Proto-anabolic networks consisting of the reductive acetyl-CoA

63 pathway together with the complete/incomplete rTCA cycle as primordial synthetic

64 pathways are therefore more logical3,15-18. However, the initial driving force for the

65 rise of proto-anabolic networks is still unclear. Moreover, the roles of N and S bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

66 geochemical transformation in the origin of life have been largely ignored because

67 computational analysis has implied that N and S are essential for thermodynamically

68 feasible phosphate-independent metabolism before the rise of last universal common

69 ancestor (LUCA)19. The proto-anabolic networks might be driven by C, H, O, N, S,

70 and/or Fe coupling transformation at their earliest stage of the transition from

71 geochemistry to biochemistry. The co-evolution of the metabolism of those elements

72 may provide strong explanatory power for the origin of life and explain why the

73 structure of metabolic networks is as it is. We therefore speculate that

74 thermodynamically feasible sulfurous species reduction coupling with anaerobic

75 ammonium oxidation reaction20,21,22,23,24 (Eqs. 1, 2; pH=8.0), with/without the

76 catalysis of Fe-S minerals and thioesters, may have been the primordial power force

77 for the rise of proto-anabolic networks.

78 8NH+2-- +2SO +2HCO→ 4N +2HS - +CH COO - +12H O+5H +∆G = -20.0 kJ mol-1 44 3 2 3 2 r (1)

79 4NH+2-- +S O +2HCO→ 2N +2HS- +2HCOO - +5H O+4H +∆G = -13.4 kJ mol-1 (2) 423 3 2 2 r

80 On early Earth, elemental sulfur, sulfite, and thiosulfate were produced

81 abundantly from volcanic and hydrothermal SO2 or from H2S oxidation by iron oxides

82 in sulfide-rich hydrothermal fluid2,15,25. In simulated hydrothermal systems with

26 83 conditions of 300 °C and high pressure, nitrite was readily converted to NH3 , and

84 Ni-Fe metals and alloys were also effective catalysts of N2 reduction to NH3 in

27 85 Hadean hydrothermal systems . There would have been much higher CO2

86 concentrations in the oceans on early Earth because there was perhaps up to 1000

28 87 times more CO2 in the atmosphere than that today . bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

88 Hence, when sulfurous species and NH3 in Hadean hydrothermal systems made

89 contact with CO2, there were spontaneous electron and proton transfers for energy

90 generation and organic molecule synthesis via sulfurous species reduction coupling

91 with ammonium oxidation. As sulfate would have been severely limiting in ancient

92 oceans2,29, we termed this process Sammox, and a Sammox microbe is more likely

93 use thiosulfate, elemental sulfur and sulfite as electron acceptor besides sulfate,

94 distinguishing it from previous studies of the coupling elemental biogeochemical

95 cycles1,30. A prebiotic reaction should occur with the ability to branch out into S and

96 N biochemistry, which could contribute to the autocatalysis and evolution of

97 primordial metabolic networks. Thus, Sammox may facilitate the synthesis of

98 thioesters and amino acids, which are essential for the self-amplification of

99 phosphate-independent metabolic networks before the rise of LUCA. We expect that

100 prebiotic chemical evidence of Sammox-powered CO2 fixation, thioesters, amino acid

101 and co-factors synthesis, combined with genetic analysis of a representative Sammox

102 microbe, will provide profound insights into the earliest origins of life and fill in the

103 missing link of the emergence of biochemistry from geochemistry.

104

105 Results and discussion

106 Sammox drives the combination of abiotic reductive acetyl-CoA pathway and

107 incomplete rTCA cycle

108 First, we aimed to verify the feasibility of Sammox-driven abiotic reductive

109 acetyl-CoA pathway, rTCA cycle, and off-cycle reactions (reductive amination and bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

110 co-factors synthesis) and to determine how methanethiol and Fe-S mineral catalysis

111 enhanced the reactions in both thiosulfate- and sulfate-fueled Sammox systems.

112 Formate and acetate were the products of Sammox-driven abiotic reductive

113 acetyl-CoA pathway in both thiosulfate- and sulfate-fueled Sammox systems with

114 bicarbonate as the carbon source (Fig. 1). We have not determined methanol but

115 qualitatively identified methyl acetate as a product in both thiosulfate- and

116 sulfate-fueled Sammox systems (Extended Data Fig. 1), implying that methanol

117 should be an intermediate of Sammox-driven abiotic reductive acetyl-CoA pathway.

118 More importantly, this result confirmed the capacity for Sammox-powered ester bond

119 formation, which was critical for the synthesis of complex organic molecules, such as

120 lipids and RNA31.

0.10 a formate acetate 0.08 succinate α-ketoglutarate 0.06

0.04

0.02 Concentration (mM)

0.00 0.10 b formate acetate 0.08 succinate α-ketoglutarate 0.06

0.04

0.02 Concentration (mM)

0.00

ls S ls S N S io Fe io N/ th S/ th S/ N/ S/ Fe N/ S/ 121 N/ 122 Figure 1. Products of simulated prebiotic Sammox-driven proto-anabolic

123 networks with bicarbonate as the sole carbon source under hydrothermal

124 conditions. Treatments were as follows (a, thiosulfate; b, sulfate), from left to

125 right: (i) sulfurous species, ammonium, Fe-S minerals, and methanethiol; (ii) bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

126 sulfurous species, ammonium, and Fe-S minerals; (iii) sulfurous species,

127 ammonium, and methanethiol; (iv) sulfurous species and ammonium; (v)

128 ammonium; and (vi) sulfurous species. The bar chart shows the yields of formate,

129 acetate, succinate, and α-ketoglutarate in each treatment group. Error bars

130 represent standard deviations of three replicates.

131

132 We haven’t detected pyruvate as the end-product of Sammox-driven abiotic

133 reductive acetyl-CoA pathway. When pyruvate was added into Sammox systems, it

134 quickly entered the incomplete rTCA cycle and was consumed as a reaction substrate

135 in both thiosulfate- and sulfate-fueled Sammox systems (Fig. 2, Fig. 3a, b). We

136 quantitatively identified succinate and α-ketoglutarate (Figs. 1, 2, Extended Data Fig.

137 2), and qualitatively identified polypyrrole (Extended Data Fig. 3) and glutamate

138 (Extended Data Fig. 4) as products of Sammox-driven abiotic incomplete rTCA cycle.

139 To further prove the feasibility of Sammox-driven abiotic incomplete rTCA cycle, we

140 adopted oxaloacetate, malate, fumarate and succinate, respectively, as substrates in

141 both thiosulfate- and sulfate-fueled Sammox systems, and all eventually produced

142 α-ketoglutarate and glutamate (Extended Data Fig. 5).

143 It is worth noting that the concentrations of α-ketoglutarate in N/S/FeS and N/S

144 groups were lower than the detection limit in thiosulfate-fueled Sammox systems with

145 bicarbonate as carbon source shown in figure 1 and 2. This was due to the reductive

146 amination of α-ketoglutarate to glutamate (Fig. 3b, Extended Data Fig 4). The

147 generation of polypyrrole implied that there should be pyrrole derived from succinate. bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

148 This might provide the possibility of biocatalysis emergence, as pyrrole was the

149 co-factors of enzyme. The above results implied that Sammox-driven abiotic

150 reductive acetyl-CoA pathway and abiotic incomplete rTCA cycle were combined and

151 tended to accumulate C4 and C5 products. We haven’t detected other intermediates of

152 the incomplete rTCA cycle, potentially because of their low productivity. There were

153 no reductions of pyruvate to lactate and α-ketoglutarate to α-hydroxyglutarate in both

154 thiosulfate- and sulfate-fueled Sammox systems, which would disrupt the incomplete

155 rTCA cycle16, therefore allowed Sammox-driven proto-anabolic networks to produce

156 major organic macromolecular products effectively.

1.0 a pyruvate 0.9 succinate 0.8 α-ketoglutarate 0.7 0.6 0.5 0.4

0.06

0.04

Concentration (mM) Concentration 0.02

0.00 1.0 pyruvate b succinate 0.8 α-ketoglutarate

0.6

0.4

0.06

0.04

Concentration (mM) 0.02

0.00

ls S ls S N S io Fe io N/ th S/ th S/ N/ S/ Fe N/ S/ 157 N/ 158 Figure 2. Products of simulated prebiotic Sammox-driven incomplete rTCA cycle

159 with bicarbonate and pyruvate as carbon sources under hydrothermal

160 conditions. Treatments were as follows (a, thiosulfate; b, sulfate), from left to

161 right: (i) sulfurous species, ammonium, Fe-S minerals, and methanethiol; (ii)

162 sulfurous species, ammonium, and Fe-S minerals; (iii) sulfurous species,

163 ammonium, and methanethiol; (iv) sulfurous species and ammonium; (v) bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

164 ammonium; and (vi) sulfurous species. The bar chart shows the concentrations

165 of pyruvate, succinate, and α-ketoglutarate in each treatment group. Error bars

166 represent standard deviations of three replicates.

a CO2

CO2

2e- 4e- H+ 4H+ CO2 H2O

b

- Aconitate Isocitrate - O O O O O O O O Oxalosuccinate - -Citrate - - O O O O Lipids OH O acetyl-CoA Glutamate -ketoglutarate Other amino acids pathway AcCoA CO2 rTCAcycle

Sammox Sammox Succinate CO2 Pyruvate Pyrroles Sugars Oxaloacetate Fumarate Alanine Sammox Sammox Malate

Pyrimidines Aspartate Sammox Other amino acids

c

S O n- NH + 167 x y 4

168 Figure. 3 Hypothetical prebiotic Sammox-driven reductive acetyl-CoA pathway

169 (a). Hypothetical proto-anabolic networks, a combination of acetyl-CoA pathway

170 together with incomplete rTCA cycle driven by Sammox, and the role of its bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

171 intermediates as universal biosynthetic precursors. Bar represents the final step

172 of incomplete rTCA cycle (b). Conceptual model of Sammox-driven the origin of

173 not free-living organic precursor on the surface of Fe-S minerals in Hadean

174 alkaline hydrothermal systems. Cubes represent Fe-S minerals/metals alloy.

175 Globes represent the not free-living organic precursor, ligand sphere (c).

176 Sulfate-fueled Sammox was more effective than thiosulfate-fueled Sammox

177 (Figs. 1, 2), due to the higher energy yield of the former reaction (Eqs. 1, 2). Both

178 Fe-S minerals and methanethiol could enhance the efficiency of the thiosulfate-fueled

179 Sammox reaction but not sulfate-fueled Sammox reaction (Fig 1), logically implying

180 that the thiosulfate-fueled Sammox reaction should be the primordial reaction driving

181 the construction of proto-anabolic networks, as it could be catalyzed. Note that

182 methanethiol was more effective than Fe-S minerals for promoting the

183 thiosulfate-fueled Sammox reaction (Fig. 1a). It incorporates into C, H, O and N

184 biochemical transformation via thiol-thioester exchange9, transforming itself and

185 other organic reaction products into new organic products10, therefore resulted in the

186 coupling biochemical transformation of C, H, O, N, and S and the expansion of

187 primordial metabolic networks. We deduced that when reaction conditions were met

188 in Hadean hydrothermal systems, Sammox-generated proton gradient could

189 modulate reduction potential similar to applying a voltage, leading to electrocatalytic

32,33 190 CO2 reduction with/without catalysis of Fe-S minerals and thioesters . Thus,

191 almost all kinds of the essential metabolic precursors for biosynthesis could be

192 provided in one geological setting. The organics formed within the vent pores in bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

193 Hadean hydrothermal systems, where they should concentrate via processes, such as

194 thermophoresis, and potentially form structures, such as lipid membranes lining

195 hydrophobic walls7,15,34. Thereby, Sammox gave rise to the origin of not free-living

196 organic precursor on the surface of Fe-S minerals35 (Fig. 3c).

197 As one of the products of Sammox, sulfide began to increase in the Hadean

198 oceans, facilitating thiols synthesis on Fe-S mineral surface32 and leading to de

199 Duve’s thioester world and the boom of S biochemistry. As a result, the S isotope

200 geochemical evidence can be traced back to 3.8 Ga and has been well

201 preserved4,25,36,37. Biogenic sulfide reacted with soluble Fe2+ and/or Ni2+/Co2+/Se to

202 maintain a continuous self-supply of freshly precipitated FeS and/or metals alloys and

203 even subsequent enzyme and redox protein active centers, such as ferredoxin38. This

204 may represent a route for S and Fe assimilation, which would facilitate escape of the

205 not free-living organic precursor from the surface of Fe-S minerals. As catalytic

206 properties improved, the yield of proto-anabolic networks increased, ultimately

207 extending the pathway through to C6 tricarboxylic acids. Sulfur biogeochemistry

208 prompted iron redox geochemistry, therefore, ferrous iron was released due to the

209 abiotic reduction of iron oxyhydroxide by biogenic sulfide, thus resulting in phosphite

210 liberation from ferruginous sediments39. In this study, we will not discuss how P

211 incorporated into primordial metabolic networks, but infer that Sammox-driven

212 proto-anabolic networks could facilitate self-evolution through optimizing the

213 ambient environment to make it sufficiently stable and habitable for life.

214 bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

215 Discovery of Sammox microbe

216 To prove the existence of Sammox process in modern microbes, we successfully

217 isolated a pure Sammox culture (designated GX3-BWBA) that could conserve energy

218 from the reduction of sulfurous species (thiosulfate, sulfate, sulfite, and elemental

219 sulfur) coupled to anaerobic ammonium oxidation. Strain GX3-BWBA appeared as

220 bean-shaped cells with a diameter of 0.6 μm and a length of 1.0 μm and harbored a

221 single flagella with a bacterial capsule (Fig. 4a, b, Extended Data Fig. 6).

222 Comparative sequence analysis of 16S rRNA genes revealed a 99% query cover and

223 99% sequence identity between strain GX3-BWBA and a series of uncultured

224 Ralstonia bacterial clones (Fig. 4c). Thus, strain GX3-BWBA was phylogenetically

225 clustered within the Ralstonia genus (Fig. 4c). GX3-BWBA grew to a maximal

226 density of 1.6 × 105 CFU ml-1 in anaerobic freshwater mineral medium containing

227 bicarbonate, ammonium, and thiosulfate/sulfate as the sole carbon and energy sources,

228 suggesting chemoautotrophy (Fig. 4b). Approximately 8 μM ammonium was

229 sufficient to support the chemoautotrophic growth of strain GX3-BWBA,

230 demonstrating high affinity for ammonium (data not shown). Thiosulfate was the

231 preferred electron acceptor over sulfate by GX3-BWBA (Fig. 4b). There was no

232 significant growth of GX3-BWBA under conditions without ammonium or sulfurous

233 species (Fig. 4b); thus, ammonium and sulfurous species were essential factors for

234 GX3-BWBA chemoautotrophic growth. We provisionally classified this Sammox

235 bacterium as “Ralstonia thioammoniphilus” (thi.o.am.mo.ni′ phi.lus. Gr. n. thion

236 sulfur; L. neut. n. sal ammoniacum salt of Ammon (NH4Cl); Gr. adj. phylos loving; bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

237 N.L. neut. adj. thioammoniphilus sulfur-ammonium-loving).

) 18 2- +

4 S O +NH +cell a 2 3 4 SO 2-+NH ++cell b 16 4 4 NH ++cell 4 14 S O 2-+cell 2 3 2- 12 SO4 +cell 10 8 6 4 2

Colony-forming-unit (CFU/ml, 10 (CFU/ml, Colony-forming-unit 0 0 5 10 15 20 25 30 35 40 Time (d)

) 5 ) 5 -1 -1 d d -1 -1 c d 4 4

3 3

2 2

1 1 production rate (pmol ml production rate (nmol ml (nmol rate production 2 2 N N 30 29 0 0 S O 2- 2- 2- 2- 2 3 SO3 S2O3 SO3

Ralstonia oxalatica AF155567.1 Ralstonia taiwanensis strain LMG 19424 AF300324.2 Ralstonia sp. AU3313 AF500583.1 Ralstonia gilardii AF076645.1 Ralstonia basilensis strain DSM 11853 AF312022.1 Ralstonia metallidurans strain DSM 2839 D87999.1 Ralstonia paucula AF085226.1 Ralstonia campinensis strain WS2 AF312020.1 Ralstonia pseudosolanacearum strain UQRS KC757037.1 Ralstonia solanacearum strain LMG 2299 EF016361.1 e Ralstonia thomasii strain LMG6866 AJ270258.1 Ralstonia syzygii strain ATCC 49543 AB021403.1 Ralstonia syzygii strain R001 U28237.1 Ralstonia syzygii subsp. celebesensis strain UQRS KC757073.1 Ralstonia syzygii subsp. indonesiensis strain UQRS KC757057.1 Ralstonia sp. AU2944 AF488779.1 Ralstonia sp. RS2 AB503703.1 Ralstonia pickettii strain ATCC 27511 Y741342.1 Ralstonia pickettii strain 4902 KT933222.1 Ralstonia sp. W7 KF560393.1 Ralstonia sp. S1SM82 KT183537.1 Uncultured bacterium clone E39 HQ827943.1 Ralstonia thioammoniphilus strain GX3-BWBA QMKS01000000 Uncultured bacterium clone SW1112-15 KM269680.1 Uncultured soil bacterium clone TIIF1 DQ297956.1 Ralstonia pickettii 12J CP001068.1 Ralstonia pickettii 12D CP001645.1 Ralstonia pickettii 12J P001069.1

0.1 238 bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

239 Figure. 4 Characters of R. thioammoniphilus GX3-BWBA. TEM image of cells

240 (a). Scale bar represents 0.2 μm. Growth of GX3-BWBA under combined

241 condition of ammonium and sulfurous species, with bicarbonate as sole carbon

242 sources (b). Production rate of isotopically labeled dinitrogen for thiosulfate and

243 sulfate-fueled Sammox by GX3-BWBA (c, d). Phylogenetic tree constructed by

244 the maximum likelihood method, using 1478 nucleotides from 16S rDNA

245 sequences, showing the position of strain GX3-BWBA in relation to members of

246 the genus Ralstonia. The bar represents 0.1 changes per nucleotide position (e).

247

248 As a probable ancient metabolism, Sammox microbe preferred other sulfurous

249 species, except for sulfate, potentially because elemental sulfur, thiosulfate, and

250 sulfite were produced abundantly from ancient volcanic and hydrothermal SO2 or

251 from sulfide oxidation by iron oxides in sulfide-rich hydrothermal fluid2,15,25.

252 Thiosulfate constituted 68–78% of the immediate HS- oxidation products and was

- 2- 40 253 involved in a dynamic HS -S2O3 cycle in anoxic marine and freshwater sediments .

254 In contrast, sulfate would have had only very limited, localized significance2.

255 We further tested thiosulfate- and sulfate-fueled Sammox metabolism using R.

256 thioammoniphilus, and the end product N2 generation rate was measured (Fig. 4c, d).

257 The generation rate of N2 in the thiosulfate treatment group was slightly higher than

29 258 that in the sulfate treatment group (Fig. 4c, d). The production rate of N2 was about

-1 -1 30 259 2.5 nmol ml day , which was significantly higher than N2 (approximately 2.5 pmol

260 ml-1 day-1; Fig. 4c, d). Figure 5 shows the coupled dynamic transformation of bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

261 sulfurous species and ammonium during Sammox process. In thiosulfate-fueled

262 Sammox, sulfide was the product of the thiosulfate reduction, whereas sulfite was an

263 intermediate (Fig. 5a). Ammonium oxidation accompanied by thiosulfate reduction

264 was significant compared with that in the control group. As intermediate of

265 ammonium oxidation, nitrite significantly increased on day 27 (Fig. 5b). Based on the

266 low concentration, we did not determine the dynamic generation of dinitrogen to

267 avoid leakage. There was weak but significant sulfate reduction in the sulfate-fueled

268 Sammox group, accompanied by sulfide generation (Fig. 5c). The concentration of

269 nitrite also showed a slight increase during the ammonium oxidation process (Fig. 5d).

270 The concentrations of sulfurous species and ammonium showed no significant

271 changes in the control group (Fig. 5). The assumed net thiosulfate-/sulfate-fueled

272 Sammox reactions are shown in Eqs. 1 and 2. The generation of precursors of CO2

273 fixation was thermodynamically favorable by Sammox. bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

a 3.2 b 1.0 2.8

2.4 Control S O 2- 2 3 2.0 2- 0.8 S2O3 - 1.6 Control HS Control NH + NH + - 4 4

HS - - 1.2 2- 0.15 Control NO2 NO2 Control SO3 2- 0.8 SO3 0.10 Concentration (mM) Concentration Concentration (mM) 0.4 0.05 0.0 0.00 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40

Time (d) Time (d) c 3.2 d 1.0

3.0

0.8 Control SO 2- 2.8 4 SO 2- + + 4 Control NH4 NH4 - - -

HS Control NO NO 0.045 2 2

0.05 0.030 Concentration (mM) Concentration (mM) 0.00 0.015

-0.05 0.000 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40 Time (d) Time (d) 274

275 Figure. 5 Thiosulfate- and sulfate-fueled Sammox by R. thioammoniphilus

276 GX3-BWBA. (a) thiosulfate and reduction products. (b) ammonium and its

277 oxidation products in thiosulfate-fueled Sammox. (c) sulfate and reduction

278 products (sulfide and sulfite as products of sulfate in control group are below

279 detection limit). (d) ammonium and its oxidation products in sulfate-fueled

280 Sammox. Error bars represent standard deviations of three biological replicates.

281

282 Notably, quite a few iron oxides could be abiotically reduced by sulfide

283 generated from Sammox in an iron-rich environment, then an illusion of ferric iron

284 reduction coupled to anaerobic ammonium oxidation (Feammox) arises41-43. Indeed, it

285 is still unclear to what extent Sammox overlaps with Feammox; however, we can

286 infer that Sammox may be widely distributed in different anoxic environments based

287 on the ubiquity of Ralstonia spp. and its high affinity for ammonium. Moreover, as an bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

288 ancient type of metabolism, there may be a variety of microbes that still preserved this

289 living strategy.

290

291 Genomic and N, S, and C metabolic gene analyses

292 To elucidate the mechanisms of Sammox reaction in modern microorganisms,

293 we sequenced the genome of R. thioammoniphilus (Extended Data Table 1, Extended

294 Data Fig. 7). There were denitrification genes in the R. thioammoniphilus genome,

295 including copper-containing nitrite reductase (nirK), nitric oxide reductase (norB),

296 and nitrous oxide reductase (nosZ) (Fig. 6a). This result confirmed that nitrite was the

297 expected intermediate product of Sammox and that the thiosulfate-/sulfate-fueled

298 Sammox reactions in R. thioammoniphilus contained two stages: oxidation of

299 ammonium to nitrite (Eqs. 3 and 4)21, followed by the final step of the denitrification

300 pathway which is reduction of nitrite to dinitrogen (Eq. 5).

301 4NH+2 +3S O-→ 4NO -- +6HS +H O+8H + (3) 423 2 2

302 4NH+ +3SO2- → 4NO-- +3HS +4H O+5H+ (4) 44 2 2

303 +2NO-+- +8H +6e→ N +4H O 222 (5)

304 We did not find any other anaerobic ammonium oxidation-related functional

305 genes in the genome of R. thioammoniphilus (Fig. 6a). Thus, the ammonium oxidation

306 in Sammox was different from that in complete ammonium oxidation (Comammox)

307 and anaerobic ammonium oxidation (Anammox)30. bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

a

b

1 308

309 Figure. 6 Representative metabolic gene clusters from R. thioammoniphilus

310 strain GX3-BWBA (a). The standard configuration of metabolic gene clusters

311 (MGCs) in primary metabolism located in two scaffolds. MGCs contain enzymes

312 of CO2 fixation genes, including 31 rTCA cycle, and 30 reductive

313 acetyl-CoA pathway genes, 6 dissimilatory sulfur metabolic genes, and

314 5 denitrification genes. Gene size and spacing are not to scale. Each gene was not

315 condensed lie in genome. Protein phylogenetic tree derived from 26443 amino

316 acid sequences of sulfite/nitrite reductase (b). Scale bars represent estimated

317 sequence divergence or amino acid changes. Archaea, Bacteria,

318 Eukaryota.

319

320 Our main concern is how ammonium was oxidized by thiosulfate/sulfate into bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

321 nitrite. Sammox in R. thioammoniphilus may still need an iron-sulfur catalyst that

322 inherited from proto-anabolic networks. As lacking of genetic evidence about

323 Sammox, we predicted that ferredoxin oxidoreductase (e.g., nitrite/sulfite reductase

324 ferredoxin domain protein and ferredoxin-nitrite reductase) or other iron-sulfur

325 enzymes encoded by genes with unknown sequences may mediate sulfur reduction

326 coupling with ammonium oxidation to nitrite (Eqs. 3–6, Extended Data Schematic 1).

327 Here, thiosulfate should be first transformed into sulfite, which could be reduced by

328 ferredoxin. Hence, Sammox reversed the nitrite reduction process. This process was

329 quite similar to a nitrite generation pathway in Feammox process (Eq. 7)30,41,42,

330 implying its theoretical feasibility. Electrons and protons generated in the first stage of

331 Sammox reaction were used to drive nitrite reduction to dinitrogen. As a highly

332 exergonic reaction, energy released from nitrite reduction drove CO2 fixation.

333 NH+ +6 oxidized ferredoxin+2H O→ NO- +6 reduced ferredoxin+8H+ 422 (6)

334 NH+ +6Fe3+ +2H O→ NO- +6Fe 2+ +8H+ 422 (7)

335 However, so far we do not clearly know the molecular mechanisms of sulfate

336 reduction coupling with ammonium oxidation to nitrite. R. thioammoniphilus contains

337 no ATP sulfurylase (Sat)-coding gene (Fig. 6a), suggesting that sulfate reduction in

338 sulfate-fueled Sammox is different from dissimilatory sulfate reduction. Dissimilatory

339 sulfate reduction may be later than sulfate reduction in sulfate-fueled Sammox,

340 according to isotope geochemical evidence25,44,45. That may be the reason sulfite

341 reductase (Extended Data Table 2, Extended Data Fig. 8) showed a much more

342 extensive distribution than Sat (Extended Data Table 3, Extended Data Fig. 9) and, bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

343 also the evidence for ancient metabolic networks led to the emergence of living

344 systems prior to the incorporation of P19.

345 Our results thus far suggest that the transition from geochemistry to biochemistry

346 of S and N may have started from the same point. Sulfite and nitrite reductases belong

347 to the only class of enzymes that share a common architecture as well as a

348 requirement for a siroheme cofactor46. The phylogenetic tree indicated that the

349 putative multifunctional sulfite/nitrite reductase of archaea and eukaryota gathered

350 into one and three clusters, respectively. Putative interdomain lateral gene transfer

351 may result in the distribution of some archaea and eukaryota multifunctional

352 sulfite/nitrite reductases into bacterial clusters (Fig. 6b). Similar to sulfite reductase,

353 nitrite reductase is widely distributed in all types of modern organisms (Extended

354 Data Table 4, Extended Data Fig. 10), and indeed, sulfite reductases from some

355 sources can catalyze the reduction of both sulfite and nitrite46. This phenomenon

356 suggested that S and N biochemistry may have a common evolutionary origin derived

357 from Sammox.

358 Similar to the sulfite/nitrite reductase, some biomolecules as remnants of

359 Sammox and ancient metabolism may be hidden in the architecture of the metabolic

360 networks of R. thioammoniphilus. In theory, such “metabolic fossils” for the origin of

361 the coupled transformation of elements should be widely distributed in all types of

362 modern organisms as well. Thiosulfate may be the most preferential electron acceptor

363 for Sammox in Hadean hydrothermal systems, leading to the wide distribution of

364 multifunctional rhodanese (thiosulfate sulfurtransferase) in prokaryotes, eukaryotes, bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

365 and archaea (Extended Data Table 5, Extended Data Fig. 11). Rhodanese has been

366 proposed to have an assimilatory role using dithiol dihydrolipoate as the sulfur

367 acceptor and acting as a sulfur insertase involved in the formation of prosthetic groups

368 in iron-sulfur proteins, such as ferredoxin47,48. This implies that rhodanese may be the

369 primary mechanism for the formation of the iron-sulfur center of primordial enzymes

370 to catalyze proto-anabolic networks in LUCA, thus indicating its ancient nature.

371 Rhodanese can also catalyze a sulfur dissimilatory metabolic reaction which is

372 thiosulfate cleavage to sulfite48 (Scheme 1), thereby facilitate thiosulfate involve in

373 the first stage of Sammox.

374 2––→ 22 – – S23 O +2R-S SO 3 +R-S-S-R+S

375 Scheme 1

376 The structural similarity between rhodanese and Cdc25 phosphatases indicated

377 the common evolutionary origin of the two enzyme families49; alternatively,

378 phosphatases may originate from the rhodanese family because P incorporation

379 occurred later than S incorporation in metabolic networks. This evidence implied the

380 relationship between S and P metabolic evolution.

381 Six candidate genes encoding components of the reductive acetyl-CoA pathway

382 were identified as formate dehydrogenase (fdoI), 5,10-methenyl-H4 folate

383 cyclohydrolase (folD), 5,10-methylene-H4 folate dehydrogenase (folD),

384 5,10-methylene-H4 folate reductase (metF), CO dehydrogenase/acetyl-CoA synthase

385 (acs), and pyruvate:ferredoxin oxidoreductase (PFOR) (Fig. 6a). Three gene

386 homologs 10-formyl-H4 folate synthetase, methyl-H4 folate: corrinoid iron-sulfur bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

387 protein methyltransferase, and corrinoid iron-sulfur protein appeared to be missing. It

388 is possible that the present genomic analysis was insufficient to distinguish whether R.

389 thioammoniphilus could facilitate CO2 fixation via the incomplete reductive

390 acetyl-CoA pathway or there were functional related enzymes encoded by genes with

391 unknown sequences50.

392 In annotations of the rTCA cycle, two essential gene homologs appeared to be

393 missing (Fig. 6a). One of the missing genes, α-oxoglutarate synthase, catalyzes

394 reductive carboxylation from succinate to α-ketoglutarate, and the other is citryl-CoA

395 synthetase. Two potential reversible ATP-dependent citrate lyase subunits (ACL)

396 (citE, mdcC) were identified (Fig. 6a), which was a key indication for autotrophic

51 397 CO2 fixation via the rTCA cycle instead of citrate synthase . These results suggested

398 the potential of R. thioammoniphilus for carbon fixation via the incomplete reductive

399 acetyl-CoA pathway and rTCA cycle.

400

401 Final remarks: from metabolic innovation to evolution

402 In this study, we provided strong evidence of abiotic Sammox-driven

403 proto-anabolic networks and a Sammox microbe, R. thioammoniphilus. Fe-S mineral

404 catalysis and thiols/thioesters as energy couplers enhance the efficiency of prebiotic

405 proto-anabolic networks in thiosulfate-fueled Sammox reaction systems but not in

406 sulfate-fueled reaction systems under simulative hydrothermal conditions. Results

407 implied that the prebiotic thiosulfate-fueled Sammox reaction should be the

408 primordial reaction driving proto-anabolic networks, as it could be catalyzed. As a bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

409 probable ancient metabolism, Sammox microbe prefer thiosulfate to sulfate,

410 potentially because thiosulfate and/or elemental sulfur, and sulfite were produced

411 abundantly from ancient volcanic and hydrothermal SO2. Our study was partially

412 supported by the theory of a chemoautotrophic origin of life supported by surface

413 metabolism and a primordial iron-sulfur world9,11. Here, Fe-S minerals and/or metal

414 alloy catalysis enhanced product generation of prebiotic thiosulfate-fueled Sammox

415 rather than providing hydrogen for CO2 reduction. The Fe-S mineral surfaces, where

416 transition metal ions, such as Ni2+, Co2+, or Se, are catalytically active52, and also the

417 very place facilitated prebiotic Sammox-driven amnio acids, thioesters and co-factors

418 synthesis. Therefore, a mixture of Fe-S minerals and transition metal ions may have

419 been the evolutionary precursor of the enzymatic active center, when chelated by

420 amnio acids and/or co-factors, converted to biocatalysts33,38. In this regard, our study

421 could explain the occurrence of enzymes with Fe-S reaction centers in the electron

422 transport chains of most known extant bioenergy flows in all three kingdoms of life.

423 Thioesters derived during prebiotic Sammox reaction provided the energetic and

424 catalytic framework of prebiotic Sammox, and feedback and feedforward to prebiotic

425 Sammox-driven proto-anabolic networks, ultimately leading to metabolic

426 reproduction and innovation52. Thus, prebiotic Sammox drove the emergence of the

427 primordial structure and function of the not free-living organic precursor, a ligand

428 sphere, held together by bonding to the surfaces of Fe-S minerals (Fig. 3c). The

429 feedback coupling between the primordial metabolic networks and their environment

430 could shape the evolution of both53. As a self-regulating system, C, H, O, N, S, and/or bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

431 Fe coupling metabolic networks could facilitate self-evolution through optimization

432 of the ambient environment, making it stable and habitable for life. The first and

433 foremost goal of Sammox-driven primordial metabolic networks was to obtain

434 sufficient phosphorus. Prebiotic Sammox-driven sulfur biochemistry prompted iron

435 redox geochemistry; therefore, phosphite was liberated from ferruginous sediments39,

436 permitting biochemistry feedback to geochemistry. When P reached a certain

437 concentration in the Hadean oceans, sulfur biochemistry could lead to the emergence

438 of phosphorus biochemistry, and the RNA world could have replaced the thioester

439 world.

440 Our findings regarding Sammox-driven proto-anabolic networks suggested that

441 primordial metabolic networks might rise from coupling transformation of C, H, O, N,

442 S, and/or Fe at their earliest stage of the transition from geochemistry to biochemistry.

443 The elements of the chemical components that supporting prebiotic Sammox-driven

444 proto-anabolic networks ultimately became essential or trace elements of life,

445 providing a strong explanation for the unique features of life. Exploration of Sammox

446 provides a new perspective for understanding the emergence of biochemistry from

447 geochemistry and highlights the fundamental significance of Sammox for the origins

448 of life in hydrothermal environment on planetary systems.

449

450

451

452 bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

453 References

454 1. Falkowski, P. G., Fenchel, T. & Delong, E. F. The microbial engines that drive

455 earth's biogeochemical cycles. Science 320, 1034–1039 (2008).

456 2. Sousa, F. L. et al. Early bioenergetic evolution. Philos. Trans. R. Soc. B-Biol. Sci.

457 368, (2013).

458 3. Weiss, M. C. et al. The physiology and habitat of the last universal common

459 ancestor. Nat. Microbiol. 1, (2016).

460 4. Moore, E. K. et al. Metal availability and the expanding network of microbial

461 metabolisms in the Archaean eon. Nat. Geosci. 10, 629–636 (2017).

462 5. Ljungdah, L., Irion, E. & Wood, H. G. Total synthesis of acetate from CO2. I.

463 Co-methylcobyric acid and CO-(methyl)-5-Methoxybenzimidazolylcobamide as

464 intermediates with Clostridium thermoaceticum. Biochemistry 4, 2771–2780

465 (1965).

466 6. Kelley, D. S. et al. A serpentinite-hosted ecosystem: the lost city hydrothermal

467 field. Science 307, 1428–1434 (2005).

468 7. Lane, N. & Martin, W. F. The origin of membrane bioenergetics. Cell 151,

469 1406–1416 (2012).

470 8. Wächtershäuser, G. Pyrite formation, the first energy source for life: a

471 hypothesis. Syst. Appl. Microbiol. 10, 207–210, (1988).

472 9. Wächtershäuser, G. Evolution of the first metabolic cycles. PHAS J. 1, 200–204

473 (1990).

474 10. de Duve, C. Blueprint for a cell: the nature and origin of life. (Neil Patterson , bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

475 1991).

476 11. Evans, M. C. W., Buchanan, B. B. & Arnon, D. I. A new ferredoxin-dependent

477 carbon reduction cycle in a photosynthetic bacterium. Proc. Natl. Acad. Sci. U. S.

478 A. 55, 928–934 (1966).

479 12. Smith, E. & Morowitz, H. J. Universality in intermediary metabolism. Proc.

480 Natl. Acad. Sci. U. S. A. 101, 13168–13173 (2004).

481 13. Muchowska, K. B. et al. Metals promote sequences of the reverse krebs cycle.

482 Nat. Ecol. Evol. 1, 1716–1721 (2017).

483 14. Varma, S. J. et al. Native iron reduces CO2 to intermediates and end-products of

484 the acetyl-CoA pathway. Nat. Ecol. Evol. 2, 1019–1024 (2018).

485 15. Martin, W. & Russell, M. J. On the origin of biochemistry at an alkaline

486 hydrothermal vent. Philos. Trans. R. Soc. B-Biol. Sci. 362, 1887–1925 (2007).

487 16. Orgel, L. E. The implausibility of metabolic cycles on the prebiotic earth. PLoS.

488 Biol. 6, 5–13 (2008).

489 17. Braakman, R. & Smith, E. The emergence and early evolution of biological

490 carbon-fixation. PLoS Comput. Biol. 8, (2012).

491 18. Camprubi, E. et al. Iron catalysis at the origin of life. IUBMB Life 69, 373–381

492 (2017).

493 19. Goldford, J. E. et al. Remnants of an ancient metabolism without phosphate. Cell

494 168, 1126–1134 (2017).

495 20. Fdz-Polanco, F. et al. Combining the biological nitrogen and sulfur cycles in

496 anaerobic conditions. Water Sci. Technol. 44, 77–84 (2001). bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

497 21. Amend, J. P. et al. Energetics of chemolithoautotrophy in the hydrothermal

498 system of Vulcano Island, southern Italy. Geobiology 1, 37–58 (2003).

499 22. Schrum, H. N. et al. Sulfate-reducing ammonium oxidation: A thermodynamically

500 feasible metabolic pathway in subseafloor sediment. Geology 37(10), 939–942

501 (2009).

502 23. Rios-Del, T. E. E. et al. Anaerobic ammonium oxidation linked to sulfate and

503 ferric iron reduction fuels nitrogen loss in marine sediments. Biodegradation 29,

504 429–442 (2018).

505 24. Cai, J. et al. Isolation and identification of bacteria responsible for simultaneous

506 anaerobic ammonium and sulfate removal. Sci. China Chem. 53, 645–650

507 (2010).

508 25. Philippot, P. et al. Early archaean microorganisms preferred elemental sulfur, not

509 sulfate. Science 317, 1534–1537 (2007).

510 26. Brandes, J. A. et al. Abiotic nitrogen reduction on the early earth. Nature 395,

511 365–367 (1998).

512 27. Smirnov, A. et al. Abiotic ammonium formation in the presence of Ni-Fe metals

513 and alloys and its implications for the hadean nitrogen cycle. Geochem. Trans. 9,

514 (2008).

515 28. Sleep, N. H. The hadean-archaean environment. Cold Spring Harbor Perspect.

516 Biol. 2, (2010).

517 29. Crowe SA, et al. Sulfate was a trace constituent of Archean seawater. Science

518 346(6210):735–739 (2014). bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

519 30. Kuypers, M. et al. The microbial nitrogen-cycling network. Nat. Rev. Microbiol.

520 16, 263–276 (2018).

521 31. Eigen, M. Self-organization of matter and the evolution of biological

522 macromolecules. Naturwissenschaften 58, 465–562 (1971).

523 32. Heinen, W. & Lauwers, A. M. Organic sulfur compounds resulting from the

524 interaction of iron sulfide, hydrogen sulfide and carbon dioxide in an anaerobic

525 aqueous environment. Orig. Life Evol. Biosph. 26, 131–150 (1996).

526 33. Yamaguchi, A. et al. Electrochemical CO2 reduction by Ni-containing iron

527 sulfides: how is CO2 electrochemically reduced at bisulfide-bearing deep-sea

528 hydrothermal precipitates? Electrochim. Acta 141, 311–318 (2014).

529 34. Martin, W. F., Sousa, F. L. & Lane, N. Evolution energy at life's origin. Science

530 344, 1092–1093 (2014).

531 35. Wächtershäuser, G. Before enzymes and templates-theory of surface metabolism.

532 Microbiol. Rev. 52, 452–484 (1988).

533 36. Shen, Y. A., Buick, R. & Canfield, D. E. Isotopic evidence for microbial sulphate

534 reduction in the early Archaean era. Nature 410, 77–81 (2001).

535 37. Mojzsis, S. J. et al.. Mass-independent isotope effects in Archean (2.5 to 3.8 Ga)

536 sedimentary sulfides determined by ion microprobe analysis. Geochim.

537 Cosmochim. Acta 67, 1635–1658 (2003).

538 38. Eck, R. V. & Dayhoff, M. O. Evolution of the structure of ferredoxin based on

539 living relics of primitive amino acid sequences. Science 152, 363–366 (1966).

540 39. Herschy, B. et al. Archean phosphorus liberation induced by iron redox bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

541 geochemistry. Nat. Commun. 9, (2018).

542 40. Jorgensen, B. B. A thiosulfate shunt in the sulfur cycle of marine-sediments.

543 Science 249, (1990).

544 41. Clement, J. C. et al. Ammonium oxidation coupled to dissimilatory reduction of

545 iron under anaerobic conditions in wetland soils. Soil Biol. Biochem. 37,

546 2323–2328 (2005).

547 42. Yang, W. H., Weber, K. A. & Silver, W. L. Nitrogen loss from soil through

548 anaerobic ammonium oxidation coupled to iron reduction. Nat. Geosci. 5,

549 538–541 (2012).

550 43. Bao, P. & Li, G. X. Sulfur-driven iron reduction coupled to anaerobic ammonium

551 oxidation. Environ. Sci. Technol. 51, 6691–6698 (2017).

552 44. Postgate, J. R. The sulphate-reducing bacteria (Cambridge University, 1984).

553 45. Wacey, D. et al. Microfossils of sulphur-metabolizing cells in

554 3.4-billion-year-old rocks of Western Australia. Nat. Geosci. 4, 698–702 (2011).

555 46. Crane, B. R. & Getzoff, E. D. The relationship between structure and function

556 for the sulfite reductases. Curr. Opin. Struct. Biol. 6, 744–756 (1996).

557 47. Bonomi, F., Werth, M. T. & Kurtz, D. M. Assembly of [FenSn(SR)]2-(n=2, 4) in

558 aqueous media from iron salts, thiols and sulfur, sulfide, thiosulfide plus

559 rhodonase. Inorg. Chem. 24, 4331–4335 (1985).

560 48. Lefaou, A. et al. Thiosulfate, polythionates and elemental sulfur assimilation and

561 reduction in the bacterial world. FEMS Microbiol. Lett. 75, 351–382 (1990).

562 49. Bordo, D. & Bork, P. The rhodanese/Cdc25 phosphatase superfamily- bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

563 sequence-structure-function relations. EMBO Rep. 3, 741–746 (2002).

564 50. Zhuang, W. Q. et al. Incomplete wood-ljungdahl pathway facilitates one-carbon

565 metabolism in organohalide-respiring Dehalococcoides mccartyi. Proc. Natl.

566 Acad. Sci. U. S. A. 111, 6419–6424 (2014).

567 51. Mall, A. et al. Reversibility of citrate synthase allows autotrophic growth of a

568 thermophilic bacterium. Science 359, 563–567 (2018).

569 52. Huber, C. & Wächtershäuser, G. Activated acetic acid by carbon fixation on

570 (Fe,Ni)S under primordial conditions. Science 276, 245–247 (1997).

571 53. Lovelock, J. E. The Ages of Gaia (Oxford University, 1995).

572

573

574 Acknowledgements This research was financially supported by the National Natural

575 Science Foundation of China (General Program Nos. 41571240 and 41571130063).

576 Author contributions Peng Bao conceived the study, designed the experiment and

577 wrote the manuscript. Hu Li performed gas analysis. Guo-Xiang Li carried out all

578 other experiments and analysis. Peng Bao and Guo-Xiang Li contributed to

579 interpreting the data.

580 Competing interests The authors declare no competing interests.

581 Additional information

582 Extended data is available for this paper at

583 Supplementary information is available for this paper at

584 bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

585 METHODS

586 Enrichment and isolation of Sammox bacterium.

587 Surface paddy soil mud (0–20 cm depth) in the upper section of a paddy field

588 was collected from Guangxi Dahuanjiang region (24°53'52" N, 108°17'43" E), in the

589 southwest of China. Soil materials were air-dried, ground with a mortar and crushed

590 to pass through a 2.0 mm sieve for character determination. Soil organic carbon

591 (19.81 g kg-1) was determined by potassium dichromate oxidation titration, and soil

592 total Fe (31.9 g kg-1) was determined by ICP-OES. Plant available sulfur (29.1 mg

593 kg-1) was determined by turbidimetry. Ammonium was 120.5 mg kg-1.

594 For enrichment, approximately 0.5 g of paddy soil was transferred to 100 ml of

595 double distilled water. After shaking, five milliliters of the suspension was inoculated

596 into 100 ml of anaerobic freshwater medium43,54 with additional sulfur (thiosulfate or

15 597 sufate) (3.0 mM) and ammonium ( NH4Cl, 1.0 mM) and incubated statically for one

-1 598 week. The basal medium contained (g l ): KH2PO4 0.2, MgCl·6H2O 0.4, CaCl2·2H2O

599 0.1, KCl 0.5. The trace elemental mixture contained (l-1): double distilled water 987

600 ml, HCl (25%) 12.5 ml (100 mM), FeCl2·4H2O 957 mg (7.5 mM), H3BO3 30 mg (0.5

601 mM), MnCl2·4H2O 100 mg (0.5 mM), CoCl2·6H2O 190 mg (0.8 mM), NiCl·6H2O 24

602 mg (0.1 mM), CuCl2·2H2O 2 mg (0.01 mM), ZnSO4·7H2O 144 mg (0.5 mM),

-1 603 Na2MoO4·2H2O 36 mg (0.15 mM). Selenite-tungstate solution (l ): NaOH 0.4 g,

-1 604 Na2SeO3·5H2O 6 mg, Na2WO4·2H2O 8 mg. Bicarbonate solution NaHCO3 84 g l 30

605 ml. The above mentioned stock solutions or aliquots were aseptically added to the

606 basal medium (l-1): trace element solution (1.0 ml), selenite-tungstate solution (0.1 ml), bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

607 bicarbonate solution (10.0 ml), biotin, 4-aminobenzoic acid, 10 mg l-1 pantothenate,

-1 -1 608 pyridoxamine, nicotinic acid and 20 mg l thiamine, 5 ml l ; vitamin B12 solution

609 with 50 mg l-1, 1.0 ml l-1. Sodium sulfate plus thiosulfate solution (1.0 ml, 3.0 mM

610 final concentration) and ammonium (0.5 ml, 1.0 mM final concentration) were added

611 into medium if necessary. The pH was adjusted to 8.0.

612 A volume of 0.5 ml of a positive Sammox culture, in which sulfur reduction

613 coupled to anaerobic ammonium oxidation, was sub-inoculated onto Acidovorax

614 complete agar medium under 5% CO2 atmosphere. Single colony was selected and

615 inoculated back into anaerobic freshwater medium. Positive Sammox samples were

616 continuously sub-inoculated using the Acidovorax complete agar medium. A single

617 Sammox bacterium (designated GX3-BWBA) was isolated after sub-cultivation for

618 five times, and was continuously cultivated for one year in anaerobic freshwater

619 medium. The purity of GX3-BWBA was confirmed by promoting the growth of

620 heterotrophic bacteria by addition of peptone and yeast extract to the defined

621 freshwater mineral medium (data not shown)55. Finally, the purity of GX3-BWBA

622 was confirmed by deep Illumina sequencing.

623 This study was performed using a series of experiments, for Sammox by

15 624 GX3-BWBA: (I) 3.0 mM thiosulfate + 1.0 mM NH4Cl, (II) 3.0 mM thiosulfate + 1.0

15 15 625 mM NH4Cl + GX3-BWBA, (III) 3.0 mM sulfate + 1.0 mM NH4Cl, (IV) 3.0 mM

15 626 sulfate + 1.0 mM NH4Cl + GX3-BWBA. Experiments started and were sampled at

627 the sampled at the 0, 3, 6, 9, 12, 15, 18, 21, 24, 27, 34 day. Experiments were

628 performed anaerobically in 100-ml serum bottles in the dark at 30 oC. Each batch of bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

629 experiments was established in triplicate, inoculated with 1.0 ml GX3-BWBA

630 exponential phase culture if necessary.

631 Genome sequencing and assembly

632 The genome of R. thioammoniphilus (strain GX3-BWBA) was sequenced using

633 an Illumina HiSeq 4000 system (Illumina, SanDiego, CA, USA) at the Beijing

634 Genomics Institute (Shenzhen, China). Genomic DNA was sheared randomly to

635 construct three read libraries with lengths of (150:150) by a Bioruptor ultrasonicator

636 (Diagenode, Denville, NJ, USA) and physico-chemical methods. The paired-end

637 fragments libraries were sequenced according to the Illumina HiSeq 4000 system’s

638 protocol. Raw reads of low quality from paired-end sequencing (those with

639 consecutive bases covered by fewer than five reads) were discarded. The sequenced

640 reads were assembled using SOAPdenovo v1.05 software. This Whole Genome

641 Shotgun project has been deposited at DDBJ/ENA/GenBank under the accession

642 QMKS00000000. The version described in this paper is version QMKS01000000.

643 Genome Component prediction

644 Gene prediction was performed on the R. thioammoniphilus genome assembly by

645 glimmer3 (http://www.cbcb.umd.edu/software/glimmer/) with Hidden Markov models.

646 tRNA, rRNA and sRNAs recognition made use of tRNAscan-SE56, RNAmmer, and

647 the Rfam database. The tandem repeats annotation was obtained using the Tandem

648 Repeat Finder (http://tandem.bu.edu/trf/trf.html), and the minisatellite DNA and

649 microsatellite DNA selected based on the number and length of repeat units. The

650 Genomic Island Suite of Tools (GIST) used for genomic islands analysis bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

651 (http://www5.esu.edu/cpsc/bioinfo/software/GIST/) with Island Path-DIOMB,

652 SIGI-HMM, Island Picker method. Prophage regions were predicted using the PHAge

653 Search Tool (PHAST) web server (http://phast.wishartlab.com/) and CRISPR

654 identification using CRISPRFinder.

655 Protein sequences will be downloaded from NCBI, and get sequences in the

656 taxonomy of species classification information; The sequences using mafft software

657 for multiple sequence alignment. After comparing the sequences, using Fasttree

658 software construct phylogenetic tree (neighbor joining algorithm). Using R language

659 ggtree package for visualization mapping.

660 Gene annotation and protein classification

661 The best hit abstracted using Blast alignment tool for function annotation. Seven

662 databases which are KEGG (Kyoto Encyclopedia of Genes and Genomes), COG

663 (Clusters of Orthologous Groups), NR (Non-Redundant Protein Database databases),

664 Swiss-Prot57, and GO (Gene Ontology), TrEMBL, EggNOG are used for general

665 function annotation. Representative metabolic gene clusters were displayed with

666 Easyfig58.

667 General procedure for prebiotic Sammox-driven CO2 fixation

668 Each 50 ml basal solution was transferred into 60 ml serum bottles and sealed

669 with butyl rubber stoppers and aluminium crimp caps. The solution was autoclaved

670 and cooled as room temperature after washed by He gas (purity=99.999%). Additional

671 trace elemental mixture solutions were filter sterilized or autoclaved individually.

672 Additional sulfur (thiosulfate or sufate) (3.0 mM) and/or ammonium (NH4Cl, 1.0 mM) bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

673 was added into serum bottle. The mixed solution was termed as Sammox reaction

674 system.

675 The basal solution contained (l-1): KCl 0.1 g. The trace elemental mixture

-1 676 contained (l ): double distilled water 987 ml, HCl (25%) 12.5 ml (100 mM), H3BO3

677 30 mg (0.5 mM), MnCl2·4H2O 100 mg (0.5 mM), CoCl2·6H2O 190 mg (0.8 mM),

678 NiCl·6H2O 24 mg (0.1 mM), CuCl2·2H2O 2 mg (0.01 mM), FeCl2·4H2O 957 mg (7.5

679 mM), ZnCl2 68 mg (0.5 mM), Na2MoO4·2H2O 36 mg (0.15 mM). Selenite-tungstate

-1 680 solution (l ): NaOH 0.4 g, Na2SeO3·5H2O 6 mg, Na2WO4·2H2O 8 mg. Bicarbonate

-1 681 solution NaHCO3 84 g l 30 ml. The above mentioned stock solutions or aliquots

682 were aseptically added to the basal medium (l-1): trace element solution (2.0 ml),

683 selenite-tungstate solution (0.2 ml), bicarbonate solution (10.0 ml). Sodium sulfate,

684 thiosulfate (1.0 ml, 3.0 mM final concentration), ammonium solution (0.5 ml, 1.0 mM

685 final concentration), and fresh precipitated Fe-S mineral (1.0 mM) were added into

686 medium if necessary. Serum bottles were kept 100 oC in a water bath in the dark for

687 24 h, then maintain at 70 oC in the dark for another 24 h, and sampling at 48 h to

688 determinate organic products.

689 This study was performed using a series of experiments: (I) 3.0 mM thiosulfate +

- 690 1.0 mM NH4Cl + 20 mM HCO3 , (II) 3.0 mM thiosulfate + 1.0 mM NH4Cl + 20 mM

- - 691 HCO3 + 1.0 mM Fe-S, (III) 3.0 mM thiosulfate + 20 mM HCO3 , (IV) 3.0 mM sulfate

- 692 + 1.0 mM NH4Cl + 20 mM HCO3 , (IV) 3.0 mM sulfate + 1.0 mM NH4Cl + 20 mM

- - 693 HCO3 + 1.0 mM Fe-S, (V) 3.0 mM sulfate + 20 mM HCO3 , (VI) 1.0 mM NH4Cl +

- 694 20 mM HCO3 . bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

695 General procedure for thiol/thioester promoted Sammox-driven CO2 fixation.

696 A solution of methanethiol (0.8 mM, final concentration) was added to Sammox

697 reaction system. The reaction systems were heated at 100 oC in a water bath in the

698 dark for 24 h, then maintain at 70 oC in the dark for another 24 h, and removed from

699 the water bath and allowed to cool to room temperature before derivatization and gas

700 chromatography–mass spectrometry analysis.

701 This study was performed using a series of experiments: (I) 3.0 mM thiosulfate +

- 702 1.0 mM NH4Cl + 20 mM HCO3 + 0.8 mM methanethiol, (II) 3.0 mM thiosulfate +

- 703 1.0 mM NH4Cl + 20 mM HCO3 + 1.0 mM Fe-S + 0.8 mM methanethiol, (III) 3.0

- 704 mM sulfate + 1.0 mM NH4Cl + 20 mM HCO3 + 0.8 mM methanethiol, (IV) 3.0 mM

- 705 sulfate + 1.0 mM NH4Cl + 20 mM HCO3 + 1.0 mM Fe-S + 0.8 mM methanethiol.

706 General procedure for verification of Sammox-driven combination of reductive

707 acetyl-CoA pathway and incomplete rTCA cycle.

708 We design this experimental set to verify if Sammox-driven reductive

709 acetyl-CoA pathway could go into Sammox-driven incomplete rTCA cycle. The

710 general procedure is the same as above. Pyruvate (1.0 mM, final concentration) was

711 added into Sammox reaction system as substrate. Serum bottles were heated at 100 oC

712 in a water bath in the dark for 24 h, then maintain at 70 oC in the dark for another 24 h,

713 and allowed to cool to room temperature before derivatization and gas

714 chromatography–mass spectrometry analysis.

715 This study was performed using a series of experiments: (I) 3.0 mM thiosulfate +

- 716 1.0 mM NH4Cl + 20 mM HCO3 + 1.0 mM pyruvate + 1.0 mM Fe-S + methanethiol, bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

- 717 (II) 3.0 mM thiosulfate + 1.0 mM NH4Cl + 20 mM HCO3 + 1.0 mM Fe-S + 1.0 mM

- 718 pyruvate, (III) 3.0 mM thiosulfate + 1.0 mM NH4Cl + 20 mM HCO3 + 1.0 mM

- 719 pyruvate, (IV) 3.0 mM thiosulfate + 20 mM HCO3 + 1.0 mM pyruvate, (V) 3.0 mM

- 720 sulfate + 1.0 mM NH4Cl + 20 mM HCO3 + 1.0 mM Fe-S + 1.0 mM pyruvate +

- 721 methanethiol, (VI) 3.0 mM sulfate + 1.0 mM NH4Cl + 20 mM HCO3 + 1.0 mM Fe-S

- 722 + 1.0 mM pyruvate, (VII)3.0 mM sulfate + 1.0 mM NH4Cl + 20 mM HCO3 + 1.0

- 723 mM pyruvate, (VIII) 3.0 mM sulfate + 20 mM HCO3 + 1.0 mM pyruvate, (IX) 1.0

- 724 mM NH4Cl + 20 mM HCO3 + 1.0 mM pyruvate.

725 Chemicals

726 All reagents and solvents were purchased from commercial suppliers and used

727 without further purification unless otherwise noted.

728 Sampling analytical methods

729 (i) Enumeration of microbe viable counts. The spread-agar-plate method was used

730 for enumeration of bacterial counts in GX3-BWBA culture59. The Acidovorax

731 complete agar medium, pH 7.0, contained 5.0 g l-1 peptone (Difco), 3.0 g l-1 beef

732 extract and 15 g l-1 agar (Difco). A 50 µl sample was taken from GX3-BWBA culture

733 at the 0, 3, 6, 9, 12, 15, 18, 21, 24, 27 and 34 days and inoculated onto plates for cell

734 counting. Visible colonies produced on the agar plate were counted at 48 h. Bacterial

735 numbers were expressed as colony-forming units (CFU) per milliliter of culture59.

736 (ii) Electron microscopy. Morphology of GX3-BWBA was investigated by

737 transmission electron microscope (TEM) (TEM, HITACHI H-7500). Cells from the

738 culture were collected by centrifugation, washed and diluted with phosphate buffer bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

739 (pH 7.5), and dropped on to specimens. The material was examined after dry by

740 airing.

741 (iii) Sulfide. The analysis of sulfide in filtered samples was performed

742 photometrically by the methylene blue method60. Absorption at 660 nm was measured

743 (Infinite 200PRO; TECAN).

744 (iV) Sulfate, sulfite, nitrite, acetate, formate and ammonium. To determine sulfate,

745 sulfite, nitrite, acetate and formate, 0.5 ml of the sample was filtered (0.22 μm) to

746 remove particulates that could interfere with ion chromatography. The ion

747 chromatography system consisted of an ICS-5000+ SP pump (Thermo Fisher

748 Scientific Inc. Sunnyvale, CA, USA), a column oven ICS-5000+ DC, an

749 electrochemical detector DC-5. The ion chromatography column system used a

750 Dionex Ionpac AS11-HC column. The operating condition was with an eluent of 30

751 mM KOH at a flow rate of 1.0 ml min-1. For determination of ammonium, the ion

752 chromatography column system used a Dionex Ionpac CS12A column.

753 (V) Thiosulfate. Thiosulfate was determined by an Agilent 1260 infinity HPLC

754 system, equipped with a quaternary pump (Agilent, USA). Thiosulfate was separated

755 by a Zorbax SB-C18 column (150×4.6 mm, 5 μm), and detected by using of

756 DAD detector at 215 nm. All analyses were performed at 40 °C with a flow rate of 1

-1 757 ml min . Na2HPO4 was used as solvent. The pH of the solvent was adjusted with 1.0

758 M HCl. Samples were filtered through 0.45 μm Cosmonice Filters (Millipore, Tokyo,

759 Japan) and immediately injected into the HPLC system61.

760 (Vi) Gas Analysis. At each sampling time point, each serum bottle was shaken bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

761 vigorously to equilibrate the N2 between dissolved and gas phases. Dinitrogen

762 sampling and determination was performed according to a literature procedure41.

763 (Vii) Derivatization procedure and product identification

764 For optimal GC resolution, the carboxylic acids were converted to ethyl esters

765 using a mixture of ethanol/ethyl chloroformate (EtOH/ECF). Derivatization of

766 carboxylic acids to esters was performed according to a literature procedure13.

767 Reaction products derivatized to ethyl esters of carboxylic acids were identified by

768 comparing the mass spectra and retention times against analogously derivatized

769 authentic samples. ECF derivatization was preferred for small molecule substrates

770 (pyruvate, lactate, malate, fumarate, succinate, α-ketoglutarate, amino acid). The

771 carboxylic acids might also convert to methyl esters using a mixture of

772 methanol/methyl chloroformate (MeOH/MCF), following the same procedure to ECF

773 derivatization to detect cis-aconitate, tricarballylate, isocitrate and citrate.

774 (Viii) Gas chromatography–mass spectrometry (GC–MS) analysis for rTCA

775 metabolites determination.

776 GC–MS analysis was performed on a GC System 7890B connected to a MSD

777 block 5977A, using Agilent High Resolution Gas Chromatography Column: PN

778 19091S–433, HP–5MS, 28 m×0.25 mm, 0.25 Micron, SN USN 462366H. Samples

779 were prepared in ethyl acetate (200 μl sample volume). The analysis was carried out

780 on a splitless 1 μl injection volume with an injection port temperature 250 °C.

781 Column oven temperature program was as follows: 60 °C for 1.0 min, ramped at

782 30 °C min−1 to 310 °C with 3.0 min hold, with a total running time of 12.33 min. The bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

783 mass spectrometer was turned on after 3 min and was operated at the electron

784 ionization mode with quadrupole temperature of 150 °C. Data acquisition was

785 performed in the full-scan mode (50-500). Hydrogen (99.999 % purity) was used as

786 carrier gas at a constant flow rate of 1.5 ml min−1.

787 (iX) liquid chromatography-mass spectrometry (LC-MS) method for polypyrrole

788 determination.

789 For qualitative detection of polypyrrole, Waters ACQUITY UPLC system with

790 an online coupled SYNAPT G2 mass spectrometer Q-TOF was used. Sample

791 separation was achieved on a ACQUITY UPLC HSS T3 column (1.8 μ m, 2.1 mm ×

792 100 mm; column temperature, 30 °C). Solvent A contained 2.5% methanol, 0.2%

793 formic acid in UPLC-grade water, and Solvent B was 97.5% UPLC-grade water with

794 0.2% formic acid. Injection of 2.0 μl of sample onto the column at 0.2 ml min−1 was

795 followed by gradient elution. Mass spectrometric qualitative detection of polypyrrole

796 was performed in resolution mode. Polypyrrole was identified by comparing the mass

797 spectra and retention times against pure commercially available pyrrole (100 μl ml-1)

798 heated at 100 oC in a water bath in the dark for 24 h, and maintain at 70 oC in the dark

799 for another 24 h. Tandem mass spectrometry data were analyzed using MassLynx

800 v4.1.

801

802 54. Strous, M. et al. Deciphering the evolution and metabolism of an anammox

803 bacterium from a community genome. Nature 440, 790–794 (2006).

804 55. Konneke, M. et al. Isolation of an autotrophic ammonia oxidizing marine bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

805 archaeon. Nature 437, 543–546 (2005).

806 56. Lowe, T. M. & Eddy, S. R. tRNAscan-SE: a program for improved detection

807 of transfer RNA genes in genomic sequence. Nucl. Acids Res. 25 (5),

808 0955–0964 (1997).

809 57. Torto-Alalibo, T., Collmer, C. W. & Gwinn-Giglio, M. The plant-associated

810 microbe gene ontology (PAMGO) consortium: community development of

811 new gene ontology terms describing biological processes involved in

812 microbe-host interactions. BMC microbiol. 9 (suppl 1), (2009).

813 58. Sullivan, M. J., Nicola K. P. & Scott A. B. Easyfig: a genome comparison

814 visualizer. Bioinformatics 27 (7), 1009–1010 (2011).

815 59. Kataoka, N. et al. Enrichment culture and isolation of slow growing bacteria.

816 Appl. Microbiol. Biotechnol. 45, 771–777 (1996).

817 60. Cline, J. D. Spectrophotometric determination of hydrogen sulfide in natural

818 waters. imnol. Oceanogr. 14 (3), 454−458 (1969).

819 61. Kondo, R. et al. Determination of thiosulfate in a meromictic lake. Fisheries

820 Sci. 66, 1076–1081 (2000).

821

822

823

824

825

826 bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

827 Legends of figures

828 Figure 1. Products of simulated prebiotic Sammox-driven proto-anabolic

829 networks with bicarbonate as the sole carbon source under hydrothermal

830 conditions. Treatments were as follows (a, thiosulfate; b, sulfate), from left to

831 right: (i) sulfurous species, ammonium, Fe-S minerals, and methanethiol; (ii)

832 sulfurous species, ammonium, and Fe-S minerals; (iii) sulfurous species,

833 ammonium, and methanethiol; (iv) sulfurous species and ammonium; (v)

834 ammonium; and (vi) sulfurous species. The bar chart shows the yields of formate,

835 acetate, succinate, and α-ketoglutarate in each treatment group. Error bars

836 represent standard deviations of three replicates.

837

838 Figure 2. Products of simulated prebiotic Sammox-driven incomplete rTCA cycle

839 with bicarbonate and pyruvate as carbon sources under hydrothermal

840 conditions. Treatments were as follows (a, thiosulfate; b, sulfate), from left to

841 right: (i) sulfurous species, ammonium, Fe-S minerals, and methanethiol; (ii)

842 sulfurous species, ammonium, and Fe-S minerals; (iii) sulfurous species,

843 ammonium, and methanethiol; (iv) sulfurous species and ammonium; (v)

844 ammonium; and (vi) sulfurous species. The bar chart shows the concentrations

845 of pyruvate, succinate, and α-ketoglutarate in each treatment group. Error bars

846 represent standard deviations of three replicates.

847

848 Figure. 3 Hypothetical prebiotic Sammox-driven reductive acetyl-CoA pathway bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

849 (a). Hypothetical proto-anabolic networks, a combination of acetyl-CoA pathway

850 together with incomplete rTCA cycle driven by Sammox, and the role of its

851 intermediates as universal biosynthetic precursors. Bar represents the final step

852 of incomplete rTCA cycle (b). Conceptual model of Sammox-driven the origin of

853 not free-living organic precursor on the surface of Fe-S minerals in Hadean

854 alkaline hydrothermal systems. Cubes represent Fe-S minerals/metals alloy.

855 Globes represent the not free-living organic precursor, ligand sphere (c).

856

857 Figure. 4 Characters of R. thioammoniphilus GX3-BWBA. TEM image of cells

858 (a). Scale bar represents 0.2 μm. Growth of GX3-BWBA under combined

859 condition of ammonium and sulfurous species, with bicarbonate as sole carbon

860 sources (b). Production rate of isotopically labeled dinitrogen for thiosulfate and

861 sulfate-fueled Sammox by GX3-BWBA (c, d). Phylogenetic tree constructed by

862 the maximum likelihood method, using 1478 nucleotides from 16S rDNA

863 sequences, showing the position of strain GX3-BWBA in relation to members of

864 the genus Ralstonia. The bar represents 0.1 changes per nucleotide position (e).

865

866 Figure. 5 Thiosulfate- and sulfate-fueled Sammox by R. thioammoniphilus

867 GX3-BWBA. (a) thiosulfate and reduction products. (b) ammonium and its

868 oxidation products in thiosulfate-fueled Sammox. (c) sulfate and reduction

869 products (sulfide and sulfite as products of sulfate in control group are below

870 detection limit). (d) ammonium and its oxidation products in sulfate-fueled bioRxiv preprint doi: https://doi.org/10.1101/461707; this version posted January 30, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

871 Sammox. Error bars represent standard deviations of three biological replicates.

872 Figure. 6 Representative metabolic gene clusters from R. thioammoniphilus

873 strain GX3-BWBA (a). The standard configuration of metabolic gene clusters

874 (MGCs) in primary metabolism located in two scaffolds. MGCs contain enzymes

875 of CO2 fixation genes, including 31 rTCA cycle, and 30 reductive

876 acetyl-CoA pathway genes, 6 dissimilatory sulfur metabolic genes, and

877 5 denitrification genes. Gene size and spacing are not to scale. Each gene was not

878 condensed lie in genome. Protein phylogenetic tree derived from 26443 amino

879 acid sequences of sulfite/nitrite reductase (b). Scale bars represent estimated

880 sequence divergence or amino acid changes. Archaea, Bacteria,

881 Eukaryota.

882

883

884

885

886

887

888

889

890