Extracellular Electron Transfer May Be an Overlooked Contribution to Pelagic Respiration in Humic-Rich Freshwater Lakes

Humic lakes and ponds receive large amounts of terrestrial carbon and are important components of the global carbon cycle, yet how their redox cycling influences the carbon budget is not fully understood. Here we compared metagenomes obtained from a humic bog and a clear-water eutrophic lake and found a much larger number of genes that might be involved in extracellular electron transfer (EET) for iron redox reactions and humic substance (HS) reduction in the bog than in the clear-water lake, consistent with the much higher iron and HS levels in the bog.

bog and a clearwater eutrophic lake, and found a much larger number of genes that might 23 be involved in extracellular electron transfer (EET) for iron redox reactions and humic 24 substance (HS) reduction in the bog than in the clearwater lake, consistent with the much 25 higher iron and HS levels in the bog. These genes were particularly rich in the bog's 26 anoxic hypolimnion, and were found in diverse bacterial lineages, some of which are 27 relatives of known iron oxidizers or iron/HS reducers. We hypothesize that HS may be a 28 previously overlooked electron acceptor and EET-enabled redox cycling may be 29 important in pelagic respiration and greenhouse gas budget in humic-rich freshwater 30 lakes. Inland lakes receive allochthonous carbon (C) fixed in their catchment areas, and 42 play an important role in the cycling of terrestrial C and affect global C budgets. Many 43 northern freshwater lakes are experiencing a "browning" process, and this trend may 44 continue with changes in precipitation patterns and atmospheric deposition chemistry (1-45 3). A leading factor contributing to the brownification is the increasing inputs of 46 allochthonous dissolved organic C (DOC) (4). A major component of terrestrially derived 47 on the outer membrane to form the EET conduit. Most Fe(III) reducers can also reduce 110 HS (28, 29), and probably use the same EET systems to transfer electrons to HS. For 111 example, in Geobacter sulfurreducens, a number of outer membrane MHCs that are 112 important in the reduction of Fe(III) are able to reduce extracellular AQDS and HS (30), 113 and in Shewanella oneidensis, the porin and periplasmic MHC components of its Fe(III)-114 reducing PCC are essential for AQDS and HS reduction (31, 32). These findings suggest 115 that reduction of the quinone moieties in HS is a non-specific redox process by EET 116

systems. 117
In this study, we searched for putative EET genes (including PCC, outer surface 118 MHCs not associated with PCC, and Cyc2) in MAGs and metagenomes to examine if 119 these genes are indeed more abundant in the humic bog than in the clearwater lake. 120 Method details on the identification and quantification of putative EET genes were 121 described in Supplemental Methods. All (meta)genome data are publicly available at the 122

RESULTS 130
MHCs are important components of EET systems involved in Fe redox reactions 131 and HS reduction. In particular, MHC with large numbers of hemes may be able to form 132 molecular "wires" for conducting electrons from the periplasmic space across the outer 133 membrane (33, 34). We therefore estimated the normalized abundance of MHCs with at 134 least five heme-binding sites in the metagenomes. In general, TH had the highest 135 abundance of MHCs, followed by TE and ME, and such differences were even more 136 pronounced for MHCs with at least eight heme-binding sites (Fig. 1A). Some of these 137 MHCs are components of other redox enzyme complexes, such as the pentaheme and 138 hexaheme MHCs in alternative complex III (ACIII), and octaheme MHCs in tetrathionate 139 reductases and hydroxylamine oxidoreductases. Putative EET MHC components (i.e. 140 MHCs in PCC and outer surface MHCs not associated with PCC, as listed in Table S2) 141 were much more frequently found in MHCs with large heme binding sites (e.g. >9), and 142 these putative EET genes were more abundant in TH than TE, and nearly absent in the 143 ME metagenome (Fig. 1B). This may indicate that MHC-based EET potential was more 144 significant in the anoxic layer than in the oxic layer of the humic bog, and was minimal in 145 the oxic layer of the clearwater lake with low Fe and HS concentrations. Notably, the 146 largest number of heme-binding sites (i.e. 51) was found in an MHC component of a 147 putative PCC, encoded in an un-binned contig in the TE metagenome (Table S2). 148 149

Porin-cytochrome c protein complex (PCC) genes 150
The best studied PCC system, MtrABC (consisting of a porin, a periplasmic 151 decaheme Cyt c, and an extracellular decaheme Cyt c), was first identified in S. respectively. The more recently discovered PCC proteins in G. sulfurreducens are not 157 homologous to MtrABC, but are also encoded in operons with genes encoding a porin 158 (OmbB), a periplasmic octaheme Cyt c (OmaB), and an outer-membrane dodecaheme 159 Cyt c (OmcB) (40). This suggests that multiple PCC systems evolved independently, and 160 may provide a clue to search for new types of PCC by examining genome-level 161 organization. For example, putative novel PCC genes not homologous to previously 162 identified PCCs were found in some Fe(II) oxidizer genomes by searching for the unique 163 genetic organization of porin-and periplasmic MHC-coding genes (41). 164 Nearly all MtrAB/MtoAB/PioAB homologs were recovered in Trout Bog, and 165 mostly from TH (Table S2). They are present in MAGs affiliated with the Proteobacteria, 166 including the Fe(II)-oxidizing Gallionella and Ferrovum, Fe(III)-reducing Albidiferax, 167 Fe(III)-and AQDS-reducing Desulfobulbus and genera not known for EET, such as 168 Polynucleobacter, Desulfocapsa, and Methylobacter (Fig. 2). Interestingly, among the 46 169 Polynucleobacter genomes available at IMG/M (https://img.jgi.doe.gov/m), 170 MtrAB/MtoAB/PioAB homologs were only found in Polynucleobacter recovered from a 171 wetland and two humic lakes (including Trout Bog and Lake Grosse Fuchskuhle located 172 in Brandenburg, Germany), suggesting that this PCC might be an acquired trait of some 173 Polynucleobacter spp. adapting to humic-rich environments. 174

Homologs of another studied PCC (represented by OmbB-OmaB-OmcB in 175
Geobacter spp.) were present in MAGs affiliated with relatives of known Fe(III) (and 176 Based on the unique genetic organization of PCC-encoding genes, we found a 179 number of putative PCC that do not share a significant sequence homology with known 180 PCCs, probably representing novel PCC types. These putative PCC genes were present in 181 Fe(III) (and HS) reducers (Geothrix, Albidiferax, and Geobacteraceae) and bacteria not 182 known for EET, including Methylotenera, Methylobacter, Methyloversatilis, and a 183 number of Bacteroidetes and Verrucomicrobia (Fig. 2). Among them, Verrucomicrobia 184 with putative PCC genes were previously found in humic-rich environments, such as 185 soils and lake sediment, in addition to the Verrucomicrobia MAGs from Trout Bog (42). Here, we found a number of non-PCC-associated outer surface MHCs in the 194 metagenomes (Table S2)  several members in the Bacteroidetes and Verrucomicrobia phyla (Fig. 2). In particular,197 seven genes predicted to encode MHCs located on the cell wall were found in a  positive actinobacterial MAG classified to Solirubrobacterales from TH, and four of 199 these genes are located in the same gene cluster with up to 15 heme-binding sites in a 200 single MHC (Table S2) Cyc2 is an outer membrane c-type cytochrome with one heme-binding motif in 204 the N-terminus and a predicted porin structure at the C-terminus, and was therefore 205 proposed as a fused PCC (45). Cyc2 was originally identified as the Fe(II) oxidase in 206 acidiphilic Acidithiobacillus ferrooxidans (23)  Similar to EET MHC genes, the normalized abundance of total Cyc2-like genes 210 was much higher in the TH than in the TE metagenome, and Cyc2-like genes were 211 largely absent in the ME metagenome (Fig. 1C). Cyc2 homologs were present in 29 212 MAGs exclusively from Trout Bog (Table S3) With the ongoing brownification of surface water due to increasing inputs of 221 terrestrial C and Fe on a large scale, elucidating the roles and contribution of HS and Fe 222 in redox and C cycling becomes even more relevant to C budgets at an ecosystem level. 223 Here we inspected EET genes/organisms involved in HS and Fe redox processes in two freshwater lakes with contrasting HS and Fe levels to examine if these genes/organisms 225 were more abundant in the humic lake, particularly in its anoxic layer. All together, a 226 total of 103, 36, and 66 MAGs were recovered from the ME, TE, and TH metagenomes, 227 respectively. Among them, putative EET genes were found in 7, 12 and 31 MAGs from 228 ME, TE and TH, respectively (Fig. 2). Therefore, a larger fraction of MAGs might 229 encode EET function in Trout Bog, especially in its hypolimnion, than in Mendota. This, 230 together with the normalized abundance of putative EET genes in the three metagenomes 231 ( Fig. 1), suggests that the genetic potential of EET was more significant in the anoxic 232 layer than in the oxic layer of the humic bog, and was the lowest in the oxic layer of the 233 clearwater lake. This distribution pattern is consistent with the availability of the 234 thermodynamically more favorable electron acceptor, i.e. oxygen, between the two layers 235 and the much higher concentrations of HS and Fe in the bog than in the clearwater lake. 236 It was not surprising to find putative EET genes in relatives of bacteria that are 237 known to be capable of Fe redox reactions and HS reduction in anoxic lake waters. 238 However, finding putative EET genes in taxa not known for EET functions is intriguing. 239 Like many known EET organisms, some of these bacteria (e.g. Bacteroidetes and 240 Verrucomicrobia) contain multiple sets of putative EET genes. In particular, some 241 Methylotenera and Methylobacter contain both Cyc2 and putative PCC genes. If these 242 methylotrophs are indeed capable of EET, this may enable insoluble or high-molecular 243 weight substrates, such as Fe(III) and HS, to be used as an electron acceptor to oxidize 244 the methyl-group in methanol and methylamine. Such EET processes, if they occur, hand, HS may be directly used as an electron acceptor to respire the more labile organic 261 C (Fig. 3). The anaerobic respiration of organic C with Fe(III) and HS are both 262 thermodynamically more favorable than methanogenesis, therefore promoting the 263 transformation of organic C towards CO 2 , not CH 4 . This may lower the overall global 264 warming potential of greenhouse gas emissions from humic lakes, as CH 4 is a much more 265 potent greenhouse gas than CO 2 . Because of lake seasonal mixing and more frequent 266 micro-mixing, such as wind-driven turbulence and convectively derived diurnal oxycline 267 fluctuations (20, 47), reduced HS and Fe can be re-oxidized through mixing-introduced 268 oxygenation to regenerate their EAC, which makes these anaerobic respiration processes 269 sustainable in the anoxic layer (Fig. 3). In these redox processes, oxygen is the ultimate 270 electron acceptor, and Fe and HS "recharge" the EAC with oxygen for subsequent use 271 when oxygen becomes unavailable in stratified hypolimnia. Hypothetically, such 272 recharging process would increase the effective EAC of humic water and shunt more 273 organic C to anaerobic respiration. Therefore, we hypothesize that HS may be a 274 previously overlooked electron acceptor and EET may be an important contribution to 275 pelagic respiration in humic-rich freshwater lakes. Coupled with C metabolism, EET-276 enabled HS and Fe redox dynamics can significantly influence C cycling and greenhouse 277 gas emission in humic lakes that experience recurrent oxic-anoxic conditions. The 278 overrepresentation of EET genes/organisms potentially involved in HS and Fe redox 279 processes in the humic lake strongly support this hypothesis, given that the energetic presence of putative EET genes was indicated with "+". 517 Oxygenation in the hypolimnion through seasonal mixing and more frequent micro-520 mixing (such as wind-driven turbulence and convectively derived diurnal oxycline 521 fluctuations) regenerates the electron accepting capacity of reduced HS and Fe to enable 522 these anaerobic respiration processes sustainable in the hypolimnion. 523 524 525

List of Supplemental Information 526
Supplemental Methods 527 Table S1. Lake characteristics 528 Table S2. List of MHCs that might be involved in EET in the metagenomes 529 Table S3. List of Cyc2-like genes in the metagenomes 530