Salinity Impact on Composition and Activity of Nitrate-Reducing Fe(II)-Oxidizing Microorganisms in Saline Lakes

ABSTRACT

Nitrate-reducing Fe(II)-oxidizing (NRFeOx) microorganisms contribute to nitrogen, carbon, and iron cycling in freshwater and marine ecosystems. However, NRFeOx microorganisms have not been investigated in hypersaline lakes, and their identity, as well as their activity in response to salinity, is unknown. In this study, we combined cultivation-based most probable number (MPN) counts with Illumina MiSeq sequencing to analyze the abundance and community compositions of NRFeOx microorganisms enriched from five lake sediments with different salinities (ranging from 0.67 g/L to 346 g/L). MPN results showed that the abundance of NRFeOx microorganisms significantly (P < 0.05) decreased with increasing lake salinity, from 7.55 × 10³ to 8.09 cells/g dry sediment. The community composition of the NRFeOx enrichment cultures obtained from the MPNs differed distinctly among the five lakes and clustered with lake salinity. Two stable enrichment cultures, named FeN-EHL and FeN-CKL, were obtained from microcosm incubations of sediment from freshwater Lake Erhai and hypersaline Lake Chaka. The culture FeN-EHL was dominated by genus Gallionella (68.4%), while the culture FeN-CKL was dominated by genus Marinobacter (71.2%), with the former growing autotrophically and the latter requiring an additional organic substrate (acetate) and Fe(II) oxidation, caused to a large extent by chemodenitrification [reaction of nitrite with Fe(II)]. Short-range ordered Fe(III) (oxyhydr)oxides were the product of Fe(II) oxidation, and the cells were partially attached to or encrusted by the formed iron minerals in both cultures. In summary, different types of interactions between Fe(II) and nitrate-reducing bacteria may exist in freshwater and hypersaline lakes, i.e., autotrophic NRFeOx and chemodenitrification in freshwater and hypersaline environments, respectively.

IMPORTANCE NRFeOx microorganisms are globally distributed in various types of environments and play a vital role in iron transformation and nitrate and heavy metal removal. However, most known NRFeOx microorganisms were isolated from freshwater and marine environments, while their identity and activity under hypersaline conditions remain unknown. Here, we demonstrated that salinity may affect the abundance, identity, and nutrition modes of NRFeOx microorganisms. Autotrophy was only detectable in a freshwater lake but not in the saline lake investigated. We enriched a mixotrophic culture capable of nitrate-reducing Fe(II) oxidation from hypersaline lake sediments. However, Fe(II) oxidation was probably caused by abiotic nitrite reduction (chemodenitrification) rather than by a biologically mediated process. Consequently, our study suggests that in hypersaline environments, Fe(II) oxidation is largely caused by chemodentrification initiated by nitrite formation by chemoheterotrophic bacteria, and additional experiments are needed to demonstrate whether or to what extent Fe(II) is enzymatically oxidized.

INTRODUCTION

Microbial redox transformation of iron (Fe) is a fundamental component of the biogeochemical Fe cycle and is mediated by Fe(III)-reducing microorganisms (FeRM) and Fe(II)-oxidizing microorganisms (FeOM), connecting the Fe cycle with other elementary cycles (e.g., C, N, and S) (1–3). Under neutral anoxic conditions, FeOM are classified into three physiological groups, i.e., microaerobic, phototrophic, and nitrate-reducing Fe(II)-oxidizing microorganisms (4, 5). Nitrate-reducing Fe(II)-oxidizing (NRFeOx) microorganisms are able to oxidize Fe(II) and reduce nitrate to N₂ (see equation 1) or to ammonium (see equation 2) (6, 7) under autotrophic or mixotrophic conditions.

$$10{\mathrm{Fe}}^2{}^{+}+{\mathrm{2NO}}_3{}^{-}+{\mathrm{24H}}_2\mathrm{O\to 1}0\mathrm{Fe}{\left(\mathrm{OH}\right)}_3+{\mathrm{N}}_2+{\mathrm{18H}}^{+}$$ (1)

$$8{\mathrm{Fe}}^2{}^{+}+{\mathrm{NO}}_3{}^{-}+{\mathrm{21H}}_2\mathrm{O\to 8Fe}{\left(\mathrm{OH}\right)}_3+{\mathrm{NH}}_4{}^{+}+{\mathrm{14H}}^{+}$$ (2)

Autotrophic NRFeOx microorganisms can fix CO₂ to build biomass and enzymatically mediate Fe(II) oxidation coupled with nitrate reduction without amendment of organic carbon (8, 9). In contrast, mixotrophic NRFeOx microorganisms require additional organic carbon (e.g., acetate or lactate) in addition to Fe(II) as a cosubstrate (8). This can be explained by the fact that Fe(II) oxidation provides an energetic benefit (additional electrons for energy generation) during mixotrophic growth using Fe(II) and the organic cosubstrate (10). With the additional electrons from Fe(II), mixotrophic NRFeOx strains need to oxidize less organic cosubstrate for energy generation, so more organic matter can be used for biomass production. Therefore, the addition of Fe(II) promoted the cell growth of mixotrophic NRFeOx microorganisms, indicating that Fe(II) oxidation is a beneficial metabolic process for them (11, 12). It has to be noted that heterotrophic denitrifying bacteria can also cause Fe(II) oxidation abiotically by the reactive intermediate products (e.g., NO₂⁻ and NO) of nitrate reduction (13). They were referred to as “chemodenitrifiers” and do not belong to the group of true nitrate-reducing Fe(II) oxidizers (8). Therefore, growth advantage can be used to simply distinguish between mixotrophic NRFeOx microorganisms and chemodenitrifiers. Several pure isolates, such as Pseudogulbenkiania sp. strain 2002 (14), Microbacterium sp. strain W5 (15), and Citrobacter freundii strain PXL1 (16), and enrichment cultures, such as culture KS (named after the scientists isolating the culture, Kristina Straub) (6, 17, 18), culture BP (Bremen Pond) (19), and culture AG (Altingen Groundwater) (20), were described to have the capability of autotrophic nitrate-reducing Fe(II) oxidation, while several strains were capable of catalyzing nitrate-reducing Fe(II) oxidation with additional organic carbon such as Acidovorax sp. strain BoFeN1 (21, 22) and Acidovorax sp. strain BrG1 (6, 23). Recent studies have revealed that NRFeOx microorganisms can be found in various types of environments, such as freshwater sediments (21, 24–26), groundwater (20), marine coastal sediment (27), paddy field soils (28), and even shallow submarine hydrothermal systems (29) under circumneutral pH conditions. Hence, most if not all, NRFeOx microorganisms were found in freshwater and marine ecosystems with a salinity range up to 35 g/L (salinity of seawater). However, NRFeOx microorganisms have not been investigated over a large range of salinities (higher than seawater salinity).

Saline and hypersaline lakes are globally widespread and constitute almost half of the total inland water surface area (30, 31). With a wider salinity gradient from seawater salinity (35 g/L) to salt saturated (salinity > 300 g/L), salt lakes represent extreme habitats for microbial life (32, 33). The microorganisms living in high-salinity environments, with various adaptation strategies, need to spend more energy coping with osmotic stress, resulting in a high energy burden (34, 35). Thus, it is not surprising to observe the decrease in metabolic diversity with increasing salinity (36). However, metabolic processes which are thermodynamically favorable are expected to occur in hypersaline environments. For example, the metabolic process of nitrate-reducing Fe(II) oxidation is thermodynamically favorable, which has a proposed change in Gibbs free energy of −96.23 kJ mol⁻¹ for standard conditions at pH 7 (37). Consequently, the existence of the nitrate-reducing Fe(II) oxidization process has been demonstrated in hypersaline Lake Kasin (salinity > 300 g/L) sediments (38) by performing most probable number (MPN) enumeration experiments. However, the above-mentioned hypersaline NRFeOx cultures could not maintain nitrate-reducing Fe(II)-oxidizing activity after repeated transfers (38). Therefore, our current knowledge on the salinity boundary of culturable NRFeOx microorganisms remains scarce. In addition, an increasing number of studies have indicated that salinity is one important factor in nature in shaping the community compositions of Fe(II)-oxidizing bacteria (FeOB) (39, 40). For example, previous studies found that Betaproteobacteria-like FeOB were prevalent in freshwater, while Zetaproteobacteria-like FeOB were found in marine environments (39, 41, 42). Nevertheless, these previous studies rarely involve the response of FeOB, especially nitrate-dependent FeOB, to a large range of salinities (i.e., from freshwater to salt saturation) and determine their abundance, community compositions, as well as activity in saline lakes.

The Qinghai-Tibetan Plateau (QTP) hosts thousands of lakes (area > 1 km²), and these lakes possess an extensive salinity gradient (from 0.1 to 426.3 g/L), resulting in many saline and hypersaline lakes (43–45). Previous studies have shown that diverse iron minerals (46, 47) and high content of nitrate (48) are present in these QTP lakes. Therefore, QTP lakes with salinity gradients are ideal field sites to study the salinity impact on the community composition and activity of NRFeOx microorganisms. Therefore, the goals of the present study were to (i) determine the abundance and community composition of NRFeOx microorganisms in saline lakes with a full range of salinity from freshwater to salt saturation, (ii) examine the relationship between NRFeOx community composition and lake salinity, and (iii) characterize the microbial community compositions, the kinetics of nitrate reduction and Fe(II) oxidation, cell-mineral interactions, and mineralogy of the enrichment cultures.

RESULTS

Geochemical parameters and Mössbauer spectroscopy analysis of the investigated lakes.

The basic geochemical parameters varied widely in the studied lakes (Table 1). Salinity varied from 0.67 g/L to 346 g/L with pH ranging from 6.98 to 9.14. The water content of the sediments (determined by weight loss after heating the sediments) ranged from 16.54 to 62.11%. The content of total phosphorus (TP) and nitrogen (TN) in the lake water ranged from 0.025 to 0.579 mg/L and 1.36 to 8.81 mg/L, respectively. The concentrations of NO₃⁻-N and NH₄⁺-N were 0.11 to 1.69 mg/L and 0.16 to 2.11 mg/L in the lake water, respectively, while they were 2.49 to 8.27 mg/kg and 2.22 to 9.48 mg/kg in the sediment, respectively. Bioavailable and crystalline iron constituted 0.01 to 0.18% (wt/wt) and 0.69 to 2.29% (wt/wt) in the studied sediment, respectively (Table 1).

TABLE 1.

Locations and geochemical properties of lake sediments on the Qinghai-Tibetan Plateau

Location or geochemical property	Data for:
	EHL	QHL	GHL	XCDL	CKL
Geographic Position (N/E)	36.33/100.43	36.64/100.37	37.00/100.32	37.27/95.30	36.45/99.04
pHa	9.14	9.11	9.12	8.58	6.98
Salinitya (g/L)	0.67	13	33	40	346
Lake type	Freshwater	Polysaline	Polysaline	Polysaline	Hypersaline
Water content of sediment column (%)	53.78	35.78	27.61	16.54	62.11
Total phosphorusa (mg/L)	0.034	0.028	0.034	0.025	0.579
Total nitrogena (mg/L)	3.18	1.36	4.60	4.23	8.81
NO3−-Na (mg/L)	1.62	0.45	1.49	0.11	1.69
NH4+-Na (mg/L)	0.51	0.08	2.11	2.05	0.16
NOO3−-Nb (mg/kg)	2.54	8.27	2.49	2.59	NA
NH4+-Nb (mg/kg)	9.48	2.30	2.32	2.22	NA
Bioavailable Fe (wt % dry sediment)	0.11	0.10	0.08	0.01	0.18
Crystalline Fe (wt % dry sediment)	0.69	1.04	0.94	1.02	2.29

^aDetermined from water column.

^bDetermined from sediment column. Bioavailable Fe, 0.5 M HCl-extractable iron; Crystalline Fe, 6 M HCl-extractable iron. The concentrations of different iron fractions were reported as an average. EHL, Erhai Lake; QHL, Qinghai Lake; GHL, Gahai Lake; XCDL, Xiaochaidan Lake; CKL, Chaka Lake; NA, not available.

Mössbauer spectra showed that all lake sediment samples contained Fe(II) and Fe(III) mineral phases (see Fig. S1 and Table S1 in the supplemental material). The observed doublet signals indicated that Fe(II) and Fe(III) existed in clay or sheet silicate minerals. The sextets indicated that some lake samples (e.g., Qinghai Lake [QHL], Xiaochaidan Lake [XCDL], and Chaka Lake [CKL]) contained poorly crystalline ferric oxyhydroxides, namely, ferrihydrite or poorly crystalline goethite with small particle sizes. There was the possibility of vivianite in CKL sediments (Fig. S1; Table S1).

Abundance and community composition of NRFeOx microorganisms in the five lake sediments with different salinities.

The average MPN-based abundances of NRFeOx microorganisms were 7.55 × 10³, 3.07 × 10³, 3.16 × 10², 81.2, and 8.09 cells/g dry sediment for Erhai Lake (EHL), QHL, Gahai Lake (GHL), XCDL, and CKL, respectively (Fig. 1). The MPN-based abundance of NRFeOx microorganisms decreased with increasing lake salinity, as indicated by a significant correlation (R² = 0.85, P = 0.03) (Fig. S2).

FIG 1.

Most probable number (MPN) counts of nitrate-reducing Fe(II)-oxidizing (NRFeOx) microorganisms enriched from lake sediments with different salinities (EHL, Erhai Lake; QHL, Qinghai Lake; GHL, Gahai Lake; XCDL, Xiaochaidan Lake; CKL, Chaka Lake). Error bars indicate 95% confidence intervals determined from seven replicate samples.

The predominant phyla in the NRFeOx MPN-enrichment cultures were distinct among the five lakes with different salinities (Fig. S3). For example, Proteobacteria, Chloroflexi, and Bacteroidetes were dominant (relative abundance > 5%) in the EHL (lake salinity, 0.67 g/L) and QHL (lake salinity, 13 g/L) enrichment cultures, while Proteobacteria was the most abundant phylum in the enrichment cultures of GHL (lake salinity, 33 g/L) and XCDL (lake salinity, 40 g/L), whose relative abundance was up to 87.02% (the highest). Enrichment culture CKL (lake salinity, 346 g/L) mainly consisted of Firmicutes (accounting for 61.73%), Halanaerobiaeota, Proteobacteria, and Euryarchaeota. Such a difference in microbial compositions of the NRFeOx MPN enrichments among the studied lakes was further supported by cluster analysis, which showed that the community compositions of the MPN enrichments were clustered according to the lake salinity (Fig. S3).

Principal-coordinate analysis (PCoA) results showed the community compositions of the studied five NRFeOx enrichment cultures obtained from the MPN experiments were separated and formed clusters according to their lake salinity (Fig. 2). One-way permutational multivariate analysis of variance (PERMANOVA) also indicated that the compositions of the overall NRFeOx communities in the enrichments obtained from the MPN experiments differed significantly (R² = 0.74, P = 0.004) among the five studied lake sediments (Fig. 2).

FIG 2.

PCoA analysis showing microbial community dissimilarities of MPN-based NRFeOx enrichment cultures obtained from lake sediments with different salinities. Values on PCoA axes indicate the percentages of total variation explained by each axis. EHL, Erhai Lake; QHL, Qinghai Lake; GHL, Gahai Lake; XCDL, Xiaochaidan Lake; CKL, Chaka Lake.

Community composition and kinetics of NRFeOx metabolic processes of two stable enrichment cultures.

After a few consecutive transfers, the NRFeOx activity of enrichment cultures obtained from Qinghai Lake (FeN-QHL; FeN indicates the enrichments obtained from microcosm incubations), Gahai Lake (FeN-GHL), and Xiaochaidan Lake (FeN-XCDL) could not be maintained, so they were removed from subsequent analysis. In contrast, the enrichment cultures obtained from freshwater Erhai Lake (FeN-EHL) and hypersaline Chaka Lake (FeN-CKL) still maintained normal NRFeOx activity. After more than 10 consecutive transfers, the microbial community composition in the FeN-EHL and FeN-CKL enrichment cultures was identified by Illumina sequencing. The relative abundance of the dominant genera was distinct between these two enrichment cultures (Fig. 3). The FeN-EHL enrichment culture was mainly composed of genera of Gallionella (relative abundance, 68.4%), Rhodoferax (9.08%), Thiobacillus (1.12%), and others (21.66%) (Fig. 3), while the FeN-CKL enrichment culture mainly consisted of genera of Marinobacter (relative abundance, 71.2%), Thiobacillus (2.01%), Adurb.Bin120 (belonging to the family Anaerolineaceae, 1.11%), and others (25.66%) (Fig. 3).

FIG 3.

Community composition (at genus level) of two stable enrichment cultures of FeN-EHL (day 9) and FeN-CKL (day 16).

Culture FeN-EHL was cultivated autotrophically with Fe(II) as the electron donor and nitrate as the electron acceptor, while culture FeN-CKL was cultivated mixotrophically with an additional organic substrate (acetate). In the FeN-EHL enrichment culture, Fe(II) oxidation and nitrate reduction began right after incubation and slowed down after day 6 (Fig. 4A). The culture FeN-EHL consumed 2.22 mM nitrate and 9.21 mM Fe(II) on average within 9 days, resulting in the ratio of reduced nitrate to oxidized Fe(II) being 0.24, which was slightly higher than, but still close to the expected theoretical stoichiometry ratio (0.2). Simultaneously, there was no obvious accumulation of nitrite and/or ammonium, and the cell number increased from 1.31 × 10⁵ to 2.02 × 10⁷ cells/mL after 9 days (Fig. 4A and B). In the enrichment culture of FeN-CKL, Fe(II) oxidation, nitrate reduction, and acetate oxidation started after a 3-day lag phase. After 16 days, culture FeN-CKL had consumed 5.30 mM nitrate, 6.00 mM Fe(II), and 1.67 mM acetate on average (Fig. 4C and D), with the ratio of reduced nitrate to oxidized Fe(II) (and acetate) being 0.69, which was much higher than the expected theoretical stoichiometry ratio (0.2), suggesting that more nitrate was reduced, probably fueled by electrons stemming from acetate oxidation but also by abiotic Fe(II) oxidation by nitrite stemming from the heterotrophic nitrate reduction. Total cell numbers increased rapidly from 1.63 × 10⁶ cells/mL on day 3 to 1.25 × 10⁷ cells/mL on day 6 (Fig. 4D) and reached 1.17 × 10⁷ cells/mL at the end of incubation (day 16). The accumulation of nitrite (2.31 mM) was observed (Fig. 4C), while no change was observed in ammonium concentration (Fig. 4D). Notably, Fe(II) oxidation did not cease even after the 16-day incubation, and nitrite was not completely removed from the culture FeN-CKL. In our setups without added iron, the FeN-CKL culture reduced 5.55 mM nitrate and oxidized 1.54 mM acetate (Fig. 4C and D). Total cell numbers increased from 1.98 × 10⁵ to 1.05 × 10⁷ cells/mL gradually within 16 days (Fig. 4D). In the abiotic controls of enrichment cultures of FeN-EHL and FeN-CKL, no changes were observed in the concentrations of Fe(II), nitrate, nitrite, acetate, ammonium, and cell numbers (Fig. 4).

FIG 4.

Oxidation of Fe(II), reduction of nitrate, and increase of cell numbers in the FeN-EHL and FeN-CKL enrichment cultures. (A and C) Time course variations of Fe(II), nitrate, and nitrite concentrations. (B and D) Time course variations of cell numbers and concentrations of ammonium and acetate. The filled squares represent the biotic groups, and the open squares represent the abiotic groups. The triangles and dashed lines represent the FeN-CKL treatments without addition of iron. All data points are mean values of samples from three replicates; error bars represent standard deviations calculated from triplicates.

Minerals produced by the two enrichment cultures.

X-ray diffraction (XRD) results showed that the Fe(III) precipitates in the FeN-EHL and FeN-CKL enrichment cultures obtained from microcosm incubations were mainly composed of amorphous or poorly crystalline minerals (Fig. S4). The flow cytometry data confirmed that the cells were detectable (bright green fluorescence) in the enrichment cultures (Fig. S5A and B). Scanning electron micrographs of the two enrichment cultures showed that some cells were partially associated with the Fe(III) precipitates (Fig. 5A and B), whereas most cells were fully encrusted by minerals after Fe(II) oxidation (Fig. 5C and D).

FIG 5.

Scanning electron micrographs (SEM) of cells in the FeN-EHL and FeN-CKL enrichment cultures. (A and B) Partial encrustation; (C and D) Full encrustation of cells at day 7. White arrows indicate cells.

DISCUSSION

Occurrence and speciation of iron in QTP lake sediments.

A substantial fraction of Fe(II) and Fe(III), as well as specific Fe(II)- and Fe(III)-containing minerals (e.g., vivianite, ferrihydrite, goethite), was identified by Mössbauer spectroscopy in the sediments of all five studied lakes, which implied the potential occurrence of an active iron cycle in these QTP lakes. A similar mixed-valence iron speciation in Erhai Lake (the same lake used in our study) sediments was reported in a previous study using Mössbauer spectroscopy (47). Based on our mineral analyses, vivianite seems to be present in Chaka Lake sediments (49, 50). While generally less common in lake sediments, we attributed the presence of this mineral phase to the high salinity and relatively high phosphate concentration. Apart from the QTP saline lakes investigated here, other salt lakes also contain diverse iron minerals. For example, goethite, alunite, and jarosite were identified in hypersaline Lake Tyrrel sediments (51); akageneite, green rust Fe(II) phases, and lepidocrocite were reported in the hypersaline Lake Kasin based on Mössbauer spectra (38); hematite, goethite, jarosite, and pyrite were tentatively identified in the hypersaline Lake Whurr sediments, while goethite and unidentified paramagnetic Fe(III) phase were reported in hypersaline Lake Orr sediments (52). The occurrence and formation of Fe(II) and Fe(III) minerals in saline lakes have been related to microbial iron oxidation and reduction. Specifically, the bioavailable iron fraction can serve as electron donor and acceptor for iron-oxidizing and iron-reducing microorganisms, respectively (38). Therefore, combined with the high concentrations of NO₃⁻-N and NH₄⁺-N in the QTP lakes, the presence of active nitrogen and iron biogeochemical cycles, including nitrate-dependent Fe(II)-oxidizing microorganisms, can be expected in these lake sediments with different salinities.

Salinity impact on abundance and composition of NRFeOx communities in the MPN enrichments of the studied lake sediments.

Salinity had a significant (P < 0.05) effect on the abundance of NRFeOx microorganisms in the MPN enrichments (Fig. 1; see Fig. S2 in the supplemental material). This finding was consistent with previous studies that microbial abundance decreased with increasing salinity (53, 54). A possible explanation for this negative correlation could be the extracellular osmolarity caused by salinity. Many NRFeOx microorganisms may fail to adapt to high osmotic stress, thus leading to a lower microbial abundance (34, 54). Similarly, the variation in community compositions of the MPN-based NRFeOx enrichment cultures was also attributed to changes in salinity of these sediments. Specifically, we found that Proteobacteria were dominant in freshwater and low-salinity NRFeOx enrichment cultures, whereas Firmicutes and Halanaerobiaeota became abundant in the hypersaline NRFeOx enrichment culture CKL (Fig. S3). Such community compositional differences were expected based on a previous study showing distinct community compositions of iron-oxidizing bacteria between freshwater and marine environments (39). These distinct microbial communities may be explained by specific salinity-induced physiological constraints associated with osmoregulation, energy, and low water availability (55–57). Therefore, salinity will select microbial communities that are more tolerant to osmotic stress.

Distinct community composition and physiology of the two enrichment cultures.

Independent of the MPN enrichments, we obtained two stable enrichment cultures from two of the investigated lakes (EHL and CKL). The autotrophic freshwater NRFeOx enrichment culture FeN-EHL was dominated by the genus Gallionella (Fig. 3), belonging to the Gallionellaceae family. This family is a typical group of microaerobic Fe(II)-oxidizing bacteria capable of autotrophic growth with Fe(II) as electron and energy source (58). The Gallionellaceae family also comprises nitrate-reducing Fe(II) oxidizers and represents major members in other autotrophic NRFeOx enrichment cultures, such as enrichment culture KS (17, 59), enrichment culture BP (19), and enrichment culture AG (20). The genus Thiobacillus (relative abundance, 1.12%) is also known as a Fe(II) oxidizer and nitrate reducer, such as the chemolithotrophic iron-oxidizing bacterium Thiobacillus ferrooxidans (60) and Thiobacillus denitrificans (61). Notably, the second most abundant bacterium present in the culture FeN-EHL was affiliated with the genus Rhodoferax, which was described before as a Fe(III) reducer (62, 63). The potential electron donor for Fe(III) reduction in this system may be organic compounds derived from autotrophic nitrate-reducing Fe(II)-oxidizing bacteria Gallionellaceae (leading to cross-feeding) (20). Our findings indicate that these microorganisms found in the culture FeN-EHL may perform iron oxidation and iron reduction, forming an active iron redox cycle in the existing enrichment culture as well as in the freshwater sediments where the culture was obtained.

In the hypersaline enrichment culture FeN-CKL, the dominant genus was Marinobacter, belonging to the class Gammaproteobacteria (Fig. 3). Most members of the genus Marinobacter were able to grow anaerobically through denitrification coupled with the oxidation of a suitable carbon substrate (64, 65). Interestingly, previous studies found that Marinobacter-related isolates processed iron transport function (66) and were capable of oxidizing iron with the formation of oxidized iron bands in semisolid gradient tubes (67). However, Marinobacter subterrani JG233 was recently shown to be a culture of oligotrophic bacteria that did not appear to gain any metabolic benefit from iron oxidation (68). In other words, the Fe(II) oxidation phenotype of Marinobacter may be the result of chemodenitrification of Fe(II) and nitrite, rather than a biologically mediated process. Therefore, Marinobacter in the FeN-CKL enrichment may play an important role in nitrate reduction (64) and possibly oxidize Fe(II) simultaneously under mixotrophic conditions (66, 69). Apart from the genus Marinobacter, bacteria of the genera Thiobacillus and Adurb.Bin120 (belonging to the family Anaerolineaceae) were the minor components of the FeN-CKL enrichment culture and were shown to utilize iron as the electron donor for denitrification in previous studies (61, 70).

The community compositions were distinct between the FeN-EHL and FeN-CKL enrichment cultures. Such difference can be explained by distinct species-specific ecological niches of the two cultures enriched from lake sediments with different salinities. Our autotrophic freshwater FeN-EHL NRFeOx enrichment culture was dominated by the genus Gallionella. The Gallionellaceae groups were typically found in low-salinity environments, such as freshwater sediments (6), steams (71), aquifers (20), and ponds (19). This implies that the Gallionellaceae spp. were prevalent in freshwater or low-salinity environments. In contrast, the genus Marinobacter was dominant in our hypersaline FeN-CKL enrichment culture. Marinobacter species are typical halophilic or halotolerant bacteria and are widely distributed throughout marine and saline environments (64, 72–75). Thus, the distinct community compositions identified in the two enrichment cultures obtained from two sediments with different salinities were supported by the existing knowledge about the environmental distribution of the two key microbial [putative Fe(II)-oxidizing] genera of strains in these enrichments, i.e., Gallionella and Marinobacter.

In addition to microbial compositions, the interaction type between Fe(II) and nitrate-reducing bacteria differed between the two enrichment cultures, i.e., autotrophic NRFeOx and chemodenitrification [reaction of Fe(II) with nitrite] in the freshwater and hypersaline environments, respectively (Fig. 4). Specially, freshwater culture FeN-EHL can grow autotrophically, with a ratio (0.24) of reduced nitrate to oxidizing Fe(II) similar to the theoretical stoichiometry (0.2) of microbial nitrate-reducing Fe(II) oxidation (equation 1). Similar ratios of reduced nitrate to oxidizing Fe(II) were also shown in autotrophic NRFeOx culture KS (0.21 to 0.24) (17) and a marine sediment enrichment (0.22 to 0.28) (27). In contrast, the hypersaline culture FeN-CKL required an additional organic substrate (acetate in this study), with the ratio of reduced nitrate to oxidized Fe(II) (and acetate) being much higher than the theoretical stoichiometry (0.69 versus 0.2). This suggests that the FeN-CKL culture may perform not only microbially mediated nitrate-reducing Fe(II) oxidation under mixotrophic conditions. The results of the treatments without addition of iron further confirmed that iron oxidation did not provide remarkable growth benefits for cells in the FeN-CKL culture (Fig. 4C and D). Furthermore, nitrite production was observed in the FeN-CKL culture, and nitrite accumulation was found until the end of incubation (Fig. 4C). This observation is consistent with previous studies. For example, nitrite accumulation, even at concentrations up to several millimolars, has been described in NRFeOx cultures of Acidovorax strain BoFeN1 (21) and Pseudogulbenkiania sp. strain 2002 (14). It should be noted that the by-products (e.g., NO₂⁻, NO) of heterotrophic denitrification can abiotically oxidize Fe(II) (76, 77). Therefore, most so-called mixotrophic NRFeOx cultures may not be able to oxidize Fe(II) enzymatically and are considered chemodenitrifiers (8). Hence, based on the kinetic results and stoichiometric analysis in this study, the Fe(II) oxidation phenotype observed in the FeN-CKL culture may be the result of chemodenitrification (or at least chemodenitrification was dominant) rather than a biologically mediated process in hypersaline environments. While some of the published NRFeOx strains may be true “mixotrophs” and they oxidize Fe(II) and organic compounds enzymatically (for example, Acidovorax sp. BoFeN1; see reference 22). Their mixotrophy has ever been supported by isotope experiments or iron(II) oxidase analyses (as a genomic indicator) (2, 78). Therefore, real hypersaline (autotrophic) NRFeOx microorganisms await further exploration.

Mineralogy and cell-mineral interactions of the two enrichment cultures.

The main Fe(III) mineral of the two enrichment cultures was amorphous or poorly crystalline Fe(III) hydroxides (Fig. S4), with a morphology similar to ferrihydrite (79), as shown in the scanning electron micrographs (SEM) (Fig. 5). Our observations are in line with the NRFeOx culture AG enriched from a nitrate-contaminated groundwater aquifer. Combined Mössbauer data and XRD results showed that ferrihydrite was the product of microbial Fe(II) oxidation (20). The NRFeOx culture KS also produced poorly crystalline Fe(III) hydroxides (ferrihydrite) as its Fe(II) oxidation product (77). However, in the NRFeOx cultures enriched from a hot spring, magnetite and/or siderite were formed after Fe(II) oxidation with nitrate reduction within 1 week (80). Furthermore, green rust, magnetite, lepidocrocite, or goethite were formed during Fe(II) oxidation by the mixotrophic nitrate-reducing Acidovorax sp. strain BoFeN1 under different-solution chemistry (81–83). The differences in the formed Fe(III) minerals may be attributed to the geochemical and physical conditions (e.g., pH, temperature, organic matter, and nutrients) at the sampling sites as well as in the growth medium and the strains involved in Fe(II) oxidation (2, 84). It is intriguing that no obvious difference was observed in the formed Fe(III) mineral precipitates between the freshwater culture FeN-EHL and the hypersaline culture FeN-CKL (Fig. 5). This implies that salinity plays a minor role in controlling the identity of microbially formed Fe(III) minerals.

Additionally, the formed Fe(III) minerals were associated with the cell surface or even fully encrusted the cells in the two enrichment cultures (Fig. 5). This finding is inconsistent with the NRFeOx culture AG, in which most cells were able to avoid mineral precipitation at their cell surface (20). However, encrustation of cells in Fe(III) minerals has been observed for several mixotrophic NRFeOx microorganisms (21, 77, 85), which is expected to be harmful to cells and hinder metabolism (79). Therefore, cells have developed several possible mechanisms to avoid encrustation, such as modification of the cell surface charge (79), establishment of an acidic microenvironment (86), and the release of organic ligands to keep the Fe(III) at least close to the cell environment in solution (87). The extent of cell-mineral associations depends on the identity of the strains and the extent of abiotic or biotic Fe(II) oxidation (2). For instance, the autotrophic NRFeOx culture KS proceeds without significant cellular encrustation in Fe(III) minerals, while significant cell encrustation was observed in the presence of abiotic Fe(II) oxidation (chemodentrification) in the presence of nitrite formed by heterotrophic nitrate reduction (77). Finally, the extent of cell encrustation is probably also dependent on the present Fe(II) concentrations, and it has to be determined whether the lab-observed encrustation observed typically at millimolar concentrations of Fe is also occurring at more environmentally relevant submillimolar Fe concentrations.

In summary, the abundance and community structure of NRFeOx microorganisms in the studied saline lakes varied with lake salinity. The abundance of NRFeOx microorganisms was negatively correlated with lake salinity. The dominant microbial groups changed from Proteobacteria, Chloroflexi, and Bacteroidetes in freshwater and low-salinity lakes to Firmicutes (accounting for 61.73%), Halanaerobiaeota, Proteobacteria, and Euryarchaeota in hypersaline lakes. Salinity also affected the type of interactions between nitrate reduction and Fe(II) oxidization: there was autotrophic growth in the freshwater FeN-EHL culture versus chemodenitrification in the hypersaline FeN-CKL culture. Collectively, this study provides new insights into the abundance, community compositions, as well as activity of NRFeOx microorganisms in saline lakes. Future investigation is required to explore the genomic and metabolic information of real mixotrophic nitrate-dependent Fe(II)-oxidizing microorganisms in hypersaline environments.

MATERIALS AND METHODS

Site description and sample collection.

Five lakes with different salinities on the Qinghai-Tibetan Plateau (QTP), China, were selected for the enrichment of microorganisms capable of nitrate reduction coupled with iron oxidation (Table 1). Erhai Lake (EHL) is a freshwater lake with a maximum water depth of 2.0 m; Qinghai Lake (QHL), Gaihai Lake (GHL), and Xiaochaidan Lake (XCDL) are polysaline lakes with maximum water depths of 27.0 m, 9.0 m, and 1.0 m, respectively; and Chaka Lake (CKL) is a hypersaline lake with average water depth of 20 to 30 cm. A detailed description of the lake geochemistry can be found in our previous study (48). Field measurements and sampling of EHL, QHL, GHL, and XCDL were conducted in June 2019, while CKL samples were collected in June 2018. In the field, a portable GPS unit eTrex H (Garmin, China) was applied to determine the geographical locations of the sampling sites. For each sampled lake, surface sediments (about 1 to 5 cm deep) were collected in biological triplicate using a grab bucket collection sampler XDB0201 (New Landmark, China). The sediments were processed for microcosm experiments and physicochemical measurements. Specifically, the sediment samples for microcosm experiments were immediately transferred to sterile Ziploc bags, their air removed, and sealed. These samples were shipped to the laboratory at 4°C as soon as possible and distributed in a glove box (100% N₂). The triplicate sediments from each lake were mixed into unique representative and homogenized samples in the glove box and then transferred into sterile serum bottles. Serum bottles were sealed with rubber stoppers and aluminum caps and were then stored at 4°C refrigerators in the dark until enrichment cultures were established. Meanwhile, sediments for physicochemical measurements were mixed homogeneously in the field and then collected into 50-mL sterilized Eppendorf tubes using a sterile spatula. About 500 mL of lake surface water was collected into 2-L autoclaved polycarbonate bottles (Nalgene, USA) for subsequent geochemical analysis. The sediment samples for geochemical analysis and water samples were stored on ice in the field and during transportation and were then stored at 4°C refrigerators in the laboratory until further analysis.

Laboratory geochemical analyses and Mössbauer spectroscopy.

The pH and salinity of lake waters were measured with portable meter SX711 (Sanxin, China). The water content of the sediments was determined in triplicate by weighing wet sediment at the beginning, drying it in the oven at 105°C for 8 h to constant weight, and then recording the dry weight. Total phosphorus (TP) and total nitrogen (TN) in lake water samples were measured with colorimetric methods (88, 89). Determination of NO₃⁻-N and NH₄⁺-N in water and sediment columns was performed using the UV spectrophotometric method and indophenol blue method as previously described (90).

To quantify the concentration of different iron fractions in the sediment samples, 0.5 M HCl was chosen for extracting bioavailable iron, and 6 M HCl was chosen for extracting crystalline iron. Fe extractions were performed in an anoxic glove box by using the sequential extraction protocol (91). The Fe concentrations were subsequently analyzed in technical triplicate by the spectrophotometric ferrozine assay (92).

Fe speciation of five lake sediments was analyzed by ⁵⁷Fe Mössbauer spectroscopy. For this purpose, about 1 g of homogenized lake sediments was loaded into plexiglass holders (area, 1 cm²), forming a thin disc. Samples were stored at −20°C until measurement. Absorption spectra were collected at 77 K and 5 K using a constant acceleration drive system (WissEL) in transmission mode with a ⁵⁷Co/Rh source. All spectra were calibrated against an α-⁵⁷Fe foil (7 μm thick, room temperature) and then fitted using the Voigt-based fitting (VBF) routine in the Recoil software (University of Ottawa) (93, 94). The half width at half maximum (HWHM) was constrained to 0.123 mm/s during fitting.

MPN counts and enrichment of NRFeOx microorganisms.

Most probable number (MPN) counts were performed to enumerate anaerobic nitrate-reducing Fe(II)-oxidizing (NRFeOx) microorganisms from lake sediments with different salinities according to the procedure described previously (38, 95). Briefly, 1 g of homogenized wet sediment was inoculated anoxically into tubes containing 9 mL medium (see Table S2 in the supplemental material) with aliquots of 1 mL each of a 7-vitamin solution, a trace element solution, and a selenite-tungstate solution (96). The pH of the medium was adjusted to the pH measured in the original lake water. The medium was prepared with a headspace of N₂/CO₂ (90:10) and buffered with bicarbonate. The final medium contained 10 mM FeCl₂ and 4 mM NaNO₃. Sterile anoxic Na₂MoO₄ solution was also added to a final concentration of 2 mM to inhibit microbial sulfate reduction (27, 97). Dilution series of sediment suspensions (10⁻¹ to 10⁻¹²) were set up in 96-well deep plates (1 mL volume each) and screened after 6 weeks of dark incubation (28°C). In the glove box, a seal (with a sticky side) was used to seal the deep-well plate. Then, the plate was put in an anaerocult bag (with wet pad plus wet oxygen indicator) according to the instructions and closed with 2 clips to keep anoxic environments. Positive wells were determined by a visual color change from green-gray to orange-brown and NRFeOx activity. These results were analyzed using the software program KLEE (98). Positive wells from MPN plates were transferred into fresh media to pursue further enrichment cultures. After two transfers, two NRFeOx enrichment cultures obtained from the MPNs of each lake were used for microbial community composition analysis (Fig. S6). However, NRFeOx enrichments from MPN plates could not be maintained after continuous transfers. Therefore, microcosm enrichments were set up.

Setup of microcosm enrichments and chemical analyses.

The homogenized lake sediments (approximately 2.5 g) were added into 58-mL sterile serum bottles containing 25 mL of anoxic medium (the same as for the MPN experiments) under an N₂ atmosphere in the glove box (Fig. S6). Serum bottles were sealed with butyl rubber stoppers and aluminum crimps, and the headspace was replaced by N₂/CO₂ (90:10 [vol/vol]) gas. After this step, the serum bottles were amended with 10 mM FeCl₂, 4 mM NaNO₃ and 2 mM Na₂MoO₄ in the presence or absence of 5 mM acetate as organic substrates. Enrichments for each lake were set up in duplicate. Blank setups were used as controls. The bottles were incubated at 28°C in the dark for 42 days. Enrichment cultures were consecutively transferred into the fresh medium as soon as NRFeOx activity was observed. Two NRFeOx enrichment cultures, named FeN-EHL and FeN-CKL, were obtained from microcosm incubations of sediments from freshwater lake Erhai (EHL; lake salinity of 0.67 g/L) and from the hypersaline lake Chaka (CKL; lake salinity, 346 g/L) over repeated transfers. Of the two enrichment cultures, the former could be cultivated under autotrophic conditions (FeN-EHL, 10 mM FeCl₂ and 4 mM NaNO₃) in freshwater medium, while the latter required an additional organic substrate (FeN-CKL, 10 mM FeCl₂, 4 to 6 mM NaNO₃, and 1 to 5 mM sodium acetate) in marine medium. Although the salinity of marine medium (approximately 50 g/L) was lower than lake salinity (>300 g/L), microorganisms (e.g., Marinobacter) in the culture FeN-CKL have the potential ability to withstand low-salt conditions due to their broad salinity range adaption and the carbon substrates in medium (99–101).

The kinetics of microbial nitrate reduction coupled with Fe(II) oxidation were determined in the two enrichment cultures, FeN-EHL and FeN-CKL, obtained from microcosm incubations. The enrichment cultures FeN-EHL and FeN-CKL used for kinetics experiments were transferred with 10% (vol/vol) inoculum over 20 subsequent transfers under autotrophic and mixotrophic conditions, respectively. Sacrificial setups (killed cells) were used as abiotic controls. Killed cells were obtained by heating the enrichment cultures in a water bath at 80°C for 30 min. All treatments and abiotic controls were performed in triplicate and cultivated at 28°C in the dark. The medium was prepared as follows: enrichment culture FeN-EHL was inoculated in freshwater medium with 10 mM FeCl₂ and 4 mM NaNO₃; enrichment culture FeN-CKL was inoculated in marine medium (see Table S2) with 10 mM FeCl₂, 6 mM NaNO₃, and 2 mM sodium acetate. Additionally, in the FeN-CKL enrichment culture, treatments without addition of iron (6 mM NaNO₃ and 2 mM sodium acetate) were conducted to demonstrate the mixotrophic growth advantage. At each sampling point, the concentrations of Fe(II), Fe(total), NO₃⁻, NO₂⁻, NH₄⁺, and acetate, as well as cell numbers, were determined. Samples were taken from the two enrichment cultures with sterile syringes in the glove box under N₂ atmosphere without opening the bottles. The concentrations of Fe(II) and Fe(total) were determined using the revised ferrozine protocol for nitrite-containing samples (76, 102). The supernatants were analyzed in technical triplicate by the spectrophotometric ferrozine assay (92). A 400-μL sample suspension was used for cell counts. The remaining suspension samples were centrifuged at 14,000 × g for 10 min to remove both cells and the Fe oxides. Subsequently, the supernatant (100 μL) was diluted with anoxic Milli-Q water (900 μL) before measuring the concentrations of NO₃⁻, NO₂⁻, and NH₄⁺ by using a flow injection analysis (FIA) system (3-QuAAtro; Bran+Lübbe, Norderstedt, Germany). Other supernatant samples were placed out of the glove box, fully exposed to air, and centrifuged at 14,000 × g for 10 min before acetate analysis. Acetate concentrations were measured using high-performance liquid chromatography (HPLC; Shimadzu) following protocols previously described (17).

DNA extraction, data processing, and statistical analyses.

The MPN-based NRFeOx enrichment cultures (after two transfers) from the five lakes (two of each lake) and two stable enrichment cultures (after more than 10 times of transfers) obtained from separate microcosm incubations were collected and used for DNA extraction. Total DNA was extracted using the FastDNA Spin kit for soil (MP Biomedical, USA) according to the manufacturer’s instructions. The 16S rRNA gene of the extracted DNA was amplified with a uniquely tagged primer set of 515F (5′-GTGYCAGCMGCCGCGGTAA-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) (103). Meanwhile, a unique 12-bp barcode sequence was added to the reverse primer (806R) to differentiate among samples. Each sample was amplified in triplicate in a 50-μL reaction system. PCR was performed at 94°C for 3 min, followed by 30 cycles at 94°C for 40 s, 56°C for 1 min, 72°C for 1 min, and a final cycle of 10 min at 72°C (104). The triplicate PCR products were purified using a DNA gel extraction kit (Axygen, USA). Cleaned PCR products were quantified using a NanoDrop spectrophotometer (NanoDrop Products, Wilmington, DE, USA) and pooled with equimolar concentrations followed by library construction. Sequencing libraries were typically created by fragmenting DNA with specialized adapters attached. Amplicon sequencing was carried out on an Illumina MiSeq platform, with paired-end sequencing of 2 × 250 bp (105).

The obtained raw reads were performed by using the UPARSE pipeline (106). The paired-end sequences were assembled with FLASH (v1.2.11) following default settings (107). The joined reads were demultiplexed, denoised, clustered, and quality filtered using QIIME2 v2018.6 software (https://qiime2.org/). High-quality reads were clustered into amplicon sequence variants (ASVs) at the threshold of 100% identity level (108). The data set of representative sequences from each ASV was aligned using the ribosomal database project (RDP) classifier algorithm at a bootstrap cutoff of 80% (109) against the SILVA 123 database (110). The ASV table was rarefied to equal sequence number (n = 11,472) for each sample enriched from MPN plates. The resulting rarefied ASV table was employed in subsequent statistical analyses. All statistical analyses of NRFeOx community structures obtained from MPN plates were carried out in the R program (http://cran.r-project.org/) implemented with various packages. Regression curve (power function) was performed to analyze the relations between the lake salinity and the abundance of NRFeOx microorganisms by using the basicTrendline package (model, log2P). NRFeOx community structures were evaluated by cluster analysis using the vegan package and hclust function (Bray-Curtis dissimilarity; permutations, 9,999; algorithm, UPGMA; method, “average”). Principal-coordinate analyses (PCoAs) were conducted to illuminate the community compositional differences among the NRFeOx microorganisms in the QTP lakes based on the Bray-Curtis dissimilarity, using the vegan and ape packages. One-way permutational multivariate analysis of variance (PERMANOVA) with the adonis function (permutations, 999) was performed to test the significance of NRFeOx community compositional differences among all samples.

Cell counts and scanning electron microscopy.

A mineral dissolution step was conducted before cell counting in the two enrichment cultures, FeN-EHL and FeN-CKL, obtained from microcosm incubations (19). The cells were stained with BacLight green stain (Thermo Fisher Scientific; 1 μM stain/1-mL sample). The cells were counted in triplicate by the Attune NxT flow cytometer (Thermo Fisher Scientific) equipped with a 488-nm laser as an excitation source, and the results were reported as an average. Cell numbers were determined with the instrument settings as described previously (19). Cells were distinguished from debris by gating based on their properties in the side-scatter (SSC) and fluorescence parameters.

The morphology of the cells and mineral precipitates in the two enrichment cultures, FeN-EHL and FeN-CKL, obtained from microcosm incubations was investigated with scanning electron microscopy (SEM). For SEM, each sample (900 μL) was fixed with 2.5% glutaraldehyde (100 μL) out of the glove box and left at 4°C overnight. The samples were dehydrated in ethanol (30, 75, 95, and twice at 100%) before embedding in hexamethyldisilazane (HMDS; Sigma-Aldrich, St. Louis, MO, USA) (111). The samples were platinum coated before imaging. Micrographs were examined by a Jeol JSM-6500F field emission SEM with a Schottky field emitter (working distance, 10 mm; acceleration voltage, 5.0 kV).

Mineralogical analysis of two enrichment cultures.

Liquid-suspended mineral precipitates (approximately 20 mL) of two enrichment cultures obtained from microcosm incubations were withdrawn with sterile syringes and were centrifuged at 14,000 × g for 10 min in an anoxic glove box (100% N₂). The supernatant was discarded, and the residual precipitates were air-dried anoxically for several days in the anoxic glove box. To remove the remaining salts (to avoid artifacts), the precipitates in enrichment culture FeN-CKL were washed with anoxic Milli-Q water three times in the glove box and analyzed before and after washing. The X-ray diffraction (XRD) analysis was carried out on a two-dimensional microdiffractometer (Bruker D8 Discover with GADDS, XRD²) equipped with a standard sealed tube with a cobalt anode (Co-Kα radiation; λ = 0.17903 nm) at parameters of 30 kV and 30 mA. The samples were not rotated, and reflection patterns were collected for 240 s at two detector positions (15° and 40°). The resulting diffractograms were analyzed using the software Match! v3.6.2.121.

Data availability.

Raw sequencing data obtained from this study have been deposited at the NCBI Sequence Read Archive (SRA) under BioProject accession number PRJNA769796.