*Corresponding author. Present address: Department of Systems Ecology, Institute of Ecological Science, de Boelelaan 1085, 1081 HV Amsterdam, The Netherlands. Tel.: +31-20-4446964; fax: +31-20-4447123, E-mail address: email@example.com
Observed inhibition of methanogenesis under Fe(III)-reducing conditions is usually explained by competition of methanogens and Fe(III)-reducing bacteria for the common substrates acetate and hydrogen. However, substrate competition alone cannot explain the strong inhibition of methanogenesis during Fe(III)-reduction. We demonstrate direct inhibition of methanogenesis by amorphous Fe(OH)3 at concentrations between 0 and 10 mM in experiments with pure cultures of methanogens. The sensitivity toward Fe(III) was higher for Methanospirillum hungatei and Methanosarcina barkeri grown with H2/CO2 than for Methanosaeta concilii and Methanosarcina barkeri grown with acetate. Cultures of Methanosarcina barkeri grown with H2/CO2 and methanol demonstrated a capacity for Fe(III) reduction, which suggests that Fe(III)-reduction by methanogens may also contribute to Fe(III) inhibition of methanogenesis. Our results have important implications for kinetic modelling of microbial redox processes in anoxic soils and sediments.
In anoxic soils and sediments, the pathway of organic matter degradation depends on the types and quantities of electron acceptors present. Ferric iron (Fe(III)) reduction and methanogenesis are dominant processes in most freshwater environments [1,2]. Understanding the biogeochemical controls on methane production is important to predict spatial and temporal patterns of the emission to the atmosphere of this important greenhouse gas. Several studies have shown that methane production is severely inhibited under iron reducing conditions [2,3]. This phenomenon is often explained by competition between Fe(III)-reducing and methanogenic microorganisms for common substrates such as acetate and hydrogen. Fe(III)-reducers are able to utilize acetate and H2 at concentrations far below levels that can be metabolised by methanogens [4,5]. However, experimental data of Ratering and Conrad  and recently of Lueders and Friedrich  and Roden  indicated that substrate competition cannot completely explain the strong inhibition of methanogenesis during Fe(III)-reduction. In addition, a mechanistic model of microbial redox processes showed that, although the higher affinity for acetate and H2 gave Fe(III)-reducers some competitive advantage over methanogens, this process could not explain the temporal dynamics of methane production . Instead, such dynamics could only be explained by the introduction of an empirical constant threshold reducible Fe(III) concentration above which no methane production occurred. Inhibition of methane production by Fe(III) was an important cause for acetate accumulation measured in anoxic sediments [2,6]. Moreover, previous investigations have shown that methanogens may transfer electrons to pathways other than methanogenesis, e.g. to reduce molecular sulphur , humic substances  and Fe(III) . Collectively, these studies suggest that some factor(s) other than substrate competition plays a significant role in inhibition of methanogenesis during Fe(III)-reduction.
We conducted experiments with pure cultures of methanogens to evaluate two nonmutually-exclusive hypotheses related to potential mechanisms for inhibition of methane production by Fe(III): (i) methanogens are directly inhibited by the presence of Fe(III) (also in absence of Fe(III)-reducers); and (ii) methanogens use the electron flow generated by acetate or H2 preferentially to reduce Fe(III), as long as Fe(III) is available. Concomitantly with these conditions, no methane is produced.
2Materials and methods
Three methanogenic pure cultures were used for the experiments: Methanosaeta concilii (DSM 3671) an obligate aceticlastic methanogen, Methanospirillum hungatei (DSM 864) a methanogen that uses H2/CO2 and formate and Methanosarcina barkeri (DSM 800) that can use H2/CO2, methanol and acetate. These model cultures cover the most important methanogenic conversions in freshwater sediments. All cultures were pregrown without Fe(III) and were transferred when the cells were growing in the exponential phase. Methanogenesis resumed immediately after transfer – in absence of Fe(III). All experiments were carried out at 30 °C.
Sterile 120-ml bottles were filled with 50 ml fresh bicarbonate-phosphate-buffered, 1 mM sodium sulphide reduced anoxic medium  that did not contain (apart from sulphide) cysteine or other components that may act as electron carrier that may contribute to abiotic Fe(III) reduction  and were closed with butyl stoppers. The bottles were inoculated with 5 ml of a culture of exponentially growing methanogens. M. hungatei and M. barkeri were grown with 1 mM acetate and 140 kPa 20% CO2/80% H2 and M. concilii was grown with 20 mM acetate and 100% N2 at a final pressure of 140 kPa. Freshly inoculated cultures were amended with 0, 1, 5 and 10 mM Fe(III). Fe(III) was added as neutralized FeCl3. Added FeCl3 hydrolysed rapidly to produce a colloidal suspension of amorphous Fe(OH)3. At each iron concentration, the experiment was carried out in triplicate for all species and a control of uninoculated amended medium. Total ionic concentration was kept the same in all treatments by NaCl addition, thus adding 30, 27, 15 and 0 mM NaCl to the treatments with 0, 1, 5 and 10 mM Fe(III), respectively. The bottles were incubated in the dark at 30 °C while shaken continuously at 125 rpm. At t= 0, 1, 2, 5, 9, 12 and 16 days, 0.1 ml liquid samples were taken quickly using 0.8-mm thick needles after shaking vigorously to ensure an even, representative, distribution of the colloids. Samples were extracted with water and 0.5 N HCl, respectively and analysed for Fe(II) and Fe(III). In addition, 0.3 ml gas samples were taken for analysis of CH4 and CO2. All syringes were flushed in bottles with a 100% N2 atmosphere and fresh sodium dithionite solution before sampling.
2.2Experiments 2 and 3
Experiments 2 and 3 differed in a few aspects from the first experiment: The treatment of 1 mM Fe(III) was omitted. M. barkeri was grown both with 1 mM acetate and 140 kPa 20% CO2/80% H2 and at 20 mM acetate and 100% N2 at 140 kPa. For both conditions a control of uninoculated amended medium was added. Experiment 3 contained additional treatments of M. barkeri at 5 mM Fe(III), 1 mM acetate and 140 kPa 20% CO2/80% H2 with the addition of 10 mM bromoethane sulfonate (BrES), an analogue of methyl-coenzyme M and inhibitor of methanogenesis, of M. barkeri grown with 25 mM methanol and 100% N2 at 140 kPa and of M. barkeri grown with 5 mM Fe(III), 25 mM methanol and 100% N2 at 140 kPa. Finally, an uninoculated treatment with 5 mM Fe(III) with the filter-sterile addition of 5 ml cell free extract from actively growing M. barkeri cultures was included. Samples were taken at t= 0, 1, 2, 5, 7, 10, 14 and 19 days, 0.1 ml liquid samples were taken for acid volatile sulphur (AVS), Fe(II) and Fe(III) and 0.2 ml liquid samples were taken for analysis of SO32−, SO42− and S2O32−. Iron was extracted by water (to quantify dissolved iron), 0.5 N HCl, 1 M ammonium acetate and 50 g/l sodium dithionite in 0.35 M acetic acid and 0.2 M sodium citrate. 0.5 N HCl is supposed to extract alone and only reduced Fe . Ammonium acetate is a commonly used extraction method in geochemistry to extract amorphous iron oxides , which is presumably the predominant source for iron reduction . Dithionite extractable iron, on the other hand, represents both amorphous and crystalline iron oxides and amorphous FeS  and would in this case represent all iron present. In addition, 0.2 ml gas samples were taken for analysis of CH4, H2S and CO2. At t= 5, 10, 14 and 19 days, 1 ml liquid samples were taken for S0 analysis. At t= 0 and t= 19 days, 2 ml liquid samples were taken for analysis of chromium reducible sulphur (CRS). Microscopic observations on the cultures were performed by phase contrast microscopy.
All liquid samples were centrifuged (after extraction, if any) for 5 min at 16,000g and supernatant was analysed. Fe(II) was analysed colorimetrically with ferrozine as reagents . After measurement, 0.25 M hydroxylamine in 0.25 M HCl was added and after 30 min absorbance was analysed again (for the determination of Fe(total) in solution). SO32−, SO42− and S2O32− were determined as described previously . AVS was determined as described by Trüper and Schlegel . S0 was analysed indirectly by conversion of S0 with SO32− to S2O32−. A 7.5% Na2SO3-solution in 30 mM mannitol was prepared anoxically. In a reaction tube, 1.5 ml of this solution was added with 1.5 ml 1 M NaOH to 1.0 ml liquid sample. The tubes were incubated for 48 h at 65 °C, while shaking. Converted S0 was measured as S2O32− as described above. CRS was determined according to the single step titration described by Fossing and Jörgensen . CH4 was analysed at 70 °C by GC on a molecular sieve column, coupled to a FID. H2S and CO2 were analysed on a Porapak Q column coupled to a TCD.
Effects of Fe(III) concentrations on methane production and differences in iron dynamics and dynamics of sulphurous compounds among species or between species and uninoculated media were analysed by a paired student's t-test. Correlations between methane production and 0.5 N HCl extractable Fe(III) were calculated using the Pearson's correlation coefficient. Backward multiple regression analysis was carried out for each species and substrate with methane production as dependent variable and 0.5 N HCl extractable Fe(II) and Fe(III) concentrations as independent variables. For this purpose, Fe(II) and Fe(III) concentrations were log-transformed to approach normal distribution. Probabilities of a difference in means separate for each point in time, as depicted in the figures, are based on Tukey's post-hoc tests following ANOVA analyses using sqrt-transformed methane data and log-transformed 0.5 N HCl extractable Fe(II) concentrations to approach normal distribution of the data.
Treatment specific results did not significantly depend upon the experiment (P>0.10) for any variable measured. The three experiments could thus statistically considered to be replicates and were thus combined, resulting in 3–9 replicates depending on the treatment.
In the absence of Fe(III), methanogens grown on H2/CO2 produced more methane than those grown on acetate (Fig. 1). In most cases, methane production was inhibited by the presence of Fe(III) (Fig. 1). Sensitivity to Fe(III) varied with methanogenic substrate and species (Fig. 1). M. hungatei was highly sensitive to Fe(III) and was almost completely inhibited by 1 mM Fe(III). M. barkeri, when grown on H2/CO2, showed a significantly (P < 0.05) decreased methane production at 5 and 10 mM Fe(III). When M. barkeri was grown on acetate, methane production was also significantly (P < 0.05) reduced at 5 and 10 mM Fe(III), although the effects were less than when grown on H2/CO2. Effects of Fe(III) on M. concilii, which also grew on acetate, were minor and not statistically significant (P>0.05). In controls without methanogens, methane was never produced.
Rates of methane production (in mol CH4 l−1 medium day−1) by M. barkeri and M. hungatei grown on H2/CO2 showed a strong negative correlation with 0.5 N HCl extractable Fe(III) concentrations (P= 0.008 and P < 0.001, respectively). Such correlations were insignificant for M. barkeri and M. concilii grown an acetate (P= 0.48 and P= 0.32, respectively).
The potential iron reducing activity of methanogens was determined from the dynamics of different iron species. The total iron balance, determined from dithionite iron extracts, did not change in time in any treatment, indicating that all iron was retrieved. During the incubation, chemical changes in iron speciation occurred as was shown by a significant decrease (P < 0.05) in time for ammonium acetate extractable total iron and for H2O extractable total iron. These changes were not significantly (P>0.10) different for methanogenic cultures and uninoculated media, indicating some chemical crystallisation of iron phases.
In most methanogenic cultures, the change in Fe(II) in the 0.5 N HCl extracts, which represents Fe(III) reduction best, was not significantly (P>0.10) different from uninoculated media. However, for M. barkeri cultures grown with H2/CO2, the increase in Fe(II) was significantly higher (P < 0.01) than in any of the other cultures (Fig. 2). Also a distinct colour change was detected in the H2/CO2-grown M. barkeri culture grown at 5 mM Fe(III)-which had the highest methane production rates of all cultures at 5 mM Fe(III). Initially all cultures amended with 5 mM Fe(III) had a reddish colour, caused by colloidal Fe(OH)3. After 7–10 days of incubation, the colloidal Fe(OH)3 was converted into a black precipitate. None of the other cultures or controls showed such behaviour (Fig. 3).
Also M. barkeri grown with methanol and 5 mM Fe(III) showed significantly higher (P < 0.01) increases in Fe(II) in time than any of the other cultures (Fig. 4(a)). Fe(II) production with methanol was similar to the one with H2/CO2, although methane production was slightly higher (Fig. 4(b)). No significant increases in Fe(II) were found with the addition of BrES or if cell free extracts of M. barkeri were incubated with Fe(III) (Fig. 4(a)).
Given that both Fe(II) and Fe(III) concentrations varied during the incubations, a backward multiple regression analysis was carried out for each species and substrate with methane production as dependent variable and 0.5 N HCl extractable Fe(II) and Fe(III) concentrations as independent variables. This analysis allows separating the effects of Fe(II) and Fe(III) on the inhibition of methanogenesis. Neither Fe(II) nor Fe(III) were significant (P>0.10) in case of M. barkeri and M. concilii grown an acetate, which is in agreement with the correlation analysis. Methane production of M. barkeri and M. hungatei grown on H2/CO2 was significantly (P= 0.002 and P= 0.004) decreased in presence of Fe(III), while Fe(II) dropped as an insignificant variable out of the regression equation.
Apart from H2/CO2 or methanol, sulphide was added as potential electron donor in all media. However, no Fe(III)-reduction was detected in M. barkeri cultures grown with acetate that also included sulphide. In media with Fe(III), sulphide was not detected in the gas phase and was always below 0.05 mM in the liquid phase without a clear trend in time. Moreover, no significant differences (P>0.10) between M. barkeri cultures and inoculated media were found for any of the potential oxidation products of S2−; S0, sulphate, sulphite and thiosulphate. CRS was for all cultures constant in time and around 1 mM, indicating that all sulphur was retrieved.
4.1Inhibition of methanogenesis by Fe(III)
Our experiments clearly demonstrate a direct inhibiting effect of Fe(III) on methanogenesis. This inhibition may have been the result of an increased redox potential (Eh) of the medium caused by ferrihydrite addition to the medium. Part of the Fe(III) will have reacted with the added 1 mM sulphide, because sulphide concentrations were strongly decreased in media to which Fe(III) was added. The stoichiometry of the reactions indicates that Fe(III) will always have been left, thus increasing the Eh. The Eh could not be measured given the risks of oxygen intrusion involved in such measurement. In our bicarbonate buffered system with a pH of 6.8  the E0′ for the prevailing couple, (Fe(OH)3+HCO3−)/Fe2+, is +200 mV at pH=7.0 . From this, in combination with the measured dissolved Fe(II) and Fe(III) concentrations, the Eh was calculated to vary in a small range between +50 and +135 mV. Fetzer and Conrad  found no inhibition effects of Eh on methanogenesis at Eh values below +420 mV. In our experiments, Eh remained far below this value and direct Eh-effects thus seem unlikely. Microbial modification of the dynamics of sulphurous compounds, e.g. through Fe(III), can also be ruled out because changes in any sulphurous compound compared to uninoculated media were not significant. The abiotic reactions between Fe(III) and S2− decreased S2− toxicity which would have stimulated methanogenesis , while the sulphate in the medium served as an S-source (unpubl. results). Because no Fe(III)-reducers were present in the cultures, competition between species for common substrates cannot explain the results either. Initial Fe(II) concentrations were very low and because methanogenesis at natural conditions occurs after Fe(III)-reduction – thus while potentially Fe(II) accumulated, it seems unlikely that Fe(II) inhibits methanogens (although this could not be proven directly by adding Fe(II), because this would have strongly affected the Eh). Our multiple regression analysis, in which Fe(II) dropped out of the equations as insignificant variable, confirmed the non-inhibiting effects of Fe(II) and suppression effects could be explained by Fe(III) alone. Fe(II) thus does not seem the inhibiting compound. Hence, direct suppression by Fe(III) is the most likely explanation for the observed suppression of methanogenesis.
Sensitivity toward Fe(III) was much stronger for methanogens grown with H2/CO2 than for methanogens grown with acetate. Fe(III) was present in a colloidal suspension. Therefore effects through direct Fe(III) uptake into the cell seem unlikely. Sensitivity differences may, however, be explained from the different enzymatic pathways involved in aceticlastic vs. hydrogenotrophic methanogenesis and especially from Fe(III) adsorption to co-factors and/or proteins at the outer membrane. Factor F420, an important electron carrier in methane production from H2/CO2 at the outer membrane , might have been inactivated at high Fe(III) concentrations, as suggested previously . Sensitivity of M. barkeri grown on acetate was greater than that of M. concilii. The pathway of acetate conversion in these species only differs in the activation of acetate to acetyl-CoA, which thus might have played a role in the inhibition. However, the precise biochemical mechanism(s) of either inhibition remains unknown. More research will be needed to elucidate these aspects.
Our findings may explain a previously poorly understood phenomenon in anoxic sediments; the changes in predominance of aceticlastic vs. hydrogenotrophic methanogenesis. After a domination of methane production by the hydrogenotrophic pathway during the first day after flooding , hydrogenotrophic methanogenesis quickly dropped to 10–20% of the total methane production for 1 week to a month [24–26], matching exactly the period dominated by iron reduction. After 2–3 months of flooding, hydrogenotrophic methanogenesis gained again in importance, up to 30%, what theoretically should be their maximum contribution to methanogenesis . These changes cannot be explained from changes in the population size, because the numbers of aceticlastic and hydrogenotrophic methanogens remained rather constant during this whole period . Instead, relative activities must have been changed. These changes might be explained from the difference in sensitivity toward Fe(III) as described above, which is more likely than an explanation based on H2 production by aceticlastic methanogens . Sensitivity differences toward Fe(III) may also explain why rRNA levels of Methanosaeta increased at ferrihydrite amendment, while Methanosarcina spp. showed suppressed dynamics .
4.2Fe(III)-reduction by methanogens
No significant increase in 0.5 N HCl-extractable Fe(II) was observed in most Fe(III)-amended methanogenic cultures compared to uninoculated controls. However, M. barkeri cultures grown with H2/CO2 and with 25 mM methanol in the presence of 5 or 10 mM Fe(III) formed significantly more Fe(II) than any other culture. No methane or Fe(II) was formed if BrES was added to the culture. Bond and Lovley  found no methane formation, but a continuation of Fe(III) reduction at the addition of both BrES for M. barkeri. However, their medium  contained cysteine and AQDS, which may have led to continued (abiotic) electron shuttling leading to Fe(III) reduction as discussed in detail by Doong and Schink .
Based on the redox reactions, it was calculated that on average 22%, 28% and 71% of the sum of electrons used within a treatment by M. barkeri went to Fe(III)-reduction at 5 mM Fe(III) with methanol, 5 mM Fe(III) with H2/CO2 and 10 mM Fe(III) with H2/CO2, respectively. The diversion of electrons to Fe(III)-reduction compared to the total metabolic activity was surprisingly constant with time for a given treatment. The total metabolic activity, calculated as the sum of electrons used relative to control treatments without Fe(III) addition, was more dynamic (Fig. 5). During the first two days, Fe(III) reduction led to metabolic activity on top of continuing methane production (Fig. 1(c)). After this initial stimulation, total metabolic activity decreased compared to treatments without Fe(III) addition, which means that not only quantitatively electrons were diverted from methanogenesis to Fe(III) reduction, but that also total metabolism was inhibited by Fe(III), as discussed above.
Purity of our culture was confirmed by microscopic observations, showing only the typical pseudosarcina morphology of M. barkeri. In addition, no growth occurred with glucose or lactate with and without Fe(III). No methane or Fe(II) was formed either, if cell free extracts from actively growing M. barkeri cultures in 5 mM Fe(III) were transferred into fresh medium with 5 mM Fe(III) (Fig. 4). These results indicate that Fe(III)-reduction is most likely mediated directly by living cells and not indirectly, e.g. through extracellular products. Cytochrome-c present in Methanosarcinales – but not in other methanogenic orders  and known to be involved in dissimilatory iron reduction  of which otherwise little is known – may play a key role in Fe(III)-reduction by M. barkeri.
Fe(III) reduction by M. barkeri was confirmed by colour changes in the cultures (Fig. 3) and microscopic observation showed black Fe-precipitates around M. barkeri clusters. The black precipitate formed in the H2/CO2-grown M. barkeri cultures was likely magnetite and/or mixed Fe(II)–Fe(III) compounds which are known to be produced during dissimilatory amorphous Fe(OH)3 reduction . These results are in agreement with results on reduction of soluble Fe(III)-citrate by hyperthermophilic methanogens  and of insoluble Fe(III)-oxide with hydrogen by some mesophilic methanogens . The Fe-precipitates may have impaired the efficiency of ferrihydrite reduction, which may explain why only 20–25% of the added Fe(III) had been converted into Fe(II) in 14 days.
The ability of M. barkeri to reduce Fe(III) may explain why this organism was less inhibited by Fe(III) than M. hungatei, because Fe(III) reduction decreases the inhibiting effects of Fe(III). The biochemical mechanism of Fe(III)-reduction in M. barkeri remains unknown. However, the results for M. barkeri suggest that reduction of Fe(III) by methanogenic archaea may also be partly responsible for suppression of methanogenesis in Fe(III)-reducing environments. Bond and Lovley  suggest that H2-oxidation, rather than methanogenesis per sé, was linked to Fe(III)-reduction by hydrogenotrophic methanogens. This process represents a type of intracellular competition for electron acceptor utilisation, which is analogous to the ability of certain sulphate-reducing bacteria to reduce Fe(III) in combination with increased efficiency of H2 scavenging . The resulting lower H2 concentrations by Fe(III)-reducing methanogens such as M. barkeri may contribute directly to Fe(III) suppression of H2-dependent non-Fe(III)-reducing methanogens. All together, this shows that kinetic and competition experiments (at electron donor and/or acceptor limiting conditions) between Fe(III)-reducing and non-Fe(III)-reducing methanogens, and Fe(III)-reducing methanogens and Fe(III)-reducing bacteria are needed to understand more adequately the interactions between these important physiological groups.
Our results have significant implications for understanding the spatial and temporal distribution of redox processes in anoxic sediments given that the applied reducible Fe(III) contents are not unreasonable for natural conditions. Studies that found inhibition of methane production at natural iron reducing conditions had similar, 1–4 mM, reducible Fe(III) contents measured by 0.5 N HCl extractions as used in the present study [3,9,34]. Our results thus imply that direct inhibition of methanogenesis by Fe(III) is likely an important ecological factor in controlling the redox sequence in freshwater environments which has thus far always been neglected. In addition, M. barkeri, when grown with H2/CO2, is able to reduce Fe(III). Even with this adaptation, the sensitivity to Fe(III) is larger for hydrogenotrophic methanogens than for aceticlastic methanogens. Taking these phenomena into account may substantially improve the ability of kinetic models to predict microbial redox processes in sediments, thereby increasing the utility of such models for various environmental purposes.
This work was supported financially by the Dutch National Research Program on Global Air Pollution and Climate Change. Eric Roden is thanked for the inspiring discussions that led to this study.