Magnetite NPs modulate microbial nitrate reduction pathways. Previous studies have shown that strain 2002 is a common denitrifying bacterium which can utilize low-molecular-weight organic substrates (e.g., acetate used herein) as an electron donor for nitrate reduction33,34. In our study, denitrification by strain 2002 is also observed in the absence of acetate (Fig. 2a), suggesting that endogenous organic substrates (e.g., glycogen) are capable of supplying the fuel for nitrate reduction. In the denitrification pathway, cells of strain 2002 first transfer the electrons sourced from exogenous or endogenous organic substrates to NO3− to form intermediate NO2− and finally generate N2O or N25,6. Therefore, NO2− and the gaseous phase (N2O/N2) are commonly detected as the N product of denitrifiers when organic substrate is used (Fig. 2a-2c)34. In our study, the major N products are the gaseous phase and NO2− in the magnetite-free systems (Fig. 2a-2c). The addition of acetate remarkably enhances the production of gaseous phase but reduces the NO2− accumulation. In these magnetite-free systems, the NH4+ yields remain low, and the proportion of NH4+ within the N products is less than 7.62% (Table 1), which is consistent with previous studies5,6.
In addition to these exogenous or endogenous organic substrates, there is a growing body of evidence showing that strain 2002 can also utilize soluble or solid-phase Fe(II) [e.g., structural Fe(II) in phyllosilicates] as an electron donor for microbial nitrate reduction5,6,35. In those relevant studies, NH4+ was not produced or generated as a trace by-product. In our study, we also observed that the nitrate reduction was enhanced with the addition of magnetite NPs (Fig., 2). Intriguingly we observed that the production of NH4+ is significantly enhanced in the bio-reactors with magnetite NPs amendments, with the proportion as high as 36.6% (Table 1). These results suggest that the pathway of microbial nitrate reduction by strain 2002 in the presence of magnetite NPs is partially switched from denitrification to DNRA.
The abiotic reaction between NO3− and Fe(II) (neither in dissolved nor solid forms), although thermodynamically feasible, hardly proceeds at room temperatures without catalyst19,30,36. By contrast, NO2− has long been recognized as a chemical oxidant for aqueous Fe2+ ions37. This abiotic reaction between dissolved Fe(Ⅱ) and NO2− is termed “chemodenitrification”, because N2O and N2 are the predominant products, and only a trace amount of NH4+ can be generated21. Noticeably, the abiotic nitrate or nitrite reduction to ammonium (“chemo-DNRA”) can proceed in the presence of green rust38. Our present study further shows that chemo-DNRA is also initiated by magnetite NPs, especially for those smaller than 25 nm (Fig. 3 and Table 1). Therefore, it is reasonable to speculate that magnetite NPs modulate microbial nitrate reduction pathways via an interplay between the magnetite and the intermediate product NO2−. Compared with dissolved Fe(Ⅱ), Fe(Ⅱ)-bearing minerals normally have a higher activity19-22. Specifically for magnetite, electrons from its particle interior easily flow towards the surface at circum-neutral pH39. This so-called electron hooping process creates electronchemical active sites on magnetite NPs, perhaps accounting for the abiotic nitrate or nitrite reduction observed herein. It is interesting to note that higher NO2− reduction rates and more NH4+ yields are observed in the biotic systems than in the abiotic systems (Table 1). These results indicate that nitrite ammonification by magnetite in our biotic systems is not a simple abiotic process. Instead, the interaction between magnetite and strain 2002 is likely a key player in nitrite ammonification. Indeed, more oxygen vacancies are observed on the surface of magnetite NPs when exposed to strain 2002 (Fig. S3). It has been suggested that oxygen vacancies can accelerate the electron hooping process within magnetite40. As such, bio-activated nano-magnetite can serve as a enzyme-like catalyst in extracelluar nitrite ammonification. However, further experiments are warranted to verify the aforementioned hypothesis.
The reactivity of mineral nanoparticle is mainly controlled by the finite-size effects41. The results of Brunauer-Emmett-Teller (BET) analysis show that the pristine CM has larger surface areas (71.1±0.47 m2/g) than HM (15.3±0.16 m2/g) (Table 2). Indeed, a higher rate and larger extent of ammonification are observed in the abiotic and biotic systems with CM. In addition, the lattice defect is another important factor affecting the NPs reactivity42. Our HRTEM data reveal that HM has a better crystallinity (i.e., fewer lattice defects) than CM, thereby exhibiting a less reactivity. It should be noted that the bio-treated HM displays more lattice dislocations than its pristine counterparts (Fig. 5 and S3), indicating that microbial activity can affect the reactivity of magnetite NPs, perhaps via increasing the lattice defects.
Alteration of magnetite upon microbial nitrate reduction. The unit cell of magnetite contains eight Fe(II) and eight Fe(III) octahedrally-coordinated ions (B site), which are coupled in antiparallel magnetic orientation to eight Fe(III) tetrahedral coordinated ions (A site)32,43. The fully oxidized phase of magnetite at room temperatures is maghemite (γ-Fe2O3), in which the B sites are completely occupied by Fe(III) ions and vacancies43,44. In nature, there exist solid solution series between maghemite and magnetite45. According to our XRD data (Fig. 4 and Table 2), magnetite is maghemitized by strain 2002 to produce a maghemite-magnetite solid solution.
For the Fe(II)-involved microbial nitrate reduction, the iron oxidation reaction is a complex interplay of chemical and biological processes5. Nevertheless, the oxidation of magnetite would break the balance between Fe(II) and Fe(III) in its B sites and result in more Fe(III) ions occupying these sites. Given that Goldschmidt radius of Fe(III) is smaller than that of Fe(II) (0.785 Å vs. 0.920 Å), such solid incorporation of Fe(III) could lead to changes in mineralogy and magnetic properties of magnetite toward maghemite32,44, as evidenced by the smaller lattice constants, interplanar spacing (Fig. 4 and Fig. 5), and “harder” magnetism (i.e., decreased χ and increased Mrs/Ms and Bc, Table 2 and Fig. 6) of the bio-treated solid phases32,43,44,46.
CM has a similar particle size to the supper paramagnetic (SP, < 25 nm) particles, while the magnetic parameters of HM are mainly controlled by the single domain particles (SD, 20-80 nm) (Table 2)47. Consistent with its SP state, CM exhibits a higher χfd% value and lower Mrs/Ms and Bc values (Table 2)31,43,47. In comparison to HM, the microbially-mediated incomplete oxidation of CM is more significant, reflected by a higher Fe(II)/FeT (Chemical) decrease of the bio-treated CM (Table 2). As a result of magnetite partial oxidation, the hopping of electrons between Fe(II)-Fe(III) octahedral pairs in the B sites would decrease48, evidenced by a decrease of the LT-SIRM loss during Verwey transition region in ZFC and FC31, as observed in the bio-treated HM (Table 2). However, despite a higher oxidization degree, the above phenomenon is not observed in CM due to its SP characteristics (no obvious Verwey transition region in ZFC and FC).
Implications for N cycle in the ferruginous oceans. Our experimental results demonstrate that DNRA is highly promoted in the presence of magnetite NPs, especially in an organic substrate-deficient condition. The maximum N retention efficiency of magnetite-involved DNRA observed herein is 36.6%. This is much close to that of the dissolved Fe2+-initiated DNRA reported by Michiels et al3, which is about 40%. Although assessing the specific abundance of magnetite in Precambrian oceans is challenging, indirect evidence from some ferruginous environments (modern analogues for the Precambrian ocean) shows that magnetite NPs can exist as a significant primary mineral in water column17,19,25-28. For instance, the mineralogical and geochemical data from ferruginous lakes Matano and Towuti (Indonesia) revealed that primary magnetite accounted for 16-24% of Fe(II) species in water column16. Therefore, we and other researchers show that DNRA coupled to the oxidation of Fe(II) species, either in its soluble or solid form, is an important but overlooked mechanism of maintaining N levels in the Precambrian oceans (Fig. 1).
According to our mineralogical analyses, magnetite is partly oxidized as a result of microbial nitrate reduction, leading to the formation of a maghemite-magnetite solution. This observation suggests that magnetite NPs are susceptible to microbially-mediated oxidation reaction. Upon diagenesis, maghemite-magnetite solid solutions might convert to a more stable hematite-magnetite intergrowth, which is commonly found in the Paleoproterozoic rocks49.