Fungal transformation of natural and synthetic cobalt-bearing manganese oxides and implications for cobalt biogeochemistry

Manganese oxide minerals can become enriched in a variety of metals through adsorption and redox processes, and this forms the basis for a close geochemical relationship between Mn oxide phases and Co. Since oxalate-producing fungi can effect geochemical transformation of Mn oxides, an understanding of the fate of Co during such processes could provide new insights on the geochemical behaviour of Co. In this work, the transformation of Mn oxides by Aspergillus niger was investigated using a Co-bearing manganiferous laterite, and a synthetic Co-doped birnessite. A . niger could transform laterite in both fragmented and powder forms, resulting in formation of biomineral crusts that were composed of Mn oxalates hosting Co, Ni and, in transformed laterite fragments, Mg. Total transformation of Co-doped birnessite resulted in precipitation of Co-bearing Mn oxalate. Fungal transformation of the Mn oxide phases included Mn(III,IV) reduction by oxalate, and may also have involved reduction of Co(III) to Co(II). These ﬁ ndings demonstrate that oxalate-producing fungi can in ﬂ uence Co speciation in Mn oxides, with implications for other hosted metals including Al and Fe. This work also provides further understanding of the roles of fungi as geoactive agents which can inform potential applica-tions in metal bioremediation, recycling and biorecovery.


Introduction
Mn oxides are ubiquitous in terrestrial and aquatic environments and occupy a central role in a range of biogeochemical processes. Mn oxide minerals are extremely redox active and also possess high sorptive capacities, which can lead to considerable enrichment by a range of secondary metals (Post, 1999;Tebo et al., 2004). Interactions between microorganisms and Mn oxides have been studied extensively, particularly the precipitation of Mn oxides oxidation of Mn(II) and the reductive dissolution of Mn oxides (Lovley, 1991;Lovley et al., 2004;Tebo et al., 2004;Miyata et al., 2004Miyata et al., , 2006aMiyata et al., , 2006b. However, less attention has been paid to the fate of the secondary metals that may be associated with the Mn oxide minerals during these processes. In particular, little information exists for cobalt which has a well-established geochemical association with Mn and is commonly hosted in Mn(III,IV) oxide minerals (Manceau et al., 1997;Nicholoson and Eley, 1997;Dzemua et al., 2013;Kuhn et al., 2017). Investigation of the fate of Co during Mn(III,IV) oxide transformations is therefore required to develop a more complete understanding of Co biogeochemistry. Newsome et al. (2020) showed that Co mobility in anaerobic laterite sediment microcosms was closely related to Mn release which was primarily attributed to indigenous Mn(IV) and Fe(III) reducing bacteria. This points to a direct relationship between Mn(III,IV) oxide mineral dissolution and Co mobility, and that anaerobic prokaryotes occupy an important role in such processes. However, little information exists on the fate of Co as a result of fungal interactions with Mn oxides, or indeed fungal interactions with Co generally. This is surprising given the capacity of several fungal species to precipitate (e.g. Acremonium sp., Stagonospora sp. and Pyrenochaeta sp.) or solubilize Mn oxides (e.g. Aspergillus niger, Serpula himantioides), and the common residence of Co within Mn oxide minerals (Post, 1999;Tebo et al., 2004;Miyata et al., 2004Miyata et al., , 2006aMiyata et al., , 2006bSantelli et al., 2011;Wei et al., 2012;Ferrier et al., 2019).
Geoactive fungi play an important role in a range of biogeochemical cycles and their involvement in the dissolution of mineral substrates, as well as in the formation of secondary biomineral phases, is well-established (Gadd 1999(Gadd , 2007(Gadd , 2010Gadd et al., 2014). Fungi can mediate the mobilization of nutrients contained within minerals, e.g. phosphate, sulfate, as well as essential, inessential and potentially-toxic metals, e.g. Na, K, Mg, Ca, Mn, Fe, Cu, Zn, Co, Ni, Al, Cs, Cd, Hg, Pb, U) (Burford et al., 2003;Fomina et al., 2007;Gadd, 2007Gadd, , 2010Rhee et al., 2012;Ceci et al., 2015;Liang et al., 2015;Yang et al., 2019;Mendes et al., 2020). While mineral bioweathering by fungi can be mediated by often synergistic interactions between biophysical and biochemical mechanisms, the secretion of mycogenic organic acids is central to such processes (Gadd, 1999). Oxalic acid secretion is of particular significance and is implicated in the transformation and solubilization of a range of mineral phases including carbonates, phosphates, arsenates, silicates, sulfides and oxides (Sayer and Gadd, 2001;Fomina et al., 2004Fomina et al., , 2007Burford et al., 2006;Wei et al., 2012Wei et al., , 2013Ceci et al., 2015;Ferrier et al., 2019;Suyamud et al., 2020;Mendes et al., 2020). Oxalic acid can attack minerals by both acidolysis and complexolysis (ligand mediated dissolution) as the oxalate anion (C 2 O 4 2À ) is a bidentate ligand capable of forming soluble complexes with a range of mono, di, tri and tetravalent cations (Gadd, 1999(Gadd, , 2007(Gadd, , 2010Gadd et al., 2014;Verma et al., 2019). In many cases, mineral dissolution and metal release is followed by the precipitation of oxalate biominerals entrapping solubilized metals. Factors that determine soluble complex formation or oxalate precipitation include the valence state of the cation, the relative abundance of oxalate anions and metals in solution, pH and stability constants of the oxalate complexes (Gadd, 1999;Verma et al., 2019) For example, simple oxalates containing divalent cations are sparingly soluble or insoluble and readily precipitate over a wide range of pH values (e.g. Ca, Mg, Mn, Co, Ni, Cu) but may also form complexes in the presence of excess oxalate, while trivalent metals such as Al(III) and Fe(III) only exist as soluble complexes (Gadd, 1999;Gadd et al., 2014;Verma et al., 2019;Kang et al, 2019Kang et al, , 2020Kang et al, , 2021Mendes et al., 2020). In addition, oxalate is a reductant which can effect the reduction of some metal species, including Mn(III,IV) to Mn(II), through oxalate precipitation (Wei et al., 2012;Verma et al., 2019). Oxalic acid has previously been shown to be able to transform natural, biogenic and synthetic Mn oxides. Wei et al. (2012)  Mn oxide minerals generally fall into two categories based on the arrangement of edge-or corner-sharing MnO 6 octahedra, the basic structural unit (Post, 1999;Tebo et al., 2004). These include phyllomanganates, composed of layers of edge-sharing MnO 6 octahedra (e.g. buserite and birnessite), and tectomanganates, which are composed of chains of edge-sharing MnO 6 octahedra the ends of which share corners with other chains to form tunnel structures (e.g. cryptomelane and todorokite) (Post, 1999;Tebo et al., 2004). Due to the presence of vacant octahedral sites, manganates possess large negatively charged surface areas. As a consequence of this, manganates in soils and sediments can incorporate a range of charge-balancing metal cations via adsorption including Ca 2+ , Co 2+ , Cu 2+ , Ni 2+ , K + , Na 2 + and Mg 2+ (Post, 1999;Tebo et al., 2004). Additionally, surface hydroxyl groups also provide sites for adsorption of metal cations, forming covalent bonds and preventing desorption (Nicholoson and Eley, 1997). In some cases, adsorbed metals are structurally incorporated into the oxides via oxidation and isomorphic substitution. For example, adsorbed Co(II) can be oxidized by layerresiding Mn(III), which then migrates to the interlayer space to be replaced by Co(III) (Manceau et al., 1997).
In this work, the fungal transformation of a Co-rich manganiferous laterite from Nkamouna, Cameroon and a synthetic Co-doped birnessite was investigated. The unexploited Nkamouna laterite represents a significant future Co resource, with sufficiently high ore grades to be one of the few operations that can produce Co as the primary metal, with significant quantities of Ni also present (Petavratzi et al., 2019) (Dzemua et al., 2011(Dzemua et al., , 2013. Lithiophorite is a phyllomanganate composed of alternating stacked layers of MnO 6 and Al(OH) 6 octahedra, with Li(OH) 6 commonly filling vacant sites in the Al(OH) 6 layers, while asbolane is structurally related to lithiophorite but with Al replaced by other transition metals, commonly Ni and Co (Chukhrov et al., 1982;Post, 1999). Cryptomelane is part of the hollandite group of tectomanganates, composed of tunnel structures formed by 2 Â 2 chains of MnO 6 octahedra and containing K + as the dominant charge-balancing cation (Post, 1999;Cheney et al., 2009). Birnessite ([Na,Ca,Mn (II)]Mn 7 O 14 Á2.8H 2 O) is a fine-grained poorly crystalline phyllomanganate composed of layers of edge-sharing MnO 6 octahedra separated by 7 Å hydrated interlayer spaces (Post, 1999;Tebo et al., 2004). Birnessite-type phases are ubiquitous in soils and sediments, and major components of seafloor manganese nodules (Post, 1999;Hein et al., 2013). While birnessite is known to occupy a central role in soil redox, sorption and ion-exchange reactions, it can be difficult to isolate due as it is generally present in low quantities and associated with other Mnand Fe-(oxyhydr)oxides, silicates and/or cell biomass (Toner and Sposito, 2005;Händel et al., 2013). Despite this, synthetic birnessites, e.g. K-birnessite, Na-birnessite and acid birnessite can be easily prepared to accommodate geochemical and biogeochemical investigations (Jokic, et al., 2001;Tebo et al., 2004;Toner and Sposito, 2005;Händel et al., 2013;Wang et al., 2013). A range of studies have used metal-doped birnessites, including those containing Co, Cu, Al and Ni (Lucht and Medoza-Cortez, 2015;Yadav et al., 2017;Arias et al., 2019;Legutko et al., 2019). In this work, a synthetic Co-doped birnessite was used to provide a general example for the fate of Co during Mn oxide transformations, to complement examination of the Nkamouna laterite. A. niger was chosen for this investigation as it can produce substantial quantities of oxalate and has a known capacity to effect manganese oxide transformations (Wei et al., 2012;Milov a-Žiakov a et al., 2016;Ferrier et al., 2019). It was hypothesised that Co-bearing Mn oxide minerals would be transformed via fungal oxalate secretion and that the composition of the secondary biominerals produced would be correlated with the chemical and mineralogical composition of the original mineral resource.

Results
SEM images of Nkamouna laterite fragments following 14 days incubation with A. niger showed extensive crusts of shard-like biominerals ( Fig. 1C and D). Control EDXA spectra showed peaks for Mn, O and Al, with smaller peaks also evident for C, Co, Ni and Fe ( Fig. 2A). The EDXA spectra obtained for the biomineral crust reported Mn, Co and Ni as before, although Fe and Al peaks were absent, and an additional Mg peak was present. Additionally, the C peak was increased relative to the uninoculated control (Fig. 2B). Table 1 summarizes pH, tolerance index and growth rate data for A. niger grown on MEA amended with 0.2% (w/v) Nkamouna laterite or synthetic Co-birnessite. The pH under A. niger biomass grown on laterite-amended media was pH 4.09, which is markedly higher than the pH 2.85 resulting on unamended, A. niger inoculated MEA. The pH value for uninoculated laterite-amended media was pH 5.92 after 14 days, while that of uninoculated MEA was pH 5.31. The tolerance index for A. niger on laterite-amended MEA was 97.7%, showing there was little decrease in biomass yield compared with A. niger grown on MEA. The growth rate ratio (R m :R c ) was 1.42, which showed that the rate of colony expansion on laterite-amended MEA was markedly higher than that of the control. On synthetic Co-birnessite amended media, the average pH under the biomass was pH 2.17,    Fig. 3. Transformation was not total, with many laterite particles not showing morphological changes compared with the control mineral, although some biomineral formation was observed including the formation of shard-like crusts ( Fig. 3B-D). These biomineral crusts displayed a morphology similar to that observed on the surface of Nkamouna laterite fragments, although they appeared poorly formed in comparison. Additional bipyramidal biominerals were also frequently observed ( Fig. 3B and C). The uninoculated control EDXA spectra for mineral debris displayed peaks for Mn, O, Al and Co, with smaller peaks for C, Ni, Si and Fe (Fig. 4A). In comparison, biomineral crusts ( Fig. 4B) showed increased peaks for Mn and Co, while the Al peak was reduced and Fe was not detected. EDXA of bipyramidal biomineral structures showed major peaks for calcium and smaller peaks for Mn, O and Al (Fig. 4C). XRPD analysis of the control Nkamouna laterite mineral debris showed patterns associated with lithiophorite [(Al, Li)]MnO 2 (OH) 2 ] which was the dominant mineral phase, with additional patterns for cryptomelane (K 1-1.5 (Mn 4+ , Mn 3+ ) 8 O 16 ) (Fig. 5A). The transformed laterite also showed patterns for lithiophorite, with a reduced pattern for cryptomelane and an additional pattern for manganese oxalate dihydrate (Fig. 5B). Figure 6 shows SEM images obtained for synthetic Cobirnessite following 14 days incubation with or without A. niger. Complete transformation was observed, with the fine, amorphous particles of Co-birnessite recovered from the uninoculated control (Fig. 6A) replaced with large    rosette structures (Fig. 6B-D). These rosette structures were up to approximately 100 μm in diameter (Fig. 6D). The EDXA spectra obtained for the uninoculated control Co-birnessite material showed clear peaks for Mn, O, K and Co (Fig. 7A). The spectrum obtained from the transformed material (Fig. 7B) also showed peaks for Mn, O and Co, although K was not detected and there was an additional C peak compared with the control. Uninoculated control material showed XRD patterns for birnessite (Fig. 8A), while the transformed mineral debris showed patterns corresponding to manganese oxalate dihydrate (lindbergite, MnC 2 O 4 Á2H 2 O) (Fig. 8B).  (Yang et al., 2020;Suyamud et al., 2020). As Al(III) and Fe(III) do not precipitate but form soluble oxalate complexes, it is likely that these elements were solubilized and removed. Furthermore, given the high sorptive capacity of Mn oxide minerals, it is also possible that Al(III) and Fe(III) were re-adsorbed onto the Nkamouna laterite as either aqueous or hydrolysed cations (Nicholson and Eley, 1997). The composition of biominerals encrusting Nkamouna laterite powder was also similar, showing a decrease in Al relative to the control, an absence of Fe, the presence of Mn, Co and Ni and the presence of Mn(II) oxalate dihydrate. It is unclear whether lithiophorite or cryptomelane were equally susceptible to attack by oxalic acid although, given the absence or decrease in Al content following transformation, it can be assumed that lithiophorite was certainly affected. This is particularly significant as lithiophorite is known to host particularly high Co concentrations (Post, 1999) and is the principle Co host phase in Nkamouna laterite (Yongue-Fouateu et al., 2006;Dzemua et al., 2013). The presence of bipyramidal Ca-bearing biominerals characteristic of Ca oxalate dihydrate (weddellite) further illustrated the role of oxalate (Gadd et al., 2014). Taken together, these results demonstrate that an oxalate-producing fungus can effect the transformation of Mn oxides in a laterite ore, and that hosted Co, Ni and Mg can be incorporated into the resulting Mn(II) oxalate to form a mixed oxalate biomineral phase. Synthetic Co-birnessite was also transformed by A. niger, with total solubilization of the original material observed. SEM revealed large rosette biomineral structures, similar to those observed following oceanic manganese nodule colonization by A niger  Moreover, EDXA showed the absence of K and the presence of C, O, Mn and Co, confirming solubilization of the synthetic birnessite, release of K, and precipitation of a Co-bearing Mn oxalate, the latter being confirmed by XRD. As Mn is hosted as Mn(III,IV) in lithophorite, cryptomelane and birnessite, oxalate excretion involves reduction to Mn(II) (Ehrlich and Newman, 2009;Wei et al., 2012). Moreover, given that Co is structurally incorporated into MnO 6 layers as Co(III), it is reasonable to suggest that reduction to Co(II) by oxalate occurred prior to oxalate precipitation in both the natural and synthetic material (Manceau et al., 1997). The increased pH of mineral-amended media (Table 2) and the buffering effect of manganese oxides are likely to be responsible for the increased growth ratio (R m :R C ) on this substrate since A. niger exhibits stable growth between pH 2 and pH 8 (Andersen et al., 2009). Our results demonstrate that transformation of Mn oxide minerals by mycogenic oxalate results in the mobilization and immobilization of secondary metal and nonmetal components based primarily on ionic species. Divalent metals that readily form insoluble oxalates can be co-hosted in mixed oxalate biomineral phases, while mono-and trivalent cations are solubilized, presumably as free cations and/or soluble oxalate complexes (Gadd, 1999;Gadd et al., 2014;Verma et al., 2019). In anaerobic conditions, the mobility of metals hosted in Mn oxides has been shown to be influenced by the action of Mn(IV)-and Fe(III)-reducing prokaryotes (Newsome Fig 6. Synthetic Co-birnessite transformation following incubation with A. niger. The SEM images show the extent of synthetic Co-birnessite transformation by A. niger following 2 weeks incubation at 25 C in the dark. (A) synthetic Co-birnessite debris recovered from an uninoculated control, (B, C, D) mineral debris recovered from inoculated plates. Images were obtained using a JEOL SM-7400F field emission scanning electron microscope operating at an accelerating voltage of 5 keV. Scale bars represent (A) 1 μm, (B) 100 μm, (C, D) 10 μm. Typical images are shown from several examinations. et al., 2020). The data presented here clearly demonstrate that aerobic oxalate producing fungi, which are ubiquitous in terrestrial environments, can act in an opposing manner to extract and immobilize metals as insoluble oxalates. Moreover, the presence of fungal genera capable of oxalate excretion, e.g. Penicillium spp., were confirmed in a range of lateritic deposits shown to also host Mn(IV)-and Fe(III)-reducing prokaryotes and containing

Discussion
Co-bearing Mn oxides (Newsome et al., 2020). It is interesting to speculate that the speciation of Co in terrestrial Mn oxide deposits could be influenced by both Mn(IV) and Fe(III) reducing prokaryotes and organic acid producing fungi depending on environmental factors such as oxygen and nutrient availability.

Experimental procedures
Organism, media and growth conditions Experiments were conducted using A. niger ATCC 1015 which was maintained on 90 mm diameter malt extract agar (MEA) plates (LabM, Bury, UK) at 25 C in the dark. Media for powder transformation experiments was prepared by adding sterile mineral powder to molten MEA at $60 C to a concentration of 0.2% (w/v) and mixing before adding $25 cm 3 of the resulting media to 90 mm diameter Petri dishes. For Nkamouna fragment transformation, sterile fragments were set into wells in MEA plates which were prepared by melting the agar using forceps that were heated briefly over a Bunsen burner. All mineral-amended MEA plates were inoculated with one 5 mm diameter plug of A. niger taken from the edge of a colony actively growing on MEA using a sterile cork borer, and incubated for 14 days. Growth measurements on mineral powder amended MEA plates were taken by measuring two bisecting colony diameters at appropriate time intervals.

Minerals
Nkamouna laterite was originally obtained directly from Nkamouna, Cameroon and kindly provided by Dr The washing process was repeated three times before drying the precipitate in a desiccator for at least 7 days to constant weight. Once dry, the Co-birnessite precipitate was added to 200 ml 0.1 HNO 3 and shaken at 125 rpm at 25 C for 2 days in order to exchange interlayer K + for H + , as described by Peng et al. (2017). The precipitate was then recovered and dried as described above. Minerals were sieved to a particle diameter of <90 μm using soil sieves (Endecotts, London, UK) and sterilized in an oven at 105 C for at least 24 h prior to experimentation.

pH measurements
Following 14 days incubation on solid media, fungal biomass was carefully removed using a scalpel and the pH of the agar underneath the mycelium measured using a flat-tipped pH electrode (VWR International Ltd, Lutterworth, UK), which enables good contact with the agar surface, and an Orion 920+ pH meter (ThermoFisher Scientific, Loughborough, UK).

Mineral recovery
Following pH measurement, minerals were recovered from the agar by gently homogenizing in Milli-Q water heated to approximately 70 C in a glass crystallizing dish. The mineral debris were allowed to settle and the fluid removed, this process being repeated twice. Finally, mineral debris was dried in a desiccator for at least 48 h.

X-ray fluorescence (XRF)
Partial elemental composition of the original mineral samples was determined using a Philips 'Zetium' PW5400 sequential X-ray fluorescence spectrometer filtered with a RhKα source (Malvern Panalytical, Malvern, UK). Samples were compacted under a load of 75 kN for 5 min, followed by 150 kN for 10 min before analysis. The results are expressed as oxides unless reported otherwise.
Scanning electron microscopy (SEM) and energy dispersive X-ray analysis (EDXA) Transformed mineral samples were mounted on 25 mm Â 5 mm aluminium electron microscopy stubs using double-sided adhesive carbon tape before coating with gold and palladium to a thickness of 25 nm using a Cressington 208HR sputter coater (Cressington Scientific Instruments, Watford, UK). SEM and EDXA examination was conducted using a field emission scanning electron microscope (JEOL SM-7400F), operating at an accelerating voltage of 5 kV for imaging and 20 kV for EDXA. Any adjustments to contrast and/or brightness made after the image was taken were carried out using Adobe Photoshop CS5.1 and were applied equally to the whole image.

X-ray diffraction (XRD)/X-ray powder diffraction (XRPD)
Transformed Nkamouna samples were analysed using XRPD and transformed Co-birnessite samples were analysed using XRD. XRPD analysis was conducted using an Enraf-Nonius FR590 powder diffraction system (Enraf-Nonius, Rotterdam, Netherlands). Samples were prepared by grinding using a mortar and pestle before applying to a quartz substrate via an acetone smear. Diffraction patterns were then recorded from 0 to 80 2-theta using an INEL 120 curved position sensitive detector, with Co K-alpha radiation. The samples were continuously spun in the plane of the sample surface for a 2 h data collection period. XRD analysis was conducted using a Siemens D5000 powder X-ray diffractometer (Siemens Healthineers, Henkestraße 127, 91052 Erlangen, Germany). Samples were ground to a fine powder using a mortar and pestle before applying to PVC slides. Diffraction patterns were recorded using angular increments of 0.1 2-theta, from 3 to 60 2-theta, at a rate of 1 2-theta/min. A Cu-Kα source was used, operating at 40 mA and 40 kV, with a scintillation detector.