Abstract
The molybdenum cofactor (Moco) is a 520 dalton prosthetic group that is synthesized in a multi-step enzymatic pathway present in most Archaea, Bacteria, and Eukarya. In animals, four oxidases (among them sulfite oxidase) use Moco as a prosthetic group. Moco is essential in animals; humans with mutations in genes that encode Moco-biosynthetic enzymes display lethal neurological and developmental defects. Moco supplementation seems a logical therapy, however the instability of Moco has precluded biochemical and cell biological studies of Moco transport and bioavailability. The nematode Caenorhabditis elegans can take up Moco from its bacterial diet and transport it to cells and tissues that express Moco-requiring enzymes, suggesting a system for Moco uptake and distribution. Here we show that protein-bound Moco is the stable, bioavailable species of Moco taken up by C. elegans from its diet and is an effective dietary supplement in a C. elegans model of Moco deficiency. Diverse purified Moco:protein complexes from bacteria, bread mold, green algae, and dairy cows were able to support the growth of otherwise Moco-deficient C. elegans mutants grown on Moco-deficient E. coli. We show that these Moco:protein complexes are very stable, suggesting they may provide a strategy for the production and delivery of therapeutically active Moco to treat human Moco deficiency.
Introduction
The molybdenum cofactor (Moco) is an ancient coenzyme that was present in the last universal common ancestor and that continues to be synthesized in all domains of life (1, 2). Moco is a pterin-based organic prosthetic group that is comprised of a C6-substituted pyrano ring, a terminal phosphate, and a dithiolate group binding to molybdenum (Fig. 1A) (3). In humans and other animals, Moco is required for the activity of 4 enzymes: sulfite oxidase, xanthine oxidase, aldehyde oxidase, and mitochondrial amidoxime reducing component (4). There are 2 forms of eukaryotic Moco, the sulfite oxidase form and the xanthine oxidase form (Fig. 1A). These Moco species differ in the third Mo-S ligand which is provided either by an enzyme-derived cysteine residue (sulfite oxidase form) or an inorganic sulfur (xanthine oxidase form) (4). The xanthine oxidase form of Moco is synthesized from the sulfite oxidase form via the enzyme Moco sulfurase (Fig. 1A) (5).
Both forms of Moco are synthesized by a highly conserved biosynthetic pathway (Fig. 1A) (6). The genes necessary for Moco biosynthesis were first elucidated by genetic studies of chlorate resistance in bacteria (7). The importance of Moco biosynthesis to human health is highlighted by Moco deficiency (MoCD), a rare inborn error of metabolism. MoCD is caused by loss-of-function mutations in genes encoding any of the human Moco-biosynthetic enzymes and results in severe neurological dysfunction and neonatal lethality (8, 9). MoCD patients with mutations in MOCS1 (orthologous to bacterial moaA and moaC) can be treated with cyclic pyranopterin monophosphate (cPMP), a stable intermediate in Moco biosynthesis immediately downstream of MOCS1 (10). However, cPMP treatment is not effective for patients with mutations in any of the downstream Moco-biosynthetic enzymes. Purification and delivery of mature Moco would be an ideal therapeutic strategy for treating all forms of MoCD, however free Moco is too unstable and oxygen-sensitive to be purified and therapeutically administered. Furthermore, it is unclear whether mature Moco can cross cellular membranes.
Genetic evidence demonstrates that the nematode C. elegans retrieves Moco as well as cPMP from its bacterial diet, E. coli (11). However, nothing was known about the biochemical mechanism of Moco transfer between these 2 highly divergent organisms. (11). Here we propose that Moco bound to protein is the stable and bioavailable Moco species that is being harvested by C. elegans. We demonstrate that supplementation of purified protein-bound Moco rescues the lethality of Moco-deficient C. elegans feeding on Moco-deficient E. coli. We show that Moco bound to diverse Moco-containing proteins originating from bacteria, algae, fungi, and mammals, is bioavailable to C. elegans, and that this supplementation does not require Moco biosynthetic enzymes in C. elegans or its bacterial diet. This work suggests future mammalian therapeutic studies of supplemental protein-bound Moco and highlights the existence of a pathway for Moco transport and harvest.
Results and Discussion
C. elegans acquires Moco from dietary E. coli
Due to its instability, Moco has long been thought to be synthesized and utilized cell autonomously with no evidence for transport between cells, tissues, and organisms. So far only C. elegans has been described to have 2 pathways by which it can obtain Moco: endogenous Moco biosynthesis from GTP or dietary uptake of Moco, which demands other genetic pathways of intestinal uptake from the bacterial diet and transport of Moco to client tissues (11). Moco-biosynthetic enzymes are conserved in all domains of life; in C. elegans these enzymes are encoded by moc-1, moc-2, moc-3, moc-4, and moc-5 that mediate sequential steps in Moco biosynthesis (Fig. 1A). Using mutations in the C. elegans moc genes (i.e. the moc-1(ok366) null mutation) Moco biosynthesis can be interrupted in all cells. In the laboratory, C. elegans feed on a monoculture of E. coli. Thus, we can also use mutations in any of the genes of the E. coli Moco-biosynthetic pathway to eliminate dietary Moco (i.e. theΔmoaA null mutation). Either endogenous Moco synthesis in C. elegans or Moco produced by the diet E. coli and then consumed by C. elegans can support growth, development, and reproduction of C. elegans. However, when C. elegans cannot synthesize their own Moco and cannot obtain Moco from their diet, they arrest in larval development and ultimately die due to inactivity of sulfite oxidase, the key Moco-utilizing enzyme in animals (11).
To test how much wild-type, Moco-producing bacteria was required to support growth and development of C. elegans defective in Moco biosynthesis, we mixed wild-type and ΔmoaA mutant E. coli at various ratios and tested for the ability of these mixtures to support the viability of moc-1 mutant C. elegans. We speculated that if Moco functions like a vitamin, only trace amounts of wild-type, Moco-producing bacteria would be needed to support moc-1 mutant viability. However, we found that about 30% of the bacterial diet needed to be wild type (Moco producing) to support growth and development of moc-1 mutant animals (Fig. 1B).
Diverse protein-bound Moco is taken up and utilized by C. elegans
We hypothesized that C. elegans harvest bacterial Moco that is stably bound within the E. coli Moco-utilizing enzymes. E. coli YiiM (EcYiiM) is one such Moco-utilizing enzyme and mediates the reduction of N-hydroxylated substrates (12, 13). To test whether Moco bound to EcYiiM can be absorbed by C. elegans, we purified recombinant EcYiiM protein from E. coli and used it to supplement the diet of Moco-biosynthetic mutant C. elegans feeding on Moco-deficient E. coli, growth conditions that would otherwise result in 100% larval arrest and death. Consistent with the model that C. elegans utilizes Moco from E. coli Moco-utilizing enzymes, moc-1 mutant animals grown on Moco-deficient E. coli grew and developed well when their diet was supplemented with EcYiiM-bound Moco. (Fig. 2A,B). Thus, EcYiiM-bound Moco is bioavailable and can support the viability of otherwise Moco-deficient C. elegans.
To test if the ability of C. elegans to harvest Moco from protein was more general to other Moco-binding proteins, we recombinantly expressed and purified two additional Moco-binding proteins in E. coli; nitrate reductase from the red bread mold Neurospora crassa (NcNR) and Moco-carrier protein from the green algae Volvox carteri (VcMCP) (14–16). We also utilized the commercially available Moco-using enzyme xanthine oxidase (BtXO) purified from bovine milk (17). Each Moco-binding protein was supplemented to moc-1 mutant C. elegans fed ΔmoaA mutant E. coli. Similar to EcYiiM, supplementation with Moco bound to either VcMCP or NcNR supported the growth of moc-1 mutant C. elegans in the absence of any other dietary Moco (Fig. 2). To a lesser extent, BtXO supplementation also supported the growth of moc-1 mutant animals cultured on Moco-deficient E. coli (Fig. 2A,B). A possible explanation for the reduced efficacy of supplemental BtXO compared to EcYiiM, VcMCP, or NcNR might be the form of Moco that is bound by these proteins. BtXO binds the xanthine oxidase form of Moco while EcYiiM, VcMCP, and NcNR bind the sulfite oxidase form of Moco (Fig. 1A) (12, 15, 16, 18). In C. elegans and other animals sulfite oxidase (SUOX-1) is the key Moco-requiring enzyme necessary for viability; suox-1 null mutant animals arrest development similar to Moco-deficient animals (11). It is reasonable that protein supplementation with the sulfite oxidase form of Moco can supply the appropriate Moco to support C. elegans SUOX-1 activity compared to a xanthine oxidase form of Moco. Alternatively, supplementation with the sulfite oxidase form of Moco may result in the partial conversion, via Moco sulfurase (encoded by C. elegans mocs-1), of that supplemental Moco into the xanthine oxidase form (Fig. 1A). Thus, by providing the sulfite oxidase form of Moco we may be providing both forms of eukaryotic Moco making it a more effective treatment for complete Moco deficiency in C. elegans. Supplementation with the xanthine oxidase form of Moco would likely not result in synthesis of the sulfite oxidase form of Moco as there is no known enzyme that desulfurates the xanthine oxidase form of Moco.
To further demonstrate that the growth of C. elegans moc-1 mutant animals was conferred by supplementation of the Moco prosthetic group and not by the supplemental purified proteins, we purified apo-VcMCP from bacteria unable to synthesize Moco. Supplemental apo-VcMCP, did not support the growth of moc-1 mutant C. elegans fed ΔmoaA mutant E. coli (Fig. 2A,C,D). Taken together these data demonstrate that the animal C. elegans is able to acquire and harvest the Moco prosthetic group when it is provided as a dietary supplement in complex with Moco-binding proteins. These proteins have diverse structures and functions and originate from both prokaryotes and eukaryotes. As such, the acquisition of protein-bound Moco by C. elegans is not specific to certain Moco-binding proteins and may reflect a general strategy for acquisition of functional Moco from the animals’ diet or microbiome. Furthermore, as Moco biosynthesis and utilization are ancient processes conserved in all domains of life, we believe it is unlikely that a novel biochemical pathway for Moco transfer across cell membranes has evolved exclusively in the nematode C. elegans. However, it remains to be tested whether protein-bound Moco can permeate the cells of other organisms. The remaining experiments were all performed with supplemental Moco bound to VcMCP due to its well-characterized role in Moco binding and our established protocols for its production (14, 16).
One model for the rescue of C. elegans Moco deficiency is that supplemental protein-bound Moco is directly ingested by C. elegans. Alternatively, the protein-bound Moco may first be taken up by E. coli which may process the Moco to then be ingested by C. elegans. To distinguish between these models, we grew ΔmoaA mutant E. coli in lysogeny broth (LB) supplemented with Moco bound to VcMCP. This ΔmoaA E. coli was then separated from the culture medium by centrifugation, washed extensively, and fed to moc-1 mutant C. elegans (“Diet B”, Fig. 3). Although cultured with Moco bound to VcMCP, the washed ΔmoaA E. coli in Diet B did not support growth of moc-1 mutant animals. Importantly, the supernatant medium from the same culture supported the growth of moc-1 mutant C. elegans grown on a lawn of Δ moaA E. coli grown separately in LB alone (“Diet A”, Fig. 3B). Together, these data suggest that supplemental protein-bound Moco does not pass through a bacterial intermediate before being acquired by C. elegans (Fig. 3B).
Moco bound to protein is stable
The instability and oxygen sensitivity of Moco has limited cell biological studies of Moco transport and precluded it from therapeutic consideration (19). The VcMCP-bound Moco used in “Diet A” (Fig. 3) was incubated at 37°C overnight, and still retained its activity and bioavailability, suggesting remarkable stability. To biochemically demonstrate the stability of Moco bound to protein, we measured the ability of mature Moco to stay in complex with VcMCP, EcYiiM, NcNR, and BtXO over time (Fig. 4). Free Moco is highly unstable, however it can be oxidized to ‘Form A’, a stable and fluorescent Moco-derivative that is quantifiable via HPLC (20, 21). Using measurements of Form A and protein concentration, we first determined the initial Moco occupancy of purified VcMCP (22%) as well as EcYiiM (4%), NcNR (50%), and BtXO (50%) (Fig. 4A,B). We then assessed the stability of each purified Moco:protein complex by determining Moco retention over time at different temperatures (Fig. 4C-F). All 4 Moco:protein complexes were remarkably stable, showing no significant protein degradation and retaining between 43 and 83% of their original Moco content after 96 hours of incubation at ambient temperature (Fig. 4C-F). This stability is surprising and suggests purification of protein-bound Moco as a new strategy for the production and delivery of therapeutically active Moco to treat MoCD.
Bioavailability of recombinant protein-bound Moco does not depend on known Moco-biosynthetic enzymes in E. coli or C. elegans
We tested if the Moco-biosynthetic enzymes are necessary for the harvesting or transport of supplemental protein-bound Moco using mutants in the dietary E. coli. We tested moc-1 mutant C. elegans growth on wild-type bacteria, or mutant bacteria lacking the genes necessary for Moco biosynthesis (moaA, moaC, moaD, moaE, moeB, mog, moeA, modA, modC, or ydaV). moc-1 mutant animals were grown on mutant E. coli with and without supplemental Moco bound to VcMCP. moc-1 mutant C. elegans grew well on wild-type E. coli but displayed larval arrest on all 10 E. coli mutants defective in Moco biosynthesis (Fig. S1). Supplemental Moco bound to VcMCP supported growth and development of moc-1 mutant C. elegans on all 10 of the Moco-biosynthetic mutant E. coli demonstrating that none of these E. coli genes were necessary for bioavailability of supplemental protein-bound Moco (Fig. S1).
Alternatively, we speculated that the Moco-biosynthetic machinery of C. elegans might play a role in the bioavailability of supplemental protein-bound Moco. To test this, we used established C. elegans mutants in the Moco-biosynthetic pathway (moc-5, moc-4, moc-3, moc-2, and moc-1, Fig. 1A). Each of these C. elegans mutants was cultured on wild-type E. coli, Δ moaA E. coli, or ΔmoaA E. coli supplemented with Moco bound to VcMCP. All of the moc mutant animals grew well on wild-type bacteria that produce Moco and arrested growth on Δ moaA E. coli that lacks Moco biosynthesis (Fig. S2A-E). Each C. elegans moc-mutant displayed dramatically improved growth on ΔmoaA E. coli when their diet was supplemented with Moco bound to VcMCP (Fig. S2A-E). These results demonstrate that moc-5, moc-4, moc-3, moc-2, and moc-1 are not required for the bioavailability of supplemental protein-bound Moco. Thus, the machinery that facilitates Moco transport is distinct from the canonical Moco biosynthetic pathway.
Supplemental protein-bound Moco supports the activity of C. elegans SUOX-1
The lethality associated with Moco deficiency in C. elegans and humans is due to inactivity of sulfite oxidase (SUOX-1), a mitochondrial Moco-requiring enzyme that oxidizes the lethal toxin sulfite to sulfate. Like Moco-biosynthesis, sulfite oxidase is essential in both C. elegans and humans (11, 22). Thus, to rescue development of Moco-deficient C. elegans, supplemental protein-bound Moco must be incorporated into and support the activity of C. elegans SUOX-1. To demonstrate this, we utilized the hypomorphic suox-1 allele gk738847 (D391N) (23). Aspartic acid 391 of sulfite oxidase is highly conserved and is present in C. elegans, Drosophila melanogaster, Danio rerio, Mus musculus, and Homo sapiens. The SUOX-1 D391N amino acid substitution causes partial SUOX-1 loss-of-function that is enhanced when dietary Moco is absent. Growing suox-1(gk738847) mutant C. elegans on Moco-deficient E. coli causes a severe developmental delay compared to its growth on wild-type Moco-producing E coli (Fig. S2F) (11). Importantly, suox-1(gk738847) mutant animals are wild type for their endogenous Moco biosynthetic pathway and are able to synthesize Moco de novo. This result shows that C. elegans depends on both endogenous Moco biosynthesis as well as dietary sources of Moco to fully support the activity of SUOX-1.
We hypothesized that supplemental Moco bound by VcMCP would improve the viability of suox-1(gk738847) animals grown on ΔmoaA E. coli. To test this, we cultured suox-1(gk738847) mutant animals on wild-type E. coli, ΔmoaA E. coli, and ΔmoaA E. coli supplemented with Moco bound to VcMCP. Consistent with our rescue of C. elegans Moco deficiency, supplemental protein-bound Moco improved the growth of suox-1(gk738847) animals grown on Moco-deficient E. coli (Fig. S2F). These results suggest that exogenous protein-bound Moco is absorbed, harvested, distributed to requisite cells and tissues, and re-inserted into the C. elegans SUOX-1 enzyme.
Our data demonstrate the ability of an essential protein-packaged prosthetic group to cross cell membranes. This transfer naturally occurs between multiple organisms (i.e. from E. coli to C. elegans) and among the cells and tissues of a single organism; Moco absorbed in the intestine of C. elegans must cross multiple cell membranes to reach all of the Moco-utilizing cells and tissues. Because Moco biosynthesis is as ancient as the last universal common ancestor, it is likely that the not yet discovered Moco-transport pathway may be general to all animals, including humans. Furthermore, roughly 70% of bacterial genomes encode Moco-biosynthetic enzymes making the intestinal microbiome a potential reservoir for this cofactor (2). Similarly, the human diet might also be a source of exogenous protein-bound Moco as most plants and animals synthesize and utilize Moco. Our results with the nematode C. elegans may stimulate future exploration of the therapeutic potential of protein-bound Moco from dietary, microbiome, or recombinant sources.
Materials and Methods
General methods and strains
C. elegans strains were cultured at 20°C on nematode growth medium (NGM) seeded with wild-type Escherichia coli unless otherwise noted (24). The wild-type strain of C. elegans was Bristol N2. For each linkage group, the C. elegans mutant strains used in this work and their associated genotypes are listed. LGI: GR2253 moc-4(ok2571). LGIV: MH3266 moc-3(ku300). LGV: GR2255 moc-2(mg595). LGX: GR2254 moc-1(ok366), GR2256 moc-5(mg589), and GR2269 suox-1(gk738847).
E. coli strains were cultured using standard methods. The wild-type strain of E. coli was BW25113, the parental strain of the Keio E. coli knockout collection (25). The E. coli mutants used in this work were JW0764 (ΔmoaA::Kanr), KJW1 (Δmog), KJW2 (ΔmoaA), KJW3 (Δ moaC), KJW4(ΔmoaD), KJW5(ΔmoaE), KJW6(ΔmoeB), KJW7(ΔmoeA), KJW8(ΔmodA), KJW9(ΔmodC), and KJW10(ΔydaV). KJW1-KJW10 are bacterial strains derived from the Keio E. coli knockout collection and have been modified using established methods to remove the kanamycin resistance cassette from each locus of interest (11, 25). Strains KJW1-KJW10 were only used to produce the data in Figure S1. JW0764 was used in all other experiments with Δ moaA E. coli.
C. elegans growth assays
C. elegans were synchronized at the first stage of larval development (L1). L1 animals were then cultured on NGM seeded with wild-type or mutant E. coli. For some experiments, mutant E. coli was supplemented with various forms and amounts of protein-bound Moco (see Dietary supplementation with protein-bound Moco). C. elegans animals were then allowed to grow and develop for 48 or 72 hours (specified in Figure Legends) at 20°C. For each experiment, the sample size (n) is individual animals measured and is reported in the Figures and Figure Legends.
For all assays, live animals were imaged using an Axio Zoom.V16 microscope (Zeiss) equipped with an ORCA-Flash4.0 digital camera (Hamamatsu). Images were captured using ZEN software (Zeiss) and processed utilizing ImageJ (NIH). Animal length was measured from the tip of the head to the end of the tail. The median and upper and lower quartiles were calculated using GraphPad Prism software.
Purification and characterization of Moco-binding proteins
Moco-binding proteins were expressed and purified using standard methods (16). The full-length yiiM coding sequence was amplified from Escherichia coli DH5α (EcYiiM) and the full-length Moco carrier protein coding sequence from Volvox carteri (VcMCP) was synthesized and codon optimized for E. coli (16). The coding sequence for Neurospora crassa nitrate reductase (NcNR) was shortened to include only the Moco-binding and dimerization region (amino acids 113-592). The coding sequences for EcYiiM, VcMCP, and NcNR were inserted into the pONE-CP plasmid, producing proteins fused to a C-terminal Streptavidin tag. Streptavidin-tagged proteins were expressed using the E. coli strain TP1000 which accumulates the eukaryotic form of Moco due to a deletion in the Mob operon (26). As a negative control, VcMCP was also purified from the E. coli strain RK5204 which is unable to produce Moco due to a mutation in moaE (27). Bovine xanthine oxidase (BtXO) was purchased from Sigma-Aldrich (X1875, batch SLCB1289).
Protein concentrations were determined using absorption at 280 nm and the Pierce BCA Protein-Assay (Thermo Scientific). Absorption was measured using a Multiskan GO Microplate Spectrophotometer (Thermo Scientific). Quantification of Moco content of the proteins was conducted using HPLC-based measurements of Form A, a stable and fluorescent Moco-oxidation product (21). Stability of protein-bound Moco was assessed by incubating the Moco:protein complexes at various temperatures (4°C, 22°C, and 37°C) for 96 hours. In 24-hour intervals, protein samples were centrifuged at 4°C to remove precipitated protein and protein concentration and Moco content were determined as described above.
Dietary supplementation with protein-bound Moco
Similar to standard C. elegans growth conditions, experiments with supplemental protein-bound Moco were performed on petri dishes filled with approximately 10ml of solidified NGM agar (24). To maximize exposure to the supplemental protein-bound Moco, we resuspended 10X concentrated mutant E. coli with either 50μl of M9 minimal buffer (control) or 50 microliters of protein-bound Moco in M9. These bacterial resuspensions (with and without protein-bound Moco) were then seeded onto individual petri dishes with NGM and allowed to dry leaving a small concentrated lawn of E. coli roughly 1-2 centimeters in diameter. After drying, synchronized L1 C. elegans were seeded directly on the lawn of E. coli and we proceeded with C. elegans growth assays. To maintain consistency among the experiments with supplemental protein-bound Moco, dietary E. coli was grown, supplemented with protein-bound Moco, and seeded fresh the same day the C. elegans growth assay was to begin.
We utilize 4 different sources of protein-bound to Moco to supplement C. elegans diets: Escherichia coli YiiM (EcYiiM), Neurospora crassa nitrate reductase (NcNR), Volvox carteri Moco carrier protein (VcMCP), and bovine xanthine oxidase (BtXO). The amount of Moco in every protein preparation was determined experimentally (see Purification and characterization of Moco-bound proteins). The independent variable in our experiments with supplemental protein-bound Moco is the total amount of Moco that is being used to resuspend the dietary E. coli. Because we do not know the extent to which the protein-bound Moco diffuses throughout the NGM agar, we are limited in our ability to estimate the concentration of protein-bound Moco to which C. elegans are exposed. Thus, we report the total amount of Moco used to supplement the C. elegans diet and assume equal protein diffusion throughout our 48-72 hour growth experiments.
Culturing E. coli with Moco bound to VcMCP
ΔmoaA mutant E. coli (JW0764) was cultured in 500μl LB supplemented with 39 nanomoles of Moco bound to VcMCP. Cells were cultured overnight at 37°C rotating at 1,400 rotations per minute. This overnight culture was then concentrated and the supernatant was removed for use in “Diet A”. The concentrated cells were washed repeatedly with M9 minimal buffer, concentrated 10X, and seeded onto NGM agar plates. This diet is referred to as “Diet B” in Figure 3.
The supernatant from the overnight culture (see above) was then filtered through a 0.20μm filter (Corning) to remove any remaining bacterial cells. This filtered ‘spent media’ was then used to resuspend a separate concentrated culture of ΔmoaA mutant E. coli (grown only in LB). These resuspended bacterial cells were then seeded onto NGM agar plates. This diet is referred to as “Diet A” in Figure 3.
Acknowledgments
We thank the Caenorhabditis Genetics Center (CGC) for providing C. elegans strains and the National BioResource Project (NIG, Japan) for providing the Keio E. coli knockout collection. This work was funded by an NIH Grant (5R01GM044619-26) to G.R., a DFG grant (GRK2223/1) to R.R.M., and a Damon Runyon Fellowship (DRG-2293-17) to K.W.