Impaired mitochondrial maturation of sulfite oxidase in a patient with severe sulfite oxidase deficiency

Abstract

Sulfite oxidase (SO) is encoded by the nuclear SUOX gene and catalyzes the final step in cysteine catabolism thereby oxidizing sulfite to sulfate. Oxidation of sulfite is dependent on two cofactors within SO, a heme and the molybdenum cofactor (Moco), the latter forming the catalytic site of sulfite oxidation. SO localizes to the intermembrane space of mitochondria where both—pre-SO processing and cofactor insertion—are essential steps during SO maturation. Isolated SO deficiency (iSOD) is a rare inborn error of metabolism caused by mutations in the SUOX gene that lead to non-functional SO. ISOD is characterized by rapidly progressive neurodegeneration and death in early infancy. We diagnosed an iSOD patient with homozygous mutation of SUOX at c.1084G>A replacing Gly362 to serine. To understand the mechanism of disease, we expressed patient-derived G362S SO in Escherichia coli and surprisingly found full catalytic activity, while in patient fibroblasts no SO activity was detected, suggesting differences between bacterial and human expression. Moco reconstitution of apo-G362S SO was found to be approximately 90-fold reduced in comparison to apo-WT SO in vitro. In line, levels of SO-bound Moco in cells overexpressing G362S SO were significantly reduced compared to cells expressing WT SO providing evidence for compromised maturation of G362S SO in cellulo. Addition of molybdate to culture medium partially rescued impaired Moco binding of G362S SO and restored SO activity in patient fibroblasts. Thus, this study demonstrates the importance of the orchestrated maturation of SO and provides a first case of Moco-responsive iSOD.

Introduction

Isolated sulfite oxidase deficiency (iSOD) is an autosomal-recessive inborn error of metabolism caused by mutations in the sulfite oxidase (SO) gene (SUOX) (OMIM 606887). SO converts sulfite to sulfate representing the final step in cysteine catabolism. Loss of SO activity in iSOD leads to severe and progressive brain lesions caused by accumulation of toxic sulfite and S-sulfocysteine (SSC) (1,2). SSC originates as reaction product from sulfite and cystine, which is the oxidized and major transport form of cysteine in the vasculature. As glutamate analogue, SSC is able to bind and activate NMDA receptors, which leads to SSC-mediated exitotoxicity in neurons (3,4). Currently, iSOD is an incurable disease with usually lethal outcome in the neonatal age or early infancy (5–8). As cysteine catabolism is the major source of sulfite in humans, dietary restriction prolonged iSOD patient survival in some cases with a mild form of the disease (5,7).

The lethality of mutations within SUOX, which lead to loss-of-function of SO, indicates that SO is an essential enzyme in humans. SO is a molybdenum (Mo)-dependent enzyme in which the catalytically active Mo is coordinated via a protein-derived cysteine residue and an enedithiolate derived from a unique pterin with a C6-substituted pyrano ring, the so called Mo cofactor (Moco). A terminal phosphate moiety at C4’ of the pyrano ring is a universal feature in all eukaryotic Moco enzymes. SO forms a homodimer in which each subunit comprises three domains (9). The N-terminal domain binds a b₅-type heme (9,10). This heme domain is connected via a flexible 11 residue tether to a central domain, which harbors Moco. The C-terminal domain confers dimerization. Oxidation of sulfite to sulfate takes place at the Moco active site, where two sulfite-derived electrons reduce Mo (11) accompanied by sulfate release. Re-oxidation of SO is mediated by two intramolecular electron transfers (IETs) from Moco to heme within SO followed by the release of sulfite-derived electrons from heme to the terminal electron acceptor cytochrome c (12,13). Thus, SO needs to be confined in the intermembrane space of mitochondria (IMS) to allow efficient transfer of sulfite-derived electrons to oxidized cytochrome c (10,14).

In humans, the nuclear SUOX gene translates into a 545 residue pre-mature apo-protein, which is subjected to an orchestrated multi-step maturation process to become catalytically active within mitochondria. Firstly, cytosolic pre-SO is translocated via an N-terminal leader peptide to the translocase of the outer membrane of mitochondria (14). Secondly, anchoring of the N-terminal peptide into the inner mitochondrial membrane allows cleavage by the inner membrane peptidase (IMP), which releases the C-terminal fragment of SO into the IMS. Here, SO becomes catalytically active once heme and Moco are integrated followed by homodimerization (14). In vitro Moco incorporation into apo-SO was investigated via site-directed spin-labeling to detect flexible peptide regions, which might confer cofactor binding (15). Here, a flexible loop was identified in apo-SO, which presents the Moco-binding pocket to the surrounding solute. The location of the loop implied a Moco-C4’-phosphate first incorporation process of the cofactor into SO (15).

The multitude of maturation steps and their hierarchic order suggest a vulnerability of this pathway to errors that could potentially cause iSOD. However, with exception of two cases (8,16) in all other reported iSOD patients the disease-causing mutation either leads to a total loss of protein (2), the substitution of residues that function within the active site (17–19), that participate in IET from Moco to heme (20) or that mediate homodimerization of SO (6,8,17,21). In the two exceptional iSOD cases, the affected residues (R366H and K379R) are directly involved in Moco-C4’-phosphate coordination as predicted by the crystal structure of the homologous chicken SO (9). Surprisingly, biochemical analysis of recombinant R366H as well as K379R SO variants expressed in Escherichia coli reported increased catalytic activity for both variants (8,16,20,22,23), which was in conflict with their identification in iSOD patients. Therefore, the underlying disease-causing mechanism of R366H and K379R patients remained elusive.

Here, we report an iSOD patient with homozygous c.1084G>A mutation resulting in substitution of Gly362 to Ser, a mutation that was already reported to cause iSOD (8). Gly362 is in direct vicinity to the Moco-C4’-phosphate coordinating residues including His361, Arg366 and Lys379. Although, specific SO activity was absent in G362S patient fibroblasts, surprisingly the recombinantly expressed protein had full catalytic activity. Moco incorporation was found to be strongly compromised in G362S SO in vitro and in vivo, which, together with molybdate-dependent restoration of SO activity in patient fibroblasts, suggested a maturation defect of SO in the patient as underlying disease-causing mechanism.

Results

Case report

The male patient was born as the third child of consanguineous Turkish parents. Spontaneous delivery occurred at term after an uneventful pregnancy with Apgar scores of 9/10/10. On day 3 of life the newborn was admitted to the children’s hospital due to hypopnoe and multiple seizures. Standard laboratory testing at admission showed a mild metabolic acidosis 7.28 (7.35–7.45) with pCO₂ 51 mm Hg (35–45) and elevated creatine kinase levels (5100 U/l). A brain magnetic resonance imaging (MRI) scan on admission (Fig. 1A and B) showed globally severely impaired diffusion (Supplementary Material, Fig. S1A), white matter hyperintensity in T2 (Supplementary Material, Fig. 1B) and cerebella hypoplasia (Fig. 1A and B). Lumbar puncture showed a xanthochrom cerebrospinal fluid (CSF) with 248 mg/dl protein, 50 mg/dl glucose and 3.9 mM lactate and a treatment for meningitis was initiated. A bedside urine sulfite test as part of a metabolic workup for seizures was negative at admission. The electroencephalography (EEG) showed a reduced activity and multifocal epileptic discharges as well as focal epileptic activity. Within hours, the patient became comatose and required invasive ventilation for 4 days.

Figure 1

MRI scans of homozygous SUOX c.1084G>A patient (A–F) Frontal MRI images with the T1W-IR-PREM sequences at day 3 of life (A), 3 months of age (C) and 5 months of age (E). Sagittal T2-TSE-PREM sequences at day 3 of life (B), 3 months of age (D) and 5 months of age (F). Visible is the cerebellar hyplasia at birth (A) and (B), in further progress after disease manifestation severe brain atrophy with enlarged textracerebral spaces and hygroms as well as massive cystic lesions (C–F). Enlarged ventricular spaces and megacisterna magna also in (D) and (F). The brain stem is well preserved (C–F).

After four days, the child could be extubated but remained in a poor neurological state showing hyperexitability, dyskinetic movements, spasticity, epileptic seizures and frequent vomiting. The brain MRI scan showed a diffuse supratentorial leukoencephalopathy with impaired diffusion and progressive brain edema (data not shown). Since blood/CSF cultures remained sterile the treatment for meningitis was discontinued. During hospitalization, the child remained severely disabled in a minimal responsive state with spontaneous breathing, progressive spasticity and required gastrointestinal tube feeding. Symptomatic treatment of frequent multifocal seizures was unsuccessful with vitamin B6, phenobarbitone, levetiracetame and sultiame. However, topiramate treatment showed a marked reduction of seizures. The EEG revealed continuous diffuse slowing of activity and multifocal epileptic discharges. The child developed a progressive microcephaly with large cystic defects of the brain. At three months of age another MRI scan revealed severe brain atrophy with multicystic leukoencephalopathy, epidural and subdural hygromas and large infratentorial CSF spaces (Fig. 1C and1D), those findings were diagnostic for iSOD or Moco deficiency (MoCD). This progressive brain atrophy, microcephaly and the hygromas became more evident at five months of age; however, the brainstem was relatively spared and showed signs of myelination (Fig. 1E and1F). The child is today alive, at the age of 4.5 years, but severely mentally retarded. This severe developmental retardation could be attributed to the severe brain lesions, and the child so far developed no speech or vocalization nor can he sit independently. He has epilepsy with seizures resembling myoclonic fits and dyskinesia.

Diagnosis of iSOD

At 3.5 months of age a bedside urine sulfite test was positive. Diagnosis of iSOD was possible following the determination of urine SSC concentrations of 237 μmol/mmol creatinine (Table 1). For comparison, in healthy individuals urinary SSC levels range from 0.3 to 18.8 μmol/mmol creatinine (24). Urinary biomarkers of xanthine oxidase, another Moco-dependent enzyme, remained within normal range. Consistently, low concentration of urothione, the catabolic excretion product of Moco, were detected (Table 1). These results collectively exclude MoCD confirming iSOD as underlying cause of disease. Next, sequencing of the SUOX gene coding exons and flanking introns identified a known (8) homozygous mutation NM_000456.2 c.1084G>A resulting in the substitution of Gly362 to serine. Sequence alignment of SO enzymes from different species (Fig. 2A) highlight that the affected glycine residue is highly conserved, similar to the neighboring His361, which takes part in electrostatic Moco-C4’-phosphate coordination (Fig. 2B).

Table 1

Urine concentrations of SSC, xanthine, uric acid and urothione in the c.1084G>A patient in comparison to healthy individuals (44).

	Healthy control	c.1084G>A patient
SSC (μmol/mmol creatinine)	<20	237
Xanthine (μmol/mmol creatinine)	<30	25.7
Uric acid (μmol/mmol creatinine)	220–790	323.4
Urothione (nmol/mmol creatinine)	26–189	6.2

Figure 2

Alignment of SOs’ from different species and structural proximity of affected residue to Moco-C4’-phosphate coordinating residues (A) Sequence alignment of SO enzymes of different origin. The substituted glycine is highlighted in red. The highly conserved amino acid triade coordinating the Moco-C4’-phosphate in the folded protein are highlighted in green. The conserved residues are black and non-conserved residues are gray. (B) Stick representation of Moco and residues coordinating the Moco-phosphate moiety in the chicken X-ray crystal structure labeled with human accessions. The G362 residue is highlighted in red.

SO abundance is reduced in patient fibroblasts

In order to elucidate the molecular mechanism of iSOD caused by the G362S substitution in SO, patient fibroblasts were investigated in respect to SUOX mRNA expression, SO protein abundance and subcellular localization. Total mRNA of human wild type (WT) control and patient fibroblasts was isolated and reverse transcribed, and SO cDNA was quantified via qPCR using 60S ribosomal cDNA (RPL37a) as housekeeping gene expression control. The abundance of G362S cDNA was significantly reduced in patient fibroblasts in contrast to WT control fibroblasts (Fig. 3A). Consistently, western blot analysis confirmed the presence of an SO-specific signal in WT cells, while the signal intensity was strongly reduced in patient cells (Fig. 3B). Subcellular localization confirmed again the low abundance of G362S protein, but revealed that the remaining protein was still observed in mitochondria (Fig. 3C) suggesting that the translocation of G362S SO was not impaired. Despite the presence of SO protein, sulfite-dependent cytochrome c reduction of crude protein extracts revealed the absence of any detectable SO activity in G362S fibroblasts (Fig. 3D). Therefore, we conclude that the remaining level of G362S protein lacks sulfite-oxidizing activity thus causing iSOD in the patient.

Figure 3

Characterization of c.1084G>A fibroblasts (A) Quantitative PCR of WT and G362S fibroblasts. Total RNA was isolated from WT and patient fibroblasts, reverse transcribed and SO-specific qPCR was conducted as described in methodology. (B) Immunostaining of 100 μg fibroblast crude extract protein loaded on an 8% SDS-gel. SO was detected via polyclonal SO antibody and β-tubulin as loading control with monoclonal tubulin antibody. (C) Subcellular localization of WT and G362S SO variants. Mitochondria were enriched as described in methodology. Immunostaining of either gephyrin or VDAC was used as cytosolic and mitochondrial marker, respectively. (D) Specific sulfite-dependent SO activity was measured with 100 μg crude protein and porcine cytochrome c as terminal electron acceptor using 500 μM sulfite.

Recombinant G362S variant shows full catalytic activity

For further biochemical characterization, the coding sequence of G362S SO was recombinantly expressed in E. coli under standard expression conditions (72 h expression, 0.5 mM sodium molybdate in growth medium) reported previously (25). The G362S variant was purified to homogeneity (Fig. 4A) and revealed a Moco saturation of 35.3 ± 1.7%, which was comparable to WT SO (Fig. 4B). UV-vis spectra of oxidized and sulfite-reduced G362S SO (Fig. 4D) were similar to WT SO (Fig. 4C). Surprisingly, sulfite-dependent kinetics of the G362S variant using cytochrome c as electron acceptor displayed a k_cat of 18.7 ± 0.7 s^-1, which was three times higher than that of the WT control (k_cat = 5.9 ± 0.1 s^-1; Fig. 4E). Similarly, the k_cat of G362S SO in the sulfite-dependent ferricyanide (FeCN) reduction was 232.0 ± 20.6 s^-1, which was 4.5-fold increased in comparison to the WT with a k_cat of 51.6 ± 3.7 s^-1 (Fig. 4F). It was entirely surprising that G362S substitution caused an overall increase in catalytic efficacy of the enzyme. These steady-state kinetic parameters of the G362S variant resemble those of two other pathological SO variants (R366H and K379R) that both affect Moco-phosphate coordination (16) suggesting a similar underlying disease mechanism. In summary, we conclude that G362S substitution does not impair the catalytic activity of SO and therefore other mechanisms that lead to loss of SO activity in patient cells had to be considered.

Figure 4

Comparison of recombinant WT and G362S (A) Coomassie brilliant blue staining of 12% SDS-gel of His₆-tagged human WT and G362S SO variants being recombinantly expressed in E. coli TP1004 and purified via Ni-NTA chromatography. A total of 10 μg protein was loaded. (B) Moco saturation of E-coli-derived proteins measured via Form A analysis described in methodology. (C) Visible spectra of oxidized (solid line) and 5 mM sulfite reduced (dashed line) WT SO. A total of 6 μM enzyme was used, diluted in 200 μl 100 mM Tris/Ac pH 8.0. (D) Visible spectra of G362S variant. (E) Sulfite-dependent cytochrome c reduction: Comparison of WT SO (black dots) and G362S SO (red dots) took place at 25°C and in 200 μl reaction volume in 100 mM Tris/Ac pH 8.0. Reduction of cytochrome c was recorded at varying sulfite concentrations. Results are expressed as mean ± standard deviation (n = 3). (F) Sulfite-dependent FeCN reduction: comparison of WT SO (black dots) and G362S SO (red dots) using 400 μM FeCN as electron acceptor at varying sulfite concentrations. Results are expressed as mean ± standard deviation (n = 3).

Structural alterations in apo-G362S variant

In contrast to our results obtained from recombinant SO, in patient fibroblasts, SO activity was absent, while SO protein was detectable at a reduced level. The reduction of G362S protein abundance might suggest an impaired Moco-dependent maturation, which was previously reported under conditions, when Moco is absent or Moco insertion into apo-SO has been compromised (14). Therefore, we hypothesized that the G362S SO variant might be unable to efficiently bind Moco in fibroblasts, which would then cause the observed loss of SO activity.

In order to address the hypothesis of a maturation defect, Moco-free apo-WT and apo-G362S SO variants were recombinantly expressed in the Moco-deficient E. coli strain KB5242K1 (ΔmoaA) (26). Following purification of both proteins, a total yield of 35.3 mg/L culture medium soluble apo-WT was obtained. In contrast, the yield of soluble apo-G362S variant (0.36 mg/L culture medium) was two orders of magnitude lower. SDS-PAGE analysis revealed expression of both proteins in equal amounts in the crude extract (Fig. 5A, disrupted cells). However, upon centrifugation (50 000g) the majority of apo-G362S protein was sedimented (Fig. 5A, supernatant). Determination of Moco content revealed no detectable signal for both variants expressed in the ΔmoaA strain, thus confirming the apo-state of both SO proteins (Fig. 5B).

Figure 5

Characterization of apo-SO variants and Moco reconstitution (A) Coomassie brilliant blue staining of 12% SDS-gels of His₆-tagged human apo-WT SO and apo-G362S variant purification. The proteins were recombinantly expressed in KB5242K1_ΔmoaA and purified via Ni-NTA chromatography. (B) Moco saturation measured via Form A analysis described in methodology. (C) and (D) Analytical size exclusion chromatography of 250 pmol apo- and holo-SO variants. A total of 50 μl protein was injected onto a 22 ml Superdex 200 10/300 GL. The running buffer was composed of 50 mM Tris/Ac pH 8.0 and 300 mM NaCl. The size separation was conducted with a flow rate of 0.2 ml/min at 4°C and elution was recorded at A₂₈₀. (E) Recording of A₄₁₃ corresponding to (D). (F) Immunostaining of recombinantly expressed SO variants as holo and apo-proteins. A total of 50 ng total protein was loaded on 8% SDS-gel. SO detection was achieved via immunostaining with SO antibody. (G) Changes in mean residue ellipticity of holo-WT and holo-G362S SO as function of temperature. 0.2 mg/ml protein was heated to 80°C while recording CD data at 222 nm continuously. (H) Equivalent to G but with apo-SO variants. (I) Sulfite-dependent cytochrome c reduction velocities of different Moco amounts of the WT SO (black dots) and G362S SO (red dots) based on Form A analysis. (J) Activity of apo-WT and apo-G362S variants upon Moco reconstitution in the sulfite-dependent cytochrome c reduction. The reaction from cPMP to Moco was composed of 20 mM HEPES pH 7.4, 3 nmol MoaD, 0.1 nmol MoaE, 0.1 nmol gephyrin, 10 mM MgCl₂, 5 μM sodium molybdate 10 mM ATP and 250 pmol apo-SO variants in 140 μl. The reaction was initialized by addition of 1 nmol cPMP. The reaction was incubated at 25°C, while time-dependent samples were collected. Samples were mixed with 50 μM cytochrome c and 300 μM sulfite which allowed the measurement of Moco reconstituted protein. Results are expressed as mean ± standard deviation (n = 3).

Analysis of the oligomeric state via analytical size exclusion chromatography revealed the Moco-containing holo-WT variant as predominantly dimeric protein (Fig. 5C, black line; MW = 124 kDa) with only a minor shoulder corresponding to the elution volume of the monomeric species. In contrast, holo-G362S revealed an elution profile that was different from the corresponding holo-WT profile. While the holo-G362S variant still displayed its major peak assignable to dimeric SO (Fig. 5C, red line; MW = 117 kDa), the ratio of dimer-to-monomer shoulder was found shifted toward the monomer peak indicating a reduced stability of the G362S dimer.

Previous studies have shown that Moco binding is critical for SO dimerization (14). Therefore, the reduced stability of G362S dimer might reflect differences in dimer formation upon Moco binding of G362S SO. In line with these observations, the elution profile of apo-WT confirms this variant as mainly monomeric species in solution (Fig. 5D, black line; MW = 60 kDa) with only a minor peak assignable to the dimer (Fig. 5D; MW = 127 kDa). Similar to apo-WT, apo-G362S displayed a major monomeric peak (MW = 59 kDa); however, another peak corresponding to a lower molecular weight (MW = 24 kDa) was detected. The latter could either represent a degradation product or unspecific interaction of the monomeric G362S SO protein with the column resin. Given the fact that this peak still revealed heme Soret absorption (Fig. 5E) and no significant degradation products of the apo-G362S variant were detected (Fig. 5F), we conclude that the retained elution is a result of increased hydrophobic interaction with the column resin suggesting a different folding than WT apo-SO.

In order to address the differences in oligomerization further, apo-SO as well as holo-SO variants were subjected to circular dichroism (CD) spectroscopy to identify differences in secondary structure composition (Supplementary Material, Fig. S2) and changes in protein stability (Fig. 5G and H). In the holo-state, the G362S SO fold appears different from holo-WT SO since characteristic bands for α-helical elements at 209 and 222 nm show more negative mean residue ellipticity (Supplementary Material, Fig. S2A). Temperature-dependent stability of both holo-SO variants was recorded at 222 nm to analyze the dynamics of SO unfolding. A decrease for the midpoint of unfolding during heat-denaturation from 57.6°C for holo-WT SO to 43.9°C for holo-G362S SO (Fig. 5G) indicated a significant reduction in stability for holo G362S SO variant.

In the apo-state, G362S SO described less negative mean residue ellipticity at 209 nm in comparison to apo-WT SO, which indicates reduced α-helical elements for apo-G362S SO. In line, the band at 192 nm measured for apo-WT SO decreased and shifted toward 196 nm in apo-G362S SO, which indicates higher β-sheet content (Supplementary Material, Fig. S2B). While for apo-WT SO the midpoint of unfolding transition was at 50.2°C, the midpoint of unfolding for apo-G362S SO could not be determined, which indicated a thermostable fold of apo-G362S SO (Fig. 5H). In conclusion, apo- as well as holo-state of G362S SO both differ in their secondary structure composition and structural stability in comparison to the WT SO controls, which is in line with the different oligomeric states observed in analytical size exclusion chromatography.

Apo-G362S is severely impaired in Moco incorporation in vitro

Following the expression, purification and characterization of WT and G362S apo-SO variants, Moco reconstitution was investigated in vitro. Moco was synthesized de novo starting from the Moco precursor cyclic pyranopterin monophosphate (cPMP) (27). The bacterial metal-binding pterin (MPT) synthase (MoaD/MoaE) was used to convert cPMP to MPT followed by the molybdate- and ATP-dependent conversion of MPT into Moco by the human Moco-synthetic protein gephyrin (28–30). For assay calibration, the reaction velocities of each holo-SO variant were correlated with the content of enzyme-bound Moco (Fig. 5I) and compared with the respective velocities of the reconstituted apo-variants of WT and G362S SO following different time points of reconstitution (Fig. 5J). As calculated from the slope of the time-dependent Moco reconstitution, the rate of Moco insertion into WT apo-SO was ~0.13 pmol/min. In contrast, the reconstitution of apo-G362S was 0.0014 pmol/min, which was approximately 90-fold lower than that of the apo-WT enzyme. This finding provides strong evidence that the G362S substitution severely impacts Moco incorporation in apo-SO.

Figure 6

Abundance, activity and SO-bound Moco in mitochondria of cells overexpressing G362S SO (A) Western blot of cytosolic and mitochondrial extracts of HEK293^SUOX−/− cells transiently overexpressing WT SO, full-length G362S SO and Δ79_G362S SO as cytosolic protein. Cytosolic gephyrin and mitochondrial VDAC were used as loading controls. A total of 20 μg protein were loaded on a 8% SDS-PAGE. (B) Western blot of cytosolic and mitochondrial extracts of normal HEK293 cells, HEK293^SUOX−/− cells or HEK293^SUOX−/− cells stably expressing WT or G362S SO using the Flp-in T-Rex system. CMV-driven overexpression was induced by addition of doxycycline for 24 h. A total of 20 μg total protein was loaded. (C) Sulfite-dependent cytochrome c activity of mitochondrial extracts from doxycycline-induced (24 h) cells stably overexpressing WT SO or G362S SO. Mitochondria from normal HEK293 cells and HEK293^SUOX−/− cells were used as controls. A total of 50 μg mitochondria were mixed with 50 μM cytochrome c (final). Reaction was started by addition of 100 μl sulfite (150 μM final) in 100 mM Tris/Ac pH 8.0 in 200 μl (P ≤ 0.001; n = 3). (D) Determination of mitochondria-derived Moco from doxycycline-induced overexpressing cells. A total of 100 μg mitochondria were used to oxidize Moco to Form A as described in methodology. Form A was quantified upon detection using HPLC (P ≤ 0.001; n = 3).

Abundance, activity and Moco insertion of G362S SO are reduced in vivo

In order to further elucidate whether Moco binding of G362S is compromised in cellulo, human embryonic kidney cells, which lack endogenous SO due to a CRISPR/Cas9-mediated knockout of SUOX in a Flp-In T-Rex background (HEK293^SUOX-/-; Supplementary Material, Fig. S3) were utilized. First, HEK293^SUOX−/− cells were transiently transfected with pcDNA3.1(+) encoding full-length WT SO, full-length G362S SO or a G362S SO variant, which lacks the N-terminal mitochondrial localization signal (Δ79-G362S) (Fig. 6A). Upon enrichment of mitochondria, immunodetection of expressed SO by western blot revealed the presence of WT SO in the mitochondrial fraction similar to mitochondrial VDAC control. In contrast, following transfection of full-length G362S SO no SO-specific signal could be detected in either fraction. On the other hand, transfection of cytosolic Δ79-G362S SO resulted in immunodetection of an SO-specific band in the cytosol similar to the gephyrin control. This surprising finding indicates that the absence of full-length G362S SO (neither in mitochondria nor in cytosol) depends on the presence of the N-terminal mitochondrial translocation signal (MTS) in G362S SO.

In order to exclude that the absence of mitochondrial G362S SO is a result of reduced transfection efficiency, the respective transgenes were genomically integrated by generating doxycycline-inducible cell lines in the HEK293^SUOX−/− background using the Flp-In T-Rex system. Cells were grown in absence or presence of doxycycline (24 h) to de-repress the CMV promotor, which drives genomic SO overexpression (Fig. 6B). In the absence of doxycycline, SO-specific immunostaining revealed leaky expression of cells expressing WT SO, while cells expressing the G362S variant lacked the SO-specific signal. Upon doxycycline treatment, strong induction of gene expression was found in cells overexpressing WT SO, which was restricted to the mitochondrial fraction. Strikingly, cells expressing G362S SO show a mitochondrial localized SO-specific signal, however, with markedly reduced signal intensity compared to the overexpressed WT control. This result resembles the reduced G362S SO abundance found for patient fibroblasts. Interestingly, expression levels of G362S SO in those stable cell lines exceeded the levels of endogenous SO in HEK293 cells (Fig. 6B).

Figure 7

Compromised pre-SO maturation aligns WT and G362S SO abundance (A) Cartoon of pre-SO highlighting the bipartite MTS. SO monomer is shown with its domain structures. Numerical annotations show amino acids from N- to C-terminus of the single domains. Residues 1–79 represent the MTS with the predicted transmembrane helix, the known IMP cleavage region and the predicted MPP cleavage site at Arg17. (B) Western blot of mitochondrial fractions of HEK293^SUOX−/− cells stably expressing WT, G362S and both variants as R17W variants in the MTS upon 4 h doxycycline treatment. (C) Sulfite-dependent cytochrome c activity of mitochondrial extracts depicted in (B). Measurements were conducted as stated in Fig. 6 (P ≤ 0.001; n = 3). (D) Determination of mitochondria-derived Moco from 4 h doxycycline-induced overexpressing cells. A total of 100 μg mitochondria were used to oxidize Moco to Form A as described in methodology. Form A was quantified upon detection using HPLC (P ≤ 0.001; n = 3). (E) Western blot of WT and G362S SO variants as R17W and normal MTS variants upon time-dependent doxycycline exposure.

In order to investigate whether HEK293-expressed proteins are active, mitochondria of untreated HEK293 cells, HEK293^SUOX−/− as well as HEK293^SUOX−/− stably expressing WT or G362S SO were enriched and sulfite-dependent SO activity was measured using cytochrome c as electron acceptor (Fig. 6C). In comparison to mitochondria derived from HEK293^SUOX−/− cells, where SO-specific activity was absent, mitochondria derived from normal HEK293 cells showed SO-specific activity of 11.1 ± 5.8 μM cytochrome c * min^-1 * mg protein^-1 attributable to endogenous SO. Strikingly, doxycycline-induced overexpression of WT SO resulted in almost 9-fold increased specific SO activity of 93.1 ± 3.2 μM cytochrome c * min^-1 * mg protein^-1, which aligns with increase in WT SO abundance within mitochondria (Fig. 6B and C). In contrast, doxycycline-induced overexpression of G362S SO resulted in detectable, however, significantly reduced SO-specific activity of 2.8 ± 0.5 μM cytochrome c * min^-1 * mg protein^-1. Hence, the reaction velocity of G362S SO-containing mitochondria is 33-fold reduced compared to overexpressed WT control and 4-fold reduced compared to endogenous SO activity found in normal HEK293 cells. The latter contrasts with the protein levels determined by western blot showing higher levels for G362S SO than for endogenous SO. Therefore, we conclude, that a fraction of mitochondria-localized G362S SO protein lacks specific SO activity.

To investigate whether the reduced activity detected in mitochondria containing G362S SO derives from compromised Moco insertion, we quantified protein-bound Moco via high-performance liquid chromatography (HPLC)-based detection of the fluorescent Moco-derivative Form A (Fig. 6D). For HEK293^SUOX−/−, 4.1 ± 0.2 pmol Form A/mg protein was detected, representing Moco, which was bound to mitochondrial mARC protein. HEK293 cells, which expressed endogenous SO where found to contain 6.1 ± 0.4 pmol Form A/mg protein. Hence, the additional 2 pmol Form A in comparison to HEK293^SUOX−/− cells suggest the additional presence of SO in normal HEK293 cells. Doxycycline-induced overexpression of WT SO resulted in 14.2 ± 0.3 pmol Form A/mg protein. When comparing the relative increase in Moco with SO-deficient and HEK293 cells, an approximately 5-fold increase in Moco correlates well with the previously determined increase in SO activity. In contrast, cells that overexpressed G362S SO were found to contain 5.0 ± 0.2 pmol Form A/mg protein. Hence, despite higher protein abundance of G362S SO compared to endogenous SO (Fig. 6C), only 0.9 pmol Form A/mg protein derived from G362S-bound Moco. This indicates that a fraction of G362S SO is present in the apo-state within mitochondria. From this result, we conclude that within mitochondria a significant fraction of G362S SO protein fails to bind Moco efficiently and is therefore inactive.

Compromised pre-SO processing rescues G362S abundance in vivo

Our previous results collectively demonstrate that G362S substitution impairs Moco incorporation in SO. Next we asked the question, whether the reduced G362S SO abundance within mitochondria is caused by increased proteolytic breakdown of apo-G362S SO, which either might be caused by hampered processing of pre-G362S SO or by impaired Moco binding.

A mitochondrial processing peptidase (MPP) cleavage site is predicted (31) as Arg -2 type motif at Arg17 followed by a membrane spanning helix in the bipartite MTS of SO (14) (Fig. 7A). In order to investigate whether Arg17 in the MTS of SO participates in SO maturation, doxycycline-inducible expression cell lines were generated either expressing WT or G362S SO with R17W substitution in the MTS, respectively. Following doxycycline-induced expression for 4 h, mitochondria were enriched and presence of SO variants was validated via western blot (Fig. 7B).

As shown for 24 h expression (Fig. 6A), doxycycline treatment induced WT SO as well as G362S SO expression; however, protein abundance of G362S SO in mitochondria was significantly reduced compared to the WT control (Fig. 7B). Strikingly, the signal was strongly reduced in the R17W-substituted WT SO variant when compared to unmodified WT SO. Inversely, R17W_G362S SO showed increased signal intensity comparable to the respective WT signal (Fig. 7B). These results suggest that Arg17 plays an important role in the mitochondrial maturation of SO and substitution to Trp17 slows down SO processing for the WT protein. As both R17W WT and G362S SO variants possessed similar expression levels, we exclude an effect of G362S substitution on pre-SO processing.

Next we determined sulfite-dependent cytochrome c reduction velocities of enriched mitochondria derived from cells expressing WT or G362S SO either with normal MTS or R17W_MTS (Fig. 7C). Endogenous SO activity of HEK293 mitochondria was 13.4 ± 0.6 μM cytochrome c * min^-1 * mg protein^-1, while activity of HEK293^SUOX−/− mitochondria was absent. Expression of WT SO resulted in SO activity of 24.5 ± 0.5 μM cytochrome c * min^-1 * mg protein^-1, while 0.3 ± 1.3 μM cytochrome c * min^-1 * mg protein^-1 was measure for G362S SO, respectively. In contrast, expression of R17W_WT SO resulted in an activity of 17.8 ± 0.3 μM cytochrome c * min^-1 * mg protein^-1 compared to mitochondria containing R17W_G362S SO with an activity of 0.1 ± 1.3 μM cytochrome c * min^-1 * mg protein^-1. These results demonstrate that R17W substitution reduces WT SO activity, which aligns with reduced WT SO abundance in mitochondria, while G362S activity remains unaffected despite its increased expression, which is similar to R17W_WT SO.

In order to investigate whether reduced activity of R17W_G362S SO is a result of compromised Moco binding, we determined SO-bound Moco in mitochondria expressing the individual SO variants (Fig. 7D). Upon doxycycline-induced overexpression of WT SO, we found 3.2-fold higher levels of SO-bound Moco in respect to SO-bound Moco measured for endogenous SO in HEK293 cells. Similar to previous results, SO-bound Moco was reduced in cells expressing G362S SO by 28% compared to endogenous SO in HEK293 cells. SO-bound Moco of mitochondria expressing R17W_WT SO was found to be significantly reduced compared to WT SO with normal MTS, however, increased over Moco levels measured in normal HEK293 cells. Despite comparable expression levels of R17W_WT and R17W_G362S SO, SO-bound Moco in mitochondria, which contain R17W_G362S SO, was undetectable. This result shows that R17W substitution does still allow binding of Moco to WT SO enzyme in vivo; however, for G362S SO a large fraction of SO in a Moco-free state can be assumed. As a result of compromised Moco binding to apo-G362S, the protein declines in abundance while WT SO protein accumulates with ongoing expression (Fig. 7E).

Molybdate-dependent rescue of G362S activity in vivo

In contrast to the compromised in vitro and in vivo Moco incorporation of G362S variant, E. coli-expressed G362S variant showed a Moco saturation comparable with WT SO. Therefore, the applied expression conditions might provide insights into the mechanism that contributed to the high Moco saturation of G362S variant in E. coli. Standard expression conditions for Mo-enzymes include molybdenum excess in the culture medium (0.5 mM sodium molybdate) to boost Moco synthesis under conditions of Mo enzyme overexpression (17,25,32,33). Therefore, we speculate that under Moco excess, the reduced cofactor insertion ability of G362S variant might be compensated.

In order to determine the impact of molybdate supplementation on Moco saturation in WT and G362S SO variants, both variants were expressed in absence of externally added molybdate in E.coli and proteins were purified to a high degree of homogeneity (Fig. 8A). WT SO showed 31.4% Moco saturation, which was close to the level of WT SO expressed under molybdate supplementation. In contrast, G362S displayed only 4.1% of Moco bound to the enzyme (Fig. 8A), which is an 8-fold reduction in comparison to WT SO and nearly 10-fold reduction when compared to the variant expressed with molybdate excess. This finding confirms a molybdate-dependent increase of Moco incorporation into G362S variant and raises the question to which extent these findings could be reproduced in human cells.

Figure 8

Effect of environmental molybdate on Moco saturation and SO activity (A) Coomassie brilliant blue staining of 12% SDS-gel of eluted proteins after Ni-NTA chromatography. Proteins were expressed as described in methodology for 72 h and 19°C; however, no additional molybdate was supplemented in the growth medium. (B) Moco saturation of eluted SO variants. The Moco saturation was determined via HPLC-based Form A analysis of Moco based on protein concentrations calculated from A₄₁₃ of heme Soret absorption. (C) Sulfite-dependent specific SO activity measuring cytochrome c reduction of crude fibroblast extracts treated with or without 400 μM sodium molybdate for 24 h. A total of 100 μg protein was used with 50 μM porcine cytochrome c and 300 μM sulfite in 100 mM Tris/Ac pH 8.0 (P ≤ 0.01; n = 3).

In order to probe the impact of molybdate on SO activity in human cells, WT and G362S patient fibroblasts were treated with 400 μM molybdate-containing growth medium. As control, fibroblasts cultured in standard non-supplemented medium were analyzed. Sulfite-dependent SO activities in crude protein extracts of WT control fibroblasts showed a two-fold increase in activity following 24 h molybdate exposure (Fig. 7B). Interestingly, while not detectable in patient fibroblasts grown under standard conditions, SO activity of 0.12 μM cytochrome c * min^-1 * mg^-1 protein was detected in molybdate-treated patient fibroblasts representing 22.5% and 9.5% of WT activity derived from cells grown without and with molybdate, respectively (Fig. 7B). In conclusion, molybdate-dependent increase in Moco synthesis accounted for Moco insertion into G362S SO variant in vivo.

Discussion

As most mitochondrial proteins, SO is synthesized in the cytosol and imported into the IMS of mitochondria (10). The maturation process of SO involves translocation through the outer translocator complex, inner mitochondrial membrane anchoring, subsequent proteolytic cleavage, cofactor insertion and dimerization (14). The hierarchical order of maturation and the multitude of steps require a highly orchestrated process that—if functionally impaired in one of the steps—can lead to a non-functional SO. The presented case with homozygous substitution of Gly362 to serine in SO is to our knowledge the first iSOD report with a compromised maturation process.

Comparative analysis of apo-WT and apo-G362S SO, which were produced for in vitro Moco reconstitution, revealed apo-G362S as highly instable protein in the absence of Moco since 99% of the protein was sedimented due to its aggregation in the crude extract. Aggregation is most likely promoted by misfolding of apo-G362S as revealed by CD spectroscopy. In line with these findings, immunostaining of SO in patient fibroblasts revealed a strongly reduced abundance of G362S variant in mitochondria compared to WT fibroblasts indicating a less stable protein also in human cells. Hand in hand with this, HEK293 cells, which stably overexpressed G362S SO, showed reduced enzyme abundance in mitochondria accompanied by reduced Moco binding. Since Moco incorporation is a prerequisite for SO dimerization (14), apo-WT and apo-G362S were predominantly observed as monomers in analytical size exclusion chromatography. Interestingly, the apo-G362S size exclusion elution profile revealed major conformational changes in comparison to the WT protein. The retention time of the variant pointed to an interaction with the size exclusion resin probably due to misfolded hydrophobic patches, which were not exposed in the apo-WT variant. The reduced stability in combination with a 90-fold reduction in Moco incorporation efficacy of the G362S apo-protein provides strong evidence for a maturation defect of SO in the patient as underlying cause of iSOD.

The G362S substitution localizes in close proximity to a triade of critical residues (His361, Arg366, Lys379) coordinating the phosphate moiety of Moco within the folded holo-SO. The literature reports two cases of iSOD with substitutions in Arg366 and Lys379. Interestingly, the biochemical analysis of recombinant R366H and K379R variants revealed functional proteins with up to 4-fold increased catalytic activity, which raised a critical question regarding the underlying mechanism of iSOD in both patients (16,23). Since the G362S variant also presented a 4-fold increased catalytic activity upon recombinant expression, similarities in the underlying pathomechanism in R366H and K379R can be assumed. Nevertheless, the increased catalytic turnover in these recombinant SO variants is unexplained so far. One explanation might be that the more loosely bound cofactor enables a velocity increase of a catalytic rate constant, which is limiting in the WT enzyme by reducing the energy barrier of a transition state intermediate related to the Moco active site.

Site-directed spin labeling of human SO revealed a flexible loop region (residues 322–332) that was found to participate in Moco insertion (15). Here, the use of spin-labeled residues in pairwise combination within apo- or holo-SO allowed to assess conformational changes of SO regions, which were dependent on the presence of Moco. The identified loop was found in an open conformation in the absence of Moco presenting the Moco binding pocket to the proteins’ surrounding (15), which indicated a phosphate-first incorporation mechanism of the cofactor. Accordingly, once Moco enters the cofactor binding pocket of SO, a critical role of the Moco-C4’ coordinating His-Arg-Lys triade argues for their important function in Moco orientation and/or binding. Accordingly, the 90-fold decreased Moco incorporation rate of apo-G362S underscores a deficiency of cofactor insertion during the SO maturation process due to the structural proximity of G362 toward the phosphate-binding residues. Our results further support an important role of the Moco-C4’ phosphate moiety during SO maturation and a critical function of the His-Arg-Lys triade within the pre-folded apo-enzyme for Moco incorporation.

Regardless of the strongly reduced ability of G362S variant to bind Moco, the remaining activity was sufficient to produce significant quantities of Moco-saturated active protein following prokaryotic expression. Indeed, changing the expression conditions by withholding excess molybdate from the medium significantly compromised the Moco saturation of the G362S variant. This finding indicates a molybdate-dependent pool of Moco (34) allowing G362S to reach Moco saturation levels comparable to the WT.

Similar to our observations, Moco levels of the gephyrin-deficient mouse fibroblast cell line L929, which is defective in incorporation of molybdate into the Moco precursor MPT, can be restored completely by providing excess molybdate in the medium (35). Moreover, the independence of the in vitro Moco biosynthesis on gephyrin or its bacterial homologs in the presence of molybdate excess provides evidence for non-enzymatic Moco formation (29). In general, these data suggest a molybdate-dependent increase in Moco, which might be beneficial to counterbalance a reduced efficacy of Moco-incorporation in the G362S fibroblast cell line. Following molybdate treatment of human fibroblasts, the specific SO activity doubled in WT cells and could be partially restored in G362S SO expressing patient cells.

Although molybdate treatment was revealed as an inefficient method to treat a loss of SO activity in MoCD in earlier studies (36), in the here reported case of iSOD, dietary molybdate supplementation might have had beneficial effects. Considering the similarities of recombinant G362S to other iSOD patient variants with altered Moco-phosphate coordination, a similar maturation deficit might be concluded in these cases. Thus, dietary molybdate should be re-considered as treatment option in case of a maturation deficient variant of SO. Similar cofactor responsive enzymes deficiencies have been reported in other disorders. Most prominent, the tetrahydrobiopterin-dependent phenylketonuria has been found to constitute a significant fraction of patients suffering from phenylalanine hydroxylase deficiencies and led to the clinical development of Kuvan (37–39). In light of our findings, the group of Moco-responsive iSODs might be larger than initially expected, and therefore molybdate treatment should be considered as a loading test in iSOD patients.

The collective data in this study highlight the urgent need for a fast genetic and subsequent biochemical analysis of iSOD patients. Such studies have the strong potential for future personalized medical intervention before disease progression leads to irreversible neuronal damage.

Materials and Methods

Determination of creatinine, S-sulfocysteine and purines in urine

Creatinine, hypoxanthine, xanthine, uric acid and SSC were analyzed by HPLC and diode array detector (DAD) as earlier described (40).

Creatinine was analyzed based on the previously described method (41). In brief, urine samples were diluted 1:10 in 10 mM octanesulfonic acid and 5% acetonitrile adjusted to spH 3.2 with orthophosphoric acid and centrifuged at 13 000 rpm for 5 min. The supernatant was removed and transferred into a 1.5 ml glass HPLC vial. Creatinine was separated into a 250 × 3 mm 5 μm Nucleodur HTec C18 column (Machery-Nagel, Germany). The running buffer (A) consisted of 10 mM octanesulfonic acid and 5% acetonitrile adjusted to pH 3.2 with orthophosphoric acid and buffer (B) consisted of 100% methanol. Creatinine was eluted at 1 ml/min in a buffer (B) 10–35% 0–5 min gradient, eluting at 6.58 ml. Creatinine was quantified based on area under the peak at 236 nm.

Xanthine and uric acid were analyzed based on a modified method described earlier (42). Briefly, urine samples were diluted 1:10 in 50 mM Na-format 0.1% trifluoroacetic acid pH 4.0 and centrifuged at 13 000 rpm for 5 min. The supernatant was removed and transferred into a 1.5 ml glass HPLC vial. Xanthine and uric acid were separated into a 250 × 3 mm 5 μm Nucleodur C18 COLUMN (Machery-Nagel, Germany). Xanthine and uric acid were isocratically eluted at 1 ml/min 50 mM Na-formate and 0.1% trifluoroacetic acid pH 4.0. Xanthine eluted at 5.17 ml and was quantified based on area under the peak at 267 nm. Uric acid eluted at 3.77 ml and was quantified based on area under the peak at 285 nm.

S-Sulfocysteine was pre-column derivatized with 9-fluorenylmethyloxycarbonyl chloride (FMOC-Cl) based on the derivatization described earlier (43). In brief, 175 μl of urine sample was transferred into an Eppendorf tube and centrifuged at 13 000 rpm for 5 min. The supernatant was removed. A total of 150 μl of the supernatant was mixed with 150 μl 200 mM Na₂CO₃/NaHCO₃ buffer, pH 9.5 and 300 μl 20 mM FMOC-Cl (dissolved in acetonitrile). The Eppendorf tube was incubated at room temperature for 20 min. The reaction was stopped by the addition of 50 μl 80 mM amantadine hydrochloride (dissolved in 1:1 acetonitrile/0.2 M HCl). After 5 min incubation time, the sample was centrifuged at 13 000 rpm for 10 min, and the supernatant was transferred into a 1.5 ml HPLC vial. S-Sulfocysteine was separated into a 250 × 3 mm 5 μm Nucleodur HTec C18 column (Machery-Nagel, Germany). The running buffer (A) consisted of 50 mM Na-formate and 0.05% trifluoroacetic acid with a pHof 4.5, and buffer (B) consisted of 100% methanol. S-Sulfocysteine was eluted at 1 ml/min in a buffer (B) at 35–60% and 0–7.5 min gradient, eluting at 8.81 ml. S-Sulfocysteine was quantified based on area under the peak at 265 nm.

Urothione analysis

Urothione analysis was performed as earlier described (44). Urothione was isolated from urine by a two-step solid phase extraction protocol and detected and quantified by HPLC and DAD using a modified protocol from earlier studies (45). Urine containing 25 μmol of creatinine was used for routine isolation of urothione. In brief, 1–10 ml urine containing 25 μmol of creatinine was centrifuged at 13 000 rpm for 10 minutes. The resulting supernatant was loaded into a 15 ml polypropylene column packed with 300 mg Florisil matrix 60-100 Mesh (Machery-Nagel, Germany), previously equilibrated with 3 × 1 ml 10 mM HCl. The column was washed with 1 × 1 ml 10 mM HCl, and urothione was eluted with 2 ml 50% acetone in water. The eluted urothione was concentrated to dryness on an eppendorf concentrator at 60°C. The resulting powder was resuspended in 500 μl 20 mM acetic acid and loaded into a 1 ml polypropylene column packed with 100 mg polyRP100 30 μm particles from Sepax Technologies (USA) previously equilibrated with 1 × 1 ML 100% MeOH and subsequently with 3 × 1ml 20 mM acetic acid. The column was washed with 1 × 1 ml 20 mM acetic acid and then with 400 μl 20 mM acetic acid in 40% MeOH. Urothione was eluted in 2 × 1 ml 20 mM acetic acid in 40% MeOH. The resulting eluate was concentrated to dryness on an eppendorf concentrator at 60°C, and the obtained powder was resuspended in 100 μl 20 mM acetic acid. Urothione was then separated into a 50 × 3 mm 1.8 μm Nucleodur Gravity C18 HPLC column (Machery-Nagel, Germany) assembled on a Hitachi Elite LaChrom HPLC system. Urothione was isocratically eluted at 0.8 ml/min in 90% buffer (A) 20 mM acetic acid, pH 2.0 and 10% buffer (B) 100% MeOH. Urothione eluted at 2.3 ml and was quantified based on area under the peak at 380 nm.

RNA isolation, cDNA synthesis and qPCR of human SO cDNA

Human fibroblasts were grown until 80–90% confluency on 10 cm dishes in RPMI medium containing 20% FCS. Cells were washed with 1× PBS and scraped from the plates followed by centrifugation for 2 min at 800g and RT. The supernatant was discarded and the pellet resolved in 1 ml of Trizol™. RNA isolation was performed according to the manufacturer’s protocol. The RNA was digested with DNase for 30 min at 37°C to remove possible genomic DNA contamination. After 30 min DNase was deactivated by adding 1 μl and 1 M EDTA and heating to 60°C for 10 min. The RNA was transferred to ice, and 1 μg was reverse transcribed using MMLV-RT and oligodT primers according to the manufacturer’s protocol (Promega, M1701). The resulting 25 μl cDNA were diluted by adding 175 μl ultrapure H₂O. For a qPCR reaction 2 μl diluted cDNA were pipetted in one well of a 364-well qPCR plate (Biorad). For each primer mix 0.5 μl of a 10 μM mix of forward and reverse primer was added to 2.5 μl qPCR master mix. The primer mix was added, and the plate was sealed with Microseal® Adhesive Sealer (BioRad) and spun down for 20 sec at 300g. The plate was placed in a CFRX384™ Real-Time System (BioRad), and the qPCR was performed according to the ORA HighQ qPCR manual. Acquired CT values were exported and analyzed with excel. For this the CT values of the gene of interest were first normalized to the housekeeping control (RPL37a). The fold change of a ΔCt of a given condition was then calculated by dividing the 2^ΔCT by the 2^ΔCT of the control condition. The primers, which were used for cDNA amplification, are 5′ CCCACCTCCATTTCTAATGCCT 3′ and 5′ ACACATGCCACACACATTTCTC 3′ for SO and 5′ GTGGTTCCTGCATGAAGACAGTG 3′ and 5′ TTCTGATGGCGGACTTTACCG 3′ for RPL37A.

Western blotting

Electrophoretic transfer of proteins in an SDS-gel onto a PVDF membrane was conducted via the semi-dry method using transfer buffer composed of 25 mM Tris pH 8.8, 192 mM glycin and 10% methanol for 2 h at 50 V and 1.2 mA/cm². The membrane was blocked in 5% milk TBS-T for 1 h at room temperature and subsequently incubated with the primary antibody α-SO (Eurogentech, ZDE 13058, 1:500) or α-SO (Abcam, ab129094, 1:10 000), α-VDAC (Abcam, ab15895, 1:1000), α-gephyrin (cell supernatant, 3B11) or α-tubulin (Sigma, T7816, 1:20 000) diluted in 2% milk TBS-T for 1 h, washed twice with TBS-T and once with TBS for 10 min and then incubated with secondary antibody (α-rabbit, Santa Cruz Biotechnology, SC2054, 1:5000; α-mouse, Santa Cruz Biotechnology, SC2055, 1:5000) diluted in 2% milk TBS-T for 1 h. Washing was performed as previously and horseradish peroxidase (HRP) signals were detected by adding 1 ml chemiluminescent HRP substrate (Thermo Scientific) on the membrane and photon detection with a ChemiDoc MP (BioRad) detection system.

Sulfite-dependent cytochrome c reduction

For measuring the specific sulfite-dependent SO activity porcine cytochrome c (ε₅₅₀ = 19630 l/mol*cm) (BioChemica) was used as SO substrate. Reduction of cytochrome c was recorded at 550 nm in a time-dependent manner by use of an ELISA reader (EL808 Biotec). The reaction took place in 200 μl, 100 mM Tris/Ac, pH 8.0 containing 50 μM porcine cytochrome c and recombinant SO or crude cell extract. In case of crude cell extracts 0.5 mM KCN and 1 μl of a 4 mg/ml bovine catalase solution (Sigma) was added, and the activity was normalized to the amount of protein used for the measurement determined via Bradford.

Sulfite-dependent FeCN reduction

Sulfite-dependent SO velocities were determined using the artificial electron acceptor hexacyanidoferrat (II) (FeCN), which accepts electrons directly at the Moco domain active site. The absorption decrease of FeCN at 420 nm (ε₄₂₀ = 1020 l/mol*cm) upon reduction was measured in an ELISA reader. The reaction volume was 200 μl in 100 mM Tris/Ac pH 8.0, 400 μM FeCN and purified enzyme. The reaction was started with 100 μl sulfite solution of varying concentrations (diluted in 100 mM Tris/Ac pH 8.0).

Mitochondrial enrichment

For mitochondrial enrichment fibroblasts or HEK293 cells were washed twice with pre-warmed mitochondria buffer containing 20 mM HEPES pH 7.4, 220 mM mannitol, 70 mM sucrose and 1 mM EDTA. Cells were harvested via scratching and suspended in 2 ml ice-cold mitochondria buffer. Cells were homogenized by use of a Teflon homogenizer (Sartorius) with 15 strokes at 1100 rpm at 4°C. Intact cells as well as cell debris was pelleted at 800g for 10 min. The supernatant containing intact mitochondria was transferred to a new reaction tube and centrifuged at 13000g for 15 min in order to pellet mitochondria. The supernatant was collected as the cytosolic fraction. The mitochondrial pellet was washed three times with 1 ml mitochondria buffer, and the final pellet was suspended in 200 μl 100 mM Tris/Ac pH 8.0.

Site-directed mutagenesis and cloning

For recombinant expression, SO cDNA lacking the mitochondrial leader peptide was cloned into pQE80L (Quiagen) using SacI and SalI restriction sites as described earlier (14). Fusion PCR was used in order to introduce the c.1084G>A mutation using mutagenesis primers 5′-CACGTGACCACAGCTTCCCTGTGCGT-3′ and 5′-ACGCACAGGGAAGCTGTGGTCACGTG-3′. For introduction of R17W substitution into MTS of SO, primers 5′-GCTCCAACAGGCCTGCTGGCTCAAGTCAATCCCCT-3′ and 5′-AGGGGATTGACTTGAGCCAGCAGGCCTGTTGGAGC-3′ were used. For stable genome integration, SO cDNA was cloned into pcDNA™5/FRT (Thermo Fischer) using SalI and XhoI as restriction sites using primers 5′-GGGTCGACCTA TGCTGCTGCTGCACAGAGCTG-3′ and 5′-GACTCGAGCGGCCGCTTTCATGGGG AGACATAGA-3′. Sequence was validated via Sanger sequencing (GATC Biotech AG).

Recombinant expression and purification

Escherichia coli TP1004 were transformed with pQE80L plasmids carrying SO constructs under control of T5 promoter fused to an N-terminal His₆-tag. Cells were plated on ampicillin (100 μg/ml) and kanamycin (25 μg/ml) containing agar and grown at 37°C o/n. Single colonies were picked and grown at 37°C in 50 ml LB medium (10 g/L tryptone, 10 g/L NaCl and 5 g/L yeast extract) containing same antibiotics. A total of 2 L LB medium supplemented with antibiotics and 500 μM sodium molybdate (or without) were inoculated with the pre-culture and expression of the target genes was induced at an OD₆₀₀ of 0.6 with 125 μM IPTG. Recombinant proteins were expressed for 72 h with 90 rpm at 18°C. Cells were harvested via centrifugation at 5000g for 20 min. Pellets were re-suspended in 25 ml/L culture medium Ni-NTA lysis buffer containing 50 mM Tris/Ac pH 8.0, 300 mM NaCl and 15 mM imidazole. Cell disruption was achieved by use of an Emulsiflex-C5 cell disrupter (Avestin) at 1300 bar and 4°C. Cellular debris was pelleted at 50 000g for 45 min at 4°C. A total of 2.5 ml Ni-NTA matrix/L culture medium were equilibrated with 2 column volumes of lysis buffer. The supernatant was loaded onto the column, and His₆-tagged proteins were allowed to bind to the matrix for 1 h in batch. The matrix was allowed to settle and undesired proteins were washed out by gravity flow of 15–20 column volumes of lysis buffer. Elution of bound proteins was achieved by loading 3 column volumes of elution buffer containing 50 mM Tris/Ac pH 8.0, 300 mM NaCl and 500 mM imidazole. The eluted proteins were rebuffered with PD-10 columns into 50 mM Tris/Ac pH 8.0 and 30 mM NaCl, frozen in liquid nitrogen and stored at -80°C.

HPLC-based analysis of Moco derivative Form A

Purified SO (100 pmol based on A₄₁₃; ε = 113 000 l*mol^-1*cm^-1) was diluted to a total volume of 140 μl with 100 mM Tris/HCl pH 7.2 and oxidized over night with 17.5 μl acidic oxidation mix (17.5 μl 1.5/2.0% I₂/KI solution + 1.5 μl 37% HCl) at room temperature in the dark. Samples were centrifuged at 21 000g for 15 min. Supernatant was mixed with 20 μl freshly prepared 1% ascorbic acid solution, 70 μl 1 M unbuffered Tris, 5 μl 1 M MgCl₂ and 0.5 μl alkaline phosphatase (Roche) and incubated for 1 h at room temperature in the dark. A 1:10 dilution was transferred into HPLC vials in order to quantify Form A in 90 μl on a C18 reverse phase column (XBridge 3.5 μm 4.6 * 150 mm) connected to an Agilent 1200SL HPLC system. Form A was eluted using methanol as organic phase with a flow rate of 2 ml/min. Detection was achieved by excitation at 370 nm and detection at 450 nm.

Analytical size exclusion chromatography

Determination of SO oligomeric state was achieved by analytical size exclusion chromatography (Superdex 200 10/300 22 ml). The column was equilibrated with 2 column volumes of 50 mM Tris/Ac pH 8.0 and 300 mM NaCl. A total of 50 μl SO solution (250 pmol) were injected in a 100 μl injection loop of an Äkta Prime system (GE Healthcare) at 4°C. Size exclusion chromatography was performed with a flow rate of 0.2 ml/min and a maximum pressure of 1 MPa.

CD spectroscopy

Far-UV spectra were recorded with J-715 CD spectropolarimeter (Jasco, Gross-Umstadt, Germany) at 20°C in the range of 185–260 nm using quartz cuvette with 0.1 cm path length. Final protein concentration was adjusted to 0.2 mg/ml in ddH₂O. Buffer baseline was recorded separately and subtracted from each sample spectrum. Obtained ellipticity (θ) was normalized to protein concentration, molecular mass in Da (Mr), number of amino acids (n) and path length of cuvette (l) using: [θ] = θ * Mr/10 * (n-1) * c * l. Thermal unfolding of purified proteins with a concentration of 0.2 mg/ml was performed in ddH₂O. Ellipticity was measured continuously at 220 nm in a sealed cuvette with a heating speed of 1°C/min from 10°C to 80°C and data pitch of 0.2 nm. Melting curves were fitted using Jasco Spectra Manager 2 software.

In vitro reconstitution of apo-SO

Moco synthesis took place in 140 μl composed of 20 mM HEPES pH 7.4, 3 nmol MoaD, 0.1 nmol MoaE, 0.1 nmol gephyrin, 10 mM MgCl₂, 5 μM sodium molybdate, 10 mM ATP and 250 pmol (based on A₄₁₃) apo-SO. The reaction was initialized by addition of 1 nmol cPMP. The mixture was incubated at 25°C, and at different time points samples were collected. The samples (50 nM SO final) were mixed in a volume of 200 μl with 50 μM porcine cytochrome c and 300 μM sulfite. Sulfite-dependent cytochrome c reduction of reconstituted SO was recorded as described above.

Generation of HEK293^SUOX−/− cells

SUOX-deficient HEK293 cells were generated using the double-nickase CRISPR/cas9 approach described previously (46). Two pX335 vectors (Addgene #42335), each encoding one SUOX-specific guide RNA (gRNA 1: CCTATCAGGACCATCGGTGT; gRNA 2: CACTGCACCGAGACCTAATA) were generated. HEK293 Flp-In™ T-REx™ cells (Thermo Fischer Scientific) were simultaneously transfected with both pX335 vectors using GeneJuice transfection reagent (Novagen) according to manufactures protocol. To separate putative knock out candidates, cells were detached via trypsin/EDTA treatment, sequentially diluted to a concentration of 0.5 cells/100 μL and 100 μl plated onto 96-well plates. After 72 h, 100 μL DMEM was added to wells, and cell growth was ensured for 10 days. Single colonies were transferred onto 12-well plates. HEK293^SUOX−/− cells were identified via western blot and genome sequence analysis.

Generation of stable cell lines in HEK293^SUOX−/− background

Stable cell lines were generated in the HEK293^SUOX−/− background using the Flp-In™ T-REx™ system according to the manufacturers protocol (Thermo Fischer). Briefly, HEK293^SUOX−/− cells were seeded in a 6-well plate and grown until 70–80% confluency. Cells were co-transfected with pOG44 and pcDNA™5.1/FRT (containing full-length SO constructs introduced with HindIII, EcoRV) (2500/500 ng) using GeneJuice transfection reagent (Novagene). Here, 9 μl GeneJuice was mixed in 100 μl DMEM without additives and incubated for 5 min. pOG44 and pcDNA™5/FRT was added and incubated for 15 min. After 48 h, cells were transferred to 10 cm plates, and selection was started using 100 μg/ml hygromycine B for 48 h. Medium was changed and hygromycine treatment was continued for 48 h. Single colonies were picked and transferred to 12-well plates. Positive clones were tested using western blotting.

Acknowledgements

We acknowledge Simona Jansen, Monika Laurien and Joana Stegemann for technical assistance.

Conflict of Interest statement. None declared.

Funding

Center for Molecular Medicine Cologne; Deutsche Forschungsgemeinschaft (Emmy Noether Grant Cl 218/1-1 to S.C.).