If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Sulfite oxidase (SOX) is a homodimeric molybdoheme enzyme that oxidizes sulfite to sulfate at the molybdenum center. Following substrate oxidation, molybdenum is reduced and subsequently regenerated by two sequential electron transfers (ETs) via heme to cytochrome c. SOX harbors both metals in spatially separated domains within each subunit, suggesting that domain movement is necessary to allow intramolecular ET. To address whether one subunit in a SOX dimer is sufficient for catalysis, we produced heterodimeric SOX variants with abolished sulfite oxidation by replacing the molybdenum-coordinating and essential cysteine in the active site. To further elucidate whether electrons can bifurcate between subunits, we truncated one or both subunits by deleting the heme domain. We generated three SOX heterodimers: (i) SOX/Mo with two active molybdenum centers but one deleted heme domain, (ii) SOX/Mo_C264S with one unmodified and one inactive subunit, and (iii) SOX_C264S/Mo harboring a functional molybdenum center on one subunit and a heme domain on the other subunit. Steady-state kinetics showed 50% SOX activity for the SOX/Mo and SOX/Mo_C264S heterodimers, whereas SOX_C264S/Mo activity was reduced by two orders of magnitude. Rapid reaction kinetics monitoring revealed comparable ET rates in SOX/Mo, SOX/Mo_C264S, and SOX/SOX, whereas in SOX_C264S/Mo, ET was strongly compromised. We also combined a functional SOX Mo domain with an inactive full-length SOX R217W variant and demonstrated interdimer ET that resembled SOX_C264S/Mo activity. Collectively, our results indicate that one functional subunit in SOX is sufficient for catalysis and that electrons derived from either Mo(IV) or Mo(V) follow this path.
) (Fig. 1A). The N-terminal heme domain is connected to the catalytic Moco domain by a flexible tether of 11 residues followed by a C-terminal domain responsible for homodimerization of two SOX subunits.
Within the Moco domain of SOX, the catalytic molybdenum atom is placed in the center of a square-based pyramid where it coordinates five atomic ligands (
). Three of these ligands are sulfur atoms, of which two are derived from a dithiolate moiety of the Moco scaffold. The third sulfur derives from a highly conserved cysteine residue within the SOX polypeptide chain (human Cys264) (
). The remaining two ligands are oxygen atoms, of which one is coordinated axially in respect to the molybdenum atom, whereas the second oxygen atom points toward the active site of SOX. The latter represents the reactive oxoligand that is used for oxotransfer to form sulfate from sulfite (
). This defines the reductive half reaction of SOX.
The Mo(VI) state of SOX is regenerated in a process that includes two sequential intramolecular electron transfer (IET) reactions from molybdenum to the iron of the heme cofactor and from there to the final electron acceptor cytochrome c (
) (Fig. 1A). Therefore, a closed conformation of the enzyme was proposed that enables fast IET from molybdenum to iron followed by an open conformation that allows the subsequent release of the electron from the heme iron toward cytochrome c (
). While oxidation of sulfite and the concomitant reduction of molybdenum in the reductive half reaction of SOX were measured with 541 s−1, the first IET from Mo(IV) to the heme iron was substantially slower with rates of 80 s−1 at pH 7.7 and 28 s−1 at pH 8.0 (
Loss of SOX activity leads to severe neurological symptoms, including seizures, and cumulates in early childhood death. A deficiency in SOX can be caused by a mutation of SUOX, the gene that encodes the SOX enzyme, or a mutation in one of the genes required for the synthesis of Moco (
). The resulting conditions are termed isolated SOX deficiency (ISOD) or Moco deficiency, respectively. Up to date, there are 58 patients reported in the literature, carrying loss-of-function, protein truncating, and missense mutations within SUOX (
). Amongst those, eight disease-causing missense mutations localize to the dimerization domain. Four were identified in ISOD patients, of which the G530D variant was biochemically analyzed, and its monomeric state was confirmed (
). The remaining four SOX variants found in the dimerization domain were identified among naturally occurring SUOX mutations in a machine learning approach in silico, which were proven kinetically inactive in vitro (
). These findings suggest a functional requirement of dimer formation in SOX. Moreover, all other eukaryotic Moco-dependent enzymes, except for the mitochondrial amidoxime–reducing component, are homodimeric enzymes in which each subunit is equipped with enzyme-specific cofactors (
). As dimerization is critical for SOX catalysis, we tempted to keep dimerization behavior intact by deleting catalytic functionality of only one subunit in a SOX dimer. With this tool, we asked whether SOX adopts a conformation in which IET occurs between molybdenum and iron of opposing subunits and whether this is the functional need of SOX dimerization. Therefore, a sequential purification method was established with which it was possible to purify recombinantly expressed human SOX (hSOX) as heterodimeric protein. These SOX heterodimers consisted of a full-length (FL) subunit and a truncated subunit that lacked the N-terminal heme domain (Fig. 1B). By replacing the molybdenum-coordinating cysteine 264 to serine on either side of a heterodimer, we abolished sulfite reactivity (
) in the truncated or FL subunit, thereby generating SOX heterodimers with controlled electron flow. We revealed the IET-competent conformation to be limited to a single subunit, thereby excluding electron bifurcation in a SOX dimer. Furthermore, by using a combination of different SOX variants, we clarify the complex pathway of electron flow constituting two fast ET steps within a SOX dimer followed by two slow intermolecular ET steps.
Purification of heterodimeric SOX variants
In order to investigate whether ET from molybdenum to heme in SOX takes place as intrasubunit ET or in an intersubunit trans-pathway from one subunit to the other, we generated heterodimeric SOX variants and compared them with WT SOX (Fig. 1B). Starting from the human WT SOX/SOX homodimer, we first generated a SOX/Mo heterodimer with one fully functional SOX subunit and a truncated subunit lacking the N-terminal heme domain. This SOX/Mo heterodimer has two active molybdenum centers but only one heme domain and serves as a control, given that heme activity cannot be assigned to either an intersubunit or an intrasubunit ET. The second heterodimer SOX/Mo_C264S consists of one fully functional FL SOX subunit and the heme-truncated subunit carrying the C264S substitution that abolishes sulfite oxidation at the molybdenum center. This restricts ET from molybdenum to heme to the functional FL SOX subunit. The third SOX_C264S/Mo heterodimer represents the reverse setting in which the active site C264S substitution abolishes sulfite reactivity of the FL subunit, which is, however, the only subunit harboring a heme domain. Hence, activity detected in this heterodimer would derive from electrons generated on a different Moco-active subunit.
Sequential affinity purification, utilizing first the N-terminal His-tag at one subunit followed by a second Strep-Tactin binding of a Strep-Tag at the N-terminus of the second subunit, allowed isolation of heterodimeric SOX variants that carried both affinity tags (Fig. 1C). Preparative size-exclusion chromatography (SEC) purified SOX dimers to homogeneity (Fig. 1C), whereas analytical SEC was used to reveal their heterodimeric state in solution (Fig. 1D). Here, heterodimeric SOX/Mo domain eluted with a calculated molecular weight of 105 kDa, which was in between the sizes of the SOX/SOX and Mo/Mo homodimers with calculated masses of 133 and 71 kDa, respectively (Fig. 1D). Purity of the proteins was confirmed by Coomassie brilliant blue staining of a 12% SDS gel, showing a band at approximately 60 and 50 kDa, representing the FL and heme-truncated subunit of SOX, respectively (Fig. 1E).
Distribution of Moco in SOX dimers
In order to ensure precise result interpretation, Moco saturations of the SOX/SOX, SOX/Mo, SOX/Mo_C264S, and SOX_C264S/Mo dimers were determined (
). Given that the amount of Moco could only be determined per dimer, an additional activity determination of the Moco domains within a dimer was applied, by using the artificial electron acceptor hexacyanoferrate (ferricyanide [FeCN]). Note that FeCN accepts electrons solely from an active molybdenum site. We assumed a 50% decreased Moco-dependent FeCN activity of heterodimers in cases where one subunit harbors the C264S substitution. We found for the SOX/Mo heterodimer, 103.1 ± 18.5% of the FeCN-dependent activity per incorporated Moco, which is similar to the SOX/SOX homodimer (Fig. 2A). In contrast, the SOX/Mo_C264S and SOX_C264S/Mo heterodimers showed decreased FeCN reduction rate per incorporated Moco of 58.7 ± 14.7% and 58.6 ± 19.8% for SOX/Mo_C264S and SOX_C264S/Mo, respectively (Fig. 2A). This finding is consistent with a near equal Moco incorporation among both subunits of a given dimer, despite the presence of the C264S substitution (Fig. 2A).
Sulfite-dependent cytochrome c reduction by heterodimeric SOX variants
To monitor the complete catalytic cycle of SOX, the physiological electron acceptor cytochrome c was used in sulfite-dependent steady state kinetics of the different SOX variants. Other than FeCN, cytochrome c accepts sulfite-derived electrons solely from the heme domain of SOX. Hence, IET within SOX is a prerequisite for steady-state cytochrome c reduction. We found both SOX variants that comprised at least one functional FL subunit displaying approximately 50% activity of the homodimeric WT control (Fig. 2B and Table 1). In contrast, cytochrome c reduction by heterodimeric SOX_C264S/Mo was strongly compromised. By increasing the SOX_C264S/Mo concentration, sulfite-dependent cytochrome c reduction was observed (Fig. 2C). This revealed an approximately 200-fold rate reduction in comparison to WT SOX/SOX (Fig. 2C). This indicates that IET within SOX occurs primarily within one subunit. The observed residual activity may be the result of either (i) an ET in trans within the same SOX heterodimer or (ii) electrons are passed from one heterodimer to the heme domain of another heterodimer via an intermolecular path.
Rapid reaction kinetics of heterodimeric SOX variants
In order to characterize the first ET in SOX, rapid enzyme kinetic measurements were performed using a stopped-flow instrument. Here, we recorded heme reduction rates upon ET from Mo(IV) to the heme iron using 5 μM protein and 300 μM sulfite, thus assuring pseudo–first order reaction kinetics. The reaction was monitored for 10 s and fitted (if not stated otherwise) with a triple-exponential curve with the rapid k1, occurring in about the first 0.1 s of the reaction, representing heme reduction upon the first IET from Mo(IV) (Fig. 3A). For homodimeric WT SOX/SOX, we recorded an IET rate (k1) of 24.8 ± 2.0 s−1 (Fig. 3D and Table 1). Heterodimeric SOX/Mo and SOX/Mo_C264S were found to behave similarly (Fig. 3D and Table 1).
In contrast, and as expected from the steady-state kinetics, ET rates in the SOX_C264S/Mo heterodimer were significantly compromised. Instead of an initial rapid increase, representing an IET from Mo(IV) to heme, heme reduction showed a slow and gradual increase. Accordingly, the measurement period was extended to 300 s and fitted with a double exponential curve (Fig. 3B). Compared with the homodimeric WT SOX/SOX control, k1 was approximately two orders of magnitude reduced (Fig. 3D and Table 1). The compromised heme reduction rate of SOX_C264S/Mo together with the unchanged rate of SOX/Mo_C264S implied that heme reduction preferentially occurs within one subunit. This indicates that the loss of this IET pathway predominantly accounts for the loss of activity seen in the sulfite:cytochrome c steady-state kinetics.
To determine whether residual SOX_C264S/Mo activity could be attributed to an ET between dimers, we first measured heme reduction of WT SOX/SOX using 5 μM enzyme and 300 μM sulfite for 300 s and fitted a quadruple exponential curve (Fig. 3C). The total absorption change of approximately 500 mAU was reached within 150 s with the initial rapid absorption change in the first 0.1 s accounting for a change of approximately 100 mAU. Given that only 35% of SOX/SOX homodimers and 15% of heterodimers in the reaction mixture contained Moco, it seems that the slower rates observed upon the initial fast rate represent reduction of heme cofactors in enzymes lacking Moco (Fig. 4A). This observation implies ET reactions beyond the traditional IET reaction within one subunit of SOX.
To probe intermolecular ET between different dimers, we expressed and purified the inactive SOX variant R217W as homodimeric enzyme (
). The absence of heme reduction in R217W/R217W was confirmed by rapid reaction kinetics (Fig. 3, B and C). As expected, when 2.5 μM SOX/SOX were mixed with 2.5 μM R217W/R217W and excess sulfite, the rate of intial heme reduction dropped by approximately 50% because of the replacement of 50% of the SOX by inactive R217W/R217W (Fig. 3C). However, following incubation time of 300 s, again nearly maximal heme reduction has been reached (Fig. 3C). This finding suggests that SOX/SOX can reduce the heme cofactors of R217W/R217W through interdimer ET.
Next, the experiment was repeated replacing SOX/SOX with homodimeric Mo/Mo lacking both heme domains. Since no heme reduction was recorded for R217W/R217W and Mo/Mo when present alone in the reaction, any heme reduction in a mixture of both must result from Moco-to-heme interdimer ET (Fig. 4E). Indeed, heme reduction was observed when R217W/R217W and Mo/Mo were both present in the reaction. Strikingly, the traces of heme reduction were remarkably similar to those recorded for SOX_C264S/Mo (Fig. 3B). This finding strongly indicates that in SOX_C264S/Mo, ET reactions to heme were also predominantly derived from interdimer reactions from Moco to heme (Fig. 4, D and E).
When comparing all determined ET rates in this study, one can conclude that the highly active variants (SOX/SOX, SOX/Mo, and SOX/Mo_C264S) share two fast rate constants (k1 = 22.2–24.8 s−1 and k2 = 1.8–2.2 s−1), representing the first (from Mo(IV)) and second (from Mo(V)) IET within one subunit of the SOX dimer (Fig. 4, A–C and Table 1, see Discussion section).
The much lower rates that were only observed with very long incubation times (k3 = 0.02–0.07 s−1 and k4 = 0.003–0.02 s−1) were similar to the compromised SOX_C264S/Mo variant and the mixture of R217W/R217W and Mo/Mo (Fig. 3, D–G and Table 1). Those rates can now be attributed to the first (from Mo(IV)) and second (from Mo(V)) intermolecular ET between different SOX dimers (Fig. 4, D and E). In expectation of interdimer third-order and fourth-order reaction kinetics, variations in k3 and k4 might be attributed to changes in availability of oxidized heme and/or Moco saturations that influence the ratio of available heme to Moco within the fast IET pathway.
Taken together, our results demonstrate that both the first and second IETs occur within one subunit of SOX, whereas electron bifurcation, which would involve ET from one subunit to another, can be excluded. Furthermore, we could dissect the complex network of ET in SOX, thereby shedding light onto the individual transfer rates under a single turnover condition of rapid reaction kinetics.
SOX is a homodimeric molybdoheme enzyme. While the roles of the Moco and heme cofactors in detoxification of sulfite and regeneration of the enzyme are well understood, it has remained unclear what role dimerization may play in the catalytic cycle (
). However, to date, the IET-competent conformation remains unknown. Thus, we asked the question whether SOX might adopt a conformation that allows IET between the molybdenum and iron atoms of opposing subunits. For the pathogenic G530D substitution in SOX, the monomeric state has been demonstrated in vitro (
), and we asked the question whether pathogenicity may result from loss of electron bifurcation. This would not only account for the importance of homodimerization but also may explain the different rates observed for the first and second IET, respectively (
The generation of heterodimeric SOX variants consisting of one WT and one truncated subunit allowed—based on our knowledge—for the first time the mechanistic separation of intrasubunit and intersubunit ET steps within the enzyme. By deleting the heme domain in one of the subunits and substituting the cysteine ligand of molybdenum for a serine residue (
), which has been confirmed by our study showing the formation of stable and isolatable SOX dimers with approximately 50% reduced sulfite-dependent FeCN activity.
Collectively, rapid reaction kinetic measurements and sulfite-dependent steady-state reduction of cytochrome c showed that both the first (k1) and second IETs (k2) occur within one subunit of SOX, given that the activity of SOX/Mo_C264S was comparable to SOX/Mo, whereas the activity of SOX_C264S/Mo was compromised by more than two orders of magnitude. Thus, our findings imply that one fully functional subunit is sufficient for catalytic activity, and the differences in rate determined for the first (k1) and second IETs (k2) are not attributed to electron bifurcation within an SOX dimer. The additional study of inactive SOX R217W/R217W with Mo/Mo demonstrated interdimer Mo-to-heme ET.
For SOX/SOX, a k1 of 24.8 ± 2.0 s−1 was determined, which is similar to previously reported values (
). Furthermore, a k2 of 2.09 ± 0.34 s−1 was observed, which is faster than the intermolecular rates k3 and k4 (Fig. 3, E–G). Therefore, this observed k2 value most likely represents the second intrasubunit IET from Mo(V) to heme in fully functional FL SOX (Fig. 4, A–C), requiring ET between individual heme molecules. Assuming that the liberation of such heme sites involves a much slower heme-to-heme ET, the lower rate for the electron derived from Mo(V) is not contradicting direct studies that have reported a much higher rate for the second ET (
When comparing k3 and k4 of the active SOX variants to k1 and k2 of the compromised SOX_C264S/Mo variant, we conclude that the latter represent the slow rates of fully active enzyme and that the increased variability in their value may reflect changes in cofactor saturation. Because of the strict Moco-dependent maturation of SOX in mitochondria, all fully matured enzymes carry Moco in vivo (
). It is for this reason, that in mitochondria sulfite-derived electrons would predominantly follow intrasubunit IET. However, the ability of SOX to liberate electrons to other SOX molecules might potentiate its ability to use alternative electron acceptors, like nitrite, that are limited to the use of the Mo(IV) electron (
). In conclusion, by generation of heterodimeric SOX variants, we were able to uncover that ET of SOX predominantly occurs within one subunit and that one fully functional subunit is sufficient for steady-state formation of sulfate from sulfite. Since SOX is the prototype for the family of Moco-containing enzymes, this study might provide insights into the mechanism of ET within other eukaryotic homodimeric molybdenum-dependent enzymes. For example, in the related plant nitrate reductase, each subunit of the homodimer carries a heme, a flavin adenine dinucleotide, and Moco. Therefore, we propose that in homodimeric Mo enzymes, ET occurs within one subunit.
Materials, plasmids, and bacterial strains
Oligonucleotides for PCR and sequencing were bought from Sigma–Aldrich. Restriction enzymes and T4 DNA ligase were purchased from Thermo Scientific. The pQE-80L and petDUET-1 expression vectors were used and bought from Qiagen and Novagen, respectively. For protein expression, Escherichia coli TP1004 (ΔlacU169, araD139, rpsL150, relA1, flbB, deoC1, ptsF, rbsR, non-9, gyrA219, ΔmobAB) and TP1001 (F-[araD139]B/r Δ(argF-lac)169∗&lambda-e14-flhD5301 Δ(fruK-yeiR)725 (fruA25)‡ relA1rpsL150(strR) rbsR22 Δ(fimB-fimE)632(::IS1) deoC1, ΔmobA) were used (
Site-directed mutagenesis was used to generate the C264S substitution. Two PCR fragments, which harbored the mutation at the 3′ or 5′ end, were produced and subsequently fused by PCR. All inserts were amplified by PCR using primers designed to introduce restriction sites for the target multiple cloning site (MCS). The amplicon and plasmid were digested with restriction enzymes (Thermo Fisher Scientific) and ligated with T4 DNA ligase (Thermo Fisher Scientific) according to the manufacturer. The human Mo domain (A167–P545) and mature FL hSOX (E80–P545), both with and without the C264S substitution, were cloned into MCS1 and MCS2 of petDUET-1, respectively (
). Mature FL hSOX was also cloned into pQE-80L. All constructs were sequenced (GATC Biotech AG) to confirm their identity.
Expression and purification of recombinant proteins
All proteins were expressed in LB media. The pQE-80L plasmid was expressed in TP1004 for 72 h at 18 °C, and the medium was supplemented with 250 μM IPTG and 1 mM sodium molybdate. The petDUET-1 plasmid was expressed in TP1001, the T7 RNA polymerase was stably integrated using the λDE3 lysogenization kit (Novagen), for 18 h at 30 °C, and the medium was supplemented with 25 μM IPTG and 0.5 mM sodium molybdate. Harvested cells were resuspended in lysis buffer (300 mM NaCl, 50 mM Tris(hydroxymethyl)aminomethane acetate salt [Tris–Ac] [pH 8.0]) and treated with lysozyme (VWR; 23,500 U/mg) for 30 min at RT. All further steps were carried out at 4 °C. Cells were disrupted with an Emulsiflex C-5 cell disruptor (Avestin). According to the protein tag, affinity chromatography using HIS-Select Nickel Affinity Gel (Sigma–Aldrich) and/or Strep-Tactin XT Superflow High Capacity resin (iba) was performed according to the manufacturer's instructions. For elution, 250 mM imidazole and 50 mM biotin in buffer W (1 mM EDTA, 150 mM NaCl, and 100 mM Tris/HCl [pH 7.5]) for elution, respectively. Proteins were further purified by SEC using a Superdex 200 column (120 ml; GE Healthcare) equilibrated with lysis buffer. For all proteins, buffer exchange to protein buffer (20 mM NaCl and 50 mM Tris–Ac [pH 8.0]) was performed using a PD10 column (GE Healthcare). The purity of the proteins was confirmed with 12% SDS-PAGE using Coomassie staining, and the dimerization state was confirmed using analytical SEC (Superdex 200, 25 ml; GE Healthcare). Mo domain and SOX R217W were expressed and purified as previously described (
). Briefly, 100 pmol SOX were oxidized to FormA by overnight incubation with acidic Lugol's iodine solution in the dark. After precipitated protein was removed, fresh 1% ascorbic acid was added and buffered to pH 8.3 using Tris. Form A was dephosphorylated to dephospho-form A by treatment with alkaline phosphatase (Roche; 7500 U/ml) and 20 mM MgCl2 for 30 min in the dark. The amount of dephospho FormA was determined with HPLC.
The reaction was carried out at RT in 200 μl 100 mM Tris–Ac (pH 8.0) with 268 μM potassium FeCN(III) (Sigma–Aldrich), 18.75 μg bovine catalase (Sigma–Aldrich), and 5 nM active enzyme. The change in absorption of FeCN upon addition of 300 μM sulfite was monitored at 420 nm with a TECAN Spark microplate reader. For all measurements, three technical replicates were performed and a negative control without enzyme. The initial velocity was calculated from the initial linear increase in absorption and the extinction coefficient of FeCN (1020 l mol−1 cm−1).
Sulfite:cytochrome c reduction
Measurements were performed at RT in 200 μl 100 mM Tris–Ac (pH 8.0). The reaction consisted of 120 μg porcine cytochrome c (Serva), 18.75 μg bovine catalase (Sigma–Aldrich), 1 to 5 nM active enzyme, and 2 to 300 μM sulfite. Alternatively, 1 to 35 nM SOX/SOX or 600 to 2500 nM C264S/Mo were used with 300 μM sulfite. The absorption of cytochrome c was measured at 550 nm using a TECAN Spark microplate reader. Technical triplicates were measured along with one negative control without enzyme. The initial linear increase in absorption and the extinction coefficient of cytochrome c (19,630 l mol−1 cm−1) were used to calculate the initial velocity.
Rapid reaction kinetics
Rapid reaction kinetics were measured anaerobically at 10 °C using an SX20 stopped-flow spectrometer (Applied Photophysics Ltd) as described by Kaczmarek et al. (
). Anaerobic conditions were insured through oxygen scavenging by incubating the device, syringes, and tubing for 1 h with 10 mM d-glucose, glucose oxidase (Sigma–Aldrich), and bovine catalase (Sigma–Aldrich) in 100 mM sodium acetate (pH 5.0). This was removed by washing with anaerobic assay buffer (50 mM Tris–Ac [pH 8.0]). Protein was thawed in a glove box (Coy laboratory products), and enzyme solutions were prepared using anaerobic assay buffer to give a final concentration of 5 μM. When a mix of two enzymes was measured, they were prepared to give a final concentration of 2.5 μM each. Sulfite solutions were prepared anaerobically in the same buffer to give a final concentration of 300 μM. The change in absorption of the cytochrome b5 heme at 423 nm was monitored with a photon multiplier using a path length of 1 cm. Measurements were performed as technical triplicates or quadruplicates. The data were fit as double, triple, or quadruple exponential curves with the Pro-Data SX software, which uses an iterative nonlinear least squares Levenberg–Marquardt algorithm.
Two biological replicates were performed for WT SOX. For each heterodimeric SOX variant, three biological replicates were analyzed, except in sulfite:FeCN reduction. Here, only two biological replicates were analyzed, but the experiment was conducted twice. For kinetic curves, one representative dataset is shown. GraphPad Prism 6 (GraphPad Software, Inc) was used to create figures and calculate kinetic parameters.
Experimental raw data are available to be shared upon request to the corresponding authors.
Conflict of interest
The authors declare that they have no conflicts of interest with the contents of this article.
We thank Monika Laurien, Simona Jansen, Joana Stegemann, and Julia Reich for excellent technical assistance. We thank Laura Vierbaum for initiating the project. This research was funded by grant CMMC2017-C13 from the Center for Molecular Medicine Cologne , Germany.
G. S. and D. B. conceptualization; M. E. and A. T. K. methodology; M. E. validation; A. T. K. and D. B. formal analysis; M. E. and D. B. investigation; A. T. K. data curation; M. E. and A. T. K. writing–original draft; G. S. and D. B. writing–review & editing; M. E. visualization; A. T. K., G. S., and D. B. supervision; G. S. project administration; G. S. funding acquisition.