The radical SAM protein HemW is a heme chaperone

Radical S-adenosylmethionine (SAM) enzymes exist in organisms from all kingdoms of life, and all of these proteins generate an adenosyl radical via the homolytic cleavage of the S–C(5′) bond of SAM. Of particular interest are radical SAM enzymes, such as heme chaperones, that insert heme into respiratory enzymes. For example, heme chaperones insert heme into target proteins but have been studied only for the formation of cytochrome c-type hemoproteins. Here, we report that a radical SAM protein, the heme chaperone HemW from bacteria, is required for the insertion of heme b into respiratory chain enzymes. As other radical SAM proteins, HemW contains three cysteines and one SAM coordinating an [4Fe–4S] cluster, and we observed one heme per subunit of HemW. We found that an intact iron–sulfur cluster was required for HemW dimerization and HemW-catalyzed heme transfer but not for stable heme binding. A bacterial two-hybrid system screen identified bacterioferritins and the heme-containing subunit NarI of the respiratory nitrate reductase NarGHI as proteins that interact with HemW. We also noted that the bacterioferritins potentially serve as heme donors for HemW. Of note, heme that was covalently bound to HemW was actively transferred to a heme-depleted, catalytically inactive nitrate reductase, restoring its nitrate-reducing enzyme activity. Finally, the human HemW orthologue radical SAM domain–containing 1 (RSAD1) stably bound heme. In conclusion, our findings indicate that the radical SAM protein family HemW/RSAD1 is a heme chaperone catalyzing the insertion of heme into hemoproteins.

All have in common the generation of an adenosyl radical via the homolytic cleavage of the S-C(5Ј) bond of SAM. SAM and three cysteine residues generally coordinate a [4Fe-4S] cluster, leading to the typical CX 3 CX 2 C protein sequence signature of radical SAM enzymes (1). The first crystal structure of a radical SAM enzyme was solved for an enzyme of bacterial heme biosynthesis called coproporphyrinogen III dehydrogenase (HemN) (9). Three iron atoms of the [4Fe-4S] cluster of HemN are coordinated by the three cysteine residues Cys 62 , Cys 66 , and Cys 69 of the conserved motif (9,10). A fourth cysteine (Cys 71 ) is not essential for [4Fe-4S] cluster coordination but for catalysis (10). During the catalytic reaction for the conversion of coproporphyrinogen III into protoporphyrinogen IX, the [4Fe-4S] 2ϩ cluster first gets reduced. This leads to the homolytic cleavage of the SAM S-C(5Ј) bond and the formation of a 5Ј-deoxyadenosyl radical. The generated radical then removes stereospecifically one hydrogen atom from a propionate side chain of the substrate to yield 5Ј-deoxyadenosine and a substrate radical, which in turn leads to the desired decarboxylation reaction (9,11). However, the presence of HemN proteins (also named coproporphyrinogen III dehydrogenase (CPDH)) carrying the CX 3 CX 2 CXC motif is limited to a few classes of bacteria (12). Multiple hemN-like genes encoding proteins of significant amino acid sequence homology were found in most classes of organisms with the exception of fungi and were originally annotated as coproporphyrinogen III oxidase.
Recently, the corresponding Lactococccus lactis protein was observed to bind heme and considered to play a role in maturation of the cytochrome oxidoreductase of the bacterium. It was therefore renamed HemW (13). L. lactis HemW (NP_267295.1) displays high homology to Escherichia coli coproporphyrinogen III dehydrogenase HemN (50% amino acid sequence similarity). Surprisingly, L. lactis HemW did not show CPDH activity in vitro and in vivo (13). In contrast to E. coli HemN, L. lactis HemW is missing 47 N-terminal amino acids and the fourth cysteine residue of the conserved CX 3 CX 2 CXC motif (13). E. coli possesses HemN and additionally a HemW-like protein annotated as YggW (NP_417430.1), a protein of hypothetical function. Because of the high degree of amino acid sequence identity of 36% (58% homology) to L. lactis HemW, we renamed YggW to HemW in the present work. Relatedly, the corresponding Pseudomonas aeruginosa protein (WP_003128950) with an amino acid sequence identity of 31% (50% homology) to L. lactis HemW was also renamed to HemW. Similar to L. lactis HemW, the HemWs of E. coli and P. aeruginosa display a truncated N terminus and the conserved cysteine motif lacking the fourth cysteine. A corresponding amino acid sequence alignment is shown in Fig. S1.
The only well characterized systems for the insertion of heme into proteins are the different cytochrome c biogenesis machineries (14). Cytochrome c is involved in multiple electron transport chains. For cytochrome c formation, protoheme IX and the apocytochrome are transported through the membranes of prokaryotes, mitochondria, and chloroplasts. Subsequently, a covalent thioether bond is actively formed between at least one cysteine and a vinyl group of the heme. Currently, five different systems are proposed to perform the processes of heme insertion into a c-type cytochrome; these systems differ in their level of complexity and are found in distinct organisms (14). Sporadically, reports on other heme-binding and potential heme-inserting proteins occur in the literature, as for NikA, an E. coli periplasmic nickel protein (15), or human glyceraldehyde-3phosphate dehydrogenase (16). Furthermore, a putative role for the protein Surf1 of Paracoccus denitrificans as a heme a chaperone involved in cytochrome c oxidase (COX) biogenesis was described (17). Recently, we described the heme-binding protein HemW from L. lactis, hypothesizing that HemW is involved in heme trafficking (13). Here, we provide biochemical, genetic, and biophysical evidence that the bacterial HemW proteins are heme chaperones for the insertion of heme b into enzymes of respiratory chains.

E. coli HemW has no coproporphyrinogen III dehydrogenase activity in vitro and in vivo
E. coli HemN and HemW amino acid sequences are 33% identical but differ in two major features. E. coli HemN carries an extra 46 N-terminal amino acid residues that have been proposed to be crucial for substrate binding (9). Moreover, the fourth cysteine of the HemN CX 3 CX 2 CXC motif is replaced by a phenylalanine in HemW. These differences are found in all HemW-like proteins (13). To investigate whether HemW carries CPDH activity, it was first analyzed in vitro. For this purpose, E. coli HemN and HemW were recombinantly produced and purified to apparent homogeneity. In contrast to HemN (18), E. coli HemW completely failed to catalyze the conversion of coproporphyrinogen III into protoporphyrinogen IX. In a next step, in vivo complementation experiments using the E. coli ⌬hemN strain JKW3838 under anaerobic growth conditions were performed. Due to the presence of the oxygen-dependent coproporphyrinogen III oxidase HemF, this mutant grew efficiently under aerobic conditions (data not shown). However, in the absence of oxygen, a severe growth impairment was detected. Minimal remaining growth might result from fermentative energy generation or residual HemF activity. Clearly, E. coli hemN (pET3-hemN) complemented the ⌬hemN E. coli strain to wildtype-comparable growth, whereas E. coli pGEX-hemW failed to restore anaerobic growth of the mutant (Fig. 1). Both results clearly indicate that E. coli HemW does not harbor coproporphyrinogen III dehydrogenase activity as was also observed for L. lactis HemW (13). These results are consistent with heme auxotrophy reported for a Salmonella enterica serovar Typhimurium hemF/hemN double mutant carrying an intact hemW (19). To test whether the deviating N terminus and the missing fourth cysteine residue were responsible for the observed behavior, we constructed a HemW-F25C protein carrying the fourth cysteine and a HemW-HemN hybrid protein carrying the 46 N-terminal amino acids of HemN fused to HemW-F25C (HemWF25Cϩ46N-term). Nevertheless, HemWF25Cϩ46N-term did not show any coproporphy-  hemN and hemW mutants. A, E. coli hemW does not encode a coproporphyrinogen III dehydrogenase. Cells were grown anaerobically at 37°C in LB medium containing 10 mM NaNO 3 and no antibiotics in tightly sealed anaerobic flasks. Changes in optical density at A 578 were followed spectroscopically. Values for each strain are averages of three independent experiments with three parallel cultures. The growth of wildtype BW25113 (red), E. coli ⌬hemN (green), ⌬hemN with pET-3a-hemN (E. coli hemN) (orange), ⌬hemN with pGEXhemWF25C (blue), and ⌬hemN with pGEXhemWF25Cϩ46N-term (purple) were compared. B, slight growth phenotype of the E. coli hemW mutant. The anaerobic growth in M9 minimal medium supplemented with 5 mM KNO 3 and glycerol as non-fermentable carbon source was compared. Without KNO 3 , almost no growth was observed. Growth with KNO 3 of wildtype E. coli BW25113 (red solid line) and E. coli JW2922 (⌬hemW) (blue solid line) and growth without KNO 3 of E. coli BW25113 (red dashed line) and E. coli JW2922 (⌬hemW) (blue dashed line) were compared. Changes in optical density at A 578 were followed spectroscopically. Values for each strain are averages of six independent experiments with three parallel cultures. Error bars represent S.D.

Novel heme chaperone HemW
rinogen III dehydrogenase activity in vitro. In agreement, no complementation of the E. coli ⌬hemN strain under anaerobic conditions was observed with any other of the HemW variants. Obviously, additional structural elements are required for efficient HemN activity. Consequently, HemW is not an inactivated potential coproporphyrinogen III dehydrogenase.

E. coli HemW binds a [4Fe-4S] cluster
For the biochemical and biophysical characterization, E. coli HemW was produced as a glutathione S-transferase (GST) fusion protein in E. coli BL21 (DE3). After anaerobic chromatographic purification and removal of the GST tag by PreScission protease cleavage, an apparent homogenous protein was obtained. SDS-PAGE analysis revealed a single protein band after staining with Coomassie Blue (Fig. 2A). The protein had a relative molecular weight of ϳ45,000 Ϯ 5,000, which nicely corresponds to the calculated molecular mass for the HemW monomer of 42,584 Da. Approximately 12.5 mg of purified HemW were obtained per liter of culture. To elucidate whether E. coli HemW coordinates an iron-sulfur cluster, the iron and sulfur contents of the protein were determined. For native, purified HemW, no obvious absorption around A 410 -A 425 was detectable in the UV/visible absorption spectrum. Purified HemW exhibited 0.5 mol of iron and 0 mol of sulfur/mol of HemW, and an A 420 :A 280 ratio of 0.04. Consequently, we carried out reconstitution of the obviously labile [Fe-S] cluster via treatment of the protein with iron ammonium citrate and lithium sulfide. After reconstitution, the iron and sulfur content of HemW increased to 3.8 mol of iron and 2.5 mol of sulfur/mol of HemW and an A 420 :A 280 ratio of 0.19. As a consequence, the typical absorbance for [Fe-S] clusters at 420 nm became clearly visible (Fig. 2B). To further characterize the cluster type of HemW, the iron-sulfur cluster of HemW was reconstituted with ammonium [ 57 Fe]ferric citrate, and Mössbauer spectroscopy of 57 Fe-reconstituted HemW was performed. Mössbauer spectra were recorded for samples containing HemW. Spectra without further addition revealed one dominant quadrupole doublet (83% of the total intensity) with an isomer shift (␦) of 0.49 mm/s and a quadrupole splitting parameter (⌬E Q ) of 1.00 mm/s, which are typical of [4Fe-4S] 2ϩ clusters (Fig. 2C, dashed line). Moreover, a second quadrupole doublet (17% of the total intensity) with an isomer shift (␦) of 1.48 mm/s and a quadrupole splitting parameter (⌬E Q ) of 3.30 mm/s was detected (Fig.  2C, dotted line). The solid line in Fig. 2 represents the superposition of the two quadrupole doublets. This spectrum is consistent with a coordination of the iron-sulfur cluster by three cysteine ligands and one potential nitrogen/oxygen ligand. The three cysteine residues are likely the Cys 16 , Cys 20 , and Cys 23 of the CX 3 CX 2 C motif at the N terminus of the protein sequence. The high isomer shift of the second quadrupole doublet excludes an origin from [Fe-S] clusters but reveals high-spin Fe(II) sites with six hard oxygen or nitrogen ligands; the component is therefore assigned to adventitiously bound Fe(II) in the protein, presumably remaining from the reconstitution procedure. However, various attempts to reduce the [4Fe-4S] 2ϩ cluster with different electron donor systems (e.g. sodium dithionite and titanium III citrate with redox mediators) for subsequent EPR analysis failed. In contrast, the [4Fe-4S] cluster of the related E. coli HemN could be reduced at such conditions (18) and used for successful EPR measurements. Surprisingly, cyclic voltammetry measurements clearly indicated a redox transition of the iron-sulfur cluster of HemW at around Ϫ410 mV (Fig. 3). The potential of Ϫ410 mV is in the range of values found for other radical SAM enzymes (20). At this redox potential, both dithionite and titanium III citrate should serve as efficient electron donors for HemW. Obviously, electron donor compounds are prevented from accessing the [4Fe-4S] cluster for reduction; consequently, no radical reaction can be initiated. A similar explanation has been suggested for the [Fe-S] cluster in succinate dehydrogenase subunit B, which appears to be inaccessible for oxidants and toxins (21).

Novel heme chaperone HemW The HemW [4Fe-4S] 2؉ cluster promotes protein dimerization
To study the influence of the iron-sulfur cluster on the oligomerization state of HemW, experiments using size-exclusion chromatography of anaerobically purified and reconstituted HemW and of a HemW variant (C16S/C20S/C23S) lacking the [4Fe-4S] cluster were performed (Fig. 4). For the [4Fe-4S] cluster-containing HemW, two fractions corresponding to monomeric (fraction 16) and dimeric protein (fraction 14) were detected (Fig. 4A). Calibration of the column revealed that fraction 16 for the monomeric protein corresponded to an M r of 45,000 Ϯ 5000 and fraction 14 for the dimeric protein corresponded to an M r of 87,000 Ϯ 6000. Interestingly, an increased amount of iron-sulfur cluster was spectroscopically detected at 420nmforthedimericHemWspeciescomparedwiththemonomeric form (Fig. 4A, dashed line). Nevertheless, iron-sulfur clusters were also detected in monomeric HemW, which indicated a dynamic transition between monomeric and dimeric HemW. This transition between monomeric and dimeric proteins was tested by rechromatography of the collected separated monomeric and dimeric HemWs in fraction 14 and fraction 16 on the gel filtration column. We observed that rechromatography of the dimeric HemW in fraction 14 resulted again in two equal-sized protein absorption peaks in fractions 14 and 16 (Fig. 4C). However, most iron-sulfur cluster absorption was detected for the dimeric protein in fraction 14. Analogously, rechromatography of monomeric protein in fraction 16 also generated two absorption maxima in fractions 14 and 16; however, the bigger peak in fraction 16 represented the monomeric protein (Fig. 4D). An analytical gel filtration analysis of HemW without [Fe-S] cluster, caused by the replacement of cysteines 16, 20, and 23 with serines in the iron-sulfur cluster binding motif, revealed only one single peak in fraction 16 corresponding to a monomeric protein (Fig. 4B). The HemW-C16S/C20S/C23S variant was subjected to iron-sulfur cluster reconstitution experiment analogously to the wildtype protein prior to these experiments but remained iron-sulfur clusterfree as determined spectroscopically and by iron and sulfur determinations. Obviously, an equilibrium exists between the monomeric and dimeric protein; however, formation of a dimeric HemW is favored by the incorporation of the ironsulfur cluster.

E. coli HemW binds SAM
The amino acid sequence analysis of E. coli HemW clearly revealed two binding sites for SAM similar to the radical SAM enzyme HemN from E. coli. Overall, in E. coli HemN, 19 amino acid residues are known from the crystal structure of the protein to coordinate two SAM molecules and one [4Fe-4S] cluster. Of the involved 19 HemN amino residues, 11 (Arg 184 , Gly 113 , Thr 114 , Cys 66 , Cys 62 , Cys 69 , Gln 172 , Asp 209 , Tyr 56 , Gly 112 , and Glu 145 , clockwise around the binding site (Fig. S2)) were found to be identical in HemW (Arg 138 , Gly 67 , Thr 68 , Cys 20 , Cys 16 , Cys 23 , Gln 126 , Asp 163 , Tyr 10 , Gly 66 , and Glu 95 ) and three were found to be homologous (Ile 211 in HemN and Met 165 in HemW, Phe 240 in HemN and Tyr 194 in HemW, and Phe 68 in HemN and Tyr 22 in HemW) when the amino acid sequences of the proteins were aligned. Due to the low amino acid sequence conservation at the C terminus of both proteins, an alignment of four amino acid residues of HemN (Ala 243 , Ala 242 , Phe 310 , and Ile 329 ) did not match the corresponding residues of HemW. This region of HemN is involved in coproporphyrinogen III coordination and most likely represents the heme-binding region of HemW. Only the cysteine (Cys 71 ) and the following residue (Gly 70 ) used to clearly differentiate HemN from HemW proteins were clearly different (Phe 25 and Asp 24 ). For experimentally studying SAM binding of HemW, the purified reconstituted protein was incubated with [ 14 C]SAM, and the mixture was passed over a desalting column for the removal of nonincorporated free SAM. Protein-bound [ 14 C]SAM was subsequently quantified using liquid scintillation counting. The control experiment was carried out using BSA and [ 14 6 -14). Consequently, SAM binding to HemW was clearly demonstrated. The highly conserved structure of the SAM and [4Fe-4S]-cluster binding site suggested the presence of two SAM molecules. However, due to the unknown amount of already bound SAM in the tested HemW proteins and the unknown exchange rate between bound and unbound SAM, it was not possible to determine the stoichiometry of SAM binding to HemW.

Analysis of the SAM cleavage capacity of HemW
The classical radical SAM enzyme chemistry requires the reduction of the [4Fe-4S] cluster and homolytic cleavage of the S-C(5Ј) bond of SAM with the generation of the 5Ј-deoxyadenosyl radical. HemN usually requires its substrate coproporphyrinogen III for radical formation (10). However, in the absence of the substrate, residual enzymatic SAM cleavage of usually less than 15% of the reaction with substrate was observed. To test HemW for full or residual SAM cleavage activity, reconstituted HemW protein was incubated with SAM with and without heme as a potential substrate under reducing conditions. The disappearance of SAM with the parallel forma- Novel heme chaperone HemW tion of deoxyadenosine was monitored by HPLC analysis (Fig.  5B, solid line). The same experiment was performed as a control with purified HemN in the absence and presence of substrate (Fig. 5B, dashed and dotted line). In this case, under tested conditions E. coli HemN revealed full SAM cleavage activity in the presence of the substrate coproporphyrinogen III (Fig. 5B, dotted line) and less than 5% of its SAM cleavage activity without substrate (Fig. 5B, dashed line). Comparable residual SAM cleavage capacity was observed for HemW with the addition (Fig. 5B, solid line) and without the addition of heme (not shown). Clearly, E. coli HemW revealed only the residual SAM cleavage activity comparable with HemN without substrate (Fig. 5B).

HemW is a heme-binding protein
Previous studies with the HemW homolog from L. lactis revealed heme binding to the protein (13). To test for heme binding of E. coli HemW and determine its specificity, the purified HemW protein and heme were incubated anaerobically overnight and analyzed spectrophotometrically (Fig. 6A, dotted line). As a control, the protein solution used and free heme were analyzed in parallel. HemW showed only the typical protein absorption at A 280 and little absorption for the iron-sulfur cluster (Fig. 6A, solid line). Free heme showed the typical spectrum with peaks around A 400 and A 580 (Fig. 6A, dashed line). The HemW-heme complex revealed a broad absorption peak between A 380 and A 420 besides the protein absorbance at A 280 (Fig. 6A, dotted line). To determine the specificity of heme binding, all three samples (HemW, heme, and HemW-heme complex) were subjected to extensive dialysis overnight and subsequent spectroscopic analyses. Although the spectrum for HemW did not change (Fig. 6B, solid line) and the spectrum for the HemW-heme complex only lost its small increase around A 400 (Fig. 6B, dotted line), all free heme was gone (Fig. 6B, dashed line). Identical results were obtained for the dialyzed and Superdex 200 gel-filtrated HemW-heme complex (data not shown). Interestingly, reduction ot the HemW-heme complex resulted in an increase of absorption at A 424 , generating a Soret band, and further absorption peaks at A 531 and A 559 (Fig.  6C, dashed line). In contrast, free reduced heme shows an absorption peak at around A 400 . These results demonstrate the specificity of HemW-heme interaction.
For further analysis of the nature and stoichiometry of HemW-heme interaction, complexes were analyzed via SDS-PAGE with heme staining and acidified butanone extraction. HemW and equimolar amounts of heme were incubated overnight and subjected in duplicate to SDS-PAGE analyses. Subsequently, one half of the gel was stained with Coomassie Brilliant Blue for detection of separated proteins (Fig. 7A, lane 1),

Novel heme chaperone HemW
whereas the proteins on the second half of the gel were blotted onto a nitrocellulose membrane for heme staining. The detection of the HemW-bound heme was based on its intrinsic peroxidase activity by incubation with the ECL reagent (Fig. 7A,  lane 2). The observed heme staining of E. coli HemW indicated stably bound heme. To obtain further evidence for the possible covalent nature of heme binding, butanone extraction experiments, which can result either in release of non-covalently linked heme in the organic phase or in still bound, covalently linked heme in the aqueous phase (22), were performed. Cytochrome c was used as a positive control, clearly indicating the presence of covalently bound heme in the aqueous phase (Fig.  7B, middle tube). Hemoglobin with non-covalently bound heme was used as a negative control. Here, almost all heme was extracted in the upper organic phase (Fig. 7B, right tube). Butanone extraction of HemW incubated with heme revealed a completely clear upper phase and a slightly brownish lower phase (Fig. 7B, left tube). The presence of HemW-derived heme in the lower aqueous phase indicated covalently bound heme. Although our experiment pointed toward covalently bound heme, strong binding of the heme in a tight hydrophobic pocket resistant to SDS and butanone treatment cannot be excluded. Subsequently, the heme staining assay was used to identify the binding stoichiometry of HemW and heme. For this purpose, a solution of 10 M HemW (Fig. 7C, lanes 1-5) was titrated with heme in increasing concentrations from 5 to 25 M (Fig. 7C,  lanes 6 -10). The subsequent heme staining revealed an increase of the heme bound to HemW up to a heme concentration of 10 M, indicating that apparent saturation of the signal occurred after addition of equimolar amounts of heme (Fig. 7C,  lanes 6 -10). Alternatively, the stoichiometry of heme binding by HemW was determined spectroscopically by using 20 M native HemW titrated with increasing amounts of heme. Measurements of the optical density at 416 nm revealed heme binding saturation at 20 M (Fig. 7D). These results clearly indicate a specific binding with a stoichiometry of one molecule of heme per HemW monomer.

Heme binding is independent of the presence of the iron-sulfur cluster
Aerobically prepared HemW without iron-sulfur cluster bound heme as efficiently as anaerobically prepared HemW with the cluster. Similarly, the HemW triple mutant (HemW-C16S/C20S/C23S) also bound heme with high efficiency. In agreement, the Mössbauer spectrum of HemW supplemented with non-enriched heme (natural isotope distribution, only 2.2% 57 Fe) showed the same spectrum and identical fit parameters as HemW without further additions. The presence of heme only slightly changed the observed redox potential of the iron-sulfur cluster to around Ϫ410 mV (Fig. 3). These results demonstrate that the iron-sulfur cluster was not affected by heme binding. Vice versa, the iron-sulfur cluster did not influence heme binding by HemW. Moreover, the presence of SAM did not change heme binding of HemW.

Respiratory nitrate reductase and bacterioferritin are interaction partners of HemW
To determine specific targets for the potential heme chaperone HemW, multiple interaction partners were tested using the bacterial adenylate cyclase two-hybrid (BACTH) system in P. aeruginosa. The P. aeruginosa system was used because high background noise levels, which obscured the results, were observed for similar experiments in E. coli. In this study, the hemoenzymes bacterial ferritin BfrA; bacterioferritin BfrB; catalase KatA; the last enzyme of heme biosynthesis ferrochelatase, HemH; and the heme-containing subunit of the respiratory nitrate reductase, NarI, were analyzed for their interaction with HemW. The reasons for testing BfrA and -B as well KatA and ferrochelatase HemH are obvious because all proteins are heme-binding/storing enzymes. The consecutive ␤-galactosidase assays revealed the highest Miller units for the combination of HemW with NarI, indicating their strong affinity. Furthermore, HemW interacted with BfrA and BfrB but not with HemH or KatA. In agreement, in the inverse experiment, neither HemH nor KatA interacted with HemW (Fig. 8). coli HemN with its substrate coproporphyrinogen III (dotted line). After addition of 0.6 mM dithionite as a potential electron donor, 0.6 mM SAM was added, and the mixture was incubated. The reaction was stopped with formic acid. Samples were chromatographically separated on a Hypercarb column with appropriate marker substances. SAM (indicated with a dot) and formed 5Ј-deoxyadenosine (indicated with a star) were detected at 254 nm. Background controls without protein or BSA did not yield the 5Ј-deoxyadenosine-specific peak. HemN in the presence of substrate revealed full SAM cleavage. Without substrate, 5% residual SAM cleavage was observed for HemN. HemW with heme revealed comparable residual SAM cleavage activity. To emphasize the changes at wavelengths from A 500 to A 650 , these parts of the spectra were enlarged. mAU, milliabsorbance units.

Novel heme chaperone HemW HemW transfers heme to heme-depleted quinol-nitrate oxidoreductase NarGHI
The strongest interaction of HemW was found with the heme-containing subunit NarI of the respiratory nitrate oxidoreductase NarGHI. Under anaerobic conditions and the presence of nitrate, E. coli utilizes this enzyme for energy generation by replacing oxygen with nitrate as the terminal electron acceptor. This respiratory complex has the ability to use all three natural quinones for energy generation (23,24). For our approach, ubiquinol served as electron donor, and the membrane-anchored subunit NarI provides the quinol-binding and oxidation site. Two low-spin hemes (b H and b L ) involved in the electron transfer from quinols to the subunit NarG are coordinated by NarI (25). The presence of both hemes was demonstrated to be essential for the oxidation of quinols and the overall activity of NarGHI as deduced from analysis of NarI variants having lost either heme b H or b L (26). To unambiguously identify HemW as a true heme chaperone, heme transfer analyses from HemW to the quinol-nitrate oxidoreductase NarGHI from E. coli were performed. Thus, Nar-enriched membrane vesicles from E. coli wildtype MC4100 and the heme-deficient E. coli ⌬hemA were prepared. The absence of heme in the heme-depleted membrane vesicles due to the ⌬hemA gene mutation was obvious by the visible change in color of the membrane preparation (Fig. 9B) and the corresponding UVvisible spectra (Fig. 9A). The enzymatic activity of heme-depleted nitrate oxidoreductase was first tested spectrophotometrically using a quinol analog as electron donor (27). A classical spectrophotometric activity assay for the nitrate reductase NarGHI was used. This assay is based on the absorption changes of the artificial electron donor, the menaquinol analog 2-methyl-1,4-naphtoquinol (menadiol), used. The oxi- HCl. Ice-cold 2-butanone was added and carefully mixed. The heme bound to protein found in the lower aqueous phase indicated the covalent binding of heme to HemW (B, left tube). Butanone extraction from cytochrome c with covalently bound heme (horse heart; 1 mg/ml) served as a positive control (B, middle tube), and hemoglobin with non-covalently bound heme served as a negative control (B, right tube). C, heme stoichiometry of HemW binding. For heme staining, purified E. coli HemW (10 M) was titrated with increasing amounts of heme (5,10,15,20, and 25 M. Following SDS-PAGE, the gel was stained with Coomassie Brilliant Blue for protein visualization (lanes [1][2][3][4][5], and the blotted membrane was stained with ECL reagent for bound heme (lanes 5-10). M, marker with proteins of known relative molecular masses. For spectroscopic determination of heme binding, 20 M HemW was incubated with different amounts of heme. D, titration curve for the determination of the stoichiometry of HemW and heme. Absorption was measured at A 416 and plotted against increasing concentrations of heme. The intersection of the two linear slopes indicates the saturation of heme at ϳ20 M heme. Binding of 1 mol of heme/mol of HemW monomer was concluded.

Novel heme chaperone HemW
dation from menadiol to menadione during the reduction of nitrate to nitrite was followed spectroscopically at 260 nm. As mentioned above, only heme-containing NarGHI can catalyze menadiol oxidation in the presence of nitrate. Results of the heme transfer to the NarGHI are summarized in box plots in Fig. 9C. Membrane vesicles prepared from E. coli wildtype MC4100 served as a positive control in this experiment (Fig. 9C,  lane A). Activity assays of membrane vesicles containing hemedepleted quinol-nitrate oxidoreductase revealed only low residual enzymatic activity (Fig. 9C, lane B). However, addition of HemW preincubated with heme led to a significant nitrate reductase activity (Fig. 9C, lane C). Addition of NADH further increased the observed enzyme activity (Fig. 9C, lane D). Over 50% of the potential nitrate reductase activity was restored. Full restoration might be hampered by the possible instability of the heme-free apoenzyme compared with the nitrate reductase holoenzyme. In the negative controls, E. coli ⌬hemA mutant membrane vesicles with the addition of solely heme or apo-HemW exhibited residual background activity identical to the results of the heme-free nitrate reductase (Fig. 9C, lanes E and  F). Similarly, the combination of heme, HemW, and NADH did not react with the nitrate reductase substrate used (Fig. 9C, lane  H). SAM did not influence the heme transfer reaction. To analyze the influence of the [4Fe-4S] cluster on heme transfer, a HemW triple mutant (HemW-C16S/C20S/C23S) lacking the cluster was tested. The activity of NarGHI was not restored by the triple-mutant HemW (Fig. 9C, lane G), whereas the heme binding behavior of the mutant enzyme was comparable with wildtype HemW. The observed catalytic activity of the nitrate oxidoreductase in the membrane vesicles prepared from the heme-deficient E. coli ⌬hemA mutant after addition of hemeloaded HemW clearly indicates successful heme transfer from HemW to the heme-requiring NarI subunit. These results support the function of HemW as a heme chaperone.

Slight growth phenotype of the E. coli hemW mutant
In light of the heme chaperone function of HemW for the respiratory nitrate reductase NarGHI, growth experiments with wildtype and a hemW mutant under anaerobic, nitrate respiratory conditions with the non-fermentable carbon source glycerol were performed. Under the tested growth conditions, the hemW mutant showed a slight, but highly reproducible growth phenotype (Fig. 1B, blue line). In the absence of nitrate, almost no growth was observed. The observed growth of the hemW mutant indicated the presence of intact nitrate reductase and a second heme-inserting system supplementing for the inactivated HemW. Backup systems for essential functions were observed in E. coli for catalases, ribonucleotide reductases, and pyruvate kinases to name a few.

Human HemW homologue RSAD1 binds heme
RSAD1 from Homo sapiens (30% amino acid sequence identity, 50% homology) was analyzed for heme binding via its recombinant production in E. coli, affinity purification, and heme staining (Fig. 10A). Clearly, heme bound strongly to human RSAD1. Additionally, a typical absorption spectrum was also recorded for the H. sapiens RSAD1-heme complex (Fig. 10B). Altogether, the experiments for human RSAD1 confirmed the results for bacterial HemW, suggesting the ubiquitous function of HemW/RSAD1 as a heme chaperone.

Discussion
HemW protein inserts heme into proteins with different functions in respiration. Genes encoding HemW/RSAD1 are found in the genomes of almost all organisms (bacteria, archaea, plants, and animals) with the noticeable exception of fungi. The presence of hemW genes nicely correlates with the utilization of heme-dependent aerobic and anaerobic respiration. Consequently, hemW is even found in organisms that utilize heme taken up from the environment for this process such as Lactococcus (13). In contrast, strict fermentative organisms such as Clostridium, deficient in heme biosynthesis and heme uptake due to the absence of classical heme-dependent respiratory processes, also lack HemW. How does this radical SAM protein-based heme chaperone work? As shown in the model in Fig. 11, heme can be derived from biosynthesis and heme import. Interestingly, no stable complex formation between HemW and ferrochelatase (HemH), the last enzyme of the heme biosynthesis, was observed (28). However, heme from HemW was found to interact with BfrA and BfrB. In 1999, Hassett and co-workers (29) reported that P. aeruginosa catalase A (KatA) requires bacterial ferritin A (BfrA) for full activity. They proposed that BfrA not only stores iron for incorporation into heme but also the necessary prosthetic heme group of KatA. Consequently, BfrA could function as a heme transporter between HemH, the last enzyme of heme biosynthesis, and the heme-accepting protein KatA.

Novel heme chaperone HemW
N-terminal radical SAM element of both proteins. No common heme-binding domain was detected.
Based on these observations and other data, a model for HemW (depicted in Fig. 11) was deduced. Heme produced in heme biosynthesis is transferred via bacterioferritin to HemW where it is covalently bound. In the presence of the [4Fe-4S] cluster, HemW dimerizes, becomes localized at the membrane, and interacts with its target NarI. Transfer of the heme requires an intact [4Fe-4S] cluster and might involve radical chemistry. The exact mechanism of the NADH stimulation of heme transfer remains to be determined Finally, the human RSAD1 protein was found to bind heme tightly, indicating the general importance of the HemW/RSAD1 enzyme family for the heme insertion into cellular proteins. Future experiments will focus on the biochemistry of the heme release from HemW to target proteins and the role of the radical chemistry in that process. Figure 9. HemW-mediated heme transfer to the nitrate oxidoreductase NarGHI. A, comparative spectroscopic analysis of prepared membrane vesicles from E. coli wildtype MC4100 and the corresponding E. coli ⌬hemA mutant. The recorded absorption spectrum at around A 425 indicated the absence of bound heme cofactor in NarGHI produced by E. coli ⌬hemA mutant (dashed line) in contrast to the spectrum recorded for the identical membrane vesicle preparation from wildtype E. coli. B, the decolorization of the prepared membrane vesicles (left tube, wildtype; right tube, ⌬hemA mutant) due to the depletion of heme is optically visible. C, enzyme assays were performed with membrane vesicles isolated from E. coli MC4100/pVA700 overexpressing narGHJI (labeled MC4100) and membrane vesicles with overproduced heme-depleted nitrate oxidoreductase isolated from E. coli ⌬hemA/pVA700 (labeled ⌬hemA SHSP18). The hemedepleted nitrate oxidoreductase was incubated with HemW-heme (C), HemW-heme ϩ NADH (D), and [4Fe-4S] clusterless HemW-C16S/C20S/C23S-heme (G) and as negative controls solely with heme (E), HemW (F), or HemW and NADH (H), respectively. 20 mM 2-methyl-1,4-naphtoquinol (menadiol) served as electron donor, 2 mM nitrate served as electron acceptor, and 5 mM NADH and 1.5 mM free heme were used where indicated. The range between Ϫ1 and 18 mol/min ϫ ml is shown. Error bars represent S.D.

Primers, strains, and plasmids
The primers, strains, and plasmids used in this study are listed in Tables S1-S3.

Cloning, expression, and purification of E. coli hemW
The hemW gene was PCR-amplified from E. coli genomic DNA using the primers hemW E.c. -pGEX-for and hemW E.c. -pGEX-rev harboring a BamHI and a XhoI restriction site (underlined in Table S1), respectively. The 1137-bp PCR product was cloned into the respectively restricted vector pGEX-6P-1 according to the manufacturer's instructions, yielding the plasmid pGEX-HemW E.c. . The vector encoded HemW with an N-terminal GST tag and a cleavage site for PreScission protease (GE Healthcare). GST-HemW was produced in E. coli BL21 (DE3) with the help of the plasmid pGEX-HemW E.c. . Two-liter cultures were grown aerobically in LB medium and 100 mg/ml ampicillin at 200 rpm at 37°C. hemW gene expression was induced at an attenuance at A 578 of 0.6 by addition of 0.5 mM isopropyl ␤-D-thiogalactoside, and cultivation was continued overnight at 17°C at 200 rpm. Cells were harvested by centrifugation at 4000 ϫ g for 15 min at 4°C. All following steps occurred under strict anaerobic conditions at 20°C. For HemW purification, cells were resuspended in 10 ml of buffer 1 (140 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , 5% glycerol, 1 mM DTT, pH 7.4), and cells were disrupted by a single passage through a French press at 19,200 p.s.i. Cell debris and insoluble proteins were removed by centrifugation for 60 min at 25,000 ϫ g at 4°C. The soluble protein fraction was loaded onto a glutathione-Sepharose column (Macherey-Nagel, Düren, Germany). HemW was liberated from the column by cleavage of the GST tag overnight with PreScission protease (GE Healthcare) according to the manufacturer's instructions. HemW-containing fractions were eluted, pooled, and concentrated by ultrafiltration using an Amicon membrane with a 30-kDa molecular mass cutoff (Merck Millipore, Billerica, MA). Protein concentrations were determined with a colorimetric assay using Bradford reagent with bovine serum albumin as standard according to the manufacturer's instructions. (Sigma-Aldrich). The triple-mutant hemWC16S/C20S/C23S was constructed using the Q5 site-directed mutagenesis kit (New England Biolabs, Frankfurt, Germany) according to the manufacturer's instructions. Successful construction of mutations was confirmed by DNA sequencing of the complete hemW gene variant. Production and purification of the HemW variant were performed analogously to wildtype HemW.

Absorption spectroscopy
UV-visible absorption spectra of HemW and HemW-heme complexes were recorded on a Jasco V-650 spectrophotometer in buffer 1 using the same buffer as a blank. The recording wavelengths were from A 250 to A 600 . Heme (10 mg) was dissolved in 1 ml of 0.1 M NaOH and incubated at room temperature for 1 h. After addition of 1 ml of Tris-HCl (1 M, pH 7.6) the solution was centrifuged at 12,100 ϫ g at room temperature for 10 min. The supernatant was filtered to remove insoluble residues, and the concentration was determined as A 385 X/58.44 ϭ x ϫ 500 ϭ x mmol/liter.

In vitro iron-sulfur cluster analysis
The in vitro reconstitution of [Fe-S] clusters was performed as described previously (31). After reconstitution of the [Fe-S] cluster, the excess iron and sulfide was removed by centrifugation at 12,100 ϫ g at 4°C and subsequent passage of the protein solution through a NAP-25 column (GE Healthcare) according to the manufacturer's instructions. The iron content of purified HemW was determined according to a protocol described elsewhere (32). After denaturation of the protein with 1 M perchloric acid, bathophenanthroline was used as the chelating reagent. The sulfur content was determined as described previously (33).

Mössbauer spectroscopy
The final HemW concentration used for Mössbauer spectroscopy analysis of E. coli HemW was 350 M. Sample preparation was performed under strict anaerobic conditions. The iron-sulfur cluster of HemW was reconstituted with ammonium [ 57 Fe]ferric citrate. For the sample containing HemW supplemented with heme, an equimolar ratio of heme to HemW was added, and the mixture was incubated overnight at 20°C. The solutions were transferred to 350-l Mössbauer  lanes 1 and 3) and E. coli (lanes 2 and 4) were incubated with heme overnight and separated via SDS-PAGE. The proteins were stained with InstantBlue TM (Expedeon Inc., San Diego, CA) (lanes 1 and 2) or blotted onto nitrocellulose membrane with subsequent ECL treatment for heme staining (lanes 3 and 4). A marker indicates the relative molecular mass of the separated proteins. B, UV-visible spectra of H. sapiens RSAD1 incubated with heme overnight. An absorption maximum at 410 nm indicated heme binding. Peaks at 531 and 556 nm are typical for incorporated heme.

Novel heme chaperone HemW
cups and frozen in liquid nitrogen. Mössbauer spectra were recorded on a spectrometer with alternating constant acceleration of the ␥-source. The minimum experimental line width was 0.24 mm/s (full width at half-height). The sample temperature was maintained constant in an Oxford Instruments Variox cryostat, whereas the 57 Co/Rh source (1.8 GBq) was kept at room temperature. Isomer shifts are quoted relative to iron metal at 300 K.

Cyclic voltammetry
Cyclic voltammetry measurements were performed using an Ametek Versastat 3. The measurements were carried out in a self-made anaerobic three-electrode electrochemical cell flushed with nitrogen. As the reference, a silver/silver chloride electrode was used (3 mol liter Ϫ1 KCl). All potentials in the text and figures are given versus normal hydrogen electrode (ϩ210 mV). A platinum wire was used as the counterelectrode with glassy carbon as the working electrode. Before each measurement, the platinum wire was annealed in a natural gas flame, and the glassy carbon electrode was pretreated in nitric acid, neutralized, polished with 0.05-m alumina, and annealed again in a natural gas flame. For each experiment, 20 cycles were recorded. The potential slightly drifted only over the first 10 cycles and stabilized thereafter. In this work, only the stabilized potential is discussed. The cycles were recorded with a scan rate of 1 V s Ϫ1 . All electrochemical experiments were carried out at ambient temperature in a 100-l drop. Samples contained 120 M HemW, 120 M heme, or 500 M S-adenosylmethionine (Sigma-Aldrich) in diverse combinations. Samples were prepared under anaerobic conditions in a glove box (Coy Laboratories) and transferred into HPLC vials before injecting into the cyclic voltammetry chamber directly on the glassy carbon electrode.

SAM binding and cleavage analyses
The SAM binding assays were performed as described previously (34). For this purpose, 100 M purified HemW or BSA was incubated with 0.5 Ci of S-[carboxyl- 14  For SAM cleavage, 25 M purified HemW (free or loaded with heme) was incubated with 0.6 mM sodium dithionite as potential electron donor and 0.6 mM SAM overnight at 17°C under anaerobic conditions. Reactions were stopped by adding 5% formic acid. For HPLC analysis, the samples were centrifuged at 16,100 ϫ g for 10 min. HPLC analysis was performed as described previously (20). In detail, for the separation of 5Ј-deoxyadenosine from SAM, a Hypercarb column (Thermo Fisher Scientific, Waltham, MA) in a Jasco 2000 system with a flow rate of 0.2 ml/min was used at room temperature. A 5-ml gradient of 0.1% TFA in H 2 O and 0.08% TFA in acetonitrile was applied. SAM and 5Ј-deoxyadenosine were detected at A 254 . Appropriate markers were used to calibrate the column.

Determination of the native molecular mass
An ÄKTA purifier system for gel permeation chromatography with a Superdex 200 HR 10/300 column was used (GE Healthcare). The column was equilibrated using buffer 1 and calibrated using carbonic anhydrase (M r ϭ 9,000), bovine serum albumin (M r ϭ 66,200), yeast alcohol dehydrogenase (M r ϭ 150,000), and ␤-amylase (M r ϭ 200,000) as marker proteins. A sample containing purified recombinant HemW was chromatographed under identical conditions with a flow rate of 0.25 ml/min under anaerobic conditions.

Heme binding assays
Heme (1 mg) was dissolved in 100 l of 100 mM NaOH and thoroughly mixed (34). After 30 min, 100 l of 1 M Tris, pH 7.4, was added. The solution was centrifuged at 4°C for 10 min at 12,100 ϫ g. The concentration was determined using ⑀ 385 ϭ 58.44 (mM cm) Ϫ1 . HemW was incubated with an equimolar concentration of heme under anaerobic conditions at 20°C overnight, and the formed complex was used for further analyses including spectroscopy, heme staining, and the in vitro transfer reaction to NarGHI.
For detection of the stable HemW-heme complex, the hemecomplexed protein was separated via SDS-PAGE followed by Figure 11. Working model of E. coli HemW. Heme from heme biosynthesis is transferred via the bacterioferritin Bfr (gray) as a carrier to the heme chaperone HemW (orange). Dimerized, the [4Fe-4S] cluster-containing HemW localizes to the membrane where it interacts with its target protein, NarI (yellow), a subunit of the respiratory nitrate reductase NarGHI. After heme incorporation into apo-NarI, the holo-NarGHI catalyzes the reduction of nitrate to nitrite. PPO is protoporphyrinogen IX oxidase; FC (HemH) is ferrochelatase.