Identification of a hemerythrin-like domain in a P_1B-type transport ATPase

Abstract

The P_1B-type ATPases couple the energy of ATP hydrolysis to metal ion translocation across cell membranes. Important for prokaryotic metal resistance and essential metal distribution in eukaryotes, P_1B-ATPases are divided into subclasses on the basis of their metal substrate specificities. Sequence analysis of putative P_1B-5-ATPases, for which the substrate has not been identified, led to the discovery of a C-terminal soluble domain homologous to hemerythrin (Hr) proteins and domains. The Hr domain from the Acidothermus cellulolyticus P_1B-5-ATPase was cloned, expressed, and purified (P_1B-5-Hr). P_1B-5-Hr binds two iron ions per monomer and adopts a predominantly helical fold. Optical absorption features of the iron-loaded and azide-treated protein are consistent with features observed for other Hr proteins. Autooxidation to the met form is very rapid, as reported for other prokaryotic Hr domains. The presence of a diiron center was confirmed by electron paramagnetic resonance (EPR) and X-ray absorption spectroscopic (XAS) data. The occurrence of a Hr-like domain in a P-type ATPase is unprecedented and suggests new regulatory mechanisms as well as an expanded function for Hr proteins in biology.

The P-type ATPases are a large family of integral membrane proteins that couple the energy of ATP hydrolysis to the transport of cations across cell membranes (1). Named for the formation of a phosphorylated intermediate during catalysis, P-type ATPases are classified on the basis of their substrate specificities and include the Ca²⁺, Na⁺/K⁺, and H⁺-ATPases (2). Members of the P_1B subgroup transport transition metal ions, including Zn²⁺/Cd²⁺/Pb²⁺ (3-5), Cu²⁺ (6), Cu⁺/Ag⁺ (7), and Co²⁺ (8). Widely distributed in nature, the P_1B-ATPases confer heavy metal tolerance to microorganisms (9) and are essential for the absorption, distribution, and bioaccumulation of metal micronutrients by cyanobacteria (10) and eukaryotes (11, 12). In humans, mutations in the Cu⁺ ATPases ATP7A and ATP7B lead to Menkes syndrome and Wilson disease, respectively (13).

All P_1B-ATPases have the same core architecture consisting of at least six transmembrane (TM)¹ helices, a soluble ATP binding domain (ATPBD), and a soluble actuator domain (A-domain), both located in the cytoplasm (Figure 1). The ATPBD includes the nucleotide binding site and a conserved DKTGT motif, of which the aspartic acid residue is phosphorylated in the catalytic cycle. Residues in the last three TM helices are proposed to form the metal binding site(s), and specificity is proposed to derive from the identities and positions of coordinating residues within these helices (14, 15). In particular, a three-residue cysteine-containing sequence motif in TM helix 4 or 6 is important for metal transport activity (16, 17). On the basis of sequence analysis and experimental evidence, the P_1B-ATPases have been further divided into five substrate-specific subfamilies, designated P_1B-1 through P_1B-5 (14, 15).

Figure 1.

Architecture and sequences of P_1B-5-ATPases. Left: Overall topology, including a novel Hr-like C-terminal soluble domain. Conserved residues in TM helices are shown in approximate locations. Right: Non-continuous alignments of TM helix 4 and the start of the ATPBD (top) and TM helices 5 and 6 (bottom) of ten representative P_1B-5-ATPases. Generally conserved residues that may play a role in substrate specificity are colored with cysteine in pink, proline in yellow, serine in cyan, threonine in blue, glutamine in green, aspartate in violet, glutamate in orange, and methionine in brown. The phosphorylation site in the ATPBD is highlighted in lavender. The helices below the sequences indicate putative TM helices. The sequences correspond to the following proteins: L. crispatus, GI 256849544, UnitProtKB C7Y4W5; L. jensenii, GI 260665179, UniProtKB D0DPP6; E. faecium, GI 69245746, UniProtKB C9CF69, L. lactis, GI 15672077, UniProtKB Q9CJA5; B. angulatum, GI 229816919, UniProtKB C4FCZ1; C. urealyticum, GI 172040330, UniProtKB B1VFS1;Arthrobacter spFB24, GI 116668908, UniProtKB A0JRS1; B. multivorans, GI 161522885, UniProtKB A9ASK0; M. chloromethanicum, GI 218529299, UniProtKB B7KQY2; A. cellulolyticus, GI 117927237, UniProtKB A0LQU2.

Most P_1B-ATPases include two additional TM helices as well as one or more soluble cytoplasmic metal binding domains (MBDs) at the N- and/or C-termini. Many MBDs from P_1B-1- and P_1B-2-ATPases exhibit a conserved ferredoxin-like fold and bind metal ions via conserved CXXC sequence motifs (18). Other types of histidine- and cysteine-rich MBDs found in the P_1B-3- and some P_1B-2-ATPases have also been shown to bind metal ions (19). A large body of data indicates that the MBDs play a regulatory role in ATPase function (19, 20). The P_1B-4- and P_1B-5-ATPases are the least well characterized and are distinguished by their relatively simple architecture, including only six TM helices (Figure 1) and an apparent absence of soluble MBDs (15). Whereas the P_1B-4-ATPases are associated with transport of Co²⁺ and other divalent metal ions (8, 21), the substrate for the P_1B-5 -ATPases has not yet been identified.

Here we report the identification, isolation, and characterization of a soluble domain found at the C-terminus of P_1B-5-ATPases. Surprisingly, this domain is homologous to hemerythrins (Hr), a family of proteins and domains characterized by the presence of a carboxylate-bridged diiron center housed within a four-helix bundle (22). Characterization of this Hr-like domain from the P_1B-5-ATPase of Acidothermus cellulolyticus, a gram-positive cellulolytic thermophile, reveals the presence of a diiron center. The presence of a Hr-like domain in a P_1B-ATPase is unprecedented and provides insight into P_1B-5 -ATPase function and the role of Hr proteins in biology.

MATERIALS AND METHODS

Bioinformatic Analysis and Sequence Alignments

Five P_1B5-ATPase sequences identified previously (14) were used as an initial query using protein-protein PSI-BLAST (23) (blastp) searches against the NCBI database. Both InterPro (24) and Pfam (25) analyses of many of the sequences indicated the presence of a C-terminal Hr/HHE (histidine-histidine-glutamate) cation binding domain (PF01814, IPR012312). Sections of the Streptomyces coelicolor P_1B-5-ATPase sequence (UniProtKB Q9RJ01), including the fourth TM helix and the start of the ATPBD as well as the C-terminal Hr domain, were then used for further queries. Sequences were aligned by the ClustalW method (26) with manual adjustments to ensure the best possible alignment of conserved metal binding residues. Secondary structure predictions were performed using the SOPMA Secondary Structure Prediction Method (27). The TopPred server (28) was used to predict the number and location of TM helices and SignalP 3.0 (29) was used to identify signal peptide cleavage sites

Cloning and Expression of a P_1B-5-ATPase Hr domain

The gene sequence encoding the C-terminal domain (residues 620–775) of the Acidothermus cellulolyticus 11B (30) P_1B-5-ATPase (YP_871788) was PCR amplified from genomic DNA (ATCC) with PCR Master Mix (Fermentas) using the primers 5’-CTGGTCGGTCTCGAATGCTGCCGGGTACGCGACAC-3’ and 5’-CTGGTCGGTCTCGGCGCTGCGCTGACCGGCATCCGGTAG-3’ (IDT), which introduce BsaI restriction sites. This sequence was selected by aligning the amino acid sequence of A. cellulolyticus P_1B-5-ATPase with that of DcrH-Hr (Figure 2), the C-terminal domain of a Desulfovibrio vulgaris chemotaxis protein and the best characterized prokaryotic Hr domain (31). The homologous region begins at Leu 620 and continues to Arg 775, but does not include the final 19 amino acids of the P_1B-5-ATPase. Secondary structure predictions suggest that these final residues are random coil whereas the rest of the sequence is predicted to be helical, as is characteristic of Hr domains (22, 32).

Figure 2.

The hemerythrin domain of P_1B-5-ATPases. Top: Alignment of eight representative P_1B-5-Hr sequences with the Hr domain of DcrH (DcrH-Hr, GI 887858, UniProtKB Q46583). The metal-binding residues of DcrH-Hr along with corresponding residues from P_1B-5-Hrs are colored with histidine in green, glutamic acid in orange, and aspartic acid in violet. The red helices below the sequence correspond to the known secondary structure elements of DcrH-Hr. The P_1B-5-Hr sequences correspond to the following proteins: S. viridochromogenes, GI 256799641; M. vanbaalenii, GI 120402396, UniProtKB A1T4W9; R. erythropolis, GI 226306903, UniProtKB C1A0I9; B. phymatum, GI 186471018, UniProtKB B2JWG3; B. dolosa, GI 254255483, UniProtKB A2WJ24; M. nodulans, GI 220919964, UniProtKB B8IWI5; A. cellulolyticus, GI 117927237, UniProtKB A0LQU2; R. vannielii, GI 283824493, UniProtKB D2LJH3. Bottom: The structure of the diiron center in the azide adduct of DcrH-Hr (PDB accession code 2AVK).

The purified PCR product and the plasmid pPR-IBA1 (IBA) were digested with BsaI (New England Biolabs), purified, and ligated, creating a construct containing an eight residue streptactin tag with a two residue linker (SAWSHPQFEK) fused to the C-terminus of the protein. The vector was transformed into E. cloni 10G chemically competent cells (Lucigen) following the manufacturer's protocol and spread on a Luria Bertani (LB) plate containing 100 μg/mL ampicillin. DNA sequencing of individual colonies confirmed the presence and accuracy of the gene fragment encoding the Hr-like domain, designated P_1B-5-Hr.

For protein expression, BL21(DE3) E. coli chemically competent cells were transformed with the plasmid encoding P_1B-5-Hr. Baffled flasks containing 1 LB media supplemented with 100 mg ampicillin were inoculated with an overnight culture of cells and grown at 37 °C with shaking. Protein expression was induced by adding 1 mM IPTG at an OD₆₀₀ of approximately 0.6. Cells continued to grow for an additional 3–4 h, after which they were harvested by centrifugation at 4800g for 10 min in a Sorvall SLC 4000 rotor at 4 °C and stored at -80 °C.

Protein Purification

All purification and handling steps were conducted at 4 °C. The cells were resuspended in a buffer containing 50 mM Tris, pH 7.0, 200 mM NaCl, 0.5 mM PMSF and lysed by sonication for 8 min by using 20 s pulses with a 40 s rest period between each pulse. Insoluble cell debris was removed by ultracentrifugation at 120,000g in a Beckman TI-60 rotor for 1 h. The supernatant was loaded onto streptactin resin (Qiagen) and washed with four column volumes of a 50 mM Tris, pH 7.0, 200 mM NaCl wash buffer. The protein was then eluted with 1.5 column volumes of wash buffer supplemented with 2.5 mM desthiobiotin. The eluted sample (as-isolated P_1B5-Hr) was concentrated in an Amicon 15 mL spin concentrator with a 10 kDa filter and its concentration was measured by the Lowry Assay (33) using bovine serum albumin as a standard.

Metal Loading and Analysis

The metal content was measured by inductively coupled plasma optical emission spectrometry (ICP-OES) using a Varian Vista MPX ICP-OES in the Integrated Molecular Structure Education and Research Center (IMSERC) at Northwestern University. As-isolated samples were digested in 5 mL of 5% Trace SELECT nitric acid (Sigma Aldrich) in chelexed water and then filtered if necessary. Standards of iron, chromium, nickel, cobalt, zinc, copper, molybdenum, manganese, vanadium, and cadmium (Sigma Aldrich) were prepared in 5% nitric acid as well. Iron loading was performed by the aerobic addition of two equivalents of Fe(NH₄)₂(SO₄)₂·6H₂O in 100 mM MES acid, pH 3.0, and 200 mM ascorbic acid to the protein solution followed by gentle mixing for 30 min. Protein was then exchanged into a buffer containing 25 mM Tris, pH 7.0, 100 mM NaCl using a 10DG desalting column (Bio-Rad). The protein and metal concentrations were measured by the Lowry assay and ICP-OES, respectively. To test whether P_1B5-Hr binds metals other than iron, an apo sample was prepared by adding 10 molar equivalents of desferrioxamine (Sigma Aldrich) followed by desalting. Five equivalents of different metal ions (MnCl₂, Fe(NH₄)₂(SO₄)₂, CoCl₂, NiCl₂, CuCl₂, Zn(OAc)₂, or CdCl₂) were added to this apo P_1B5-Hr, followed by desalting and Lowry and ICP-OES analysis..

Analytical Gel Filtration Chromatography

Analytical gel filtration chromatography was performed on a 22.5 mL Superdex 75 (GE Healthcare) column equilibrated with 50 mM Tris, pH 7.0, and 200 mM NaCl. 200 μL of either as-isolated or iron-loaded P_1B-5-Hr were loaded onto the column and molecular masses were determined using the following standards: blue dextran (void volume); aldolase, 158 kDa; ovalbumin, 43 kDa; chymotrypsin, 25 kDa; and RNAse A, 13.7 kDa.

Optical and Circular Dichroism Spectroscopy

All spectra were collected at room temperature. The UV-visible spectra of P_1B-5-Hr (500 μM) were recorded using a Perkin Elmer LAMBDA 1050 spectrophotometer. The azide adduct was generated by adding 50 molar equivalents of NaN₃ per diiron center and incubating for 20-30 min prior to spectroscopic measurements. Reduced and oxidized samples were obtained by adding 10 molar equivalents of NaS₂O₄, and 1 molar equivalent of K₃Fe(CN)₆ per diiron center, respectively, and gently mixing immediately before spectroscopic measurements. Circular dichroism (CD) spectra were recorded on a JASCO J-815 spectrometer using a 0.6 mm path length quartz cuvette and protein diluted to 100 μM concentration in 50 mM Tris, pH 7.0, 200 mM NaCl. An average of 3 scans was collected at 20 °C with 1 nm resolution.

Thermal Stability Assay

Apo and iron-loaded P_1B-5-Hr were concentrated to 2-5 mg/mL and mixed with thermal stability buffer (100 mM HEPES, pH 7.5, 150 mM NaCl) and 5000x SYPRO Orange (Molecular Probes, Inc) in DMSO in a 25:120:2 ratio. 1 μL of each solution was distributed into each well of a PCR plate. The wells were then filled with 10 μL of buffer containing 50 mM Tris, pH 7.0 and 200 mM NaCl with some wells containing 5 mg/mL desferrioxamine as well. The plates were sealed and transferred to a Biorad CFX384 Real-Time PCR detection system in the High Throughput Analysis Laboratory (HTAL) at Northwestern University. The temperature of the plates was increased at a rate of 0.5 °C per min over the range of 10 °C to 95 °C and fluorescence was monitored at 570 nm. The data were processed using OriginPro 6.1 for Windows. In order to determine the inflection point of the melting curves, which was assumed to equal the melting temperature (T_m), a Boltzmann sigmoidal equation was fitted to the raw data. This experiment was performed in replicate up to 20 times to ensure T_m precision.

Electron Paramagnetic Resonance (EPR) Spectroscopy

Samples for EPR were concentrated to 1 mM protein in 25 mM Tris, pH 7.0, 100 mM NaCl, 20% glycerol, transferred to Q-band EPR tubes, frozen, and stored in liquid nitrogen. Cryoreduction and signal quantitation were carried out as described previously (34). Annealing at 260 K was performed by placing the EPR sample in a salt water bath and then freezing in liquid nitrogen (35). X-band spectra were collected at 10 K using a Bruker ESP300 spectrometer with the field modulation set to 10 G.

X-ray Absorption Spectroscopy

For X-ray absorption spectroscopy (XAS), samples were prepared in 25 mM Tris, pH 7.0, 100 mM NaCl, 30% glycerol with iron concentrations in the 1-2 mM range. Two independent replicates of iron-loaded P_1B-5-Hr and P_1B-5-Hr in the presence of 50 molar equivalents of NaN₃ were prepared. Samples were loaded into Lucite cells wrapped with Kapton tape and frozen in liquid nitrogen. XAS data were collected at the Stanford Synchrotron Radiation Lightsource (SSRL) on beamline 7-3, equipped with a single rhodium-coated silicon mirror and a Si[220] double crystal monochromator detuned 50% for harmonic rejection. Samples were maintained at 10 K using an Oxford Instruments continuous-flow liquid helium cryostat. Protein fluorescence excitation spectra were collected using a 30-element Ge solid-state array detector. XAS spectra were measured as described previously (36). Data were processed using the Macintosh OS X version of the EXAFSPAK program suite integrated with the Feff v7.2 software for theoretical model generation. XAS data reduction and analysis were performed following previously reported protocols (37). EXAFS data were simulated over a k range of 1 to 14.2 Å^-1 for a spectral resolution of 0.12 Å. Data were fit using a scale factor of 0.95 and E_o values for Fe-O/N/C and Fe-Fe interactions of -10 and -15 respectively. Simulation parameters for fitting the raw unfiltered data are given along with the number of degrees of freedom weighted simulation “goodness of fit” (F’) parameter in Table 1.

Table 1.

Summary of raw Fe EXAFS simulation analysis for iron-loaded and azide-treated A. cellulolyticus P_1B-5-Hr. Values given in bold represent the best-fit simulation parameters.

Sample	Fit #	Fe-Nearest Neighbor Ligandsa				Fe-Long Range Ligandsa
		Atomb	R(Å)c	C.N.d	σ 2e	Atomb	R(Å)c	C.N.d	σ 2e	F'f
Fe	1	O/N	1.97	2.5	4.7	C	3.03	1.0	2.6	0.30
		O/N	2.09	2.5	4.6	C	3.46	2.5	1.3
	2	O/N	1.97	2.5	4.91	C	3.02	1.0	5.5	0.27
		O/N	2.10	2.5	5.3	Fe	3.39	1.0	4.8
	3	O/N	1.97	2.5	4.9	C	3.02	1.0	5.7	0.25
		O/N	2.10	2.5	5.4	Fe	3.39	1.0	4.7
						C	4.09	1.0	4.9
Azide	1	O/N	1.99	2.5	4.3	C	3.04	1.0	2.0	0.21
		O/N	2.12	2.5	4.1	C	3.46	3.5	2.8
	2	O/N	1.99	2.5	5.0	C	3.03	1.0	3.5	0.16
		O/N	2.11	2.5	5.8	Fe	3.39	1.0	4.3
	3	O/N	1.99	2.5	5.0	C	3.03	1.0	3.3	0.14
		O/N	2.11	2.5	5.7	Fe	3.39	1.0	4.2
						C	4.12	1.0	3.0

^aIndependent metal-ligand scattering environment

^bScattering atoms: O (Oxygen), N (Nitrogen), C (Carbon) and Fe (Iron)

^cMetal-ligand bond length (all standard deviations < 0.03 Å)

^dMetal-ligand coordination number (all standard deviations < 1.0)

^eDebye-Waller factor given in Ų × 10³ (all standard deviations < 0.9 Å)

^fNumber of degrees of freedom weighted mean square deviation between empirical and theoretical data

RESULTS AND DISCUSSION

Sequence Characteristics of Hr Domain-Containing P_1B-5-ATPases

A BLAST search using a 60-residue sequence from the S. coelicolor P_1B-5-ATPase, one of the previously identified P_1B-5-ATPases (14), corresponding to the fourth TM helix and the start of the ATPBD yielded 195 complete P_1B-5-ATPase sequences. This residue span was selected because it includes residues in the putative TM metal binding site proposed to confer substrate specificity, including the cysteine- and proline-containing signature motif. It is therefore ideal for distinguishing members of the P_1B-5-ATPase subfamily. Every species containing a P_1B-5-ATPase is eubacterial, but beyond this broad classification, there are no obvious unifying characteristics. P_1B-5-ATPases are found in both gram positive, gram negative, aerobic, anaerobic, facultative, and microaerobic organisms. Of the 195 identified sequences, 40 contained Hr-like domains. Membrane topology predictions indicate the presence of six to seven TM helices. Those proteins that have seven predicted TM helices, including the A. cellulolyticus P_1B-5-ATPase, contain a signal peptide cleavage site between the first two helices, consistent with a six-helix architecture. The P_1B-4-ATPases are also predicted to have six TM helices, and these six helices correspond to the final six helices of the eight helix P_1B-1-, P_1B-2, and P_1B-3-ATPases (15).

With a larger library of sequences identified, the defining characteristics of the P_1B-5 subfamily (14) can be refined (Figure 1). Alignment of the 195 P_1B-5-ATPase sequences (Supplemental Figure 1) indicates that the signature sequence in the fourth TM helix is (T/S)PCP with threonine present in 113 sequences and serine in the other 82 sequences. In 143 of 195 sequences, the fifth TM helix begins with a glutamine. The fifth helix also contains a conserved potential metal binding residue near the periplasmic face, a methionine in 171 sequences and a glutamic acid in 24 sequences. The final TM helix houses a QEXXD motif that is conserved in all but two P_1B-5-ATPases. This motif along with the (T/S)PCP sequence likely contributes key substrate binding residues. Previously noted serine and asparagine residues in the fifth and sixth helices, respectively (14), are not universally conserved in this more extensive sequence library.

Like most P_1B-4-ATPases, the P_1B-5-ATPases lack N-terminal soluble MBDs, suggesting that the two N-terminal TM helices in the other subfamilies may serve to tether and position the MBDs for regulatory interactions with the other soluble domains. The only exception, from Burkholderia glumae, contains a membrane-bound cytochrome b₅₆₁-like domain (PF01292, IP011577) fused to the N-terminus. The P_1B-5-ATPases are common to the Burkholderia genus, and a gene encoding a cytochrome b₅₆₁-like protein is sometimes found in the same operon (e.g. B. cenocepacia J2315, B. vietnamiensis G4, B. ambifaria MC40-6, B. multivorans ATCC 17616). In some organisms, these cytochrome b₅₆₁ proteins are associated with NiFe hydrogenases (38, 39), although this enzyme does not appear to be present in all Burkholderia genomes encoding a P_1B-5-ATPase.

The only C-terminal soluble domain found in the P_1B-5-ATPases is the Hr-like domain. Interestingly, all 40 Hr-containing P_1B-5-ATPases have the conserved methionine in the fifth TM helix, and 38 have a TPCP signature motif in the fourth helix. These Hr domains were not identified in recent studies of Hr diversity and classification (40, 41), probably due to the low overall sequence homology. For example, A. cellulolyticus P_1B-5-Hr is 18% identical and 40% similar to DcrH-Hr. Moreover, the metal binding residues are not completely conserved. In structurally characterized eukaryotic and prokaryotic Hr proteins, the two iron ions are bridged by an aspartic acid and a glutamic acid and coordinated by three and two histidines, respectively (22, 32) (shown for DcrH-Hr in Figure 2). In 27 of 40 sequences, the residue corresponding to DcrH-Hr Asp 946 (numbering based on the full length DcrH sequence) is glutamic acid and in 23 sequences, aspartic acid replaces DcrH-Hr Glu 886. Moreover, one of the metal coordinating histidine residues, equivalent to DcrH-Hr His 905, is completely absent in all the P_1B-5-ATPase Hr domains (Figure 2). Some of the identified P_1B-5-Hr sequences have a histidine one amino acid later, yielding HX₄H rather than HX₃H, but this only occurs in 7 of the 40 sequences, suggesting it may not be involved in metal binding. There is a conserved glutamic acid two amino acids prior to the histidine position. Thus, the consensus motif of P_1B-5-Hr is H– HX₃(D/E)–HXE–HX₄(E/D). In A. cellulolyticus P_1B-5-Hr, these residues are His 644, His 685, Glu 689, His 714, Glu 716, His 757, and Asp 762. Alterations in the metal binding residues are found in a large number of predicted prokaryotic Hrs (41). In addition, there are significant deviations between the hydrophobic substrate channel residues of classical Hr domains and the corresponding residues in the P_1B-5-Hr proteins. Of the 21 residues that comprise the substrate channel in DcrH-Hr (32), only seven are similar in P_1B5-Hr proteins.

Isolation and Characterization of P_1B-5-Hr

Overexpression of P_1B-5-Hr yielded approximately 3 mg purified protein per 1 L of media. The single-step purification on the streptactin column was sufficient for >95% purity (Figure 3). Samples were analyzed for a range of metal ions by ICP-OES, and only iron was detected. The iron stoichiometry was variable, with values ranging from 0 to 2 iron ions per protein monomer. If the stoichiometry was less than 2 iron ions per protein, two equivalents of Fe(NH₄)₂(SO₄)₂·6H₂O were added and then excess iron was removed on a desalting column. After this treatment, the stoichiometry was 2.13 ± 0.10 iron ions per protein. An apo form of P_1B-5-Hr was prepared by desferrioxamine treatment and then loaded with five equivalents of Mn(II), Fe(II), Co(II), Ni(II), Cu(II), Zn(II), and Cd(II). Stoichiometries of 0.22 ± 0.01 Mn(II), 1.93 ± 0.06 Fe(II), 0.75 ± 0.02 Co(II), 0.48 ± 0.03 Ni(II), 1.10 ± 0.01 Cu(II), 1.25 ± 0.12 Zn(II), and 0.92 ± 0.02 Cd(II) were measured. Given the consistent measurement of 2 Fe per protein and the spectroscopic data (vide infra), it seems likely that P_1B-5-Hr binds iron specifically and perhaps binds other metal ions adventitiously.

Figure 3.

SDS-PAGE analysis of purified A. cellulolyticus P_1B-5-Hr. Molecular mass markers are labeled in kDa. The predicted molecular mass of A. cellulolyticus P_1B-5-Hr is 18.5 kDa.

The oligomeric state of P_1B-5-Hr in solution was investigated by analytical gel filtration experiments (Figure S1). The as-isolated sample eluted in two peaks, one at ~34 kDa and a second beyond the calibration range but below the void volume (~200 kDa). The theoretical mass of P_1B-5-Hr, including streptactin tag and linker is 18.5 kDa, consistent with the presence of a dimer and a higher molecular mass oligomer. The same two peaks are observed for the iron-loaded sample, although with slightly altered calculated masses of ~40 kDa and > 300 kDa. In addition, an aggregate peak is present, consistent with some precipitation observed upon iron addition.

CD and Optical Spectroscopy

The CD spectra of both apo and iron-loaded P_1B-5-Hr exhibit local minima at 209 nm and 221 nm (Figure 4), indicative of a primarily helical structure, consistent with a Hr four-helix bundle (22). The similarity between the CD spectra of the iron-loaded and apo samples suggests that iron binding does not induce major structural changes. The purified, iron-loaded protein is always yellow in color and its optical spectrum (Figure 5) lacks a ligand-to-metal charge transfer (LMCT) band observed at 500 nm for the oxy form of Hr, which has a terminally bound hydroperoxide ligand (31, 42, 43). Reduction by excess dithionite bleached the yellow color, which reappeared upon exposure to air, without forming any detectable absorbance features attributable to the oxy form, indicating that autooxidation to the met form is very rapid (t_1/2 < 1 min). Rapid autooxidation was also reported for the prokaryotic Hr domain DcrH-Hr (31) and Hr from Methylococcus capsulatus (Bath) (42). A shoulder at approximately 650 nm is a d-d transition from the diiron(III) site (44). There is a broad absorbance in the 300-400 nm range attributable to LMCT transitions, but distinct peaks are not observed as in other Hr proteins and domains (31, 42, 44, 45). Addition of the oxidizing agent potassium ferricyanide did not create any new spectroscopic features (data not shown) indicating that the iron-loaded sample contains only Fe(III). The spectrum of the Hr domain from the human FBXL5 protein is similar (46). Interestingly, alignments indicate that the histidine residue absent in P_1B5-Hr is also missing in FBXL5. Thus, the broader spectroscopic features in the 300-400 nm range may be due to differences in iron coordination in P_1B-5-Hr. Addition of azide produces a new optical feature at 453 nm (Figure 5). Similar features are observed for DcrH-Hr (31), M. capsulatus (Bath) Hr (42), and Hr from marine invertebrates (44, 47).

Figure 4.

CD spectrum of apo and iron-loaded A. cellulolyticus P_1B-5-Hr (150-200 μM) in 50 mM Tris, pH 7.0, 200 mM NaCl.

Figure 5.

Optical spectra of A. cellulolyticus P_1B-5-Hr. Top: Spectra of iron-loaded, azide-treated, and background-corrected, reduced P_1B-5-Hr in 25 mM Tris, pH 7.0, 100 mM NaCl. The inset shows a shoulder at ~650 nm. The extinction coefficient is reported per diiron center as quantified by ICP OES. Bottom: Difference spectrum generated from the subtraction of the iron-loaded spectrum from the azide absorbance.

Thermal Stability Assay

To evaluate the effect of iron loading on P_1B-5-Hr stability, thermal stability assays were performed. SYPRO Orange, a dye that fluoresces when interacting with hydrophobic residues and can act as an indicator for denatured protein, was mixed with both apo and iron-loaded P_1B-5-Hr. These solutions were gradually heated while the fluorescence was monitored, and the signal normalized to generate a denaturation curve (Figure 6). Fitting of these curves yields T_m values of 65.5 ± 1.5 °C for apo P_1B-5-Hr and 74.9 ± 2.0 °C for iron-loaded P_1B-5-Hr. To verify that the increased thermostability is a direct effect of the iron bound to P_1B-5-Hr, conditions containing an excess of desferrioxamine were tested for thermal stability. Apo protein was relatively unaffected by the desferrioxamine with a T_m value of 65.8 ± 0.2 °C. The iron-loaded protein, however, exhibited a decreased T_m to 63.4 ± 0.3 °C, effectively equivalent to the apo sample. The T_m of the apo protein correlates well with temperature-dependent growth of A. cellulolyticus which has a maximum growth temperature of 65 °C.

Figure 6.

Thermal stability fluorescence spectra of apo and iron-loaded P_1B-5-Hr with Boltzmann sigmoidal fits. All fluorescence units have been normalized to account for differences in protein concentration.

EPR Spectroscopy

Iron-loaded P_1B-5-Hr is EPR silent, consistent with an antiferromagnetically coupled diiron(III) center. Initial attempts to generate an EPR signal by semi-reduction with dithionite or reduction followed by semi-oxidation were unsuccessful, perhaps due to instability of the mixed valence state. Instead, to generate a mixed valence Fe(II)Fe(III) center, a frozen, iron-loaded sample was reduced by γ-irradiation as done previously for Hr (48) and other diiron-containing proteins, including the soluble methane monooxygenase hydroxylase (34) and the ribonucleotide reductase R2 protein (48). After annealing at 260 K to eliminate radicals in the frozen matrix whose signals partly obscured that of cryoreduced P_1B-5-Hr, the spectrum from the protein shown in Figure 7 was obtained. The g-values of 1.94, 1.83, and 1.76 rule out an assignment to a mononuclear iron center, and instead definitively reveal the presence of a mixed valence Fe(II)Fe(III) diiron cluster in which the partner Fe ions are antiferromagnetically coupled to generate an S=1/2 cluster state (48, 49). Quantitation with a Cu(II)-EDTA standard indicates that the dose of 1.4 Mrad created ~0.1-0.15 mM Fe(II)Fe(III) cluster. Our experience indicates that this dose typically reduces roughly 15% of the diamagnetic parent, leading to an estimated total concentration of diiron centers of ~0.75 mM, which corresponds approximately to ~ 3/4 occupancy of the sites (sample concentration 1 mM). A signal at g = 4.3 was also present, and quantitation indicates that the concentration of adventitious Fe(III) is ~0.1 mM or about 10% of the total iron.

Figure 7.

X-band EPR spectrum of cryoreduced and annealed iron-loaded A. cellulolyticus P_1B-5-Hr. The spectrum was collected at 10 K with a microwave power of 2 mW.

X-Ray Absorption Spectroscopy

To confirm the presence of a diiron center, XAS data were collected on samples of iron-loaded P_1B-5-Hr and P_1B-5-Hr in the presence of azide. The X-ray absorption near edge spectra (XANES) indicate the presence of both Fe(II) and Fe(III) in the protein-bound Fe XANES (Figure 8). Analysis of the first inflection energies for protein-bound Fe compared to Fe(II) and Fe(II) model controls indicates that the iron-loaded sample (inflection energy of 7124.6 eV) has a 50% mixture of Fe(II) and Fe(III) whereas the azide treated sample (inflection energy of 7123.9 eV) has a 65%/35% mixture of Fe(II)/Fe(III). Immediately following exposure of the iron-loaded sample to the X-ray beam, the resulting metal appears to be a stable equivalent mixture of Fe(II) and Fe(III). The azide treated sample was highly susceptible to X-ray influenced metal photoreduction.

Figure 8.

Normalized XANES spectra of iron-loaded A. cellulolyticus P_1B-5-Hr (light solid line) and azide-treated A. cellulolyticus P_1B-5-Hr (bold solid line), compared with the model control spectra for (NH₄)₂Fe(SO₄)₂ (dotted line) and NH₄Fe(SO₄)₂ (dashed line).

The extended X-ray absorption fine structure (EXAFS) data for both protein samples are best fit with multiple shells of interacting ligands (Figure 9, Table 1). There are two O/N ligands at 1.97-1.99 Å, four O/N ligands at 2.09-2.12 Å, consistent with six-coordinate iron (Figure 2) and a highly prevalent scattering Fe-Fe signal in both samples at 3.39 Å. This Fe-Fe distance is identical to the 3.39 Å distance observed crystallographically for met DcrH-Hr and longer than the 3.25 Å distance typical of invertebrate Hr (22, 50). An increase in Fe-Fe distance is consistent with the presence of a partly photoreduced diiron center (32, 51). The two nearest neighbor O/N ligand environments likely correspond to histidine nitrogen atoms and carboxylate oxygen atoms at ~2.1 Å, as observed in other Hr structures, and perhaps exogenous oxygen-based ligands at 1.97-1.99 Å. The latter distance is too long for a bridging oxygen atom and suggests the presence of a bridging hydroxide, as observed for the deoxy form of Hr (52), or alternatively, a water molecule. The azide treated sample is very similar to the iron-loaded sample, which is not surprising since the Fe-N distance in azide adducts of Hr is approximately 2 Å (32, 50). The Fe-Fe distance is unchanged, suggesting that binding of azide does not alter the dinuclear center core structure in any appreciable manner as would be expected if azide was bridging the two metal ions. Thus, azide is likely binding in a terminal fashion, as observed for other Hr adducts. Long-range carbon scattering environments at ca. 3.0 Å and 4.1 Å are observed in both samples. Carbon scattering at these bond distances is typical in the presence of rigid imidazole scattering from histidine residues acting as metal ligands (53).

Figure 9.

Smoothed EXAFS and Fourier transforms of EXAFS data for iron bound to iron-loaded (A, B respectively) and azide-treated (C and D, respectively) A. cellulolyticus P_1B-5-Hr. Empirical data are shown in black and theoretical simulations are shown in green.

Functional Implications

The combined metal binding and spectroscopic data indicate that P_1B-5-Hr houses a diiron center, similar to that in other Hr proteins and domains. Hr domains have been implicated in a variety of biological functions. In invertebrates, Hr reversibly binds O₂ for transport (54). Prokaryotic Hr proteins and domains fall into two groups: single domain proteins or domains fused to larger proteins. Rigorous functional characterization has not been performed, but putative functions include O₂ transport (M. capsulatus (Bath) Hr) (42, 55), O₂ sensing for chemotaxis (DcrH-Hr) (31), nitric oxide sensing (56), iron storage (40, 57), and cadmium detoxification (58). In humans, iron sensing by the Hr domain of FBXL5 leads to degradation of iron regulatory proteins (46, 59).

Soluble metal-binding domains in other P_1B-ATPases play a regulatory role, likely via metal-dependent protein-protein interactions with other domains. For example, Cu⁺ binding by the N-terminal MBDs of P_1B-1-ATPases precludes protein-protein interactions with the ATPBD (60, 61). Although the substrate of the P_1B-5-ATPases has not been identified, the discovery of a soluble Hr domain provides intriguing clues. One possibility is that these ATPases transport iron. By analogy to the P_1B-1-ATPases, iron binding to P_1B-5-Hr might affect transport by modulating interdomain interactions. An alternative or additional function of P_1B-5-Hr might involve O₂ sensing. For DcrH-Hr, conformational changes upon O₂ binding to the diiron(II) site are proposed to initiate signal transduction that leads to anaerotaxis (32). For the P_1B-5-ATPases, sensing of O₂ as well as iron would have the added benefit of preventing oxidative damage: under oxidative stress conditions, the iron-loaded Hr domain would initiate iron efflux, minimizing the potentially damaging effects of iron-mediated Fenton chemistry.

Another possibility is that iron binding and/or O₂ sensing regulate transport of another metal ion. Notably, the A. cellulolyticus genome encodes a nickel-containing superoxide dismutase, and O₂ sensing could play a role in its regulation by maintaining appropriate Ni²⁺ concentrations. Another potentially O₂-regulated system present in A. cellulolyticus is NiFe hydrogenase (62). This idea is consistent with the observation (vide supra) that some Burkholderia species encode P_1B-5-ATPases and cytochrome b₅₆₁-like proteins in the same operon, and cytochrome b₅₆₁-like proteins may be associated with NiFe hydrogenase (38, 39). There could also be synergism or antagonism between iron and the transported metal substrate. Although testing of these hypotheses will require functional characterization of a full length P_1B-5-ATPase, the identification of P_1B-5-Hr suggests a novel regulatory mechanism for P_1B-ATPases and expands the known functional repertoire of Hr domains.