Structural Properties of 2/2 Hemoglobins: The Group III Protein from Helicobacter hepaticus

Summary

The ε-proteobacterium Helicobacter hepaticus (Hh) contains a gene coding for a hemoglobin (Hb). The protein belongs to the 2/2 Hb lineage and is representative of Group III, a set of Hbs about which little is known. An expression and purification procedure was developed for Hh Hb. Electronic absorption and NMR spectra were used to characterize ligation states of the ferric and ferrous protein. The pK_a of the acid/alkaline transition of ferric Hh Hb was 7.3, an unusually low value. NMR analysis of the cyanomet complex showed the orientation of the heme group to be reversed compared to most Group I and II 2/2 Hbs. Ferrous Hh Hb formed a stable cyanide complex that yielded NMR spectra similar to those of the carbonmonoxy complex. All forms of Hh Hb self-associated at NMR concentrations. Comparison was made to the related Campylobacter jejuni 2/2 Hb (Ctb), and the amino acid conservation pattern of Group III was re-inspected to help in the generalization of structure–function relationships.

INTRODUCTION

The 2/2 lineage of the hemoglobin (Hb) superfamily comprises several hundred bacterial proteins, few of which have been characterized in detail. Since their discovery, 2/2 Hbs have offered new perspectives on hemoglobin history (1). Their documented roles thus far include NO detoxification (2), protection from reactive oxygen species (ROS) (3) and reactive nitrogen species (RNS) (4), and dioxygen scavenging (5). The diversity of physiological and environmental contexts in which 2/2 Hbs are found suggests that additional enzymatic activities and insights into heme chemistry are yet to be discovered.

The 2/2 Hb lineage is distinguished by a ~120-residue globin domain and a 2-on-2 helical topology. It is divided into three phylogenetic groups (I, II, and III) (6). Most 2/2 Hb studies have concentrated on Group I and Group II proteins; among Group III proteins, only Ctb, the 2/2 Hb from Campylobacter jejuni, has been investigated (7–10). The X-ray structure of Ctb (11) illustrates major differences from the canonical globin fold: absence of the A and D helices, altered B–E interhelical angle, tight C–E turn, and shortened F, G, and H helices. As detailed in a recent structural comparison (12), Ctb deviates from Group I and Group II 2/2 Hbs in several helix and loop lengths and the composition of the heme binding site. Although Ctb crystallizes as a dimer, the quaternary structure is viewed as an artifact (11), and gel filtration data confirm the protein to be a monomer under physiologically realistic conditions (7).

To broaden our understanding of the relationship between structure and function in 2/2 Hbs, we initiated the characterization of the Group III protein from Helicobacter hepaticus (Hh), an ε-proteobacterium that thrives in the microaerobic environment of the mouse intestinal tract (13). Hh Hb, the sequence of which is 70% identical to that of Ctb, was originally chosen to capture common features of Group III. Hh Hb was expected to be sufficiently related to Ctb to benefit from previous studies of this protein and sufficiently distant from it to deemphasize protein-specific properties. We applied optical and NMR methods to investigate the active site of Hh Hb in solution and we performed alignments over a much larger number of sequences than available at the time of our last analysis (14). The new information allows for a critical inspection of the properties of Group III 2/2 Hbs, including heme orientation in the heme cavity, affinity for anionic ligands, and propensity for self-association.

EXPERIMENTAL PROCEDURES

All protocols were standard except where noted. Expanded procedures are provided in the Supporting Information.

Gene isolation and protein expression and purification

H. hepaticus ATCC51449 genomic DNA was supplied by Dr. S. Suerbaum (University of Würzburg). The gene for conserved hypothetical protein AAP76761 (SwissProt Q7VJS9, Hh Hb hereafter) was cloned by PCR. Overexpression in E. coli BL21(DE3) cells grown in M9 medium yielded apoprotein inclusion bodies, which were solubilized with 6 M guanidine HCl. Refolded holoprotein was obtained by dialysis in the presence of excess hemin to circumvent apoprotein aggregation. Purification was completed by anion exchange chromatography followed by gel filtration. The yield was 20–25 mg/L growth medium. A truncated variant lacking the two C-terminal histidines (referred to as Δ129–130) was also prepared. The extinction coefficient of the ferric wild-type protein, determined by the hemochromogen assay (15), was ε_413nm = 144,000 M⁻¹cm⁻¹ at pH 5 and 108,000 M⁻¹cm⁻¹ at pH 10. Values of ε_413nm at intermediate pH were calculated using an apparent pK_a of 7.34 (see Results). Protein concentrations are reported on a heme basis.

NMR spectroscopy

Data were collected at a ¹H frequency of 600 MHz (Bruker DRX or Avance spectrometers) or 800 MHz (Varian Inova spectrometer). Homonuclear NOESY, DQF-COSY, TOCSY data, ¹H-¹⁵N HSQC spectra, long-range ¹H-¹⁵N HMQC spectra, ¹⁵N-separated NOESY data, and natural abundance ¹H-¹³C HMQC spectra were acquired as reported in past studies of Synechococcus sp. PCC 7002 Group I Hb (GlbN) (16–18). TRACT data (19) were obtained as described in the original reference at concentrations between 1.6 mM and 0.06 mM. NMR data were processed with TopSpin or NMRPipe (20) and analyzed with Sparky (21) or TopSpin (TRACT data). ¹H and ¹⁵N chemical shifts were referenced indirectly to 2,2-dimethyl-2-silapentate-5-sulfonic acid.

BLAST searches

Blastp (22) searches were performed using Hh Hb as a query sequence against the non-redundant protein sequence database, an expect threshold of 10, and a BLOSUM62 scoring matrix. To ensure accurate hits, only sequences with an expect value of 1e–10 or lower were retained. Blastp searches were repeated using the lowest scoring hit as the query sequence until only redundant hits were recorded. The results were compared to Genomic BLAST searches performed against the microbial completed genome database (23). This database also allowed for a thorough search of microbial genomes by phyla, ensuring no homologous sequences were missed. Sequences from different strains of the same species were omitted from the final alignment. Both orthologous and paralogous sequences were kept for the alignments as gene history was not the focus of this study. Sequences are listed in Supporting Information Table S1 along with gene index numbers.

Alignments

Secondary structure based profile alignments were performed against Ctb using ClustalW (24), default secondary structure parameters, and a Gonnet 250 matrix with a gap open penalty of 10 and a gap extension penalty of 0.10. Alignment biases that may have arisen from those sequences with N and C-terminal additions were removed with ClustalX in the residue range encompassed by the entire Hh Hb sequence. No further adjustment to the alignments was needed.

RESULTS AND DISCUSSION

The ferric aquo, hydroxy, and peroxide-induced complexes

At pH 3.8, ferric Hh Hb exhibited a Soret maximum at 413 nm, a visible band at 498 nm and a pronounced ligand-to-metal charge transfer band at 637 nm consistent with an aquomet (His–Fe(III)PPIX–H₂O) complex (25) (Supporting Information Fig. S1). As the pH was raised, the Soret band remained at 413 nm and bands at 545, 578, and 610 nm emerged as seen in hydroxymet (His–Fe(III)PPIX–OH⁻) complexes. Global fitting to a two-state model returned an apparent pK_a of 7.34 ± 0.01 and a Hill coefficient of 1.06 ± 0.02 for this acid-alkaline transition. Absorption maxima of these and other complexes are listed in Supporting Information Table S2.

The response to changes in pH was monitored by ¹H NMR spectroscopy. In His–Fe(III)PPIX–H₂O state, the four heme methyl groups of typical globins have an average chemical shift of 75–80 ppm (26). These are readily recognizable in Hh Hb at pH 6.8 (Fig. 1B and Supporting Information Fig. S2). In contrast, the His–Fe(III)PPIX–OH⁻ state has a typical average methyl shift of 25–35 ppm. The Hh Hb spectrum at pH 9.5 (Fig. 1A) is consistent with this complex. Variable temperature data collected at pH 6.8 (Supporting Information Fig. S3) confirmed that only one set of heme methyl signals was detectable. Thus, the protein had a strong preference for accommodating the heme group in a single orientation.

Figure 1.

Portions of the 600 MHz ¹H 1D NMR spectrum of Hh Hb (25 °C). Ferric state at A, pH 9.54; B, pH 6.83 with heme methyls marked with ●; and C, pH 5.70. D: Cyanomet complex at pH 7.3 with tentative Tyr B10 OH assignment; E: ferrous cyanide complex (20 mM phosphate, pH 7.4). F: carbonmonoxy complex (60 mM phosphate, pH 7.2). G: Apparent correlation time of cyanomet Hh Hb as a function of protein concentration (25 °C, pH 7.2) obtained from TRACT data. The ▲ indicates the value for GlbN (see text).

The NMR transition observed in the 6.8–9 pH range likely corresponded to that detected by absorbance spectroscopy. However, the heme signals attributed to the typical aquomet species weakened as the pH was decreased below neutrality, while new resonances appeared at an intermediate shift of ~40 ppm (Fig. 1C). This behavior was similar to that of elephant myoglobin, a protein with a two-state acid-alkaline transition by absorbance spectroscopy (pK_a 8.5, (27)), but two aquomet forms by NMR spectroscopy (28). A pH-driven H-bond network rearrangement involving Gln E7 (distal) was proposed to explain the observation (28). Likewise, H-bonding residues on the distal side of Hh Hb (Tyr19 (B10), His47 (E7), and Trp87 (G8)) are good candidates for the formation of an adaptable network involving the hydroxide ligand.

A possible role for Hh Hb is peroxidase-like chemistry, as proposed for Ctb (9). The guaiacol/H₂O₂ peroxidase assay (29) was performed to test the activity of ferric Hh Hb toward small phenols. Under large excess of guaiacol and high concentrations of H₂O₂, the reaction mixture, monitored as a function of time, slowly developed the color of tetrahydroguaiacol (data not shown). The modest turnover rate was not evaluated and indicated weak peroxidase activity under the chosen conditions. In addition, the susceptibility of ferric Hh Hb to oxidative damage by H₂O₂ was evident by SDS-PAGE analysis. High molecular weight species (dimers and trimers, Supporting Information Fig. S4) attributed to o,o’-dityrosine linkages (30) were detected, and addition of cyanide as a strong ligand or guaiacol as an electron donor inhibited oligomerization. Interestingly, when H₂O₂ was added in a 1:1 ratio to ferric Hh Hb, the electronic absorbance spectrum showed a Soret band at 417 nm, broad visible bands at 532 nm and 553 nm, and a shoulder at ~620 nm, whereas when 100-fold excess was used, the Soret band broadened and two resolved bands developed gradually at 545 nm and 580 nm (Supporting Information Fig. S5A,B). These spectral features were reminiscent, respectively, of a high-valent hydroxy complex (Compound II, (31)) and a ferric-superoxy complex (Compound III, (32)). Formation of Compounds II and III has also been observed in the reaction of the Group II 2/2 Hb from Mycobacterium tuberculosis (33). Overall, however, the observations cast doubt on a peroxidase function.

The ferrous carbonmonoxy complex

NMR data collected on the dithionite-reduced protein with CO added were typical of a diamagnetic complex (Fig. 1F). Several heme assignments were obtained using homonuclear data sets and standard approaches (34). The vinyl groups, identified by their J-coupling pattern and size, led to the identification of other heme signals in a sequence of NOE connectivities summarized as 8-CH₃ ↔ δ-meso ↔ 1-CH₃ ↔ 2-vinyl ↔ α-meso ↔ 3-CH₃ ↔ 4-vinyl ↔ β-meso ↔ 5-CH₃ (heme structure in Supporting Information Fig. S6). The chemical shifts are listed in Supporting Information Table S3. Protein resonances were broad and overlapped, and the spectra were not analyzed further. The optical spectra of Hh Hb deoxy (λ_max: 433 nm, 562 nm), oxy (419 nm, 548 nm, 586 nm), and carbonmonoxy (422 nm, 542 nm, 572 nm) are presented in Supporting Information Fig. S5C,D.

The cyanide complexes

The NMR spectrum of the cyanomet (His–Fe(III)PPIX–CN) complex of globins is often better resolved than the spectrum of diamagnetic counterparts; in addition, the chemical shift of heme resonances in this one-unpaired-electron (S = 1/2) complex can be interpreted in structural terms. The 1D spectrum of cyanomet Hh Hb is shown in Fig. 1D. Several hyperfine shifted resonances were assigned with natural abundance ¹H-¹³C HMQC and homonuclear NOESY, TOCSY, and COSY data (Supporting Information Figs. S7–S11). The ¹H-¹³C HMQC data identified the four heme methyl groups, two of which had ¹H shift in the diamagnetic region. The following NOE connectivities were observed: 7-propionate ↔ 8-CH₃ ↔ 1-CH₃ ↔ 2-vinyl ↔ 3-CH₃ ↔ 4-vinyl ↔ 5-CH₃ ↔ 6-propionate.

The added chemical shift experienced by a heme methyl group in the Fe(III)CN complex compared to the diamagnetic Fe(II)CO complex represents the hyperfine contribution (provided the structures are otherwise identical). This hyperfine shift is dominated by a contact effect and is sensitive to the orientation of the axial histidine ring with respect to the heme plane (26, 35). Comparison of the carbonmonoxy and cyanomet shifts (Supporting Information Table S3) indicated that the contact shifts of the 3- and 8-CH₃s were nearly identical (22 ppm) and the contact shifts of the 1- and 5-CH₃s were close to 0 ppm. The methyl ordering (from low to high field, 3-CH₃ ~ 8-CH₃ ≫ 5-CH₃ ≥ 1-CH₃) was consistent with the Cδ2 and Cε1 atoms of His73 (proximal) aligning closely with the B and D heme pyrrole nitrogens. In keeping with globin topology and feasible histidine rotameric states (36), this implied that the heme adopted a reverse orientation compared to mammalian Hbs, i.e., the 4-vinyl and 5-CH₃ occupying the positions of the 1-CH₃ and 8-CH₃ in the latter proteins. Heme reversal was also observed in the Group I protein from Nostoc commune (37).

The ¹H-¹⁵N HSQC spectrum of the Fe(III)CN complex contained ~115 cross peaks attributable to backbone amides (Supporting Information Fig. S9) out of an expected 125. Missing peaks were likely to belong to the C-terminal tail (unstructured in Ctb) and flexible loops. ¹⁵N-separated NOESY data identified a GLG triplet, unambiguously assigned as Gly28-Leu29-G30. NH_i–NH_i+1 NOE connectivities corresponding to the C helix were detected from Gly28 to Gly38 (Fig. 2 and Supporting Information Fig. S10). This stretch encompasses Ile32 and Ile37, which are in the heme pocket and provided a starting point for further analysis.

Figure 2.

Panels of a ¹⁵N-separated NOESY experiment collected on cyanomet Hh Hb (~1 mM, pH 7.1, 25 °C, 20 mM phosphate, ¹H frequency of 600 MHz, 80 ms mixing time). Residue identity and ¹⁵N chemical shifts are indicated in each panel. The gray dashed line traces sequential connectivities through the C helix. The black dashed lines indicate the diagonal in each panel. The aliphatic portion of the panels is shown in Supporting Information Figure S10.

Ile37 (topological CD5 in myoglobin) was in NOE contact with the heme 8-CH₃ and with the indole NεH of a Trp, consequently assigned as Trp44 (E4). Ile32 (C7) had NOEs to the 1-CH₃ and Phe79, in the F–G linker. A second indole NεH was attributed to Trp55 (E15) through NOEs to the heme 4-vinyl. This NεH exchanged slowly with solvent and confirmed a tightly packed local environment. The assigned residues and the Ctb structure further let to: Phe54 (E14) near the 4-vinyl and 5-CH₃, Tyr65 in the EF loop in contact with Phe54, Val22 (B13) next to Trp44 (E4), Leu29 (C4), Phe18 (B9) and Tyr19 (B10). The placement of these side chains supported the conclusion of heme reversal. Spectra recorded in ¹H₂O contained a solvent-exchangeable resonance at ~27 ppm, which remained a singlet when the protein was uniformly labeled with ¹⁵N. The large hyperfine shift attributed it to the phenol OH of Tyr19 (B10), a residue H-bonded to the cyanide ligand in Ctb (11). Other partial or complete assignments were obtained and formed a self-consistent representation of the heme environment. The structure of the distal heme pocket, modeled after that of Ctb, is shown in Fig. 3.

Figure 3.

The distal environment in cyanomet Hh Hb modeled after that of Ctb (PBD:2IG3, chain A). The proximal histidine (His73, F8) is shown in red under the heme group. Other residues are colored according to conservation: green, strict; cyan, strong; yellow, weak (see Fig. 4). Phe B9, Tyr B10, and Trp G8 are strictly conserved in Group II, but not in Group I 2/2 Hbs. Variability is observed at the other positions in both Group I and Group II sequences (12, 44).

His E7 does not serve as an axial ligand to the iron and, in cyanomet Ctb crystals, adopts two orientations, one with imidazole toward the heme (χ1 = t), the second with imidazole toward solvent (g−) and forming a H-bond to the backbone carbonyl of residue E3. ¹H-¹⁵N HMQC data (38) revealed in Hh Hb, near neutral pH, a histidine in the NεH tautomeric state and a histidine undergoing the transition between neutral and charged states. The latter ring had ¹H chemical shift sensitive to temperature (Supporting Information Fig. S12). Aside from the proximal histidine, Hh Hb contains His47 (E7, distal), His104 (H3), His129, and His130. Tentative assignments were His104 to the low-pK_a histidine in the NεH tautomeric state, and His47 to the high-pK_a histidine with temperature-sensitive lines. The absence of readily recognized C-terminal histidine signals was attributed to unfavorable dynamics in the unstructured tail. Data collected on the Δ129–130 variant supported this interpretation of the data.

Unlike the Fe(III)CN complexes, the Fe(II)CN complexes of globins are generally unstable ((39) and ref. therein). Ctb (8) and Scapharca inaequivalvis homodimeric Hb (40) are two of the few that form a long-lived His–Fe(II)PPIX–CN complex. Supporting Information Fig. S5E,F displays the electronic absorption spectrum of cyanide-bound Hh Hb after reduction of the heme iron with dithionite. A new species was obtained with Soret maximum at 434 nm and two well-resolved Q-bands at 537 nm and 567 nm in agreement with Fe(II)CN spectra of globins (40). The factors governing the association and dissociation rate constants of cyanide are complex (41). In Hh Hb and Ctb, the combination of distal histidine (accelerating HCN dissociation) and H-bonding residues embedded in a highly hydrophobic environment on the distal side (decelerating CN⁻ dissociation) appeared to be essential (8).

The Fe(II)CN complex of Hh Hb was stable for hours. The 1D NMR spectrum (Fig. 1E) exhibited no detectable hyperfine shifted resonances, in agreement with an S = 0 state. The ¹H-¹⁵N HSQC spectrum (Supporting Information Fig. S13) was similar to that of the Fe(II)CO complex (Supporting Information Fig. S14), and compared to the spectrum of the Fe(III)CN state, showed that a large number of cross-peaks were affected by the changes in oxidation and ligation states.

In all ligation states, the quality of the NMR data was not consistent with a well-folded ~16 kDa Hb. ¹⁵N relaxation data (specifically 1D TRACT, (19)) were therefore collected to evaluate the rotational correlation time (τ_c) of the protein (Supporting Information Fig. S15). As a control, the experiment was performed on ferric GlbN, a S = 1/2, monomeric, 14 kDa bis-histidine Group I 2/2 Hb (16). The apparent τ_c of this protein was ~7 ns. In contrast, the apparent τ_c of cyanomet Hh Hb was 22 ns at concentrations of 1.6 mM, in agreement with particles with the molecular weight of at least a dimer. Fig. 1G shows a steep decrease in τ_c as the concentration was lowered to ~ 60 µM. Thus, Hh Hb was not a monomer in the samples used for NMR data collection. In addition, the far-UV circular dichroism data (Supporting Information Fig. S16) reported a secondary structure content consistent with a fully folded protein, and thermal denaturation data demonstrated heme retention over a wide range of temperature (not shown). Self-association therefore involved the folded holoprotein. Truncation of the sequence to eliminate His129 and His130, which could participate in intermolecular interactions, did not sharpen the lines significantly.

Conservation patterns

Our previous Group III sequence analysis tallied 30 proteins from three phyla (proteobacteria, actinobacteria, and firmicutes) (14). The expanded and corrected bacterial genome database now contains 181 Group III 2/2 Hbs in eight phyla (although the firmicute representative has been removed from the database, examples in acidobacteria, bacteroidetes, verrucomicrobia, spirochaetes, chlorofexi, and deinococcus-thermus have been discovered). Group III 2/2 Hbs, however, are still far from ubiquitous in bacteria. 73% of the sequences occur in proteobacteria (Supporting Information Table S4), and none are found in either Archaea or Eukaryota. The mean identity for the 181 sequences was 33%. Pairwise values reached below 20% as observed in Group I and Group II 2/2 Hbs. This revised the earlier assessment that Group III proteins formed a homogeneous set (14). Nevertheless, the data continued to show that Group I and Group II 2/2 Hbs arise in a greater number of phyla than Group III 2/2 Hbs.

The richer data set provided an opportunity to relate conservation patterns to the properties of Hh Hb and Ctb. Fig. 4 shows the alignment of 30 representative sequences chosen to account for all the bacterial phyla, including proteobacterial subdivisions. As per CLUSTALX terminology (24), 7 positions were strictly conserved, 5 positions were strongly conserved, and two positions were weakly conserved. All conserved residues were in the heme pocket, except for Asp BC1 (Asp25 in Ctb), the Oδ2 atom of which is the acceptor in H-bonds with the backbone amides of two and three residues downstream. These capping interactions stabilize a nearly 90° turn between the B and C helices.

Figure 4.

Sequence alignment of Group III 2/2 globin domains. The first line contains the canonical 3/3 helix labeling. The next line refers to the sequence of Hh Hb. Strictly conserved residues (*, B9, B10, CD1, E7, E15, F8, and G8) are set on green and red (proximal histidine). Strongly conserved residues (:) are set on blue (C4, E11, E14, G12, H14), and weakly conserved residues (.) are on yellow (BC1, H11). Ctb helical segments (11) are in bold. CAMJE, Campylobacter jejuni NCTC 11168 (1–127); HELHP, Helicobacter hepaticus ATCC 51449 (1– 130); 9CHLR, Ktedonobacter racemifer DSM 44963(1–134); BDEBA, Bdellovibrio bacteriovorus HD 100 (1–128); MEISD, Meiothermus silvanus DSM 9946 (1–132); BRAM5, Brachyspira murdochii DSM 12563 (1–124); ACIC5, Acidobacterium capsulatum ATCC 51196 (1–129); BURCA, Burkholderia cenocepacia AU 1054 (7–157); MYCA1, Mycobacterium avium 104 (1–152); 9ACTO, Frankia sp. EuI1c (1–150); 9RHOD, Labrenzia alexandrii DFL-11 (15–157); CHIPD, Chitinophaga pinensis DSM 2588 (1–130); ALCBS, Alcanivorax borkumensis SK2 (10–157); HAHCH, Hahella chejuensis KCTC 2396 (1–144); METRJ, Methylobacterium radiotolerans JCM 2831 (1–138); 9SPHI, Sphingobacterium spiritivorum ATCC 33300 (1–130); 9FLAO, Chryseobacterium gleum ATCC 35910 (1–126); 9PROT, alpha proteobacterium BAL199 (1–135); 9RHOB, Rhodobacterales bacterium HTCC2083 (1–132); MESSB, Mesorhizobium sp. BNC1 (9–154); 9NOCA, Rhodococcus jostii RHA1 (1–138); CORAD, Coraliomargarita akajimensis DSM 45221 (2–136); NOVAD, Novosphingobium aromaticivorans DSM 12444 (1–127); 9GAMM, Stenotrophomonas sp. SKA14 (19–161); XANP2, Xanthobacter autotrophicus Py2 GI:154246970 (9–150); CONWI, Conexibacter woesei DSM 14684 (1–147); SPILD, Spirosoma linguale DSM 74 (1–133); 9BACT, Algoriphagus sp. PR1 (1–132); BRUSU, Brucella suis 1330 (1–136); SILPO, Ruegeria pomeroyi DSS-3 GI:56708976 (1–139). Additional information is provided in Supporting Information Table S1.

Fig. 3 emphasizes the hydrophobic character of the heme pocket distal side and Fig. 4 shows this to be a highly conserved feature of Group III globins. Heme packing against Phe CD1 (actually in the middle of the C helix and fastening it to the heme), Ile E11, Phe E14, Leu G9, and Ile H14 also suggests that restricted access to the distal binding site and a gating role for the strictly conserved His E7 (11) are common Group III globin properties. In addition, three strictly conserved residues, Tyr B10, Trp E15, and Trp G8, all place H-bonding side chains within the distal heme pocket. Tyr B10 and Trp G8 are close to the heme iron and have been implicated in direct interactions with ligands (11). Variable capping interactions (42) in the C–E junction, however, caution that secondary structure boundaries may not be conserved, and that C and E helix length, side chain placement (including His E7), and heme accessibility may differ across the Group. Variability is also observed in the EF loop, although the effect on the tertiary structure could be less pronounced than caused by changes in the C–E region.

Only small residues (Ala, Ser, or Gly) occur at position H11 (Ala112) to act as spacers between Trp E15 and Trp G8. Conservation at B9, G12, and H14 supports a set orientation of the H helix and tryptophan arrangement in the heme pocket across Group III. It is also interesting that residues of the FG corner, G helix, and H helix are identified by the program Membrane Protein Explorer (43) as forming two hydropathy segments. This character appears to be preserved in the sequences of Fig. 4. The surface of Hh Hb is modeled in Supporting Information Fig. S17.

CONCLUSION

The sequence alignments, NMR data on Hh Hb, and published Ctb results allowed us to qualify distinctive features of Group III 2/2 Hbs. (i) Conservation of bulky residues in contact with the heme indicates that a reversed orientation of the heme within its binding site compared to most globins may be the rule. (ii) As a result of heme packing, the eclipsed orientation of the proximal histidine observed in Ctb and Hh Hb may also occur in other structures. (iii) In contrast, the 11-residue and α (as opposed to 7-residue and 3₁₀) C helix and extended E helix are unlikely to be general features. (iv) The fold in Ctb and Hh Hb is sufficiently rigid to prevent ligation of the distal histidine to the heme iron, but as per (iii), the C–E junction is too variable to anticipate this to be a conserved property. (v) Likewise, the ability to form stable Fe(II)CN and Fe(III)OH complexes, and perhaps other complexes with anionic ligands, may vary among the Group, but is likely to be enhanced compared to other 2/2 Hbs. (vii) Non-functional association of many Group III proteins may occur at high concentration.

Overall, our study supported that Group III 2/2 Hbs differ from Group I and Group II 2/2 Hbs in their reactivity and in their interactions with cellular components. Furthermore, it showed Group III to be less homogeneous than originally thought and raised the possibility that diverse chemical behaviors could be exhibited by its members. Analysis of the properties of a Group III protein with altered C–E linker and EF loop will be interesting to test further the generality of Ctb and Hh Hb observations.

Abbreviations

nD

n-dimensional

DT

dithionite

DTT

dithiothreitol

GlbN

Group I 2/2 Hb from Synechococcus sp. PCC 7002

Hh

Helicobacter hepaticus

LB

Luria-Bertani

PPIX

protoporphyrin IX

RNS

reactive nitrogen species

ROS

reactive oxygen species

τ_c

apparent correlation time

TRACT

TROSY for rotational correlation times