The role of propionates in substrate binding to heme oxygenase from Neisseria meningitidis; A NMR study

Abstract

Heme oxygenase, HO, cleaves hemin into biliverdin, iron and CO. For mammalian HOs, both native hemin propionates are required for substrate binding and activity. The HO from the pathogenic bacterium Neisseria meningitidis, NmHO, possesses a crystallographically undetected C-terminal fragment that by solution ¹H NMR is found to fold and interact with the active site. One of the substrate propionates has been proposed to form a salt bridge to the C-terminus rather than to the conventional buried cationic side chain in other HOs. Moreover, the C-terminal dipeptide Arg208His209 cleaves spontaneously over ~24 hours at a rate dependent on substituent size. 2D ¹H NMR of NmHO azide complexes with hemins with selectively deleted or rearranged propionates all bind to NmHO with a structurally conserved active site as reflected in optical spectra and NMR NOESY cross peak and hyperfine shift patterns. In contrast to mammalian HOs, NmHO requires only a single propionate interacting with the buried terminus of Lys16 to exhibit full activity and tolerates the existence of a propionate at the exposed 8-position. The structure of the C-terminus is qualitatively retained upon deletion of the 7-propionate but a dramatic change in the 7-propionate carboxylate ¹³C chemical shift upon C-terminal cleavage confirms its role in the interaction with the C-terminus. The stronger hydrophobic contacts between pyrroles A and B with NmHO contribute more substantially to the substrate binding free energy than in mammalian HOs, “liberating” one propionate to stabilize the C-terminus. The functional implications of the C-terminus in product release are discussed.

Heme oxygenase, HO¹, is the widely distributed enzyme (1) that utilizes hemin as both substrate and cofactor to stereospecifically cleave a single meso position to generate biliverdin, iron and CO, proceeding via the intermediates depicted in Scheme 1.

Scheme 1.

In mammals, the three products serve as a precursor to the powerful antioxidant bilirubin, as a source of 97% of the required iron and as a neural messenger, respectively (2-6). In plants and photosynthetic bacteria, the product open tetrapyrrole serves as a precursor to light-harvesting pigments (7), while the released iron is vital to infection for some pathogenic bacteria; (2, 4). The mammalian HOs have been the most extensively studied (3) and, in spite of only limited sequence homology (4), both the characteristic α-helical topology (8-13) and the mechanism (2, 4, 6) are conserved among other HOs. The relatively loose binding of substrate by HOs, as evidenced by both numerous vacancies and variable position of structural elements (8-13), involves an axial bond to a His, two salt-bridges between the ubiquitous propionates and two suitably positioned cationic side chains as depicted in Figure 1A; the remaining interactions are relatively weak van der Waal contacts with pyrroles A and B.

Figure 1.

Orientation of native protohemin-IX within the active site of: (A) human HO (8), hHO, showing the key salt bridges of the 6-propionate with the side chain of Lys18 and the 7-propionate with the side chain of Arg183; and (B) NmHO (9, 11) showing the conserved salt bridge between the 6-propionate and Lys16, as well as two residues, Cys113, Phe123, key to determining the orientation of substrate in the active site. The contacts of two residues not detected in the crystal structure, His207 with M₁ and M₈ and Arg208 with M₈, are also shown in bold.

The stereoselectivity of the cleavage arises from the steric effect of the distal helix which tilts the active Fe³⁺-OOH toward one meso position while blocking the remaining three meso positions (9-13). The particular meso position cleaved depends on the in-plane orientation of the substrate relative to the distal helix. Only the α-meso position is cleaved in mammalian HOs. The crucial role of propionate links in determining the in-plane orientation, and hence stereoselectivity, of substrate cleavage is confirmed by studies on human HO, hHO, (14) mutants where repositioning the cationic side-chains leads to a ~90° in-plane rotation of substrate with concomitant conversion of α-meso to mixed β, δ-meso cleavage. Earlier studies of human HO, hHO, with a variety of modified substrates (15, 16), have revealed that, while there is toleration for a wide range of substituents on pyrroles A and B (Figure 1B), the two propionates at positions 6 and 7 must both be conserved to retain activity.

In spite of sharing the same folding topology and mechanism with mammalian HOs, the HOs from pathogenic bacteria (2, 4-6) exhibit some distinct properties relative to function. Three pathogenic bacterial HOs have been structurally characterized, those from Neisseria meningitidis (9, 11), NmHO, Pseudomonas aeruginosa (13), PaHO, and Corynbacterium diphtheria (12), CdHO. CdHO, which exhibits the closest sequence and structural homology (4, 6, 12) to mammalian HOs, exhibits the same stereoselectivity, sensitivity to mutation (17), and similar 90° in-plane rotation, and altered stereoselectivity, upon mutation of cationic residues(18), as mammalian HOs (14, 19). PaHO exhibits (13, 20) mixed β-,δ-meso cleavage due to an ~90° in-plane rotation of protohemin-IX relative to that in mammalian HOs, resulting from the rearrangement of cationic side chains in the wild-type protein.

NmHO exhibits numerous distinct properties. While there is normal α-meso cleavage consistent with protohemin-IX oriented as depicted in Figure 1B, the structure of the active site is more compact and has fewer vacancies(9, 11) than in other HOs. The roles of the propionates in the crystal structures are not well-defined and key C-terminal residues are undetected (9, 11). In contrast, solution NMR has demonstrated that the C-terminal fragment, His207Arg208His209, is, in fact, ordered and interacts with the active site (21-24). A molecular model based on limited energy minimization and using as constraints limited NOESY data and a series of proposed salt bridges suggested by the crystal structure, has been offered (22, 24) and is reproduced in Figure 2. We note that this model utilizes one of the two propionates (at native PHIX position 7 in the crystal (9, 11)), leaving but one propionate to make the conventional salt bridge to the protein matrix. Surprisingly, the NmHO C-terminus cleaves spontaneously upon substrate binding (22) to yield a complex with an increased rate of product release. The fact that the necessary acceleration of product release (25) in mammalian HOs occurs (3) by the binding of biliverdin reductase, BVR, at the exposed substrate site(26) that is homologous to the site of the C-terminal interaction in NmHO, demands a more quantitative understanding of the C-terminus in the latter complex.

Figure 2.

Proposed molecular model (24) for the C-terminus of substrate-bound NmHO determined by limited energy minimization based on the proposed donor-acceptor interactions (arrows) His207 (green) peptide NH H-bond donor to carboxylate of Asp27 (gray) and the Arg208 (blue) guanidyl group salt bridges to the carboxylates of the 7-propionate (red) and C-terminal His209 carboxylate, which in turn, is an acceptor to the side chains of both Lys126 and Arg208.

We have pursued detailed ¹H NMR studies (21-24, 27)of systematically perturbed interactions of the substrate with NmHO in general, and its C-terminal fragment in particular, to evaluate the molecular model (Figure 2) and ascertain the structural features that favor the facile cleavage of the C-terminus. Studies of the three sequential C-terminal deletion mutants directly confirmed (24) the substrate contacts to His207 and Arg208 and indicated the existence of a salt bridge for the C-terminal carboxylate with a cationic side chain of both Arg208 and Lys126. Systematic variation of substrate pyrroles A/B substituents revealed (27) a much tighter interaction between substrate and protein matrix, and hence much stronger descrimination between the alternate substrate orientation about the α,γ-meso axis, than in any other HO (20, 28-30), and that the C-terminal cleavage rate increased (24) as the pyrrole A substituent size decreased. Herein we complete these studies by perturbing the substrate propionates, both by replacement of individual propionates by methyls, as well as by interchanging methyl and propionate on one pyrrole, in order to ascertain whether the propionates must be conserved for activity, as found in mammalian HOs (15, 16), and identify the interactions of the propionates at each position.

To this end, we take advantage of a series of modified hemins (31, 32) depicted in Figure 3 designed to illuminate the interaction of substrates with NmHO relative to that in mammalian HOs; labeling of these substrates is based on modifications of protohemin-IX (PHIX) and 2-4-dimethyldeuterohemin-IX (DMDHIX). We emphasize the low-spin azide complex where the novel orbital ground state (33) results in strongly contact-shifted, and hence resolved, signals (24, 27, 34, 35) for the protons on the substrate edge exposed to solvent which makes contact to the C-terminus (see Figure 2). While relaxation for the azide complexes is somewhat stronger for the more frequently studied cyanide complexes (36), it does not seriously interfere with effective 2D NMR analysis (24, 27, 35), and the physiological oxy-complex is more reasonably modeled by the bent azide than the linear cyanide ligand.

Figure 3.

Structure and labeling of modified hemins relevant to the present study (M = methyl); the orientation of the vinyls groups relative to the adjacent methyl on the same pyrrole is shown as deduced by NOESY cross peaks in solution. (A) 7-de-propionate-,7-methyl-protohemin-IX, (7P→M)PHIX; (B) 6-de-propionate-,6-methyl-protohemin-IX, (6P→M)PHIX; (C) protohemin-XIII, PHXIII; (D) 2-,4-dimethyldeuterohemin-IX, DMDHIX; (E) 7-de-propionate-,7-methyl-2-,4-dimethyldeuterohemin-IX, (7P→M)DMDHIX; (F) 7-de-propionate-,7-methyl, 8-de-methyl-,8-propionate-2-,4-dimethyldeuterohemin-IX, (7P→M:8M→P)DMDHIX. The orientations of the substrates, relative to the protein matrix, are depicted as found in the active site shown in Figure 1B, and the determined vinyl orientations are shown at cis (vinyl H_β next to methyl on the same pyrrole) or trans (vinyl H_α next to methyl on the same pyrrole). The arbitrary labeling of the pyrrole positions in two-fold symmetric PHXIII and in the DMDHIX derivatives is selected so as to correspond to the same number position of PHIX in the crystal structure.

MATERIAL AND METHODS

Materials

The wild type, WT, Neisseria meningitidis heme oxygenase, NmHO, was prepared and purified as described in detail previously (21). The two derivatives of protohemin-IX with selective methyl for propionate substitutions, 6-de-propionate-,6-methyl-protohemin-IX, and 7-de-propionate-,7-methyl-protohemin-IX, labeled (6P→M)PHIX and (7P→M)PHIX, 2-,4-dimethyldeuteroheme-IX, DMDHIX, and protohemin-XIII, PHXIII, (which possesses the symmetric 1-,4-divinyl arrangement) are the same as described in detail previously (32). The two hemins with selective removal or rearrangement of propionates, based on 2-,4-dimethyldeuterohemin-IX (labeled DMDHIX, are 7-de-propionate-,7-methyl-2-,4-dimethyldeuterohemin-IX, labeled (7P→M)DMDHIX, and 7-de-propionate-,7-methyl-,8-de-methy,8-propionate-2-,4-dimethyldeuterohemin-IX, labeled (7P→M:-8M→P)DMDHIX, which were prepared as described elsewhere (31). The positions on the DMDHIX and PHXIII derivatives are labeled (see Figure 3) on the basis of the orientation of native protohemin-IX in the crystal structure, as shown in Figure 1B (9, 11). The carboxylates of native protohemin-IX were 95% ¹³C-labeled singly or doubly to yield 6-(¹³C_γO ⁻₂)PHIX and 6,7-(¹³C_γO ⁻₂)₂PHIX, respectively, as described elsewhere (37).

The 1:1 complexes of the substrate of interest with NmHO were prepared as described previously (21) for the PHIX and DMDHIX complexes, except that the mono-propionate hemins were dissolved in 95% H₂O, 5% pyridine, with the pyridine removed immediately after preparation via Sephadex column chromatography, as described previously for the reaction with globins (38). The quantitative incorporation of each substrate into NmHO was confirmed both by the characteristic uv-visible spectra as the cyanide complexes and the identification of the signals in the ¹H NMR spectrum as the azide complexes. Samples for ¹H NMR were concentrated by ultrafiltration to yield ~2mM samples in 50 mM phosphate, 75 mM azide, pH ~7.1.

UV-visible spectroscopy

Spectra were recorded on a Varian Cary 50 uv-visible spectrometer over the range 200-800nm at a rate of 50 nm/s; 32 transients were averaged. The sample NmHO substrate complexes were 30 mM in cyanide 50 mM in phosphate, pH 7.1.

In vitro enzyme assay

The activity of the various substrate complexes of NmHO was determined by following by uv-visible spectroscopy the effect of reaction with ascorbate, in the presence of the iron chelator desferrioxamine, in the conversion of hemin to biliverdin, as described in detail previously (21). Subsequently, the product biliverdin was reacted with rat biliverdin reductase, prepared as described elsewhere (39), and NADPH to generate bilirubin.

NMR spectroscopy

¹H NMR data were collected on Bruker AVANCE 500 and 600 spectrometers operating at 500 and 600 MHz, respectively. ¹H chemical shifts are referenced to 2,2-dimethyl-2-silapentane-5-sulfonate (DSS) through the water resonance calibrated at each temperature. Reference spectra were collected over a 65 ppm spectral width and a repetition rate 2.5 s⁻¹. 600 MHz NOESY spectra (40) (mixing time 40 ms; repetition rate 1.5 to 2.5 s⁻¹) and 500 MHz Clean-TOCSY spectra (41) (to suppress ROESY response; 25°, 35°C, spin lock 25 ms; repetition rate 1-2 s⁻¹) were recorded over a bandwidth of 20 and 45 ppm (NOESY) and 20 ppm (TOCSY) using 512 t1 blocks of 128 and 256 scans each consisting of 2048 t2 points. 2D data sets were apodized by 30°- or 45°-sine-squared-bell-functions and zero-filled to 2048 × 2048 data points prior to Fourier transformation.

¹H-decoupled ¹³C NMR spectra were recorded at 125 MHz on a Bruker AVANCE 500 spectrometer over a bandwidth 300 ppm at the repetition rate of 1 s⁻¹ at 25° C utilizing 32K data points. The ¹³C chemical shifts were indirectly referenced to the proton spectrum (42).

RESULTS

Structural integrity of propionate-modified substrate complexes

The reaction of (6P→M)PHIX and (7P→M)PHIX with NmHO, in the presence of excess cyanide resulted in the same characteristic changes in their uv-visible spectra as observed for native PHIX, while reaction of (7P→M)DMDHIX and (7P→M:8M→P)DMDHIX with NmHO in the presence of excess cyanide resulted in the same uv-visible spectral changes as observed for the DMDHIX complex (Supporting Information Figure S1). The near identity of the uv-visible spectra of the NmHO complexes with the propionate modified and “normal” dipropionate substrates support the incorporation of the modified substrate into an environment that is essentially the same as that for PHIX and DMDHIX. The ¹H NMR spectra of the azide complexes of (7P→M)PHIX and (6P→M)PHIX are presented in Figures 4B and 4C, respectively, where they can be compared with that of the complex with native PHIX (Figure 4A). Also included is the ¹H NMR spectrum of the azide complex for the two-fold symmetric PHXIII (Figure 4D). The NMR spectra of the NmHO azide complexes of (7P→M)DMDHIX and (7P→M:8M→P)DMDHIX are shown in Figures 5B and 5C, respectively, and can be compared with that of the previously characterized DMDHIX complex (24) shown in Figure 5A. The NMR spectra in Figures 4B, 4C, 4D, 5B and 5C display a strongly dominant single set of hyperfine shifted methyl signals which indicate an unique orientation of the modified substrates within an active site very similar to that observed for the “normal” dipropionate PHIX and DMDHIX.

Figure 4.

Resolved portions of the 600 MHz ¹H NMR reference spectra, in ¹H₂O, 50 mM phosphate, 75 mM azide, pH 7.1 at 25°C, for: (A) WT NmHO-PHIX-N₃; (B) WT NmHO-(7P→M)PHIX-N₃; (C) WT NmHO-(6P→M)PHIX-N₃; and (D) WT NmHO-PHXIII-N₃. The spectrum for the “aged” or C-terminal cleaved NmHO complex, NmHO^X-(7P→M)PHIX-N₃, is shown in (B’).

Figure 5.

Resolved portions of the 600 MHz ¹H NMR reference spectrum, in ¹H₂O, 50 mM phosphate, 75 mM azide, pH 7.1 at 25°C for: (A) WT NmHO-DMDHIX-N₃; (B) WT NmHO-(7P→M)DMDHIX-N₃; and (C) WT NmHO-(7P→M:8M→P)DMDHIX-N₃. The NMR spectra of the “degraded” or C-terminal cleaved complexes, NmHO^X-(7P→M)DMDHIX-N₃, and NmHO^X-(7P→M:8M→P)DMDHIX-N₃, are shown in (B’) and (C’), respectively.

Activity of modified substrate complexes

The complexes with modified substrates all underwent reaction with ascorbate in the presence of the iron chelator, desferrioxamine, in a standard assay (22) that yielded uv-visible spectra with characteristic broad bands at ~380 and ~670 nm (43) indicative of biliverdin formation, with the resulting spectra largely indistinguishable from those of the related substrates with the two normal propionates (Supporting Information Figure S2). Subsequent reaction of the biliverdin with rat biliverdin reductase (39) and NADPH afforded the spectra indicative of bilirubin formation. Thus, HO activity is conserved upon either deletion of one propionate or exchange between a propionate and methyl on the same pyrrole.

Substrate/active site residue assignment protocols

Substrate protons for the major isomers in solution are labeled M_i(H_i) for methyl (single proton) at pyrrole positions i=1-8 (see Figure 1). Residue protons are labeled by residue number and position; peptide protons are labeled only by position. Unambiguous assignment of the substrate resonances is obtained by characteristic NOESY connectivities about the periphery and is facilitated by the expected pattern of substrate hyperfine shifts (24, 27, 35). On the one hand, cross peaks between a pair of methyls on the same pyrrole (adjacent to a common meso position) are strong (weak). Methyl to vinyl H_βs (H_αs) NOESY cross peaks are strong for a vinyl oriented cis (trans) to its adjacent methyl (27) as depicted in Figure 3. On the other hand, the known orbital ground state for the azide complex places large τ spin-density (24, 27, 34, 35)only at positions 1, 4, 5 and 8, such that methyl and vinyl H_αs appear strongly shifted to low field, with the vinyl H_βs exhibit strongly upfield shifted and resolved resonances. Substrate chemical shifts for pyrrole α-substituents are listed in Table 1; complete substrate assignments and active site residue assignments are provided in Supporting Information, Tables S1, S2 and S3, S4.

Table 1.

Chemical shifts for variable substrates in azide complexes of WT NmHO^a)

Substrateb)	PHIXc),d)	PHIXd),e)	(6P→M)- PHIX	(7P→M)- PHIX	DMDH- IX c)	(7P→M)- DMDH- IX	(7P→M:- 8M→P)- DMDHIX
Peake)
M1(H1α)f)	18.68	19.85	20.55	18.34	21.42	20.63	22.90
M2(H2α)	(10.32)	(10.22)	(10.75)	(9.04)	10.10	10.62	10.10
M3	7.60	7.60	8.45	7.66	11.39	?	11.80
M4(H4α)	(11.78)	(18.60)	(18.65)	(11.82)	23.00	22.35	23.64
M5	18.02	20.65	21.57	17.90	18.05	17.60	17.84
M6	-		14.58	-	-	-	-
M7	-		-	12.95	-	14.29	15.70
M8	20.00	19.12	18.79	20.57	19.67	19.70	-
H6α1	9.62	-	-	9.78	8.73	8.38	9.90
H6α2	2.00	-		1.65	5.35	4.06	5.40
H7α1	9.82	10.13	9.38	-	12.17	-	-
H7α2	5.65	2.50	3.25	-	6.70	-	-
H8α1	-	-	-	-	-	-	13.81
H8α2	-	-	-	-	-	-	8.30
His23 Cβ1H	12.56	12.67	12.74	12.65	12.65	12.85	12.77
His23 Cβ2H	9.68	9.78	9.77	9.85	9.28	9.68	9.25

^a)Chemical shifts in ppm, referenced to DSS, in ¹H₂O, 50 mM phosphate, 75 mM azide, pH ~7.1 and 25°C.

^b)Substrate structures and proton labels are shown in Figure 3.

^c)Crystallographic orientation, as in Figure 1B.

^d)Orientation reversed about the α,γ-meso axis relative to that in the crystal; data taken from Ref. (24).

^e)Orientation rotated 180° about the α,γ-meso axis relative to that in the crystal.

^f)Vinyl H_α chemical shifts are given in parentheses; vinyl and propionate C_βH chemical shifts are provided in Supporting Information, Tables S1 and S2.

Active site residues are assigned sequence-specifically (44) by characteristic backbone NOESY cross peaks among TOCSY-detected residues characteristic of an α-helix (data not shown) as described in detail previously for a series of NmHO and hHO azide complexes (24, 27, 35). These studies afford the assignments for the helical fragments; proximal helix Thr19-Val26, Phe52-Lys54, distal helix Tyr112-Gly116, Phe123-Phe125, Arg140-Leu142 and Ala180-Tyr184. The two residues focal to establishing the orientation of a substrate in the active site (27) are Phe123 in contact with substituents at positions 1 and 8, and Cys113 in contact with substituents at positions 4 and 5 in the crystal structure (Figure 1B). Chemical shifts for the axial His23 are included in Table 1 and for other residues are provided Supporting Information, Tables S3 and S4.

Seating of propionate-modified substrates

To uniquely place modified substrates in the active site of NmHO we will describe the proton at a given position on the modified substrate, as labeled in Figure 3, and identify where each of those protons on the substrate is placed in the active site defined by the position of the PH proton in the crystal structure, as shown in Figure 1B (9, 11). Portions of the ¹H NMR NOESY spectrum of NmHO-(7P→M)PHIX-N₃ for assigning the substrate and active site residues relevant to substrate seating are shown in Figure 6. A weak cross peak between two low-field methyls (Figure 6B) locates the M₁/M₈ pair, and an intense cross peak of one of these to a more weakly low-field shifted methyl (Figure 6H) identifies the M₈/M₇ pair. Intense cross peaks of high-field resolved vinyl H_βs to a partially resolved methyl and a strongly low-field shifted H_α locate the 4-vinyl group and M₃ (not shown). The remaining low-field methyl exhibits cross peaks to a moderately low-field shifted pair of H αs and H_βs that identify the 5-methyl and 6-propionate. The M₁ cross peaks to another vinyl H βs locate the 2-vinyl group and establish that both vinyl groups are oriented cis, as in the crystal structure (9, 11)for the PHIX complexes, as shown in Figure 3B. The M₁ and M₈ cross peaks to Phe123 and the M₅ (and 4-vinyl; not shown) cross peak to Cys113 C_αH (Figure 6A) establish an orientation where the lone propionate is in the position of the 6-propionate in the crystal structure (Figure 1B). M₁/M₈ contacts to two residues not predicted by the crystal structure are observed to both the ring C_δH (Figure 6I) and C_βH₂ (Figure 6A) of a His and the C_αH (Figure 6B) of another residue, in a fashion very similar to that observed in PHIX and DMDHIX complexes (24, 35). The molecular model and C-terminal deletions have shown (24) that these contacts involve His207 and the C_αH of Arg208.

Figure 6.

Resolved portions of the 600 MHz ¹H NMR NOESY spectrum, in ¹H₂O, 50 mM phosphate, 75 mM azide, pH 7.1 at 25°C for WT NmHO-(7P→M)PHIX-N₃; illustrating key intra-substrate (A, B, G, H), substrate residue contacts (A, C, D, E, and I) for Phe52, Cys113, Phe123, Leu124, His207 and Arg208, and intra-residue contacts (C, F) for His207 that uniquely assign the substrate and determine the orientation of the substrate in the pocket. Noted are the His207 C_βH and C_δH contact with M₈ (E, I) and His207 with M₁ (E). Panels (A’, D’, E’ and G’) for the C-terminal cleaved complex NmHO^X-(7P→M)PHIX-N₃ exhibit the same data as panels A, B, D, E, G, H for the WT complex; except for the loss of the His207 and Arg208 cross peaks to M₈/M₁. All panels except C, F and I utilize 45 ppm bandwidth, 2.5 s⁻¹ repetition rate and 45° phase-shifted sine-bell apodization; panels C, F and I utilize 20 ppm bandwidth, 1.0 s⁻¹ repetition rate and 30° phase-shifted sine-bell apodization.

The reference spectrum for NmHO-(6P→M)PHIX-N₃ exhibits somewhat broader peaks than the other complexes but still allows the necessary assignments (Supporting Information Figure S3). The weak, resolved intermethyl cross peak that identifies M₁/M₈ and their cross peaks to the Cys113 C_αH demonstrate that the orientation of the substrate is rotated 180° about the α,γ-meso position relative to native PHIX in the crystal (as shown in Figure 3B). Strongly upfield shifted vinyl H_βs exhibit weak cross peaks to the remaining low-field methyl and to the vinyl H_α, which in turn, exhibits a cross peak to a partially resolved methyl, identifying M₅, M₃ and the 4-vinyl in the trans orientation (as shown in Figure 3C). A cross peak from M₁ to a vinyl H α shows that the 2-vinyl is similarly in the trans orientation (as depicted in Figure 3B). M₅ in the 180°-rotated orientation exhibits cross peaks to a His C_βH₂ and the C_αH of another residue that dictates their origin as His207 and Arg208.

The NOESY spectrum of NmHO-(7P→M)DMDHIX-N₃ (Supporting Information Figure S4) locates three pairs of strongly coupled methyls with weak coupling between the pairs, and another weak coupling to a seventh methyl that uniquely identify M₁-M₅ and M₇, M₈. The M₅ contacts to C_βHs locates the 6-propionate. The M₁/M₈ cross peaks to Phe123 and the M₅/M₄ cross peaks to Cys113 uniquely determine the orientation where the lone propionate occupies the same position as does the 6-propionate in the crystal structure (Figure 3E). The M₈/M₁ exhibit NOESY cross peaks to both the C_βHs and C_δH of a His and the C_αH of another residue that identify His207 and Arg208. It is noted that M₈ contacts occur to both the backbone and ring of His207; this latter contact was much weaker or undetected in the complexes with the native PHIX (or DMDHIX) possessing the normal propionates at both positions 6 and 7.

The NOESY spectrum of NmHO-(7P→M:8M→P)DMDHIX-N₃ depicts two pairs of strongly coupled methyls which collectively identify the pairs M₁/M₂ and M₃/M₄. A weak cross peak between the pairs, as well as a weak NOE from one of the pairs to a low-field methyl, identifies M₁-M₅; the M₅ contacts to a C_αH locate the 6-propionate (Supporting Information Figure S5). A less-strongly low-field shifted methyl exhibits cross peaks to a low-field shifted C αH and C_βHs of a propionate identifying the remaining M₇ and 8-propionate; H_8α exhibits a weak cross peak to M₁. Contacts of M₁/M₈ with Phe123 and M₄/M₅ with Cys113 establish the orientation with one propionate occupying the crystallographic 6-propionate position and the other in the crystallographic 8-positions (as depicted in Figure 3F). The contact of H_8α to a C_αH of one residue, and M₁/H_8α contacts to the C_βHs of an apparent His, represent the interaction of His207 and Arg208 with the active site. It was not possible to locate the His207 C_δH, but M₁ and H_8α cross peaks to C_βHs not predicted by the crystal structure indicate their origin as His207.

The NOESY spectrum of NmHO-PHXIII-N₃ (not shown) identifies all the substrate signals with the NOESY cross peak of the 1-vinyl H_1β to M₈ and Phe123 and H_1α cross peak to M₂, indicating that the vinyl group is in a trans position (as depicted in Figure 3C). The 4-vinyl group exhibits the same cross peaks as in PHIX and hence retains the conventional cis orientation. The M₈/H_1β cross peaks to a C_βH₂ group and the M₈ cross peak to a C_αH locate His207 and Arg208. All chemical shift data are provided in Supporting Information, Tables S1 and S3.

Loss of substrate-residue contacts upon NmHO “aging”

Over a period of a few days at 25°C, the resolved resonances of the WT NmHO complexes with (7P→M)PHIX, (7P→M)DMDHIX and (7P→M:8M→P)DMDHIX disappear and are replaced in the conversion to an “aged”, or homogeneously degraded, derivative which we label NmHO^X. The reference spectra for the three “aged” derivatives are illustrated in Figure 4B’ and Figures 5B’ and 5C’, respectively. Concomitant to the generation of NmHO^X are observed the appearance of two very narrow resonances in the aromatic spectral window. Similar conversion of NmHO to NmHO^X has been characterized in detail for both PHIX and DMDHIX complexes and the product shown (21, 22) to be the result of cleavage of the Arg208His209 dipeptide. In contrast, the WT NmHO complexes of (5P→M)PHIX and PHXIII exhibit only minor (<10%) “aging” over the same period of time, dictating that the terminal cleavage rate must be significantly (factor >10) slower relative to that for either the other propionate modified or native PHIX and DMDHIX substrates. Unfortunately, the very sparingly available complexes of the propionate-modified substrates were heterogeneously decomposed upon completion of 2D NMR, precluding meaningful mass spectrometric studies to characterize the initially cleaved peptide.

The NOESY spectrum of NmHO^X-(7P→M)PHIX-N₃ provides the same assignment as for the WT NmHO complex (not shown). The most prominent difference is that the M₁/M₈ contacts to the His207 and Arg208 are lost (see Figure 6A’). Similar NOESY data for the NmHO^X complexes of (7P→M)DMDHIX and (7P→M:8M→P)DMDHIX (Supporting Information Figures S6 and S7, respectively), in each case, exhibit a largely conserved active site structure but with the loss of the M₁/M₈ and M₁/H_1α contacts to His207 and Arg208. The loss of these contacts is consistent with cleavage of the C-terminus. The substrate chemical shifts are perturbed upon “aging” in a manner close to that observed (21, 23) for PHIX and DMDHIX. Active site residues chemical shifts are only inconsequentially perturbed by “aging”. Both substrate (Tables S1, S2) and active site residue (Tables S3, S4) chemical shifts for the NmHO^X complexes are provided in Supporting Information.

Effect of C-terminal cleavage on the substrate propionate carboxylates

The conversion of native NmHO–PHIX-H O to yield NmHO^X 2 -PHIX-H₂O by the loss of Arg208His209 is relatively rapid (half life 24 h) and is readily followed (22) by ¹H NMR via the relative intensities of the resolved M₈ peaks. Figure 7A provides the ¹³C NMR spectrum in the carboxylate spectral window for NmHO-(6,7-¹³CO ⁻₂)₂PHIX-H₂O (~90% NmHO and ~10% NmHO^X), which reveals two well-spaced signals, labeled a and b, uniquely attributable to the two labeled carboxylates. The spectrum of this complex upon ~25% degradation to NmHO^X is shown in Figure 7B and identifies the degraded species with peaks labeled c and d; clearly the chemical shift for one propionate is conserved upon C-terminal cleavage while the other one significantly perturbed. The ¹³C NMR spectrum of a partially degraded, singly labeled complex, NmHO-(6-¹³CO ⁻₂)PHIX-H₂O (~75% NmHO, ~25% NmHO^X), is shown in Figure 7C and confirms that the propionate sensitive to C-terminal cleavage is clearly that at the crystallographic 7 position.

Figure 7.

Portion of the 151 MHz ¹³C NMR reference spectrum for the carboxylate chemical shift window for: (A) ~90% WT NmHO-(6-,7-¹³CO ⁻₂)₂PHIX-H₂O, ~10% C-terminal cleaved NmHO^X-(6-,7-¹³CO ⁻₂)₂PHIX-H₂O; (B) ~70% WT NmHO-(6-,7-¹³CO ⁻₂)₂PHIX-H₂O, ~30% C-terminal cleaved NmHO^X-(6-,7-¹³CO ⁻₂)₂PHIX-H₂O; and (C) ~75% WT NmHO-(6-¹³CO ⁻₂)PHIX-H₂O, ~25% C-terminal cleaved NmHO^X-(6-¹³CO ⁻₂)PHIX-H₂O, which assign peaks a and b to the 7- and 6-propionate carboxylates, respectively, in the WT NmHO complex, and peaks c and d to the 6- and 7-carboxylates, respectively, in the C-terminal cleaved NmHO^X complex.

DISCUSSION

Structure of the active site

Both uv-visible and ¹H NMR spectroscopy clearly show that each of the propionate-modified substrates quantitatively incorporate into the active site of NmHO; standard in vitro assays (22) show that the complexes retain full activity. The ¹H NMR data establish that the substrates seat overwhelmingly in a single orientation that in each case dictates that the only propionate protein salt-bridge crucial to substrate binding to NmHO is that between Lys16 with the 6-propionate position in the crystal structure (9, 11). The strong conservation of active site structure, in spite of the drastic propionate deletion/rearrangement, is confirmed by not only the conserved pattern of NOESY cross peaks, but also by the conservation of the pattern of hyperfine shifts. On the one hand, non-ligated active site residues exhibit only dipolar shifts (36, 45) which sensitively reflect the anisotropy/orientation of the paramagnetic susceptibility tensor, χ, and the position of the residue. Even small changes in the orientation of χ can result in large changes in the dipolar shift pattern (23, 46). The dipolar shifts for crystallographically detected active site residues are highly conserved among the variable substrates and upon C-terminal cleavage (Supporting Information Tables S3 and S4) dictating that both the orientation of χ and the residue geometries are conserved among the complexes.

On the other hand, ligated His23 and substrate hyperfine shifts are dominated by the contact interaction (36, 45) (and to a lesser degree, by 2,4-substituents) (47). The data in Table 1 show that the pattern of the His23 C_βHs, and hence the orientation of the His imidazole relative to the protein matrix, is conserved for the various substituents. Substrate methyls very sensitively reflects the in-plane rotational orientation of the substrate relative to the axial histidine imidazole plane(36, 48), and by inference, the protein matrix. Thus the pattern of the M₁, M₃, M₅ and M₈ shifts is inconsequentially differentiated within the pair of complexes of PHIX and (7P→M)PHIX-N₃ (see Table 1); the same highly conserved pattern of M₁, M₂, M₃, M₄ and M₅ shift pattern is observed within the three complexes of DMDHIX , (7P→M)DMDHIX-N₃ and (7P→M:8M→P)DMDHIX. For the (6P→M)PHIX complex with the “reversed” orientation (Figure 3B), the pattern of methyl shifts is conserved relative to that for the equilibrium ~20% populated “reversed” orientation (27) of native PHIX (data in column 2 of Table 1). This dictates that the in-plane orientation of the substrate is inconsequentially affected by either deletion or rearrangement of one of the propionates. Lastly, the low-field bias for H-bond signals for Gln49 and His53, which are linked to the axial ligand via the catalytic water molecules (9, 11), is strictly conserved.

Structure of the C-terminus and its cleavage

Each of the NmHO complexes of these propionate-modified substrates exhibit contacts to residues at the crystallographically exposed 1 and 8 positions, (i.e., the positions of the PHIX 1CH₃ and CH₃ in the crystal structure; Figure 1B) that are not described in the crystal structure in which the C-terminal His207Arg208His209 fragment is undetected (9, 11). This absence in the crystal was attributed to structural disorder, but possible loss of the C-terminus by an unusual degradation clearly established in solution by ¹H NMR and mass spectrometry(22), has not been yet ruled out. In any case, these contacts have been assigned to His207 and Arg208, and were confirmed by C-terminal deletion (22). A molecular model (24) based on limited energy minimization using NOESY constraints and a series of three salt bridges and one H-bond involving the three C-terminal residues, the side chain of Lys126, the carboxylate of Asp27 and the propionate at the crystallographic 7-position, is reproduced in Figure 2. The observation of very similar His207 and Arg208 interactions with the active site with each of the modified substrates indicates that the architecture of the interaction of the C-terminus with the active site is qualitatively retained upon loss of the 7-propionate to Arg208 salt bridge.

The relatively modest effect on the C-terminal structure upon deleting the 7 propionate could be attributed to the failure of the 7 propionate to interact with the C-terminus, as described by our model (22, 24). However, the ¹³C chemical shift for the PHIX propionate carboxylates (Figure 7) is strongly perturbed for the 7 propionate, but not for the 6 propionate, upon C-terminal cleavage, confirming that the 7 propionate does interact with the C-terminus in the native complex. The loss of the 7 propionate salt bridge may weaken, but not significantly rupture, the interaction of the C-terminus with the active site. This confirms the relatively robust nature of this interaction of the C-terminus with the active site (24). Qualitative changes upon deleting the 7 propionate include slightly altered hyperfine shifts for His207 in spite of conserved χ, and more intense NOESY cross peaks between 8CH₃ and the His207 C_δH than in complexes with the native propionates. Inasmuch as the C-terminal interaction with the active site in 7-propionate deleted substrates is not biologically relevant, no attempt was made to model this interaction.

It is not clear whether the propionate at the crystallographic 8-position in NmHO-(7P→M:8M→P)DMDHIX-N₃ interacts with the terminus of Arg208. The molecular model suggests that the carboxylate of the 8-propionate could orient to form the salt bridge. The fact that the C-terminal cleavage leads to significant chemical shift changes for the 8-propionate C_βH chemical shifts (−1.50, −0.80 in WT NmHO to −0.80, −0.49 in NmHO^X) suggests that a salt bridge between the 8 propionate and Arg208 side chain may occur in this complex.

NMR studies (22, 24) with variably 2-,4-disubstituted substrates had shown that the rate of C-terminal cleavage correlates with the proximity of the His207 to the substrate, with the rate of cleavage significantly enhanced upon replacing the bulky vinyls with methyls or hydrogens. Both (6P→M)PHIX and PHXIII have the vinyl at the crystallographic 1 position oriented trans (H_βs close to position 8; see Figures 3B and 3C), which will sterically interfere with His207 approaching the substrate. The retarded rates of cleavage for these two modified substrates are consistent with its model.

Comparison to other HOs and functional implications

Clearly NmHO requires but a single propionate salt bridge to the protein matrix to effectively bind substrate, in contrast to mammalian HO (15, 16). Moreover, protohemin-I (the analog to (7P→M:8M→P)PHIX, but with 2,4-vinyls) fails to yield an active hHO complex (15). In contrast, the (7P→M:8M→P)PHIX complex (Figure 3F) retains full activity in NmHO. The fact that deleting one propionate abolishes substrate binding for hHO but not NmHO dictates that other substrate-enzyme interactions must be stronger in the NmHO than hHO complex. Since one salt bridge between the propionate at position 6 (Figure 1) with the side chain terminus of a Lys (Lys16 for NmHO, Lys18 for hHO) is conserved, it is reasonable to attribute the stronger interactions in NmHO to van der Waals contacts primarily with pyrroles A and B. Stronger van der Waals interactions have been implied in the much more compact active site in crystals (2, 9, 11). A similar conclusion has been reached on the basis of the much stronger orientational preference of substrates about the α-, γ-meso axis (27) when compared to other HOs (20, 28-30). Hence, the stronger substrate-enzyme van der Waals interactions in NmHO substrate complexes can be viewed as a mechanism for “liberating” one of the two propionates for a functional role unnecessary in other HOs, namely the ordering of the C-terminus to interact with the active site.

The solution NMR-detected C-terminal interaction with the active site can be expected to have functional implications for two distinct processes that most likely occur at the exposed substrate edge that, in NmHO, is “shielded” by the interaction of the C-terminus with the active site (24). On the one hand, HO-biliverdin complexes dissociate too slowly to sustain effective biological catalysis (25). For mammalian systems the product biliverdin is not released until a transient 1:1 complex is formed with BVR with contact at the exposed substrate edge (26). The BVR acts as a cooperative facilitator that significantly accelerates the rate of biliverdin release and at the same time precludes the presence of significant concentrations of the toxic biliverdin. While bilirubin may not be toxic for bacteria, there is still the need for the transient binding of an allosteric effector (the BVR analog) to accelerate biliverdin release. On the other hand, while cytochrome-P450 reductase and NADPH or ascorbate react with mammalian systems directly to yield iron-free biliverdin (3), the same reactions with NmHO complexes proceed only to the iron (III)-biliverdin complex where the iron can only be extracted by a powerful iron chelator (2, 49). The final reduction step for NmHO complexes must rely on the specific NmHO electron donor. It is possible, that once the iron biliverdin complex of NmHO is formed, the native electron donor protein may serve as both a source of the key electron for the iron as well as an allosteric effector to labilize the resulting iron-free biliverdin. Unfortunately, neither the BVR analog nor the electron donor in pathogenic bacteria is known (2). A more detailed understanding of the fate of biliverdin in N. meningitidis would be necessary to provide fruitful insight into the unique C-terminal structural role of NmHO.

CONCLUSIONS

NmHO binds substrates possessing a single propionate to form active complexes by either deleting or rearranging the position of one of the propionates of the substrate. This is in strong contrast with hHO where neither the propionate deletion nor rearrangement affords and active complex. The unique salt bridge for NmHO is between the propionate at the crystallographic 6 position and the terminus of Lys16. The second substrate propionate at the normal crystallographic 7 position (and possibly at the unprecedented 8 position), forms a salt bridge to the (crystallographically undetected) side chain of Arg208 as a portion of the stabilization of the interactions of the C-terminus with the active site. The solvent-exposed side of the substrate consisting of the crystallographic 1 and 8 positions interacts with the C-terminus in a qualitative conserved manner upon elimination of the 7-propionate, as described by an earlier molecular model (24), although altered hyperfine shifts suggest some minor rearrangement in the C-terminus architecture. Spontaneous C-terminal cleavage abolishes the characteristic contacts between the C-terminus and active site with the exception for substrates oriented with the vinyl group at the crystallographic 1 position, which interferes with the interaction of the C-terminus with the active site. The effective binding of a substrate with a single propionate by NmHO, instead of the required two from mammalian HOs, is attributed to a larger contribution to the substrate binding from peripheral van der Waals interactions, as supported by more compact structure (2, 9, 11), fewer vacancies and much stronger orientational preference about the α,γ-meso axis (27). These structural differences are proposed to have evolved, in part, to “liberate” one propionate for stabilizing interactions of the active site with the C-terminus, which, in turn, is proposed to function in product release.

Abbreviations used

BVR

biliverdin reductase

DSS

2,2-dimethyl-2-silapentane-5-sulfonate

HO

heme oxygenase

hHO

human heme oxygenase-1

NmHO

Neisseria meningitidis heme oxygenase

CdHO

Corynebacterium diphtheriae heme oxygenase

PaHO

Pseudomonas aeruginosa heme oxygenase

NOESY

two-dimensional nuclear Overhauser spectroscopy

TOCSY

two-dimensional total correlation spectroscopy

ROESY

rotating-frame two-dimensional Overhauser spectroscopy

PHIX

protohemin-IX

PHXIII

protohemin-XIII

DMDHIX

2-,4-dimethyldeuterohemin-IX

(6P→M)PHIX

6-de-propionate-,6-methyl-protohemin-IX

(7P→M)PHIX

7-de-propionate-,7-methyl-protohemin-IX

(7P→M)DMDHIX

7-de-propionate-,7-methyl-2-,4-dimethyldeuterohemin-IX

(7P→M:8M→P)DMDHIX

7-de-propionate-,7-methyl,8-de-methyl,8-propionate-2-,4-dimethyldeuterohemin-IX