Báo cáo khoa học: Hyperthermal stability of neuroglobin and cytoglobin doc

Hyperthermal stability of neuroglobin and cytoglobin Djemel Hamdane 1 , Laurent Kiger 1 , Sylvia Dewilde 2 , Julien Uzan 1 , Thorsten Burmester 3 , Thomas Hankeln 4 , Luc Moens 2 and Michael C Marden 1 1 Inserm U473, Le Kremlin-Bice ˆ tre, France 2 Department of Biomedical Sciences, University of Antwerp, Belgium 3 Institute of Zoology, Johannes Gutenberg University of Mainz, Germany 4 Institute of Molecular Genetics, Johannes Gutenberg University of Mainz, Germany Neuroglobin (Ngb) and cytoglobin (Cygb) have been identified recently [1–3] as new globins in vertebrates. Ngb is predominately expressed in certain regions of the brain as well as in some endocrine tissues [1,4] and at a higher level in the retina [5], whereas Cygb is expressed in all tissues. Although sequence analysis reveals little similarity with hemoglobin (Hb) and myo- globin (Mb), Ngb and Cygb share many of the charac- teristics of the globins such as reversible oxygen binding and the overall three-dimensional globin fold [6–8]. In the absence of external ligands such as oxy- gen, Ngb and Cygb are hexa-coordinated with a bis- histidyl heme [9]. The primary functions of Ngb and Cygb remain unknown, but some hypotheses have been suggested [9–11]. Expression of Ngb increases in cultures of neu- rons under hypoxic conditions, and it appears that Ngb may protect cells against hypoxia [12,13]. Thus Ngb may have a Mb-like function, supplying the respiratory chain of neuronal mitochondria with O 2 [1,5]. Cygb may be a signaling or sensor protein [14], and may be involved in collagen synthesis or in the protection against reactive oxygen species [15]. Other physiological roles, however, such as electron transfer, peroxidase activity, NO binding or NO detoxification, as observed for other hemoproteins, are still conceivable. Unique in the globin family, Ngb and Cygb possess cysteine residues capable of forming a disulfide bond [16]. Reduction of the disulfide bond in Ngb increases the affinity for the distal histidine by a factor of nearly Keywords cytoglobin; disulfide bond; ligand kinetics; neuroglobin; protein melting temperature Correspondence M. C. Marden, Inserm U473, 78 rue du General Leclerc, 94275 Le Kremlin-Bice ˆ tre, France E-mail: marden@kb.inserm.fr (Received 6 December 2004, revised 14 February 2005, accepted 1 March 2005) doi:10.1111/j.1742-4658.2005.04635.x Neuroglobin (Ngb) and cytoglobin (Cygb), recent additions to the globin family, display a hexa-coordinated (bis-histidyl) heme in the absence of external ligands. Although these proteins have the classical globin fold they reveal a very high thermal stability with a melting temperature (T m )of 100 °C for Ngb and 95 °C for Cygb. Moreover, flash photolysis experi- ments at high temperatures reveal that Ngb remains functional at 90 °C. Human Ngb may have a disulfide bond in the CD loop region; reduction of the disulfide bond increases the affinity of the iron atom for the distal (E7) histidine, and leads to a 3 °C increase in the T m for ferrous Ngb. A similar T m is found for a mutant of human Ngb without cysteines. Appar- ently, the disulfide bond is not involved directly in protein stability, but may influence the stability indirectly because it modifies the affinity of the distal histidine. Mutation of the distal histidine leads to lower thermal sta- bility, similar to that for other globins. Only globins with a high affinity of the distal histidine show the very high thermal stability, indicating that sta- ble hexa-coordination is necessary for the enhanced thermal stability; the CD loop which contains the cysteines appears as a critical region in the neuroglobin thermal stability, because it may influence the affinity of the distal histidine. Abbreviations Cygb, cytoglobin; GdmCl, guanidinium chloride; Hb, hemoglobin; Mb, myoglobin; Ngb, neuroglobin; T m , melting temperature. 2076 FEBS Journal 272 (2005) 2076–2084 ª 2005 FEBS 10; it has therefore been hypothesized that reduction of the SS bridge may promote oxygen release [16]. Proteins, and particularly enzymes, are generally quite sensitive to environmental changes, e.g. elevated temperatures, due to highly cooperative unfolding [17]. However, there are some exceptions such as those found in extreme thermophilic microorganisms. Com- parison of the protein structure from mesophiles and thermophiles has allowed some explanation of thermo- stability based on small solvent-exposed surface area [18], increased packing density [19–21], core hydro- phobicity [22], decreased surface loop length [21], and the generation of salt bridges or hydrogen bonds betweens polar residues [23–25]. The affinity of apo- globin for heme or the orientation of the heme in the pocket cavity may play a major role in the stability of the holoprotein [26]. In general, few proteins are stable above 80 °C; examples are calcium-binding proteins such as calmodulin or troponin C with T m >90°C for the Ca-bound form [27]. Results Spectroscopy The visible spectrum of dithionite-reduced Ngb, Cygb, and Drosophila Hb showed characteristic absorption maxima of hexa-coordinated (bis-histidyl) species [3,9, 28]. We observed enhanced absorption of the alpha band at 560 nm, a signature of the hexa-coordinated form. The far-ultraviolet circular dichroism spectrum (190– 250 nm) of ferric Ngb was typical of the globin family (Fig. 1) showing mainly an alpha helical secondary structure, in agreement with the X-ray structure [6]. The spectra for native Ngb had negative bands at 208 and 222 nm (Fig. 1), as expected for a high percentage of alpha helix. Analysis of the secondary structure of Ngb gave 78% alpha helix and 22% of other forms, similar to HbA which was used as a control. The spec- trum for Cygb showed slightly less alpha helix, as expected if the extra residues ( 20 at each extremity) are not helical. Human Ngb has cysteine residues at positions 46 (CD7), 55 (D5) and 120 (G19). The cysteines CD7 and D5 (Fig. 2), may form a disulfide bond within the CD loop [16]. However, in mouse Ngb there are only two cysteine residues (D5 and G19) and thus no intradisul- fide bond is present. A similar circular dichroism spec- trum was observed for mutant Ngb without cysteine residues (triple mutation C46G C55S C120S, which we refer to as CCC fi GSS), and for the mutant with modified distal (E7) residue (data not shown). These experiments suggest that the wild-type and mutant Ngb proteins are correctly folded to the structure typical of globins. Thermal denaturation Changes in the far-UV circular dichroism signal at 222.6 nm were used to follow the thermal unfolding. The circular dichroism spectra vs. temperature revealed a high thermal stability for Ngb and Cygb. The melt- ing profiles are shown in Figs 3–6. The melting temperature (T m ) for human Ngb was 100 °C for the ferrous form,  20 °C higher than that for horse heart myoglobin (Mb). The mutant of Fig. 1. Circular dichroism spectra in the far-UV region of human Ngb (…), human Cygb (——), and human HbA (– – –) at 25 °Cin 1m M phosphate buffer at pH 7. Fig. 2. Crystallographic structure of human Ngb mutant CCC fi GSS (6). The hexa-coordination by the E7 (65) and F8 (97) histidines helps stabilize the protein. The sites for the cysteines (CD7 and D5) are shown in green; the disulfide bond (which decreases the E7 histidine affinity) decreases the melting temperature slightly, indica- ting an indirect effect on the stability. D. Hamdane et al. Thermal stability of neuroglobin FEBS Journal 272 (2005) 2076–2084 ª 2005 FEBS 2077 human Ngb without cysteines (CCC fi GSS) or sam- ples of Ngb with dithiothreitol (to break the disulfide bond, Fig. 3) or mouse Ngb (which does not have the internal disulfide bond) had a T m value > 100 °C (Table 1). This would suggest that Ngb without the di- sulfide bond is the most stable form. Because loss of the disulfide bond in human Ngb increases the affinity of the distal histidine (Table 1), the protein stability may depend more directly on the hexa-coordination rather than the disulfide bond. The state of the iron atom may also influence T m (Fig. 4). For all species studied, we observed that the deoxy form was the most stable (Table 1). The T m value of the deoxy ferrous species was obtained after incubation of protein in dithionite under nitrogen. Note that a rapid autoxidation at high temperatures may prevent measurements on samples that remain fully ferrous. The fact that ligands CO or CN – decrease the T m also suggests that the most stable form is that in which the protein forms a sort of clamp Fig. 3. Effect of the cysteine bridge of human Ngb on the thermal stability. The melting temperature, corresponding to the peak of this curve of the first derivative of the circular dichroism signal vs. temperature, is shifted to higher values when the disulfide bond is broken with dithiothreitol or for the mutant without cysteines. Experiments were performed in 10 m M phosphate at pH 7 for ferric human Ngb (d), Ngb with 0.5 m M dithiothreitol under nitrogen (j), and the ferric mutant (CCC fi GSS) without cysteines (– –). mouse Ngb Temperature (°C) 80 85 90 95 100 105 f u 0.0 0.2 0.4 0.6 0.8 1.0 ferrous ferrous CO ferric-CN ferric Fig. 4. Melting profiles (fraction unfolded f U vs. temperature) of mouse Ngb for different ligation states. Experimental conditions were 1 m M phosphate buffer at pH 7 (at 25 °C). Smooth curves are simulations for a two state transition, as described in Experimental procedures. Temperature (°C) 60 70 80 90 100 f U 0.0 0.2 0.4 0.6 0.8 1.0 Cygb Ngb Ngb CCC->GSS Mb Drosophila Fig. 5. Melting profiles of ferric hexa-coordinated globins. The frac- tion unfolded f U vs. temperature is shown for Drosophila Hb (r), Mb (d), human Ngb (m), the mutant CCC fi GSS of human Ngb (.) and Mb (d). Experimental conditions were 1 m M phosphate buffer at pH 7 (at 25 °C). [Guanidinium-chloride] (M) 0.0 0.5 1.0 1.5 2.0 2.5 Tm (°C) 60 70 80 90 100 110 120 wt CCC->GSS 40 60 80 100 0.0 0.2 0.4 0.6 0.8 1.0 temperature (°C) f U 2M 1M 0.5M human Ngb Fig. 6. Dependence of the melting temperature T m on [guanidinium chloride] for wt human Ngb and the triple mutant CCC fi GSS. The T m (Table 1) was obtained by extrapolation to 0 M of guanidinium chloride. The insert shows the thermal unfolding curve of human Ngb at three concentrations of guanidinium chloride, in 10 m M phosphate at pH 7. Thermal stability of neuroglobin D. Hamdane et al. 2078 FEBS Journal 272 (2005) 2076–2084 ª 2005 FEBS around the heme group via the bis-histidyl binding to the heme group. Furthermore, the decrease in T m upon binding the external ligand could be underestimated at high temperature due to oxidation or loss of the exter- nal ligand. Cygb and the globin from Drosophila are also hexa- coordinated [2,3,28] and show various degrees of enhanced stability (Fig. 5, Table 1). Cygb has an affin- ity for the distal histidine 2.5-fold lower than human Ngb and exhibits a T m 5 °C lower than human Ngb. A similar, but larger, effect was observed for the glo- bin of Drosophila, in which the affinity of the distal histidine is 14 times lower and the T m is decreased by 24 °C relative to human Ngb. The very high stability requires the hexa-coordinated state; for these cases the T m may exceed 100 °C, and additional curves were measured at different concentrations of guanidinium chloride (Fig. 6) to better determine the T m value. Replacement of the distal histidine by valine, leucine or glutamine in mouse Ngb leads to a loss of the enhanced alpha absorption band in the deoxy form, characteristic of the internal residue coordination (data not shown). Relative to wild-type mouse Ngb, the sin- gle mutation E7L in mouse Ngb caused a decrease of 20 °C in thermostability, again suggesting a critical role for His E7 in the enhanced thermal stability of Ngb. Certain mutations of the distal histidine in Mb and Hb lead to instability linked to a higher autoxidation rate and ⁄ or heme loss. Note that the E7 mutants are stable with regard to O 2 binding, indicating that the mutation does not affect the pocket to a large extent. Reversibility Although the thermal denaturation was irreversible for human Hb, we observed a significant thermal reversi- bility for mouse and human Ngb, and human Cygb. The loss in helical content was 15%, estimated by the difference at 222 nm between the initial and final circu- lar dichroism spectra at 25 °C after the temperature cycle to 100 °C (data not shown). The reversibility was also tested by the absorption spectra (Fig. 7) and by flash photolysis kinetics (Fig. 8). Ngb maintains a high Table 1. Melting temperature (T m ) and histidine affinity (K His ¼ k on ⁄ k off ) for hexa-coordinated globins. T m (°C) K His K CN – ,Mb ⁄ K CN – Disulfide bond Human Ngb (yes) 100 280 Human Ngb + dithiothreitol (no) 103 3300 Human Ngb CCC fi GSS (no) 103 4500 Ferric human Ngb (yes) 97 45 Ferric human Ngb CCC fi GSS (no) 101 428 Iron state Ferrous mouse Ngb CO 95 Ferric mouse Ngb CN – 94 Ferrous mouse Ngb (His) 103 2000 Ferric mouse Ngb (His) 100 137 Variable (E7) His affinity Human Ngb CCC fi GSS 103 4500 Human Ngb (with disulfide bond) 100 280 Human Cygb 95 110 Drosophila Hb 76 18 Mouse Ngb His (E7) fi Leu 80 _ Horse heart Mb 81 < < 1 Human HbCO 71 < < 1 Fig. 7. Absorption spectra of ferrous human Ngb with dithiothreitol (to break the disulfide bridge) at 25 °C (solid line), after 5 min incu- bated at 90 °C, and finally at 25 °C after the temperature cycle (s). The spectrum for ferric human Ngb (with Soret band at 413 nm) is also shown. time (sec) 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 0.1 1 ∆ A N human Ngb-CO 25°C 50°C 70°C 90°C Fig. 8. Ligand rebinding kinetics for human Ngb at temperatures from 25 to 90 °C for samples equilibrated under 0.1 atm (100 l M) CO, in 100 m M phosphate buffer at pH 7. D. Hamdane et al. Thermal stability of neuroglobin FEBS Journal 272 (2005) 2076–2084 ª 2005 FEBS 2079 percentage ( 85%) of its initial characteristics after the temperature cycle, whereas Mb shows  70%; the fraction of functional HbA after the temperature cycle was only 20%. The shape of thermal denaturation curves of the various globins may differ, suggesting different mecha- nisms or degrees of cooperativity for the unfolding transition. Classical denaturation between two states results a cooperative denaturation with a maximum slope at T m . The enthalpy of denaturation DH m of Ngb and horse heart Mb was 72 and 110 kcalÆmol )1 , respectively. Cytoglobin and the globin of Drosophila have lower values of DH m , 60 and 53 kcalÆmol )1 , respectively. Note that human Hb and Cygb are tetra- meric and dimeric, respectively, and may involve a more complicated mechanism including subunit disso- ciation. Ligand-binding kinetics The circular dichroism spectra show that the protein is still correctly folded at elevated temperatures, but do not provide much information about protein function. We studied ligand binding using flash photolysis to see whether Ngb was functional at extreme temperatures. The kinetics after CO photodissociation showed a bi- phasic curve. The rapid phase corresponds to compet- itive CO and His E7 association, whereas the slower phase is the replacement of the E7 His by CO. The kinetics for human Ngb at different tempera- tures, up to 90 °C, are shown in Fig. 8. The kinetic curves show a steady progression vs. temperature, indi- cating that there is no major change in the basic ligand- binding properties. The increase in temperature leads to an increase in the amplitude of the slow phase, indica- ting that higher temperatures favor His vs. CO rebind- ing; that is, the histidine association rate (k His,on ) has a higher activation energy than that for CO (Table 2). Competition with the internal histidine ligand decreases the affinity for external ligands such as CO: K CO;obs ¼ K CO;penta 1 þ K His ¼ k CO;on =k CO;off 1 þ k His;on =k His;off ð1Þ From the kinetic curves vs. [CO], one can extract three of the rate parameters; the CO off rate must be deter- mined independently. Equilibrium studies allow an independent measure of the shift in observed affinity due to the histidine. Cyanide affinity The absorption difference spectrum in the visible region of ferric Ngb and cyanide derivative are shown Fig. 9. The maximum absorption of ferric Ngb occurs at 413 nm (Fig. 7). Cyanide binding to ferric Ngb leads to a red shift in the Soret band; peak absorption is seen at 416 nm for the mutant Ngb without cyste- ines, and 417 nm for species with the disulfide bond. The fraction saturation was calculated from the spec- tral difference, and the titration curve (Fig. 9 insert) gives a linear Hill plot. The affinity of cyanide for ferric Ngb was much lower than for Mb (Table 1), indicating competition by the distal histidine, as in the ferrous form. The affinity for cyanide was higher for mutant forms with- out the distal histidine. For human Ngb without cys- teine residues, the CN – affinity was lower, suggesting a higher affinity for the competing histidine, as observed in the ferrous form. Based on the shift in the CN – affinity, one can estimate the histidine affinity for the Table 2. Activation and binding energies for CO binding to human Ngb. Species His (kcalÆmol )1 ) CO (kcalÆmol )1 ) E on E off DEE on E off DE DE obs Human Ngb 11 24 13 5.5 10 4.5 ) 9.5 a Horse heart Mb 7.5 16 8.5 8.5 a A value of )10 (± 3) kcalÆmol )1 was determined from equilibrium studies. Experimental conditions were 100 m M phosphate buffer at pH 7.0, in the presence of 5 m M dithiothreitol. wavelength (nm) 300 400 500 600 ∆A -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 LOG(KCN µM) 1.6 2.0 2.4 2.8 3.2 3.6 LOG (Y/1-Y) -2 -1 0 1 2 human Ngb CCC GSS 428 nm 409 nm wt Fig. 9. Absorption difference spectra at various cyanide concentra- tions, relative to the ferric Ngb form (without cyanide). The spectra were measured at room temperature in 100 m M potassium phos- phate at pH 8. The insert shows the Hill plot of cyanide binding to ferric Ngb. The shift to a lower CN – affinity for Ngb without the disulfide bond is similar to that for oxygen in the ferrous form. Thermal stability of neuroglobin D. Hamdane et al. 2080 FEBS Journal 272 (2005) 2076–2084 ª 2005 FEBS ferric form; however, this requires a valid reference point. Combining the ferrous and ferric data, there is a definite relationship between T m and the estimated his- tidine affinity (Table 1): enhanced stability occurs only for high histidine affinities. Another consequence of the ligand competition is that the affinity of external ligands such as O 2 or CO shows a weak dependence on temperature [29]. Although the individual ligand-binding parameters show typical binding energies, the overall observed binding energy depends on the energy difference between the external (e.g. O 2 ) and the endogenous pro- tein ligands. The binding energies for histidine partially compensate those for oxygen, resulting in a lower overall temperature dependence of the observed oxy- gen affinity [29]. In the case of CO, the internal ligand more than compensates the changes vs. temperature [29,30], leading to a reverse temperature dependence; that is, the observed CO affinity is higher at higher temperatures (Fig. 10). The hexa-coordinated globins are thus characterized by a high thermal stability and a weak temperature dependence for the binding of external ligands. Discussion Ngb and Cygb show an unusually high T m relative to other hemoproteins. In addition, Ngb did not lose its functional properties or precipitate over the tempera- ture cycle. After photodissociation of CO, the ligand- binding kinetics remained biphasic and showed only a progressive change over the entire temperature range. The spectral and kinetic data indicate that Ngb remained stable and functional at such high tempera- tures. One possible explanation for the enhanced stability was the disulfide bond. However, reduction with dithiothreitol to break the bond, or mutations without cysteine residues, showed even higher stability (Fig. 3, Table 1). Furthermore, external ligands such as CO (or CN – for the ferric form) decreased the stability (Table 1). This points to the hexa-coordination by the distal histidine as the dominant factor. In fact, only the globins with a significant histidine affinity (> 10) showed enhanced stability. The hexa-coordination is thus necessary for the ther- mal stability. Considering the globins with a high histi- dine affinity (Table 1), there was a general trend of T m with K His . The histidine affinity is not known for the ferric forms; however, one can estimate this parameter based on the decrease in CN – affinity due to ligand competition; assuming this relation, the ferric globins also show higher stability with higher histidine affinity (lower CN – affinity). Note that the same mechanism appears to occur for both ferrous and ferric proteins: breaking the disulfide bond increases the histidine affinity, thereby decreasing the observed affinities for external ligands; loss of the disulfide bond leads to a CN – affinity that is lower by a factor of 7, compared with a factor of 8 for the oxygen affinity in the ferrous form. Mouse or human Ngb (without the disulfide bond) show the highest stability, with K His values > 1000 at 25 °C. Thermal unfolding does not require a significant amount of the penta-coordinated form. Based on the absorption spectra, the hexa-coordinated form remains 10 5 / T ( /K) 300 310 320 330 340 350 360 ln (K CO ) -16 -14 -12 Ngb Mb A B Fe ∆E obs ∆E Fe-His Fe-CO Fig. 10. (A) Arrhenius plot for the equilibrium binding coefficient for CO, based on kinetic parameters. Sample conditions were 100 m M phosphate buffer at pH 7.0, in the presence of 5 mM dithiothreitol. While Mb shows the classic positive value of DE (8 kcalÆmol )1 ), indicating a higher affinity at lower temperatures, Ngb shows a negative value ()10 kcalÆmol )1 ) due to the fact that the competing protein ligand (E7 histidine) more than compensates the intrinsic binding energy. (B) The energy diagram for competitive CO and his- tidine binding to human Ngb. D. Hamdane et al. Thermal stability of neuroglobin FEBS Journal 272 (2005) 2076–2084 ª 2005 FEBS 2081 the major form at the T m . The histidine is still  90% bound at 90 °C. Because the enthalpy of the dissoci- ation is greater than that of association, the histidine binding coefficient K ¼ k on ⁄ k off decreases at higher temperatures. Once a significant fraction of the penta- coordinated form is present, unfolding occurs. Based on the data shown in Table 1, one can pro- vide some elements of the relationship between T m and the histidine affinity. First, a high histidine affinity is necessary for the hyperstability. The data indicate a transition from a normal T m of  75 °C to a value of > 100 °C for forms with a tightly bound histidine. A value for K His of at least 10 is needed to see a shift in T m , indicating that unfolding occurs when a significant percentage of the penta-coordinated form is present. Between these limits the T m increases monotonically with the histidine affinity. Hexa-coordination would seem to act as a protein clamp around the heme (Fig. 2), thus stabilizing the overall structure. The mutation of the Ngb cysteines implicated in the disul- fide bridge increases the affinity of the E7 His [16], and thereby induces a 5 °C increase in T m compared with wild-type Ngb. The high thermal stability is apparently related to the affinity for distal histidine; the disulfide bridge has an effect on the stability mainly when it changes the histidine affinity. The need for thermal stability in human Ngb is not obvious, unless a slow turnover of nerve cells requires an enhanced stability of certain proteins. Other organ- isms such as plants or insects might profit from both the stability and the weak temperature dependence of the oxygen affinity. These features may have been crit- ical for the survival of ancient globins under more severe and fluctuating environmental conditions, and the high sequence homology has simply conserved these features as well. Experimental procedures Recombinant globins Expression cloning and purification of Ngb were performed as described previously [9]. Human Cygb cDNA in the expression plasmid pET3a [2] was expressed under the same conditions except that d-amino-laevulinic acid was omitted. Expressed Cygb was purified from inclusion bodies using the procedure described by Geuens et al. [14]. Reconstruc- tion of native Cygb from the apoprotein was undertaken by adding a 1.4-fold excess of hemin, followed by dialysis against 50 mm Tris ⁄ HCl at pH 8.5. After reconstruction, the solution was cleared by low (10 000 g; 30 min) and high (105 000 g; 60 min) speed centrifugation. Final purifi- cation of Cygb was performed by gel filtration using a Sephacryl S200 column equilibrated in 50 mm Tris ⁄ HCl at pH 8.5. The recombinant globins were further purified on an FPLC Akta Purifier (Amersham Bioscience, Uppsala, Swe- den) using a Hitrap DEAE Sepharose Fast Flow column [16]. The ferrous form was obtained by addition of sodium dithionite after equilibration under nitrogen. Proteins with reduced cysteines were obtained by incubation with 10 mm dithiothreitol for 24 h. Ligand-binding kinetics UV and visible spectra measurements were carried out with a HP8453 or Varian Cary 400 spectrophotometer. Laser flash photolysis and stopped-flow rapid mixing, the methods used to assess hexa-coordination and bimolecular CO and O 2 rate constants, have been described previously [9]. Photo- lysis was performed with 10 ns pulses at 532 nm. Detection of the sample absorption was in the Soret band, typically at 436 nm. Samples from 1 to 10 lm were in 4 · 10 mm quartz cuvettes. The samples were  7 lm (on a heme basis) in 100 mm potassium phosphate at pH 7. Analysis of the kinetics was performed by numerical integration of the differential equations for the transition between the three species: (penta-coordinated) Fe, Fe–CO, and Fe–His [29]. Simulation of a series of curves at differ- ent CO concentrations allows a determination of the CO on rate, and the histidine on and off rates. The CO off rate was independently determined by replacement of CO by oxygen or NO. Cyanide titration Spectrometric titrations of Ngb were carried out in quartz cuvettes on a Varian Cary 50 spectrophotometer. Cyanide titration was performed in 6 lm samples in 100 mm potas- sium phosphate buffer at pH 8 at room temperature. The titrating solution was 100 mm KCN in the same buffer. The KCN concentration was varied between 5 lm and 3mm. The spectrum was obtained after 30 min incubation after each KCN addition to achieve equilibrium. We con- trolled the variation of pH of the reaction solution, because high levels of KCN are known to increase the pH of poorly buffered solutions, which in turn will change the concentra- tion of free CN – . Circular dichroism Circular dichroism spectra were measured with a Jasco J-810 spectrometer (Tokyo, Japan). The protein samples (1–3 lm) were in 1 or 10 mm phosphate buffer at pH 7. The far ultra- violet spectra (190–260 nm) were measured in quartz cells of 0.5 or 4 mm optical pathlength, and represent an average of seven accumulations. Spectra were acquired at a scan speed Thermal stability of neuroglobin D. Hamdane et al. 2082 FEBS Journal 272 (2005) 2076–2084 ª 2005 FEBS 50 nmÆmin )1 at a resolution of 2 nm and a response time of 2 s; all spectra were corrected for the buffer baseline. Protein thermostability The proteins were subjected to the thermal melting profile by monitoring the changes of circular dichroism spectra at 222.6 nm. For thermal unfolding curves from 25 to 110 °C (heating plate temperature), samples were continuously scanned at 1 °CÆmin )1 ; scans at 2 °CÆmin )1 did not change the results. The temperature was programmed using a Jasco PTC-423S thermoelectric temperature controller. The accu- racy of the sensor was checked with a precision thermom- eter. The ferrous sample was obtained by addition of 500 lm sodium dithionite after equilibration under nitrogen. The protein without the disulfide bond was obtained by incuba- tion with 500 lm dithiothreitol during 24 h under nitrogen. The Ngb–CN – species was obtained by incubation of the fer- ric protein with 3 mm of KCN. The Ngb–CO form was obtained by equilibration of the ferric form under 1atm CO and then adding 200 lm dithionite to reduce the ferric iron. The thermal denaturation curves were analysed with a sim- ple two state model for folded (f) and unfolded (u) protein: curves for the circular dichroism signal (y) as a function of temperature (T) were fitted using a nonlinear least squares analysis [31] to the form: y obs ¼ (y f + s f T)+(y u + s u T) E ⁄ (1 + E), where E ¼ exp[(DH m ⁄ RT). (T ) T m ) ⁄ T m ], y obs is the observed circular dichroism signal; y f + s f T and y u + s u T describe the linear dependence (with slope s) of the pre- and post-transitional baselines on temperature, respect- ively. DH m is the enthalpy of unfolding at the melting tem- perature (T m ), corresponding to 50% unfolded. Curve fitting was performed using the original jasco software; values are for the average of at least three measurements. To obtain the fraction unfolded, the circular dichroism signal was normal- ized: f U ¼ (y f ) y obs ) ⁄ (y f ) y U ), where y obs is the observed circular dichroism signal and y f and y U the circular dichro- ism signal of the folded protein and unfolded, respectively, taking into account the linear baselines. For the high T m values, 100% transition was not achieved. The experiment was repeated at different concen- trations of guanidinium chloride, with a pre-incubation of 1 h at 37 °C; the curve of T m vs. [GdmCl] could then be extrapolated to estimate the protein T m in the absence of denaturants. To normalize thermal denaturation curves that do not show a complete transition, the overall change in circular dichroism signal was taken for the same protein under conditions of a lower T m (such as a different ligation state or in the presence of denaturants). Acknowledgements This work was supported by Inserm, the De ´ le ´ gation Ge ´ ne ´ rale pour l’Armement (France), the University of Paris XI, the Deutsche Forschungsgemeinschaft (Ha2103 ⁄ 3 and Bu956 ⁄ 5), the University of Antwerp, and a grant from the European Commission (Contract N° QLG3-CT-2002–01548). S. Dewilde is a postdoc- toral fellow of the Fund for Scientific Research, Fland- ers (FWO). References 1 Burmester T, Weich B, Reinhardt S & Hankeln T (2000) A vertebrate globin expressed in the brain. Nature 407, 520–523. 2 Burmester T, Ebner B, Weich B & Hankeln T (2002) Cytoglobin: a novel globin type ubiquitously expressed in vertebrate tissues. Mol Biol Evol 19, 416–421. 3 Trent JT III & Hargrove MS (2002) A ubiquitously expressed human hexa-coordinate hemoglobin. J Biol Chem 277, 19538–19545. 4 Reuss S, Saaler-Reinhardt S, Weich B, Wystub S, Reuss MH, Burmester T & Hankeln T (2002) Expression analysis of neuroglobin mRNA in rodent tissues. Neuroscience 115, 645–656. 5 Schmidt M, Giessl A, Laufs T, Hankeln T, Wolfrum U & Burmester T (2003) How does the eye breathe? Evi- dence for neuroglobin-mediated oxygen supply in the mammalian retina. J Biol Chem 278, 1932–1935. 6 Pesce A, Dewilde S, Nardini M, Moens L, Ascenzi P, Hankeln T, Burmester T & Bolognesi M (2003) Human brain neuroglobin structure reveals a distinct mode of controlling oxygen affinity. Structure (Camb) 11, 1087– 1095. 7 De Sanctis D, Dewilde S, Pesce A, Moens L, Ascenzi P, Hankeln T, Burmester T & Bolognesi M (2004) Crystal structure of cytoglobin: the fourth globin type discov- ered in man displays heme hexa-coordination. J Mol Biol 336, 917–927. 8 Sugimoto H, Makino M, Sawai H, Kawada N, Yoshiz- ato K & Shiro Y (2004) Structural basis of human cyto- globin for ligand binding. J Mol Biol 339, 873–885. 9 Dewilde S, Kiger L, Burmester T, Hankeln T, Baudin- Creuza V, Aerts T, Marden MC, Caubergs R & Moens L (2001) Biochemical characterization and ligand bind- ing properties of neuroglobin, a novel member of the globin family. J Biol Chem 276, 38949–38955. 10 Pesce A, Bolognesi M, Bocedi A, Ascenzi P, Dewilde S, Moens L, Hankeln T & Burmester T (2002) Neuroglo- bin and cytoglobin. Fresh blood for the vertebrate globin family. EMBO Report 3, 1146–1151. 11 Burmester T & Hankeln T (2004) Neuroglobin: a respiratory protein of the nervous system. News Physiol Sci 19, 110–113. 12 Sun Y, Jin K, Mao XO, Zhu Y & Greenberg DA (2001) Neuroglobin is up-regulated by and protects D. Hamdane et al. Thermal stability of neuroglobin FEBS Journal 272 (2005) 2076–2084 ª 2005 FEBS 2083 neurons from hypoxic–ischemic injury. Proc Natl Acad Sci USA 98, 15306–15311. 13 Sun Y, Jin K, Peel A, Mao XO, Xie L & Greenberg DA (2003) Neuroglobin protects the brain from experi- mental stroke in vivo. Proc Natl Acad Sci USA 100, 3497–3500. 14 Geuens E, Brouns I, Flamez D, Dewilde S, Timmer- mans JP & Moens L (2003) A globin in the nucleus. J Biol Chem 278, 30417–30420. 15 Schmidt M, Gerlach F, Avivi A, Laufs T, Wystub S, Simpson JC, Nevo E, Saaler-Reinhardt S, Reuss S, Hankeln T et al. (2004) Cytoglobin is a respiratory pro- tein in connective tissue and neurons, which is up-reg- ulated by hypoxia. J Biol Chem 279, 8063–8069. 16 Hamdane D, Kiger L, Dewilde S, Green BN, Pesce A, Uzan J, Burmester T, Hankeln T, Bolognesi M, Moens L et al. (2003) The redox state of the cell regulates the ligand binding affinity of human neuroglobin and cytoglobin. J Biol Chem 278, 51713– 51721. 17 Cooper A & Dryden DT (1984) Allostery without con- formational change. A plausible model. Eur Biophys J 11, 103–109. 18 Chan MK, Mukund S, Kletzin A, Adams MW & Rees DC (1995) Structure of a hyperthermophilic tungstop- terin enzyme, aldehyde ferredoxin oxidoreductase. Science 267, 1463–1469. 19 Anderson DE, Hurley JH, Nicholson H, Baase WA & Matthews BW (1993) Hydrophobic core repacking and aromatic–aromatic interaction in the thermostable mutant of T4 lysozyme Ser117 fi Phe. Protein Sci 2, 1285–1290. 20 Britton KL, Baker PJ, Borges KM, Engel PC, Pasquo A, Rice DW, Robb FT, Scandurra R, Stillman TJ & Yip KS (1995) Insights into thermal stability from a comparison of the glutamate dehydrogenases from Pyrococcus furiosus and Thermococcus litoralis. Eur J Biochem 229, 688–695. 21 Russell RJ, Hough DW, Danson MJ & Taylor GL (1994) The crystal structure of citrate synthase from the thermophilic archaeon, Thermoplasma acidophilum. Structure 2, 1157–1167. 22 Schumann J, Bohm G, Schumacher G, Rudolph R & Jaenicke R (1993) Stabilization of creatinase from Pseu- domonas putida by random mutagenesis. Protein Sci 2, 1612–1620. 23 Tanner JJ, Hecht RM & Krause KL (1996) Determi- nants of enzyme thermostability observed in the molecu- lar structure of Thermus aquaticus d-glyceraldehyde-3- phosphate dehydrogenase at 2.5 Angstroms resolution. Biochemistry 35, 2597–2609. 24 Yip KS, Stillman TJ, Britton KL, Artymiuk PJ, Baker PJ, Sedelnikova SE, Engel PC, Pasquo A, Chiaraluce R & Consalvi V (1995) The structure of Pyrococcus furio- sus glutamate dehydrogenase reveals a key role for ion-pair networks in maintaining enzyme stability at extreme temperatures. Structure 3, 1147–1158. 25 Spassov VZ, Karshikoff AD & Ladenstein R (1995) The optimization of protein–solvent interactions: ther- mostability and the role of hydrophobic and electro- static interactions. Protein Sci 4, 1516–1527. 26 Hargrove MS & Olson JS (1996) The stability of holo- myoglobin is determined by heme affinity. Biochemistry 35, 11310–11318. 27 Brzeska H, Venyaminov S, Grabarek Z & Drabikowski W (1983) Comparative studies on thermostability of cal- modulin, skeletal muscle troponin C and their tryptic fragments. FEBS Let 153, 169–173. 28 Hankeln T, Jaenicke V, Kiger L, Dewilde S, Ungerechts G, Schmidt M, Urban J, Marden MC, Moens L & Bur- mester T (2002) Characterization of Drosophila hemo- globin. Evidence for hemoglobin-mediated respiration in insects. J Biol Chem 277, 29012–29017. 29 Uzan J, Dewilde S, Burmester T, Hankeln T, Moens L, Hamdane D, Marden MC & Kiger L (2004) Neuroglo- bin and other hexa-coordinated hemoglobins show a weak temperature dependence of oxygen binding. Bio- phys J 2, 1196–1204. 30 Fago A, Hundahl C, Dewilde S, Gilany K, Moens L & Weber RE (2004) Allosteric regulation and temperature dependence of oxygen binding in human neuroglobin and cytoglobin. Molecular mechanisms and physiologi- cal significance. J Biol Chem 279, 44417–44126. 31 Santoro MM & Bolen DW (1988) Unfolding free energy changes determined by the linear extrapolation method. 1. Unfolding of phenylmethanesulfonyl alpha-chymo- trypsin using different denaturants. Biochemistry 27, 8063–8068. Thermal stability of neuroglobin D. Hamdane et al. 2084 FEBS Journal 272 (2005) 2076–2084 ª 2005 FEBS . T m . The enthalpy of denaturation DH m of Ngb and horse heart Mb was 72 and 110 kcalÆmol )1 , respectively. Cytoglobin and the globin of Drosophila have lower values of DH m , 60 and 53 kcalÆmol )1 , respectively Ngb had negative bands at 208 and 222 nm (Fig. 1), as expected for a high percentage of alpha helix. Analysis of the secondary structure of Ngb gave 78% alpha helix and 22% of other forms, similar. forms a sort of clamp Fig. 3. Effect of the cysteine bridge of human Ngb on the thermal stability. The melting temperature, corresponding to the peak of this curve of the first derivative of the circular