Effects of Iron Nitrosylation on Sickle Cell Hemoglobin Solubility*

One mechanism by which nitric oxide (NO) has been proposed to benefit patients with sickle cell disease is by reducing intracellular polymerization of sickle hemoglobin (HbS). In this study we have examined the ability of nitric oxide to inhibit polymerization by measuring the solubilizing effect of iron nitrosyl sickle hemoglobin (HbS-NO). Electron paramagnetic resonance spectroscopy was used to confirm that, as found in vivo , the primary type of NO ligation produced in our partially saturated NO samples is pentacoordinate (cid:1) -nitrosyl. Linear dichroism spectroscopy and delay time measurements were used to confirm polymerization. Based on sedimentation studies we found that, although fully ligated (100% tetranitrosyl) HbS is very soluble, the physiologically relevant, partially ligated species do not pro-vide a significant solubilizing effect. The average solubilizing effect of 26% NO saturation was 0.045; much less than the 0.15 calculated for the effect of 26%

Sickle cell disease is caused by a mutation in the sixth codon of the ␤-globin gene that results in the hydrophobic residue valine replacing hydrophilic glutamate in the hemoglobin (Hb) 1 ␤-chain (1,2). This mutation causes sickle cell hemoglobin (HbS) to polymerize under hypoxic conditions, deforming, and rigidifying the red blood cell (3). These changes in red cell rheology contribute to microvascular occlusion, tissue damage, and a high degree of morbidity and mortality (4).
Nitric oxide is now being considered as a therapy for sickle cell disease (5). During inhalation therapy, NO reacts with oxygenated Hb to form methemoglobin (MetHb) and nitrate, with deoxygenated Hb (deoxyHb) to form nitric oxide hemoglobin (HbNO), and a very small amount reacts with the ␤-93 cysteine to form S-nitrosohemoglobin (SNO-Hb) (6). In theory, one mechanism through which NO may benefit patients is to decrease sickle hemoglobin polymerization. It is believed that only T-state HbS polymerizes (7,8). Thus, at high oxygen pressures, like those found at the lungs, HbS molecules are in the R-state (high-affinity) and do not polymerize. In the tissues, oxygen pressure is low, and polymerization becomes more likely. The kinetics of polymerization have been described by an approximate, empirical equation relating the delay time (the time during which polymers are not detectable) to the supersaturation ratio (9) as shown in Equation 1, where t d is the delay time, c 0 is the total concentration of hemoglobin molecules, c s is the solubility of the hemoglobin, and is a proportionality factor. The exponent, n, is found to be about 30 -40 under physiological conditions. Thus, the delay time is extremely sensitive to the supersaturation ratio, c 0 /c s .
It has recently been proposed that nitric oxide (NO) binding at the heme group to form iron nitrosylated HbS (HbS-NO) may also play a role in increasing HbS solubility in vivo (10 -13). Briehl and Salhany (14) found that tetranitrosyl hemoglobin (100% saturated HbS-NO) polymerizes in the presence of inositol hexaphosphate (IHP). However, stripped 100% tetranitrosyl did not polymerize under the same conditions (14). Subsequently, McDade et al. (11) reported preliminary results indicating that even small amounts of NO can substantially reduce polymerization in purified HbS solutions and decrease sickling (12).
The difference in the effect of various ligands may be related to the difference in their abilities to produce a T-to R-state transition upon binding HbS. The quaternary state of nitrosyl hemoglobin has been investigated using combinations of absorption, electron paramagnetic resonance (EPR), and nuclear magnetic resonance (NMR) spectroscopies (15)(16)(17)(18)(19)(20). The quaternary state of tetranitrosyl hemoglobin favors the R-state when it is stripped of organic phosphates such as 2,3-diphosphoglycerate (DPG) but favors the T-state in the presence of IHP (18,21). When nitrosyl hemoglobin is in the T-state, the iron-proximal histidine bonds can break forming a pentacoordinate, NO-bound heme. This pentacoordinate species is responsible for the hyperfine splitting observed in EPR spectra. Thus EPR can be used to probe the quaternary state of nitrosylated hemoglobin.
The pH, presence of organic phosphates and other solution conditions have a large influence on the quaternary state of nitrosyl Hb (HbNO) (18,19). Since it is believed that only T-state HbS polymerizes, these conditions could also greatly affect the solubility of HbS-NO. In assessing the contribution the formation of HbS-NO may have in treating patients with sickle cell disease, it should be remembered that only very small amounts of NO can be administered (6). Yonetani et al. (20) showed that when small amounts of NO are added to deoxyHb, ␣-nitrosyl hemoglobin forms (NO binds preferentially to the ␣ hemoglobin subunits). The ␣-nitrosyl hemoglobin has an especially low affinity for oxygen and was thus referred to as in a Super T-state, which is detectable through its characteristic hyperfine splitting in its EPR spectrum (20). In this study we have examined the solubility of partially ligated HbS-NO and 100% tetranitrosyl hemoglobin under conditions designed to mimic those in vivo. We find that although 100% tetranitrosyl HbS is very soluble, partial nitrosylation is unlikely to have any significant effect on the solubility of HbS under physiological conditions. EXPERIMENTAL PROCEDURES Unless otherwise noted, all reagents were obtained from Sigma Chemical. Blood was obtained from patients with sickle cell disease (SS) with low levels of fetal hemoglobin (HbF) who had not been recently transfused following federal regulations and guidelines. Electrophoresis was used to determine the levels of HbF (3.2 Ϯ 2.1%) and HbA 2 (5.0 Ϯ 0.4%) in the samples. The hemolysate was prepared from these samples within 24 h of drawing as described previously (22). The samples were kept cold during all preparatory procedures (on ice or at 4°C).
Blood cells were washed in phosphate-buffered saline (PBS), pH 7.4 and lysed by incubation in distilled water. The membranes were spun down, and the hemolysate was then dialyzed against PBS buffer. Based on a colorimetric assay (Sigma Chemical) we found that this procedure results in a negligible change in the molar ratio of DPG to hemoglobin, consistent with earlier results (23). Thus, the concentration of the most physiologically relevant allosteric effector was the same for our studies as that found in vivo. In some cases, when the sample volume was slightly inadequate for an experiment, small amounts of previously prepared hemolysate were used to increase the sample volume. Those samples had been pelleted in liquid nitrogen for storage. No effect of using the small amounts of previously frozen samples was observed. The hemolysate was concentrated using Centriprep and Centricon concentrating devices (Millipore, Bedford, MA) to a final concentration of 16 -19 mM (in heme). After purging the samples with argon, sodium dithionite was added. Precautions in the general handing of and preserving the anaerobicity of dithionite were followed as described previously (24). Dithionite was placed in a septum-capped bottle and purged with argon for about 2 h, after which PBS buffer that had been bubbled with argon gas (also for about 2 h) was added to make a stock dithionite solution. This was then added to the HbS sample that had been purged with argon to a final concentration of 50 mM.
Iron nitrosyl hemoglobin was made by adding sodium nitrite to the deoxygenated samples as described previously (25). A stock nitrite solution was made and bubbled with argon gas. An aliquot of the stock solution was added to the HbS sample that had been prepared with excess sodium dithionite. The nitrite was reduced by dithionite to form NO, which binds quickly to the hemoglobin, yielding nearly stochiometric iron nitrosyl hemoglobin. This procedure was carried out within 1 h after addition of dithionite. We found that this method was superior to exposing the hemoglobin directly to NO gas even when the system is properly flushed with an inert gas, and the NO is passed through NaOH to remove traces of higher nitrogen oxides (26). We found that it is very difficult to remove all these unwanted nitrogen oxides and that we often would have large concentrations of nitrite in our NO-saturated buffer. Furthermore, we sometimes found a significant degree of protein denaturation when our Hb was exposed directly to NO gas. Finally, when exposing Hb directly to NO gas it was not possible to control the distribution of NO binding. One generally would need to make 100% tetranitrosyl Hb and then mix this with deoxyHb to achieve a partially saturated sample. If NO-saturated buffer is added directly to the Hb sample, it will bind faster than the sample can be properly mixed. To mimic in vivo conditions, where the concentration of NO is small, it was desirable for us to be able to make HbS-NO samples that had evenly distributed NO ligation. This is achievable using the nitrite reduction method since the rate of this reaction was measured to be 0.035 Ϯ 0.007 1/s at 4°C when using the same concentrations of nitrite and dithionite as those used in preparing HbS-NO for the solubility measurements. The rate was measured by monitoring the formation of HbS-NO when Hb ϩ dithionite was rapidly mixed with sodium nitrite using an Olis RSM-1000 stopped-flow spectrophotometer (Bogart, GA). By rigorously mixing nitrite solutions directly to the Hb ϩ dithionite solutions on ice, we could achieve evenly distributed NO ligation. This is possible because the sample can be thoroughly mixed before NO is released, so there is no extreme local concentration of NO. If NO-saturated buffer is added to a hemoglobin solution, the NO will react faster than the most rapid mixing technique and will thus form a larger quantity of tetranitrosyl hemoglobin. For some experiments, nitrite was added in slight molar excess to Hb in order to make 100% tetranitrosyl Hb. This was also occasionally then diluted with deoxyHb for other experiments where we purposely made a sample that was initially composed of a mixture of tetranitrosyl hemoglobin molecules and completely deoxygenated hemoglobin molecules.
The addition of dithionite resulted in a decrease in the pH of the PBS buffer from 7.4 to 6.8. The presence of Hb or nitrite did not affect the pH further as determined by a separate control experiment where we added HbS and nitrite to a dithionite solution. Although this pH is below physiological, it afforded us the possibility to work with slightly lower Hb concentrations (due to the reduced solubility) and thus conserve sickle cell blood. Furthermore, the same pH was used for both the HbS-NO samples and deoxyHbS controls. No significant difference in the type of iron nitrosyl ligation or the solubilizing effect of NO on HbS was found when the experiments were repeated in a Trizma buffer (pH 7.2 at 37°C) with 0.1 M KCl and 0.02 M NaCl.
For each solubility measurement, deoxygenated HbS, nitrosylated HbS, and other necessary materials were assembled in a dry hood (model Dri Lab He-43-2, vacuum/atmosphere Corporation, Los Angeles, CA). The atmosphere in the hood was cycled between vacuum and argon several times before use. Inside the hood, the HbS-NO sample was prepared along with a deoxygenated sample to be used as a control. The HbS-NO sample and deoxyHb control were each then used to fill two Quick-seal polyallomer ultracentrifuge tubes for the TLS-55 swinging bucket rotor (Beckman Coulter Inc.), two 0.01-cm path length cells, and two-0.1 cm path length cells (Hellma Inc., Plainview, NY). In addition, a portion of the HbS-NO sample was always used to fill an EPR tube (Wilmad/Lab Glass, Buena NJ, 701-PQ) and sometimes a second, screw-cap EPR tube was filled (701-TR). Thus, for all experiments, we were able to perform a variety of spectroscopic and other procedures on identically prepared HbS-NO and deoxyHb samples.
The total concentration and percent of NO ligation was determined using visible absorption spectroscopy on the 0.01-cm path length cells. The absorption spectra were fit to basis spectra normalized by their extinction coefficient using a least-squared minimization. The absence of any methemoglobin was also determined in this way. The basis spectra used and a typical fit are shown in Fig. 1. The percentage of HbS-NO was also determined by double integration of the EPR spectra (described in the next paragraph). The difference in using these two methods was only 2.2 Ϯ 2.0%. The values of Hb nitrosylation reported in the results section are the averages of those determined by EPR and absorption.
Electron paramagnetic resonance spectroscopy was conducted on a Bruker Electron Spin Resonance-ER-200 D Spectrometer with a B-VT 1000/ER 4111 VT variable temperature unit set at temperatures of 140 K using a 5-milliwatt microwave power, 5 G modulation amplitude, and 9.36 GHz microwave frequency. A 0.1-s time constant was used for 100-s scans over 500 G. The concentration of HbS-NO on samples that had previously been frozen in liquid nitrogen, was determined by double integration of the spectrum using EPRWare for Windows (Scientific Software Services, Normal, IL) and compared with a standard HbS-NO sample.
The type of nitrosyl ligation was determined by fitting EPR spectra to basis spectra shown in Fig. 2. These three basis spectra, pentacoordinate ␣-nitrosyl, hexacoordinate ␣-nitrosyl, and hexacoordinate ␤-nitrosyl represent the three types of possible NO ligations at the heme (19). The pentacoordinate ␣-nitrosyl Hb was made by incubating a 10% NO-saturated Hb sample in 0.2 M, pH 5, sodium phosphate buffer for 1 h at 37°C in the presence of a 2-fold molar excess of inositol hexaphosphate (IHP). The spectrum for the hexacoordinate ␤-nitrosyl Hb was obtained by subtracting the pentacoordinate ␣-nitrosyl component from a 100% NO-saturated sample prepared in 0.2 M, pH 5, sodium phosphate buffer with IHP. The hexacoordinate ␣-nitrosyl spectrum was obtained by subtracting the ␤-nitrosyl component from a 100% NOsaturated Hb sample that had been stripped of organic phosphates using a G-25 Sephadex column (Amersham Biosciences) and prepared in 0.2 M pH 9 Trizma (Tris base) buffer. A few iterations of these types of procedures were necessary to produce the final spectra shown in Fig.  2. The spectra shown in Fig. 2 resemble those published previously that had been prepared by other methods (19).
Linear dichroism (LD) spectroscopy was used to determine presence of HbS polymers in the 0.01-cm pathlength cells. The LD was measured on a LD/optical rotary dispersion (ORD) spectrometer developed by OLIS inc. (Bogart, GA). Light produced by a xenon lamp is sent through the patented rapid scanning monochromator that scans a spectrum every millisecond. The light then passes through a polarizer oriented at 45 o with respect to the horizontal plane. After traversing the sample, the light passes through a photoelastic modulator (Hinds Instruments, Hillsboro, OR) and then through a beam-splitting polarizer producing horizontally and vertically polarized light. The two polarized beams are collected by two photomultiplier tubes and the difference divided by the sum of the intensities that is modulated by two times the reference frequency of the photoelastic modulator will be the ORD ϩ LD. Since the ORD of hemoglobin is very small in the visible region, it can be ignored and any measured signal was attributed to LD of partially aligned sickle cell hemoglobin polymers.
The delay time for polymerization was measured by observing the turbidity of the samples at 800 nm on either a PerkinElmer Lambda 9 spectrometer or a Cary 100 Bio UV/visible spectrometer. The samples were temperature jumped from 0 to 37°C, and the turbidity was measured as a function of time. The samples were taken directly from ice to jacketed cell holders on the spectrophotometers. The temperature of the cells in the PerkinElmer machine was controlled by a Haake circulating bath (Paramus, NJ), and the temperature of the cells in the Cary machine was regulated by a Cary temperature controller.
The centrifuge tubes were sealed in the dry hood and then incubated for 1 h at 37°C to allow them to polymerize before measuring the solubility by centrifugation. As with the delay time measurements, every experiment included a HbS-NO sample and a deoxygenated HbS control prepared in the same way as the HbS-NO sample except without the addition of nitrite. The samples were spun at 214,000 ϫ g for 1 h at 37°C as described previously (7). The solubility was measured by determining the total concentration of hemoglobin in the supernatant using absorption spectroscopy as described above. LD was used to make sure that no polymers were present in the supernatant.

RESULTS
We found no evidence for polymerization of 100% tetranitrosyl HbS from nine different preparations. Fig. 3 shows the absence of any LD for a 100% tetranitrosyl HbS sample. It is compared with the LD of a deoxygenated sample and a partially nitrosylated one. For the deoxyHbS and partially nitrosylated HbS samples, one can flip the LD signal (as shown) by rotating the sample about the axis defined by the incident light. The ability to flip the signal is strong evidence that it results from LD of partially aligned polymers (27). Fig. 4 shows a typical delay time measurement for a 100% tetranitrosyl sample compared with that for a deoxyHbS sample prepared from the same hemolysate. No polymerization is observed in the 100% tetranitrosyl sample.
Our solubility measurements also confirmed that 100% tetranitrosyl HbS is very soluble. If polymers are formed, the concentration in the supernatant following a sedimentation experiment should be equal to the solubility. When no polymers are formed, the concentration in the supernatant following sedimentation should be equal to the total concentration of HbS, and one can only say that the solubility is greater than or equal to the total concentration of the HbS. We found that the solubility of 100% tetranitrosyl HbS, prepared as a hemolysate in PBS buffer is greater than or equal to 19 mM.
The solubility of partially ligated HbS-NO was measured for samples prepared in two different ways. In one method, submolar amounts of nitrite were added to HbS in the presence of dithionite. In the other method, 100% tetranitrosyl HbS was diluted with deoxygenated HbS. Although the first method should be able to achieve equally distributed NO ligation, the faster off-rate from the ␤-subunits always leads to the formation primarily of ␣-nitrosyl HbS. This is illustrated in Fig. 5, where the EPR spectrum for 100% tetranitrosyl Hb is compared with that of a partially ligated sample that was frozen after incubation at 0°C (labeled before incubation) for 1 h and another that was incubated for an additional hour at 37°C (labeled after incubation). The formation of ␣-nitrosyl HbS occurred both when 100% tetranitrosyl HbS was diluted with deoxygenated HbS and when the initial NO ligation was evenly distributed. The final amount of ␣-nitrosyl HbS was slightly less for the case when 100% tetranitrosyl is diluted with de-oxyHbS (81 versus 87% when the NO ligation is initially evenly distributed). For each solubility measurement, the presence of polymers was confirmed by LD measurements (Fig. 3) and delay time measurements. We observed that increasing NO ligation to 26% had little effect on HbS solubility. For the sample where we tried to achieve even initial NO ligation for a final saturation of 26 Ϯ 5%, we found the solubility to be 11.4 Ϯ 0.8 mM compared with a solubility of 11.0 Ϯ 1.1 for the deoxy controls. For the samples where we mixed previously made 100% tetranitrosyl HbS with deoxyHbS for a final NO saturation of 26 Ϯ 5% NO ligation the solubility was 11.6 mM Ϯ 0.7 compared with 11.0 Ϯ 0.1 for the deoxy controls. These results are illustrated graphically in Fig. 6.

DISCUSSION
As discussed below, the solubility of 100% tetranitrosyl HbS is not likely to be important physiologically. However, its solubility is of general biochemical interest. We have found that it is very soluble. Briehl and Salhany (14) found no evidence that stripped 100% tetranitrosyl HbS can polymerize, even in pH 6 Tris buffer with a concentration of 30 mM at 20°C. In the presence of IHP at pH 7.2, they found that the minimum gelling concentration of HbS was between 24 and 30 mM at 20°C. Based on their results and the fact that IHP promotes polymerization more effectively than phosphate or DPG (28), one would predict that the solubility of 100% tetranitrosyl HbS in the presence of these physiologically relevant allosteric effectors would be between that of stripped HbS and that in the presence of IHP. Our results, using physiologically relevant concentrations of DPG, are consistent with this prediction. We have not seen evidence for polymerization of 100% tetranitrosyl HbS and have found the solubility to be greater than 19 mM.
It is useful to compare the effect of NO (Fig. 6) on HbS solubility to that of oxygen. In order to calculate the solubilizing effect of oxygen we use empirical Equation 2 deduced by Eaton and Hofrichter (29) where T is the temperature in degrees celsius and y s is the solution phase saturation with oxygen, which is very close to the overall HbS oxygen saturation. To compare our results (which are for hemolysates containing small amounts of HbF and HbA 2 ) we have converted our data to a solubilizing effect, S L , defined in Equation 3 as, where C S L is the solubility of the liganded form (with NO or O 2 ) and C S D is the solubility for a fully deoxygenated sample under the same conditions with the same amount of other hemoglobins (F and A 2 ) present. The average solubilizing effect of 26 Ϯ 5% NO saturation, when the NO is added in an attempt to achieve initial equal distribution of NO ligation, was 0.04 and that for 26 Ϯ 5% NO ligation, prepared by diluting 100% tetranitrosyl HbS with deoxyHbS was 0.05. The small difference between the solubilizing effects obtained using the two preparation methods can be attributed to the formation of ␣-nitrosyl HbS during incubation at 37°C regardless of how much HbS starts as tetranitrosyl. The calculated solubilizing effect of a 26% oxygen-saturated HbS sample is 0.15. Thus, the solubilizing effect of NO is much less than that of oxygen, contrary to earlier, preliminary reports (11,12).
The poor solubilizing effect of NO can be partially attributed to its inherently poor ability to effect a T-to R-state transition. Whereas IHP can switch 100% tetranitrosyl hemoglobin from the R-to the T-state it cannot do that for oxyhemoglobin or carbonmonoxyhemoglobin (17). In addition, it takes an average FIG. 4. Time delay measurement for 100% tetranitrosyl HbS compared with deoxyHbS. The turbidity was observed using the Cary 100 Bio UV/visible spectrometer at 800 nm using 0.1-cm pathlength cells. The concentrations of the 100% tetranitrosyl and deoxy samples were 16 mM. The delay times were calculated as the time to reach one-tenth of the final optical density after subtraction of the initial absorption at time t ϭ 0. The delay time for the deoxy sample was 110 s while the 100% tetranitrosyl sample was greater than 7200 s (possibly infinity).
FIG. 5. EPR spectrum of 100% tetranitrosyl Hb and partially ligated Hb before and after incubation. The 100% tetranitrosyl Hb (16 mM) was composed of 20% pentacoordinate ␣ and 27% hexacoordinate ␣and 53% hexacoordinate ␤-nitrosyl hemoglobin as calculated by fitting to the basis spectra shown in Fig. 2. The partially ligated Hb sample was prepared by diluting 100% tetranitrosyl HbS with deoxygenated HbS. The final concentration of the sample was 13.8 mM. The percentage of HbNO was 30% as determined by absorption spectra and 32% HbNO from the EPR spectra. Before incubation for 1 h at 37°C, the sample was 40% pentacoordinate ␣ and 12% hexacoordinate ␣and 49% hexacoordinate ␤-nitrosyl. After incubation, the sample was 81% pentacoordinate ␣ and 4% hexacoordinate ␣and 15% hexacoordinate ␤-nitrosyl. of 2.7 NO molecules to cause a T-to R-state transition (30). When two NO molecules are bound to the hemes of the ␣-subunits (for which they have a higher affinity) and the ␤-subunits are not ligated, the resultant ␣-nitrosyl Hb is in a Super T-state (20). Thus, a nitrosylated hemoglobin molecule is less likely to be R-state than an equivalently oxygen-bound one, especially in the presence of allosteric effectors like DPG or phosphate.
The poor solubilizing effect can also be explained in terms of a poor ability to form tetranitrosyl Hb, which would be R-state. For a Hb sample that is 25% saturated with oxygen, 80% of the bound oxygen will be to R 4 molecules, 15% will be to R 3 molecules, and 5% will be to T 1 molecules, where the subscripts indicate the degree of ligation for each quaternary state (R or T) (31). Thus, in the case of oxygen, most of the oxygen will be bound so as to effect a T-to R-state transition. The unequal distribution of oxygen ligation is due to the larger affinity of R-state Hb for oxygen compared with T-state. On the hand, unlike in the case of oxygen, the rate of NO binding to Hb is not cooperative (30,32,33). Although the off-rate from T-state nitrosyl Hb is faster than that from R-state (34,35), the combination of the lack of cooperativity in the NO-binding rate, together with a slower off-rate from ␣-subunits compared with ␤-subunits (18,19,35) results in preferential NO binding to ␣-subunits rather than formation of a significant fraction of R 4 . Thus, the major species formed for a 25% saturated Hb-NO sample will be ␣(Fe-NO) 2 ␤(Fe) 2 and ␣(Fe-NO)␣(Fe)␤(Fe) 2 as observed here in this report and under a variety of conditions (18 -20, 36).
Our results strongly suggest that the direct solubilizing effect of NO on HbS in vivo, whether produced endogenously or therapeutically, is insignificant. Numerous studies have shown that when HbNO is formed in vivo the primary product is ␣-nitrosyl Hb (20,(37)(38)(39)(40)(41). The ␣-nitrosyl Hb was shown to transition from hexacoordinate (R-state) in arterial blood to pentacoordinate (T-state) in venous blood, (40). That ␣-nitrosyl hemoglobin was found to predominate in rapidly frozen venous and arterial blood (40) provides strong evidence that the type of HbS-NO species we have studied are relevant to those found in vivo. Due to the long incubation of our samples at 37°C to ensure complete polymerization, we can assume that the NO ligation state also reached equilibrium. The published results on venous and arterial blood (40) demonstrate that the halftime for equilibration of NO ligation from ␤to ␣-subunits (about 8 min at 25°C, Ref. 19) is sufficient to bring the NO ligation distribution between ␤and ␣-subunits to equilibrium in vivo as well. Although most of our experiments were carried out in pH 6.8 buffer, pentacoordinate HbS-NO was found to also dominate in all pH 7.2 Trizma buffer (78% pentacoordinate ␣-nitrosyl after 1 h of incubation at 37°C). The pH (6.8) used in most of our experiments is not as subphysiological as one may think since the pH in the sickle red cell has been found to be 7.1-7.2 (42,43). Most importantly, the pentacoordinate HbS-NO species has been shown to predominate in venous blood (40). Here we have shown that this form of HbS-NO, predominating in the venous circulation in vivo, has a poor solubilizing effect.
One might argue that the increased oxygen saturation in venous or arterial circulation, compared with the conditions used here, would lend HbS-NO a greater solubilizing effect by causing a T to R conversion. However, ␣-nitrosyl Hb has been shown to be a negative allosteric effector (20). This lower oxygen affinity causes the ␤-subunits of nitrosylated Hb molecules to be preferentially deoxygenated yielding a T-like quaternary state. Thus (in theory), for a given oxygen pressure, the overall oxygen saturation will be less in the presence of NO than in its absence, which could lead to an actual decrease in the solubility. Using data on oxygen affinity and quaternary state of ␣-nitrosyl hemoglobin presented by Yonetani et al. (20), we estimated the solubilizing effect S L Ј for a 25% iron-nitrosylated sample when the oxygen pressure was that producing a normal hemoglobin oxygen saturation of 25, 50, and 75%. For these oxygen pressures we found S L Ј to be 0.03, Ϫ0.01, and Ϫ0.05, respectively. 2 Further investigation is necessary to examine the solubilizing effect of iron nitrosylation under physiological oxygen pressures. However, our results are directly applicable to estimating the solubilizing effect in hypoxic tissue and, due to the likeness in the type of nitrosylated species studied here where the superscript C s NOϩO2 refers to the solubility in the presence of 25% nitrosylation and an oxygen saturation that would result when the nitrosyl Hb is at the same oxygen pressure as that when normal hemoglobin would produce a given oxygen saturation (25,50, and 75% used in these calculations). C s O2 is the solubility in the absence of iron nitrosylation, calculated for 25, 50, and 75% oxygen saturations using Equation 2. This same equation was used to calculate C s NOϩO2 where the y s then represented the total oxygen saturation and the estimated percentage of nitrosylated hemoglobin that would be hexacoordinate (or R-state). and in vivo, are also likely to pertain to venous blood.
Even if the solubilizing effect of HbS-NO were strong, the amount of iron nitrosylation achievable through (for example) NO inhalation therapy is less than 5 M (6, 44). This small amount of nitrosylation would need to have some sort of magical effect (compared with oxygen) in order to have a significant effect in vivo. NO therapy may also increase levels of SNO-Hb, where NO reacts with the ␤93 cysteine. However, the amount of SNO-Hb formed is more than 10-fold less than the iron nitrosyl Hb formed (44). NO inhalation can also results in peak levels of methemoglobin (MetHbS) formation of 80 M, or about 1% of the total hemoglobin concentration. The solubility of 100% MetHbS is much higher than that of deoxyHbS but no greater than 100% oxygenated HbS (7,45). Given that a change in oxygen saturation of 1% does not significantly affect the solubility, one can predict that no significant effect would be gained from this amount of MetHbS formation either.
Based on our understanding of sickle cell pathophysiology one can envision several mechanisms by which NO administration might have beneficial effects on patient morbidity (5). These include reactions with sickle hemoglobin to alter its solubility or oxygen affinity, reactions with sickle erythrocytes to change cell volume and intracellular hemoglobin concentration, and effects on hemoglobin synthesis to change the intracellular hemoglobin composition, i.e. the ratio of sickle to nonsickle hemoglobins. At the physiological level, these include its strong direct effects on vascular tone in general, as well as effects on specific organs such as the pulmonary vessels. Less directly, the effects of NO on platelet aggregation and endothelial adhesion molecules might be beneficial by reducing components of the disease that appear similar to ischemia-reperfusion injury. We have found a negligible effect of nitrosylation on solubility. This result argues against reduced polymerization as a mechanism for increased oxygen affinity associated with exposure to NO suggested by Head et al. (10) and is consistent with Gladwin et al. (46) inability to observe such an increased oxygen affinity. If NO does increase the oxygen affinity of sickle cell hemoglobin as reported by Head et al. (10) but contested by Gladwin et al. (46), then our results suggest that it must do so via another mechanism than through a direct solubilizing effect.
In conclusion, we have found that the direct solubilizing effect of iron nitrosylation due to endogenous amounts of NO or NO produced by inhalation therapy or other means is very likely to be negligible in sickle cell patients. A role for NObased therapies for sickle cell disease based on vasodilation, anti-platelet, or anti-adhesive activity, or hemoglobin F induction are being actively investigated (5).