Crawford and colleagues1 have studied the effects of heme oxidation, glutathione (GSH), and oxygenation (pO2) on the bioactivity of S-nitrosohemoglobin (SNOHb). They report that (1) SNOHb in the ferric or met oxidation state (SNOmetHb) is a more potent vasodilator than SNOHb in the oxygenated state (SNOoxyHb); (2) SNOoxyHb vasoactivity is not mediated by nitric oxide (NO) itself; (3) SNOHb activity is potentiated by GSH; and (4) SNOHb activity is greater at low pO2 than at high pO2. These observations confirm previous findings.2-5
Crawford and colleagues do reach one conclusion that differs significantly from previous work2-5 : they report that the behavior of SNOHb under hypoxia is independent of Hb allostery. They observed “similar degrees” of hypoxic potentiation for SNOHb and other nitrosovasodilators (actually ranging from 2- to 15-fold). Then, by assuming that all these species must act through similar mechanisms, they conclude that hypoxic potentiation of vasodilation by SNOHb cannot involve Hb allostery. However, this conclusion only revives an old controversy, resolved earlier3,6 when it became clear that biologists had been using dimers of hemoglobin (instead of tetramers) to study hemoglobin-derived vasoactivity and thus could not test for allostery. Likewise, Crawford and coworkers did not actually test the role of allostery. Decisive experimental results2-5,7 clearly illuminate the allosteric mechanism operative in Hb and differentiate it from the O2-dependent behavior of other nitrosovasodilators, in particular other S-nitrosothiols. The paper of Crawford and colleagues represents a fundamental misunderstanding of this work.
Intraerythrocytic SNOHb does not dilate blood vessels directly, but rather transfers its NO group within the red blood cell (RBC) to cysteine thiols in glutathione (forming S-nitrosoglutathione; GSNO) or band-3 protein, which transduces the NO bioactivity.2-5,7,8 Crystal structures and molecular models of SNOHb reveal that the NO group has little, if any, access to solvent in liganded (R-structured) Hb molecules, and thus its release is inhibited (Stamler et al5 and Chan et al9 ). In contrast, NO is solvent-accessible in the deoxygenated (T or high spin) structure. Thus, transfer of the NO group is linked to the allosteric state of Hb. In a series of experiments, we examined in both cell-free and RBC preparations the effect on vessel tone of the position of SNOHb's R-T equilibrium, which was modulated by varying pO2 widely (in the presence of physiologic concentrations of glutathione and band-37,8 ), or by varying widely the thiol concentration (at extremes of pO2 that strongly favor either the R or T structures),3,5 or by altering spin state.2,4 Within the physiologic range of pO2, low-mass SNOs3,5 do not exhibit significant pO2-regulated responses, in contrast to SNOHb and RBCs.5,7 That is, not only is the pO2-dependence of GSNO vasoactivity smaller in degree than that of SNOHb/RBCs, but the functional relationship between pO2 within the physiologic range and biologic response is markedly different—only SNOHb exhibits graded responses across the physiologic pO2 range within which Hb undergoes an allosteric transition.5,7 Moreover, the greater vasoactivity of SNOmetHb over SNOoxyHb can be rationalized by the influence of spin state on Hb allostery, as championed by Perutz.10
The mechanism and degree of low pO2-potentiated vasodilation by NO donors vary substantially among different donors and tissues. For example, the mechanism of potentiation can be allosteric (SNOoxyHb),3 superoxide-related (NO),11 or dependent on direct actions of O2 on the blood vessel (eg, on ion channels).12 Moreover, there is great diversity in the mechanisms through which NO donors regulate muscle tone (involving various membrane channels and enzymes).13-15 Thus, not surprisingly, Crawford et al find that potentiation by low pO2 varies from 2- to 15-fold among different classes of NO donors (Table 1 and Figure 5C are at variance).1 However, no experiments were conducted that would have tested their conclusion that potentiation of SNOHb activity at low pO2 is independent of allostery.
The challenges posed by bioassay experiments with Hb have not been adequately addressed in the literature. Aspects of experimental design and interpretation, which undermine the conclusions by Crawford et al, therefore merit further discussion. They also serve to illuminate further the limitations of the experiments of Crawford et al in testing for allostery.
The multimeric state of Hb is determined by concentration and O2 saturation. Thus, in all of the experiments examining the effect of SNOHb on vessel tone (Figures 1, 2, 5, and including the single experiments at 250 nM Hb and at 2.5 μM Hb),1 “Hb” dimers, which do not undergo the R-T allosteric transition, outnumber tetramers (in most experiments by many fold).16 The use of partially nitrosylated globins introduces heterogeneity that further obscures the position of the allosteric equilibrium. Moreover, the main analysis of pO2-dependence for assessing allostery (Figure 5C)1 was carried out with SNOmetHb, which does not bind oxygen.
A single experiment conducted with inositol hexaphosphate (IHP) was offered as a further test of allostery. IHP exerts coupled effects on both R-T and dimer-tetramer equilibria, the latter showing a biphasic concentration-dependence.17 Thus, any interpretation of the effect of IHP on the SNOHb preparations used (250 nM) must account for the composite behavior of oxygenated dimers, deoxygenated tetramers, and intermediate ligation states, as well as both partially and fully S-nitrosylated dimers and tetramers, all present in the preparation. Without such an analysis, the reported effects of IHP in the bioassays are essentially uninterpretable.
Experiments with SNOHb were carried out in the presence of glutathione to enable vasodilation. However, the high concentration of glutathione used (vis-à-vis Hb) militates against the possibility of probing the effect of R/T on the equilibrium between SNOHb and GSNO, as detailed previously.3 Indeed, Crawford et al used exactly those bioassay conditions (1-100 nM Hb/100 μM GSH) in which pO2 (∼12 vs 700 mmHg) was shown to have minimal effect3 —in bioassay, similar magnitudes of vasodilation can belie significant differences in the rate of GSNO formation.
The authors discuss differences in the behavior of GSNO versus the “bioactive product” of SNOHb in the presence of GSH (Figure 2).1 However, GSNO is, in fact, produced in the reaction between SNOHb and GSH,2,3,18 and was inevitably present in their reaction mixtures. Nonetheless, SNOHb/GSH bioactivity was ascribed, mysteriously, to some compound other than GSNO—it is unlikely that a novel compound was produced. Such a misinterpretation of the behavior of SNOHb/GSH may arise from any number of differences in protocol or assay conditions (eg, order of GSH addition, amount of GSNO formed, number of times a tissue is exposed to GSH, redox state, and amount of trace metals [free or bound to proteins]).
Based on reversibility by oxyHb, Crawford et al conclude that vasodilation by SNOmetHb/GSH (and by GSNO/GSH), but not by SNOoxyHb/GSH, is mediated by NO release. In these experiments, amounts of GSNO formed were unknown, and the operative reaction pathways are entirely unclear. Inferences about the identity of vasoactive species based on putative scavenging of NO by heme are tenuous, with multiple alternative mechanisms of relevance (eg, differential interactions of oxyHb/SNOoxyHb/metHb/SNOmetHb dimers in solution and differential access to and actions within tissue by extravasated Hbs). A role for cyclic guanosine monophosphate (cGMP) is a better indicator of dependence on NO release, and SNOmetHb does not necessarily raise cGMP or depend obligatorily on cGMP for bioactivity.4
The potentiation of SNO vasoactivity by superoxide dismutase (SOD; Figures 4, 6)1 was interpreted in terms of superoxide scavenging of NO. But the pO2-independence of the SOD effect indicates otherwise, and the conclusion (based on these data) that vasoactivity of S-nitrosothiols such as GSNO is mediated by NO overlooks extensive evidence to the contrary.19-22 In particular, GSNO-mediated relaxation involves transnitrosation (NO+ transfer chemistry). The release of NO from SNOs is also not unimolecular,23 but rather a result of reactions with nucleophiles, metals, and reductants,23 including both superoxide24 and SOD itself.25,26 Different effects of SOD on different SNOs may thus reflect differences in their reactivity toward superoxide or SOD, rather than dependence on NO for vasoactivity.
Interpretations of data from chamber bioassays are valid only if compounds are distributed evenly in solution. Crawford et al conclude that oxyHb and myoglobin (Mb) scavenge NO generated in solution by SNOmetHb and GSNO. However, dimers of SNOHb and oxyHb (and Mb monomers) extravasate into tissues, and GSNO participates in chemistry at the solution-tissue interface. Thus, their mechanistic comparisons, based on compounds variably distributed in solution and tissues, are not valid. (Hb dimers and Mb monomers may affect NO-related bioactivity via multiple mechanisms, which are beyond the scope of this discussion.)
The spectrum of “SNOHb” (Figure 1)1 includes a large absorbance at 340 nm, whereas SNOHb has no such absorbance. Thus the nature of the vasoactive species assayed remains enigmatic.
Standard bioassay chambers are not designed to maintain an exact pO2. Crawford et al selected the steepest portion of the Hb-O2 dissociation curve as their low pO2 condition (50% Hb saturation). As a result, Hb saturation would fluctuate widely during and between experiments—that is, from 35% to 65% with variation in pO2 of ± 2 mmHg—blurring allostery-dependent behavior.
In sum, the conclusions of Crawford et al with regard to the role of allostery in the pO2-dependent vasoactivity of SNOHb rest primarily on an analysis of the effects of dimers of SNO“Hb,” and dimers of oxidized “Hb,” neither of which can respond allosterically to O2; dimers of Hb will also extravasate readily into the vessel wall. Thus, these studies simply do not bear on the mechanism through which RBCs effect hypoxic vasodilation. Nor do their results provide useful insight into the mechanisms of vasorelaxation by other nitrosothiols, and in particular, readers should not take home the message that NO mediates the effects of GSNO or SNOmetHb.
Blood flow and SNOHb: an allosteric controversy
Does S-nitrosohemoglobin (SNOHb) modulate blood flow under physiologic conditions? If so, what are the mechanism(s) by which this occurs? Is oxygen-dependent (allosteric) control necessary? What is the role of low-molecular-weight thiols? How is SNOHb made in vivo and what are the physiologic concentrations in humans? These and other questions are still being asked by many investigators and lie at the heart of the ongoing controversy surrounding the SNOHb hypothesis. The concepts underlying this hypothesis integrate the mechanisms of allosteric regulation of oxygen binding to Hb with the delivery of nitric oxide (NO) from SNOHb. The proposed biologic function of this property of SNOHb is to mediate vasorelaxation in a process coupled to the R to T conformational transition of this protein.1-5 While this is an interesting and provocative idea, the reader not fully immersed in this complex field would be left with the impression from the letter by McMahon et al that this novel role of SNOHb in modulating physiologic blood flow is an established fact. In our opinion, the SNOHb controversy is very much alive, with many of the questions raised above far from resolved. The focus of the letter from McMahon et al was an article that we recently published6 describing a series of studies in which we asked how does SNOHb promote vasodilation in the presence of glutathione (GSH) and what role does oxygen play in this process.
The data showed the following: (1) the oxidation state of the heme (ferrous vs ferric) in SNOHb modulates the vasodilatory activity; (2) there are distinct differences in the vessel relaxation characteristics of SNOHb and S-nitrosoglutathione (GSNO), a putative intermediate of SNOHb bioactivity; and (3) lowering oxygen tensions enhances vasodilation stimulated not only by SNOHb, but also by other nitrosovasodilators that are not subject to allosteric regulation. We concluded therefore that such data do not “provid[e] conclusive evidence for an allosteric-based mechanism for SNOHb mediated vasodilation [italics added].”6 It is this conclusion that McMahon et al have disputed with some passion in a list of 9 points detailed in their letter.
All investigators will appreciate that subtle differences in protocols can dramatically affect the result acquired; “the devil is in the details,” as the saying goes. However, many of the points raised in their letter address detailed aspects of protocol that do not affect the validity of the experiments. For example, the “enigmatic” absorbance at 340 nm that McMahon et al's letter refers to in point 8 is due to the presence of L-NAME (N omega-nitro-l-arginine methyl ester) and indomethacin (present in the vessel bioassay experiments from which the spectra were derived). Similarly, criticism placed on our discussion of the effects of superoxide dismutase (SOD) in our study appear simply semantic in nature (point 6). In contrast it is remarkable, in our view, that the data obtained from different laboratories in this field,1,5-7 including the Stamler laboratory, show that the results from experiments investigating the vasodilatory effects of SNOHb are remarkably similar (we will discuss some of these below, but also encourage the readers to compare these data themselves). In other words, the data, and by extrapolation the experimental protocols for SNOHb vasodilation experiments, are on the whole consistent from lab to lab; what differs is the interpretation. With this in mind, detailed issues regarding other aspects of the experimental procedures (which importantly apply to all studies using similar protocols) did not further the debate and are not discussed further here. We will focus on 2 issues that were prevalent themes in McMahon et al's letter, namely the impact of Hb dimers and role of GSH/GSNO.
The proposal that deoxygenation of SNOHb (ie, allosteric control of “R” and “T” state Hb) controls its vasoactive responses stems from observations showing (1) potentiation of SNOHb vasodilation at low oxygen tensions1,5 ; (2) arterial-venous gradients of SNOHb (higher concentrations in the arterial relative to venous circulation)2,4 ; and (3) a correlation of Hb oxygen saturation and SNOHb concentrations in humans subjected to hyperoxia and hypoxia.2 We did not investigate points 2 and 3 in our study, but the reader should be aware that these issues are also subjects of vigorous debate, as other groups have not detected A-V concentration gradients of SNOHb in humans (with SNOHb levels being significantly lower than first reports).8,9 More importantly, points 2 and 3 do not provide direct mechanistic evidence of allosteric control at the level of Hb. We did however, investigate point 1 and confirmed that lowering the oxygen tension increased the efficiency of SNOHb-dependent relaxation.
To test the role of allostery, we exploited the dimer-tetramer equilibrium and modulated the allosteric state of Hb. Conditions in which dimers form a significant proportion of the Hb population, and cannot exhibit allostery, were a key component of the design of these experiments, not an oversight of the biochemical properties of Hb in solution and far from a “fundamental misunderstanding” of experimental protocols. McMahon et al state that under our experimental conditions (250 nM-2.5 μM Hb protein or 1 μM-10 μM heme), dimers would outnumber tetramers several fold. According to the citation used for this statement (see Figure 3 in Valdes et al10 ), at equilibrium, oxyHb at a concentration of 1 μM heme will contain approximately 80% dimers. However, increasing the concentration to 10 μM heme (the higher of the 2 concentrations we tested in our study) decreases the dimer content to approximately 40% of the total Hb. In this way the proportion of Hb capable of exhibiting allosteric behavior was either 20% or 60% and should reveal a contribution of allostery to vasorelaxation at low oxygen tension. However, lowering the oxygen tension to 11 torr (resulting in 40%-50% deoxygenation) potentiated vasodilation independent of the proportion of Hb capable of exhibiting allostery. As an alternative approach we promoted “T” state Hb formation by addition of inositol hexaphosphate (IHP; note the equilibrium between dimer and tetramer is significantly shifted to the latter in “T” state Hb), and observed a similar fold increase in the efficiency of vasodilation by lowering oxygen tension, compared with the responses in the absence of IHP. McMahon et al speculate in their point 2 that the aggregation state of Hb and ligation/oxidation intermediates thereof may confound data interpretation. However, it is important to note that many of the experiments in the literature from all the interested groups will have varying proportions of such intermediates (which will be dependent on the ability to control the oxygen tension [see point 9 of McMahon et al's letter]), and such parameters are usually an uncontrolled element of these studies. Nevertheless, an enhanced vasorelaxation of SNOHb at low oxygen tension is a consistent feature. Is there an alternative explanation for the oxygen-dependent SNOHb vasorelaxation that lies outside the allosteric properties of the heme protein? The simplest explanation is that the mechanisms underlying the oxygen-dependent responses are intrinsic to the vessel wall, and this prompted us to investigate the effects of oxygen on other NO-dependent mediators of relaxation.
Finally, we would like to comment on the experimental conditions as they pertain to the use of GSH (point 3 of McMahon et al's letter) and the role of GSNO (points 4-6). In our studies, we used 100 μM GSH and SNOHb concentrations (0-2.5 μM protein) with levels of S-nitrosation varying from 0.4 to 2 SNOs per Hb tetramer. The resultant ratios of GSH to SNOHb fall within the range used by McMahon et al1 in which they also observe a potentiation of relaxation responses at low oxygen tensions. Again, although these conditions supposedly “militate against the possibility of probing the effect of R/T” (point 3 of McMahon et al's letter), if this were the case, we should not have observed any oxygen dependence in SNOHb-dependent vasodilation. However, we did observe this, and once more the simplest explanation is that allosteric processes in Hb do not account for the enhanced vasodilation response at low oxygen tensions.
The role of GSNO is more difficult to ascertain from existing data. The assertion that GSNO-mediated relaxation involves transnitrosation ignores data from many laboratories demonstrating that NO scavengers inhibit GSNO-dependent relaxation of aortic and other vascular beds.6,11-14 In addition, the lowest oxygen tension used in our study was approximately 11 mmHg, which is similar to that used by McMahon et al1 (approximately 12 mmHg, see point 3 of McMahon et al's letter). This underscores the similarity in experimental conditions discussed earlier and indicates that such factors are not the basis for misinterpretation. Clearly, GSNO-dependent processes are complex and may involve transnitrosation reactions and NO-dependent and -independent processes, and require further elucidation.
In summary, we have shown that one of the key pieces of data supporting an allosteric role for SNOHb-dependent vessel relaxation can be explained by mechanism(s) independent of Hb conformational states. Such a process appears to reside in the vessel wall and its further understanding will provide important insights into the effects of hypoxia. Many questions surrounding the biologic function of SNOHb, and NO interactions with Hb in general, remain and are actively being investigated. Understanding these interactions will extend our understanding of pathophysiologic functions of NO and S-nitrosothiols in the vasculature and potentially yield novel insights for the therapeutic sector in the treatment of diseases in which NO-dependent function is compromised. Regarding the SNOHb hypothesis, we encourage the readers to derive their own conclusions by directing them to recent review articles that engage these controversies, discuss the “decisive” results, and encompass the different views in this developing field,15-17 and hope that these stimulate more experimentation in what is a topical and important area of investigation.
Correspondence: Rakesh P. Patel, Department of Pathology, University of Alabama at Birmingham, 901 19th St S, BMR-2, Rm 307, Birmingham, AL 35294; e-mail: firstname.lastname@example.org