Detection of Human Red Blood Cell-bound Nitric Oxide*

Major disparities in reported levels of basal human nitric oxide metabolites have resulted in a recent literature focusing almost exclusively on methods. We chose to analyze triiodide chemiluminescence, drawn by the prospect of identifying why the most commonly em-ployed assay in nitric oxide biology typically yielded lower metabolite values, compared with several other techniques. We found that the sensitivity of triiodide was greatly affected by the auto-capture of nitric oxide by deoxygenated cell-free heme in the reaction chamber. Potential contaminants and signal losses were also associated with standard sample purification procedures and the chemistry involved in nitrite removal. To inhibit heme nitric oxide auto-capture, we added potassium ferricyanide to the triiodide reagent, reasoning this would provide a more complete detection of any liberated nitric oxide. From human venous blood samples, we established nitric oxide levels ranging from 0.000178 to 0.00024 mol nitric oxide/mol hemoglobin. We went on to find significantly elevated nitric oxide levels in venous blood taken from diabetic patients in comparison to healthy controls ( p < 0.0001). We concluded that the lack of signals reported of late by several groups using triiodide chemiluminescence for the detection of hemoglobin-bound nitric oxide may not represent levels on the border of assay sensitivity

Traditionally, endothelial-derived nitric oxide (NO) 1 has been viewed solely as a paracrine effector (1,2), playing an important role in the local control of vascular tone and blood flow. However, more recent evidence has revealed an endocrine role for NO via the formation of metabolites that preserve bioactivity and contribute to blood flow regulation and oxygen delivery (3)(4)(5)(6)(7). Studies into the mechanisms for this effect have led to two alternative models that emphasize the significance of different metabolites, the ability of S-nitrosohemoglobin (SNO-Hb) within red blood cells to provide bio-available NO (3,5) and the potential formation of NO from nitrite (NO 2 Ϫ ) mediated by the reductive potential of deoxyhemoglobin (8). Although NO 2 Ϫ reduction by deoxyhemoglobin has been shown to represent a potential source of NO (in conjunction with methemoglobin) (9), it has yet to be convincingly demonstrated that this NO could remain free to play a physiological role in vasodilation (10,11). This would necessitate NO escaping the red blood cell and avoiding uptake/binding to ferrous heme, which is highly improbable considering the low oxygen conditions under which deoxyhemoglobin acts as a NO 2 Ϫ reductase (12,13). Further detracting from a direct role for NO 2 Ϫ in hypoxic vasodilation is the fact that nitric-oxide synthase inhibition, which acutely reduces endogenous NO 2 Ϫ levels (14), does not block exerciseinduced hypoxic vasodilation (10). Recent reports have highlighted the formation of SNO-Hb following the oxygenation of nitrosylhemoglobin (HbNO) formed from the reaction of deoxyhemoglobin with NO 2 Ϫ (NO 2 Ϫ reductase activity) (15). Intriguingly, this links NO 2 Ϫ reduction, which would occur under low oxygen conditions, to the formation of SNO-Hb via oxygenation of HbNO in the lungs, creating red blood cells able to exhibit potent vasodilatory activity at low partial pressures of oxygen in the peripheral circulation (16).
The measurement of NO metabolites in biological matrices is notoriously difficult and technically demanding due to dilute concentrations of NO and its complex interactions with various biological constituents (17,18). This is reflected in the wide ranging reports of basal levels of human red blood cell Hbbound NO. These discrepancies have fueled arguments over the existence of arterial/venous metabolite gradients and a fierce body of literature critically scrutinizing the methodology of measurements (12, 17, 19 -27). Although it is generally agreed that wide ranging metabolite values have resulted from the use of different analytical approaches with no rigorous standards to ensure specific, accurate, and reproducible measurements (12,17,20), these inconsistent findings have still diverted attention away from the potential physiological significance of a "reserve" pool of NO in the microcirculation and have even cast doubt on its whole relevance.
One technique reported suitable in terms of detection limit for the measurement of basal NO metabolites is ozone-based chemiluminescence (25)(26)(27)(28). Several chemical reductive assays have been developed for the detection of various NO species in conjunction with this technique. The most commonly reported of these has been the triiodide assay (28 -30). This assay is proposed to stoichiometrically measure NO gas from S-nitrosothiols (RSNOs) and NO 2 Ϫ (18) and also measure N-nitroso compounds (RNNO) and nitrosylheme complexes (NO-heme/ HbNO) (28). Reported NO yields are ϳ100% for low molecular weight RSNOs, ϳ78% for SNO-albumin, ϳ81% for HbNO, and ϳ67-105% for N-nitroso compounds (28). The reaction mechanisms for triiodide cleavage of RSNO have previously been outlined (29), but the underlying reactions involved in the denitrosation of RNNOs and nitrosylheme species still remain to be elucidated (28). No explanation is given as to why the NO yield from SNO-albumin is so low in comparison to that from low molecular weight RSNOs. However, others have stated that the standard sample pretreatment associated with the triiodide assay destroys some SNOs, thus affecting the NO yield (25,31).
For the reason that the triiodide assay converts most NOrelated species back to NO gas (strong reducing agents in strong acid at high temperature), it is necessary to use sample separation procedures to obtain the species of interest (11). This not only increases the chance of contamination but also susceptibility to losses of unstable NO species (25). According to Nagababu et al. (11), the standard purification procedure used in conjunction with the triiodide assay by Gladwin and co-workers (18,38) results in the oxygenation of samples and the subsequent loss of HbFe 3ϩ NO, which they suggest accounts for the majority of HbNO in human blood. It is tempting to speculate that this is one reason why recent papers from groups using the triiodide assay have failed to detect basal Hb-bound NO in human blood (23,27). In line with these reports, we also fail to detect signals from human red blood cell samples using the very same methods. However, higher levels of Hb-and plasma-bound NO from human samples have been reported from groups using a variety of different assays and techniques including photolytic cleavage of NO linked to ozonebased chemiluminescence (5), a novel potassium ferricyanide (K 3 Fe III (CN) 6 ) reagent linked to chemiluminescence (11), spectrophotometric detection (32), electrochemical detection (33,34), and mass spectrometry (35). Nevertheless, these findings have, themselves, been ascribed to artifact resulting from the generation of NO from NO 2 Ϫ and NO 3 Ϫ (11,23,27,36). Despite its widely reported use in the literature, the triiodide assay has been difficult to follow because of numerous subtle changes to the methodology. This lack of continuity is exemplified by measurements 1) with different concentrations of potassium iodide and K 3 Fe ϩ (CN) 6 ; 2) with and without potassium cyanide, Nonidet P-40, N-ethylmaleimide, diethylenetriaminepentaacetic acid, mercury chloride (HgCl 2 ), acidified sulfanilamide; 3) with markedly different incubation times; and 4) with different red blood cell, whole blood, and plasma preparations (10,18,(37)(38)(39)(40)(41). Further inconsistencies only serve to confuse the issue; for example, a method was used to measure HbNO (10), which the same group previously highlights cannot be used to measure this species (18).
Another chemical-reductive assay used in conjunction with ozone-based chemiluminescence is the cysteine/cuprous chloride assay, developed to specifically measure RSNOs (42). This assay exploits trans-nitrosation reactions between biological SNOs and the liberation of NO from RSNOs in the presence of Cu ϩ (43). The neutral pH of the cysteine/cuprous chloride reagent ensures that NO 2 Ϫ , NO 3 Ϫ , and 3-nitrotyrosine are not detected, thus maintaining specificity for RSNO. This means that sample pretreatment is not necessary, reducing the likelihood of contamination and potential losses of unstable NO species. NO yields from this assay are reportedly stoichiometric for low mass SNOs; however, the yield from SNO-albumin is ϳ33.9% (42). The low returns from this specific plasma protein are thought to result from sheltering of the SNO bond inhibiting trans-nitrosation to cysteine (42). This low yield from SNO-albumin may also reflect the unphysiological nature of SNO standards used (for example, in terms of fatty acid and copper content).
Drawn by the prospect of identifying why the chemical-reductive assays have typically yielded lower levels of Hb-bound NO than other methods, we chose to examine two of these assays and set out to establish 1) the effect of Hb in the reaction chamber on NO detection, 2) the effect of inhibiting NO Hb interaction in the reaction chamber, and 3) the consequence of sample pretreatment on the recovery of NO from human red blood cells. Initially, we looked at NO recovery from sodium nitrite (NaNO 2 Ϫ ) and S-nitrosoglutathione (GSNO) in the presence/absence of different concentrations of Hb in the reaction chamber. We then explored the potential of K 3 Fe III (CN) 6 to inhibit heme NO interaction and examined its influence on the recovery of NO from NaNO 2 Ϫ and GSNO in the presence/absence of Hb in the reaction chamber. Finally, we used our modified reagent to analyze basal venous blood samples taken from healthy individuals and diabetic patients.
We describe here several key findings implying that the current triiodide and cysteine/cuprous chloride assays both under-measure Hb-bound NO present in human red blood cell samples due to the auto-capture of NO by deoxygenated Hb/ cell-free heme in the reaction chamber. Furthermore, the current triiodide assay is also affected by signal losses and potential artifacts associated with sample pretreatment. By inhibiting heme NO auto-capture and avoiding harsh chemical pretreatment of samples, we highlight that the level of basal red blood cell-bound NO detected from human samples is significantly higher than recently reported (10,23,27,38,39).

EXPERIMENTAL PROCEDURES
Chemicals and Standards-All chemicals were purchased from Sigma, except for GSNO, which was from Alexis, and glacial acetic acid, from Fischer. Standards of NaNO 2 Ϫ and GSNO were prepared in high pressure liquid chromatography-grade NO 2 Ϫ -free water. Biological Samples-Blood was drawn from the femoral or antecubital vein of healthy and diabetic individuals. Samples were injected into 4-ml EDTA collection tubes, which were centrifuged at 600 ϫ g for 10 min at 4°C. Both the red cell fraction and plasma were snap frozen in liquid nitrogen and stored at Ϫ80°C for subsequent analysis. Upon analysis, both frozen red blood cell and plasma samples were thawed in a water bath at 37°C for 3 min. The effects of freezing and thawing have not been specifically addressed; however, all samples were treated the same. Red blood cell samples were subsequently lysed 1:4 in EDTA (0.5 mM; pH corrected to 7.0) and incubated for 5 min on ice. Purified Hb was acquired by passing 500 l of the red blood cell lysate through a prewashed (with high pressure liquid chromatography NO 2 Ϫ -free water) Sephadex G25 column, which takes ϳ3 min at room temperature (20 -22°C). Consequently, the preparation of red blood cell lysate, from the freezer to the time of injection, took ϳ8 -9 min, whereas the preparation of purified Hb took ϳ11-12 min. Preparative losses of NO species over these time periods have not been addressed, but they were the same for all samples studied.
Ozone-based Chemiluminescence Setup-Assay reagents (5 ml) were placed in a glass purge vessel with a rubber septum-covered injection inlet. Oxygen-free nitrogen gas was bubbled through the reagent mix, which was heated to 50°C (Ϯ1°C) in a water bath on a thermostatically controlled hotplate. The reaction vessel, linked to a trap containing 40 ml of 1 N sodium hydroxide, was further connected to the NO analyzer (Sievers NOA 280i, Analytix, UK).
Cysteine/Cuprous Chloride Assay-A stock solution of reagent (1 mM cysteine, 100 M cuprous chloride in 400 ml of distilled water) was prepared fresh each day and pH corrected to 7.0 by the addition of sodium hydroxide. Only one Hb sample was injected into each assay mix before being replaced to prevent saturation of Hb and exhaustion of the denitrosating reagent.
Original Triiodide Assay-A stock solution of triiodide reagent (70 ml) was prepared fresh each day as previously outlined (18). Only one red blood cell or Hb sample was injected into each assay mix before being replaced to prevent saturation of Hb and exhaustion of the denitrosating reagent. Reactions of the triiodide reagent with other redoxactive agents could not be controlled; however, the effect of endogenous redox-active agents on triiodide have previously been investigated and found not to interfere with the denitrosation process (28).
Modified Triiodide Assay-The original triiodide assay was modified by adding K 3 Fe III (CN) 6 (25 mM final) to limit Hb/cell-free heme (Fe(II)) auto-capture of NO. Fresh K 3 Fe III (CN) 6 was added to the triiodide reagent just prior to its use, rather than to the master mix. The assay mix was replaced after every blood injection to prevent saturation by Hb and exhaustion of the denitrosating reagent. Direct injection of red blood cell lysate or purified Hb (200 l) into the modified mix was taken to reflect total red blood cell-bound NO (i.e. SNO-Hb, HbNO, and any NO 2 Ϫ bound to Hb). NO Scavenging Experiments-One hundred l of either NaNO 2 Ϫ (250 or 500 nM) or GSNO (500 nM or 1 M) was injected into the reaction chamber, followed by different volumes of plasma (25 l-1 ml) or different concentrations of Hb; lysed red blood cells or purified Hb (1.25-160 M Hb final, i.e. 5-660 M heme final). Finally, a repeat 100-l injection of the same NaNO 2 Ϫ or GSNO standard was made. Experiments were performed in the original triiodide and cysteine/ cuprous chloride reagents to identify the scavenging potential of cellfree heme/plasma and in the modified triiodide reagent to illustrate the inhibition of NO scavenging.
Human Red Blood Cell-bound NO Measurement (Original versus Modified Triiodide Reagents)-Blood taken from the antecubital vein of healthy individuals was treated as outlined above (see "Biological Samples"). Red blood cell lysate from the same blood sample was injected into the original triiodide reagent and then into the modified triiodide reagent to compare chemiluminescence signals.
Human Red Blood Cell-bound NO Measurement (Healthy versus Diabetic Blood)-Blood taken from the femoral vein of healthy individuals and diabetic patients was treated as outlined above (see "Biological Samples"). Red blood cell lysate was subsequently injected into our modified triiodide reagent mix to measure total red blood cell-bound NO (i.e. SNO-Hb, HbNO, and any NO 2 Ϫ bound to Hb). Data Presentation and Statistics-All data are presented either as original recordings or as means Ϯ S.E. An unpaired t test was used to compare differences in means between groups. A two-tailed p value Ͻ 0.05 was considered significant throughout.

RESULTS
Calibration, Detection Limit, and Reproducibility-With the original triiodide reagent, the relationship between NO added from NaNO 2 Ϫ (Fig. 1A) and GSNO (Fig. 1B) and NO detected (total area under curve) was linear across a broad range of concentrations (10 nM-5 M); the limit of detection was ϳ1-5 pmol for both standards (Fig. 2). Similarly, with the cysteine/ cuprous chloride reagent, a linear relationship was observed with the amount of NO detected (total area under curve) over a wide range of GSNO concentrations (50 nM-5 M). GSNO detection by the triiodide and cysteine/cuprous chloride reagents compared directly (Fig. 1C). The addition of K 3 Fe III (CN) 6 to the cysteine/cuprous chloride reagent totally inhibited the detection of GSNO (data not shown); however, the addition of K 3 Fe III (CN) 6 to the triiodide reagent had no influence on the signals derived from NaNO 2 Ϫ (Fig. 1A), GSNO (Fig. 1B), or plasma samples (data not shown).
Detection of NO in the Presence of Hb-Chemiluminescent signal amplitudes arising from NaNO 2 Ϫ in the original triiodide reagent (Fig. 3), GSNO in the original triiodide reagent (Fig. 4), and GSNO in the cysteine/cuprous chloride reagent (data not shown) were inversely related to the concentration of heme in the reaction chamber (Fig. 5). The cysteine/cuprous chloride reagent was more sensitive to the addition of heme than the triiodide reagent (note the x-axis scales, Fig. 5). The inverse relationships of signal amplitude with the increasing addition of heme were similar with both reagents, whether red blood cell lysate or purified Hb (Figs. 3 and 4 for the triiodide assay; data not shown for the cysteine/cuprous chloride reagent) was present in the reaction chamber. The broadening of signals derived from NaNO 2 Ϫ or GSNO in the presence of red blood cell lysate or purified Hb was inhibited following the addition of K 3 Fe III (CN) 6 to the triiodide reagent (Figs. 6 and 7 respectively). K 3 Fe III (CN) 6 could not be used in conjunction with the cysteine/cuprous chloride reagent, because it inhibited the detection of GSNO (as outlined above).
Detection of NO in the Presence of Plasma-The addition of plasma into the reaction chamber had no influence on either  2. Sensitivity of nitric oxide measurement using the original triiodide reagent. Raw chemiluminescence signals observed from sodium nitrite standards (100-l injections) of the following concentrations (right to left: 5 M, 1 M, 500 nM, 100 nM, 50 nM, 10 nM, and water) in triiodide (5 ml). The inset is a magnified view of the last four injections. Note that high pressure liquid chromatography-grade nitrite-free water produces a signal when exposed to room air, attributed to the hydrolysis of absorbed ambient N-oxides (28). signal amplitude or the area under curve analysis of NaNO 2 Ϫ (Fig. 8) or GSNO (data not shown) standards in the triiodide reagent. Plasma was not tested with the cysteine/cuprous chloride reagent because of the known limitations of this reagent in measuring albumin SNO (ϳ33.9% NO yield) (42).
Total Red Blood Cell-bound NO (Original versus Modified Triiodide Reagents)-Although chemiluminescence signals were observed from untreated plasma samples injected directly into the original triiodide reagent (Fig. 8), nothing was detected from either red blood cell lysate or purified Hb samples (Figs. 6 and 7, A and C). However, signals were observed upon injection of either purified Hb (Fig. 6, B and D) or red blood cell lysate (Fig. 7, B and D) directly into the modified reagent mix containing K 3 Fe III (CN) 6 . To clarify this finding, total red blood cell-bound NO measurements were compared from the same venous human-lysed red blood cell samples injected into the original triiodide reagent (Fig. 9A, left) and into our modified triiodide mix (Fig. 9A, right). Whereas the chemiluminescence signal was indistinguishable in the original reagent, in our modified mix, the area under curve equated to 0.00024 Ϯ 0.00001 mol nitric oxide/mol Hb, which corresponds to 558.49 Ϯ 12.43 nM NO in whole blood (three repeat experiments using different frozen aliquots of the same red blood cell sample).
Total Red Blood Cell-bound NO (Healthy versus Diabetic Blood)-Comparison was made in terms of total red blood cell-bound NO from the blood of healthy individuals and diabetic patients. The NO:Hb ratio from healthy blood samples (0.00022 Ϯ 0.00001; n ϭ 4) was on average 1.6ϫ lower than that measured in the blood from the diabetic patients (0.00036 Ϯ 0.00002; n ϭ 5) (p Ͻ 0.0001) (Fig. 10). DISCUSSION Our data show that previous reports of low levels of Hbbound NO, as detected of late by several groups using chemical-reductive assays, do not represent levels on the border of assay sensitivities but rather may underestimate values because of methodological limitations (10,23,27,38,39). By adapting one of the assays to overcome the major methodological limitations, so as to avoid auto-capture of NO by free heme in the reaction chamber and chemical pretreatment of biological samples, we measured significant amounts of NO bound inside red blood cells in line with other techniques and assays (5,11,32). However, we do not presently advocate the use of our modified reagent, as it still remains to be adequately validated. Nevertheless, we counsel strongly that future assay systems overcome the limitations we have identified so that accurate conclusions can be drawn regarding physiological levels of human NO metabolites.
The effect of heme NO capture has previously been acknowledged but not postulated as a shortcoming of the chemicalreductive assays (28,29). This issue arises, because cell-free Hb in the reaction chamber interacts with free NO (ϳ10 7 M Ϫ1 s Ϫ1 ) (44), preventing its passage in the carrier gas through the reaction mix to the NO analyzer. Because of the anaerobic conditions in the reaction chamber, the interaction between cell-free heme and NO is likely to form HbNO. Typical yields of NO from this species in the original triiodide assay are known to approximate ϳ81%, thus ϳ20% of the signal may be lost.
The auto-capture of NO by cell-free heme in the reaction chamber alters the amplitude of chemiluminescence signals observed from NO standards; however, the area under curve analysis of total NO remains fairly consistent (until the signal amplitude reaches a level where the low signal to noise ratio affects the accuracy of the area under curve measurement). We speculate that the area under curve remained fairly constant because 1) relatively large (500 nM-1 M) amounts of NaNO 2 Ϫ and GSNO were used in comparison to NO detected from human red blood cell samples and 2) that free NO from standards is less likely to be auto-captured in contrast to NO attached to Hb, which is cleaved adjacent to the heme sink. Despite these two factors, increased levels of free heme in the reaction chamber significantly impaired signal amplitude from NO standards (by ϳ80%). Although area under curve analysis may not be affected under conditions where the amount of NO is high relative to the amount of free heme, in situations of lower NO (as in a physiological blood sample), this occurrence is likely to be of much greater relevance, with practical detection being severely hampered. Considering that the final heme concentration from a 200-l red blood cell sample (1:4 EDTA lysed) injected into 5 ml of reagent approximates 160 M and that the final concentration of heme injected with the standard preparative protocol used in many laboratories to detect SNO-Hb approximates 40 M (18), it is little wonder that Hb-bound NO measurements in reagents that do not inhibit heme NO autocapture suffer from reduced amplitude and broadening of the chemiluminescence signal to such an extent that they are likely lost in the noise of the base line, a fact we have demonstrated with red blood cells in the original triiodide reagent (Fig. 9A).
In the present study, we rationalized that K 3 Fe III (CN) 6 added to the original triiodide reagent worked by oxidizing heme from its ferrous Fe(II) to its ferric Fe(III) form, producing methemoglobin that binds NO with lower affinity and is thus a less potent scavenger of NO (45). Therefore, K 3 Fe III (CN) 6 allows NO released from metabolite stores to escape auto-capture in the reaction chamber and be released into the carrier gas for detection by the NO analyzer. Furthermore, we also proposed that the ability of K 3 Fe III (CN) 6 to cleave NO from HbNO (38) may act to improve the yield of NO observed from this species (which is ϳ81% in the original triiodide assay), thus improving overall detectability of total Hb-bound NO species. However, until this method has been fully characterized and the efficiency of cleavage from all NO species against physiological standards established, we cannot accurately determine the yield of NO from blood samples using this method. In particular, we cannot rule out the possibility that K 3 Fe III (CN) 6 addition to triiodide could result in HbFe III NO formation in the red blood cell sample. K 3 Fe III (CN) 6 has previously been used individually and in combination with potassium cyanide to pretreat red blood cell lysate and purified Hb samples to selectively remove HbNO and/or stabilize SNO-Hb (10,18,34,38,39,41). Purification of the sample is then necessary to remove the reactants. We found that the pretreatment of Hb samples with K 3 Fe III (CN) 6 and K 3 Fe III (CN) 6 in combination with potassium cyanide resulted in significant retardation of the Hb fraction on the Sephadex G25 column, resulting in poor separation and significant contamination of the Hb fraction with NO 2 Ϫ . Furthermore, both K 3 Fe III (CN) 6 and potassium cyanide injected into the triiodide reagent generate artifactual chemiluminescence signals, as reported in passing by other workers (18). Consequently, it may not be acceptable to derive levels of NO metabolites by difference from two parallel red blood cell samples (treated versus untreated), as one cannot be certain of the total removal of these reagents from biological samples across the Sephadex column.
Sulfanilamide in hydrochloric acid is another sample pretreatment process that has been extensively used to remove NO 2 Ϫ from samples (10,18,27,28,41,46). In our hands, this treatment of red blood cell lysate, purified Hb, or plasma samples resulted in a time-dependent loss of chemiluminescence signal. This was initially attributed to the significant component of NO 2 Ϫ in the sample as proposed by Bryan et al. (27). However, on closer analysis, we observed that hydrochloric acid alone (across a range of concentrations) was capable of completely reducing the signal. Thus, the validity of sulfanilamide hydrochloric acid pretreatment to selectively remove NO 2 Ϫ must be questioned.
We used our modified triiodide assay to measure total red blood cell-bound NO, which negated the need for the harsh chemical pretreatment of samples. Instead, we lysed red blood cells prior to injection into the assay reagent. With this methodology, we found levels of total red blood cell-bound NO in venous blood samples from healthy individuals ranging from 0.000178 to 0.00024 mol nitric oxide/mol Hb (equating to ϳ445-600 nM in whole blood). These levels of total red blood cell-bound NO are significantly higher than values recently published by groups using the triiodide assay (10,23,27,38,39,41) to measure individual NO species.
In addition, we found significantly higher levels of total red blood cell-bound NO in blood from diabetic patients in comparison to healthy individuals (p Ͻ 0.0001). This is in agreement with previous findings demonstrating altered NO metabolism in diabetic individuals associated with the preferential binding of NO to Hb in conjunction with elevated glycosylation of Hb (HbA1c) (47). We have also shown the fate of vascular NO in poorly controlled diabetics (i.e. those with high HbA1c) to be related to elevated HbNO/SNO-Hb and correlated with abnormal blood vessel relaxation (34).
During the revision of this manuscript, another assay system has been developed to overcome the issue of heme NO autocapture in the reaction chamber. Utilizing ozone-based chemiluminescence and the cysteine/cuprous chloride reagent, Doctor et al. (48) have added carbon monoxide (CO) to the carrier gas. CO is suggested to out-compete NO for free heme due to the fact that it is in such excess. As long as CO does not influence the sensitivity of the nitric oxide analyzer by reacting with ozone or contain any impurities to create an artifactual signal, it would appear a clean method by which to inhibit heme NO auto-capture.
In conclusion, this work emphasized the need for future assays of Hb-bound NO to 1) incorporate a method to prevent free heme rebinding NO in the reaction chamber and to 2) avoid harsh chemical pretreatment of samples. This should limit the possibility of sample contamination, prevent preparative losses of unstable nitric oxide species (e.g. HbFe 3ϩ NO), and also ensure minimal disturbance of the biological sample avoiding disruption of the NO metabolite equilibrium. We hope that this work, while not only advancing our understanding of true NO metabolite levels in human blood, also leads to further technical advances in their assessment.