Hemoglobin Oxygen Affinity in Patients with Cystic Fibrosis

In patients with cystic fibrosis lung damages cause arterial hypoxia. As a typical compensatory reaction one might expect changes in oxygen affinity of hemoglobin. Therefore position (standard half saturation pressure P50st) and slope (Hill’s n) of the O2 dissociation curve as well as the Bohr coefficients (BC) for CO2 and lactic acid were determined in blood of 14 adult patients (8 males, 6 females) and 14 healthy controls (6 males, 8 females). While Hill’s n amounted to approximately 2.6 in all subjects, P50st was slightly increased by 1mmHg in both patient groups (controls male 26.7±0.2, controls female 27.0±0.1, patients male 27.7±0.5, patients female 28.0±0.3 mmHg; mean and standard error, overall p<0.01). Main cause was a rise of 1–2 µmol/g hemoglobin in erythrocytic 2,3-biphosphoglycerate concentration. One patient only, clearly identified as an outlier and with the mutation G551D, showed a reduction of both P50st (24.5 mmHg) and [2,3-biphosphoglycerate] (9.8 µmol/g hemoglobin). There were no differences in BCCO2, but small sex differences in the BC for lactic acid in the controls which were not detectable in the patients. Causes for the right shift of the O2 dissociation curve might be hypoxic stimulation of erythrocytic glycolysis and an increased red cell turnover both causing increased [2,3-biphosphoglycerate]. However, for situations with additional hypercapnia as observed in exercising patients a left shift seems to be a more favourable adaptation in cystic fibrosis. Additionally when in vivo PO2 values were corrected to the standard conditions they mostly lay left of the in vitro O2 dissociation curve in both patients and controls. This hints to unknown fugitive factors influencing oxygen affinity.


Introduction
Cystic fibrosis (CF) is the most frequent genetic disease in Caucasians [1][2][3]. Mutations on chromosome 7 (location 7q31.2) reduce the effectiveness of the cystic fibrosis transmembrane conductance regulator (CFTR), which is essential for the secretion of chloride (Cl 2 ) and consequently water in many glands. The clinical manifestation with heaviest impact is the progressive pulmonary disease. Because of the resulting deteriorated lung function in patients with cystic fibrosis causing hypoxia and partly also hypercapnia one might expect compensatory reactions in concentration and oxygen affinity of hemoglobin (Hb) to secure oxygen loading in spite of the reduced oxygen pressure (PO 2 ) in pulmonary capillaries.
There are various strategies of defense against arterial hypoxia [4][5][6][7][8]. In addition to hyperventilation most healthy humans, except partly Tibetans and Ethiopeans [9][10][11], react to hypoxia with an increase in Hb concentration ([Hb]) which facilitates sufficient binding of oxygen at lowered PO 2 in the lungs. Furthermore a right shift of the oxygen dissociation curve (ODC) under standard conditions (pH 7.4, PCO 2 40 mmHg, 37uC) partly compensates for the reduced diffusion pressure in the tissues because of the low oxygen saturation (SO 2 ) in the capillaries; the shift is caused by more 2,3-biphosphoglyrate (BPG) in the red cells. In contrast typical altitude animals like llamas, guinea pigs and partly birds possess left-shifted ODCs securing oxygen loading in the lungs and rather low [Hb] reducing circulatory resistance. In addition small red blood cells and a dense capillary net in the tissues diminish the diffusion distance and thus compensate for the decreased capillary PO 2 [12]. The human fetus exists also at very low arterial PO 2 but the concentration of the high affinity fetal Hb (HbF) is increased. Recent in vivo determinations of the ODC in adults point to a possible left shift of its upper part at altitude [13,14].
Astonishingly few studies investigated the combined effect of hypoxia and hypercapnia. The fetal conditions with higher arterial PCO 2 than in maternal blood point to an advantage of a left shifted ODC. Moles living in earth holes with reduced air exchange inspire hypoxic/hypercapnic gas and possess also Hb with increased oxygen affinity [15]. Huckauf et al ( [16]) describe a left shifted ODC in patients with chronic obstructive lung disease and in a review Morgan [17] mentions that [BPG] is often reduced in critically ill patients.
Patients with cystic fibrosis often show normal or even anemic [Hb] (e. g. [18][19][20]. Interestingly, however, they may possess an increased red cell volume masked by a concomitant rise in plasma volume [21,22]. Compensatory reactions of oxygen affinity in cystic fibrosis have been investigated rarely. Slight right shifts of the standard ODC, characterized by a rise of P 50 st and caused by increased [BPG], were detected by some authors [18,23], while others found unchanged [BPG] or P 50 st [19,24]. However, there are various additional mechanisms for the regulation of oxygen affinity. Besides phosphates other anions like lactate, chloride and glutathione (e. g. [25,26]) bind to Hb. Depending on the binding site these substances also influence the cooperativity of the subunits visible as change in the slope of the ODC (Hill's n); additionally they may modify the intraerythrocytic pH. The Bohr effect, which in the physiological pH range causes an increase of PO 2 at constant saturation by acidification (essential in working muscles), may vary depending on various factors like oxygen saturation [27,28], type of acid [27,28], substance concentrations and age of the erythrocytes (e. g. [29]); in altitude residents a tendency to lowered Bohr coefficients (BC = DlogPO 2 / DpH) has been observed [30,31]. Also sex differences in oxygen binding properties have been described: women [32][33][34] as well as children [35] tend to higher P 50 st than men. Finally in vivo variations of the ODC in venous blood of anemic patients as well as of trained subjects have been observed during exercise which were no more detectable after in vitro equilibration of blood [36][37][38]. The underlying mechanisms are not yet clarified.
Previous studies on oxygen affinity in cystic fibrosis were performed on rather heterogeneous groups of patients. Differences in the severity of the illness are almost inevitable but possible effects of age and sex were not considered. Also control groups were small or not clearly defined or even lacking. To our knowledge neither cooperativity (Hill's n) nor the Bohr effect have ever been studied in cystic fibrosis.
Considering all these factors it seemed worthwhile to perform a systematic study of mechanisms influencing blood O 2 affinity as possible facilitation of oxygen uptake in cystic fibrosis.

Study Participants
Measurements were performed in 14 adult patients and 14 controls; anthropometric data are presented in Table 1. The patients (8 males, 6 females) showed severely reduced lung function but were in a stable clinical condition. One male subject was bearer of the Class III G551D mutation which is one of five mutations with a frequency .0.1% accounting for 2 to 3% mutations worldwide. It impairs CFTR-mediated Cl 2 transport by limiting channel gating at the cell surface [3,39].
The patients were the members of a group with exercise therapy. They usually lived at home but were under continuous supervision by physicians of the pediatric clinic of the faculty. Twice a week they performed a disease status tailored exercise program addressing endurance, strength, coordination and flexibility supervised by staff of the Institute of Sports Medicine and received individual advice for additional daily exercises at home. Occasionally some patients used short term oxygen supplementation, but not on the test day. The nonsmoking controls (6 males, 8 females) were physically active but not specifically or regularly training staff members and students. One female was slightly anemic ([Hb] 11.2 g/dl), but all other measurements yielded clinically normal values within the range of the group. The study protocol was approved by the ethics committee of the faculty (Ethikkommission, Charité -Universitä tsmedizin Berlin, Ethikausschuss CBF, No. ek.185-13b) and written informed consent was obtained from all participants.

Study Procedure
The subjects arrived at the laboratory between 9.00 and 10.00 a.m. Lung function (forced vital capacity FVC, forced expiratory Table 1. Subjects. Means and standard errors (SE). BMI body mass index, FVC forced vital capacity, FEV1 forced expiratory volume during 1 s; PEF peak expiratory flow. % of expected values for age and sex [40] or of individual FVC. All anthropometrical and lung function values are significantly (P,0.001) reduced in the patients. doi:10.1371/journal.pone.0097932.t001 volume during 1 s FEV1, peak expiratory flow PEF) was measured with a spirometer system (Oxycon gamma, Mijnhardt, Bunnik, The Netherlands). Percent of expected values for age and sex [40] or of individual FVC are presented in Table 1. Blood was sampled in supine position. Acid base status at 37uC (ABL 500 or 510 with no systematic difference between apparatus, Radiometer Copenhagen, Denmark), oxygenation status (PO 2 , SO 2 , COHb, MetHb) and [Hb] (OSM 3; Radiometer Copenhagen, Denmark) were measured in heparinized blood samples taken from hyperemized ear-lobes. Values for PO 2 are slightly lower than in arterial blood [41], but this is of negligible importance for saturations above 90% in the flat part of the ODC. Fifty ml of venous blood were drawn without stasis using heparinized vaccutainers and stored in an icewater mixture. Oxygenation status, [Hb], hematocrit (Hct, microhematocrit method) and [Cl 2 ] in plasma (EML 100, Radiometer Copenhagen) were determined immediately. Aliquots were deproteinized and stored at 220uC for duplicate measurements of ATP and BPG concentrations (enzymatic kits, Sigma Diagnostics) on the next day. Five ml each were equilibrated 20 min in sphere tonometers at 37uC with air/CO 2 or nitrogen/CO 2 mixtures (3, 6 or 10% CO 2 ). Lactic acid (13.5 mmol/l blood) was added to an additional sample equilibrated thereafter with 6% CO 2 in air or N 2. After taking aliquots for additional ATP and BPG measurements 0.2 ml of oxygenated blood were successively added 8 to 10 times to 1 ml deoxygenated blood using 2 connected syringes and mixed. After measurement of SO 2 , COHb, MetHb, [Hb], pH, PCO 2 and PO 2 , ODCs were drawn in the Hill plot (log SO 2 /100-SO 2 ) versus log PO 2 ).
Samples of native blood as well as of blood equilibrated with N 2 /6% CO 2 and with air/6% CO 2 were centrifuged for 10 min (3500 rpm, 4uC). Part of the red cell sediment was hemolyzed by repeated freezing and thawing and used for measurement of pH and [Cl 2 ] in the erythrocytes.
Twelve patients (7 males, 5 females) performed an incremental test (initially 0.3 W/kg, plus 0.3 W/kg every 2 min) until exhaustion on a cycle ergometer (Lode Excalibur, The Netherlands) during exercise therapy. Blood gases and lactate concentration (Ebio plus, Eppendorf, Germany) were measured in ear lobe blood and used to calculate P 50 at exhaustion.

Calculations
The slope n of the oxygen dissociation curves linearized in the Hill plot served as measure of cooperativity. For 5% steps of SO 2 between 15 and 90% logPO 2 values were calculated from the regression equations and the corresponding pH values obtained by interpolation. Comparison of the ODCs for 3 and 10% CO 2 yielded Bohr coefficients for CO 2 (BCCO 2 ), comparison of the 6% CO 2 and the 13.5 mmol/l lactic acid curves yielded Bohr coefficients for fixed acid (BCLa) at each saturation step. P 50 st were calculated from the curves of blood equilibrated with 6% CO 2 by use of the corresponding individual BCCO 2 . Mean cellular hemoglobin concentrations (MCHC) calculated from [Hb] and Hct were corrected for 2% trapped plasma. [Cl 2 ] ery were corrected for 10% in the sediment after centrifugation with 3500 rpm; because of the large buffer capacity of red cells this is not necessary for pH ery . Electrodes in the electrolyte analyser measure concentrations in water [42]; therefore [Cl 2 ] ery is given per l cell water. Values for the control subjects coincide with titrimetric measurements [43]. In vitro blood buffer capacities (2 D[acid]/DpH) for CO 2 and lactic acid were calculated from the measurements in the corresponding equilibrated samples.

Statistics
All data are presented as means6standard errors (SE). Dependent on the number of comparisons, t-tests or analysis of variance (ANOVA) were used for significance calculations. The probability that an outlier does not belong to a sample was tested eventually [44]. Differences with P,0.05 were considered as significant.

Results
Anthropometry and Pathology Table 1 shows marked reduction in both body height and body mass in the patients compared to healthy subjects. Their lung function was substantially impeded by restrictive as well as obstructive damage visible from low vital capacity and expiratory flow (FEV1, PEV); FEV1 ranged between 22 and 74%.

Blood Gases and Acid Base Status
The impaired lung function of the patients caused a reduction of ear-lobe PO 2 and SO 2 ( Table 2). Generally these values were also slightly lower in males than in females. Correspondingly, PCO 2 tended to higher values in males and in patients. However, the pH was equal in all subgroups because of non-respiratory compensation visible as increased base excess in males and in patients. Venous blood pH scattered more, but there were also no systematic differences among groups (means between 7.35 and 7.38); red cell pH showed no influence of sex or illness as well (means about 7.16). In vitro buffer capacities of blood tended to higher values in all males; this was significant for acidification with CO 2 as well as lactic acid (both P,0.05; latter not shown in

Blood Composition
[Hb] and Hct were higher in males than females, but there were no significant differences between controls and patients (

Oxygen Dissociation Curves
In the Hill plot ( Fig. 1) all curves were linear (correlation coefficients better than 0.98, not corrected for the slightly decreasing pH with rising saturation) and the slopes amounted to approximately 2.6 with very little scattering in all groups ( Table 4).
The standard half saturation pressures (Table 4) corresponded to known normal values in the controls. In patients P 50 st was significantly increased by 1 mmHg (with slightly but not significantly higher values for females). When corrected to arterialized pH and PCO 2, all means were 0.8 mmHg lower. The patient with the G551D mutation presented a markedly lowered P 50 st of 24.5 mmHg (arterialized blood 23.3 mmHg) clearly identified as  (Fig. 2). The male patient with the extremely low P 50 st value fell, however, far outside of his group with a correspondingly low [BPG].

Bohr Coefficients
The Bohr coefficients (Fig. 3) for CO 2 corresponded to published data: The value was about 20.5 and decreased numerically with higher saturations (P,0.01 for all subjects); there was also a tendency to lower values in women. No influence of the disease was visible. The Bohr coefficients for lactic acid (Fig. 4) were generally lower numerically than for CO 2 in all groups up to 45% saturation (20.40 to 20.45). Differences between males and females at higher SO 2 disappeared in the patients (interaction sex-illness P,0.01). Among the patients the subject with the G551D mutation presented the highest BC for both acids between 70 and 90% SO 2 (approx. 20.54).

Exercise Tests in Patients
At exhaustion SO 2 dropped in all patients resulting from reduced PO 2 and both respiratory and non-respiratory acidosis which caused a rise of P 50 (Table 5). Again the subject with the G551D mutation showed the lowest P 50 value (29.6 mmHg).

In vivo Effects
When the PO 2 values in non-equilibrated venous blood (fresh or stored in ice until measurement) were corrected with the corresponding Bohr coefficients (BCCO 2 ) to pH 7.4, they should have fallen on the individual standard ODC. However, in the range between 45 and 90% SO 2 there was a tendency for a deviation to the left (Fig. 5) in controls (21.860.4 mmHg, P, 0.05) as well as in patients (22.260.4 mmHg, P,0.001). Some samples with higher values of SO 2 were not considered, because the BCs were not measured for SO 2 .90%. In addition there is large scattering of PO 2 in the flat part of the ODC. There was no correlation between PO 2 differences and [BPG] differences for native and equilibrated blood.

Synopsis of Results
Our results confirm former investigations that there is a small right shift of the standard ODC in most patients with cystic fibrosis probably caused by slightly increased intraerythrocytic concentrations of organic phosphates [18,23]. This is accompanied by a constant slope of the ODC and only small changes of the Bohr coefficients. In spite of the lacking hypocapnia this reaction is similar to the typical human acclimatization to altitude but seems to be attenuated. During exercise the right shift of the ODC is enforced by hypercapnia in CF patients and a clear drawback for arterial oxygen loading. Interestingly there seems to exist an additional mechanism in controls as well as in patients: The in vivo standard ODC falls slightly left of the in vitro curve.

Blood Gases and Acid Base Status
The deterioration of lung function in the patients results in hypoxia visible in arterialized blood; a tendency to a slightly higher PCO 2 than in the controls is not significant probably because of the low number of measurements and the resting situation; when exercising the increase in PCO 2 is more marked. Measurements in 69 patients in our laboratory showed corresponding results; PaCO 2 increased with the severity of the illness at rest as well as during exercise [45]. This is different to healthy subjects who always show a decrease of PaCO 2 at high work load. The arterial oxygen saturation in patients at rest is as low as in highlanders [46,47] living 2600 m above sea level (inspiratory PO 2 approx. 120 mmHg). But in spite of a similar reduction of spirometric values the female patients show higher PO 2 and lower PCO 2 than the male patients like their healthy counterparts. Probably the long-known stimulation of respiratory brain centres by female  hormones important for fetal oxygen supply is the cause (reviewed in [47]). The fact that arterialized pH is equal in all subgroups in spite of differences in PCO 2 demonstrates the importance of acidbase homeostasis for physiological functions. Non-respiratory compensation is mainly done by renal excretion/reabsorption of bicarbonate. In the patients the osmotic effect of the rise of [HCO 3 2 ] is counteracted by a decrease of [Cl 2 ]. Also the loss of chloride via sweat glands might play a role. The slightly increased in vitro buffer capacity in both male groups is obviously caused by the higher Hb concentration. In cystic fibrosis the slight rise in bicarbonate concentration as well as the possibly increased Hb mass [21,22] help to attenuate the extracellular pH changes during exercise [48] caused by CO 2 retention.

Blood Composition
Hb concentrations showed typical sex differences but no sign of anemia in the patients. The latter might be expected in CF because of frequent problems with iron resorption. However, in  our patients iron metabolism was routinely checked and deficiency was treated. One explanation for normal [Hb] in other studies might be the counteracting effects between iron deficiency and hypoxia [49]. Christoforou et al. [19] described a negative correlation of [erythropoietin] with FVC and FEV1. Such a dependency is probably the cause of the correlation between [Hb] and FEV1 in our male patients. Some authors [21,22] have even observed an increase in red cell volume in CF probably stimulated by erythropoietin which might be explained as a typical hypoxia reaction. However, only in a fraction of the corresponding studies [19,20,50,51]http://www.ncbi.nlm.nih.gov/ pubmed?term = Mc-Colley%20SA%5BAuthor%5D&cauthor = true&cauthor_uid = 21365780 erythropoietin concentration was increased. Own unpublished measurements support the idea of chronic stimulation of erythropoesis in CF patients based on elevated erythropoietin as well as soluble transferrin receptor concentrations in a cohort of 79 CF patients. Also a low MCHC like in the patients is often related to an increased water content typical for young erythrocytes. Furthermore in patients the high level of the soluble transferrin receptor [20] might be indicative for an increased red cell production and thus a reduced erythrocytic age. However, also a link between CFTR and the function of the hypoxia inducible factor has been put forward [52] which may serve as one potential reason for a lack of increased [Hb] in CF patients.
Factors possibly increasing [BPG] and [ATP] are low SO 2 (reducing product inhibition because of BPG binding to Hb) and alkalosis (stimulating glycolysis and thus BPG synthesis). A probable explanation for the rather small increase of [BPG] and P 50 st in CF compared to highlanders with similarly lowered arterial SO 2 and equal pH at rest (e. g. 18 mmol/g Hb in [30]) might be the different effect of physical activity: CO 2 retention causes respiratory acidosis already during moderate physical activity in the patients while highlanders effectively hyperventilate at each exercise level. In the present patient group with normal daily life and exercise therapy physical activity was obviously a factor of some importance. Additionally a low red cell age as suggested above might lead to elevated [BPG] as well as [ATP] because of high enzymatic activity [29]. The low [BPG] in the patient with the G551D mutation possibly results from changed enzyme activities because no differences in erythrocyte physiology were detectable. CFTR is incorporated into the red cell membrane (e. g. [53]), but a relation to BPG metabolism remains speculative.
[Cl 2 ] in red cells in part follows passively changes in plasma [Cl 2 ] and therefore is lowered in patients. Generally the marked concentration difference results from the high erythrocytic content of non-diffusible anions (Hb 2 and organic phosphates) causing a Donnan equilibrium. Cl 2 crosses the cell membrane mainly through band 3 channels. The reduction of the number of CFTR molecules in patients (e. g. [53]) does not affect this exchange [24]. Cl 2 concurs with BPG for the same binding sites on Hb [54] but its affinity is lower and the small decrease of its concentration in CF is compensated for by increased [BPG].

Oxygen Dissociation Curves
The P 50 st values of controls scatter around the normal mean value (approximately 27 mmHg) without significant sex differences. The generally higher P 50 st in patients results from the increased [BPG] (change approx. 0.6 mmHg per mmole BPG/gHb according to [55]) while ATP plays only a minor role because of complexing with Mg ++ . This corresponds to the typical chronic hypoxic reaction of most humans. It allows to extract more oxygen in the tissue capillaries without lowering the diffusion pressure, but it is not helpful for oxygen loading in pulmonary capillaries. In highlanders with similar reduction of arterial SO 2 P 50 st scatters around 30 mmHg [30]. In both healthy subjects and patients the reaction (affinity change, increased ventilation and partly stimulated erythropoesis) is a sufficient compensation of moderate hypoxia at rest but maximal performance capacity is reduced. The left shift of the ODC in moles [15] living and working under comparable conditions (inspiring air with reduced O 2 and increased CO 2 content) as the patients is more reasonable but is rare in humans. Under extreme acute conditions (above 6000 m of altitude) healthy mountaineers lower their P 50 by extreme hyperventilation [56] which is not a sustainable option for CF- patients. For chronic acclimatization a reduction of [BPG] would be more appropriate. Surprisingly the male patient with the G551D mutation showed such an effect. One can estimate that a reduction of P 50 st like in his blood would raise the arterial SO 2 at exhaustion in the male patients by 4%.
Interestingly in the 3 papers with P 50 st measurements in CF patients single low P 50 st values between 23.5 and 25.5 mmHg can be found [18,19,23]. This points to a special form of hypoxia acclimatization in some patients similar to that in moles.
The magnitude of a change in P 50 st may reflect further compensating mechanisms. Rosenthal et al [18] showed that P 50 st is negatively correlated with systemic oxygen delivery which depends on arterial SO 2 , [Hb] and cardiac output. This means that low [Hb] or cardiac output favor a rise of P 50 st. Indeed Arturson [57] described a P 50 st increase with falling [Hb] in chronic pulmonary insufficiency. The present CF patients were not anemic. This might also explain why we observed a small tendency rather than a substantial change in P 50 st.
Hill's n did not deviate much from the usually expected value of 2.7 for HbA (e.g. [55]). BPG binds to Deoxy-Hb only which may therefore increase Hill's n with rising concentration. A sligtly higher n in highlanders [30] and anemic patients [58] might be explained by this mechanism. However, the [BPG] differences between controls and patients in this study are too small to cause measurable effects.

Bohr Effect and Exercise
Similar like in altitude inhabitants [30,31] the Bohr coefficients are little changed in patients with cystic fibrosis. The coefficients for CO 2 correspond to published values [27,28,33]. They are large (numerically) at low saturation because of oxygenation dependent binding of carbamate in addition to H + effects during acidification with CO 2 . They are lower at very high saturation; the Bohr effect disappears when all Hb molecules are in the R (relaxed) state. Anions (Cl 2 , La 2 and BPG) compete with CO 2 at the terminal valines (e. g. [25]). Therefore BCCO 2 increases at low [BPG]; this might be the cause for the rather high value in the patient with mutation G 551D. The fixed acid Bohr coefficients are small especially at low saturation compared to BCCO 2 . The slight unexplained influence of CF on BCLa plays only a very modest role, because the peak lactic acid concentration during exercise is rather low in the patients (Table 5) compared to healthy subjects (e. g. [59]). In the lungs the Bohr effect of CO 2 is helpful for oxygen loading, when CO 2 leaves the blood, especially during hyperventilation with resulting hypocapnia during heavy exercise.
In the patients with exercise hypercapnia and mostly high P 50 st, however, an increase of BCCO 2 would be detrimental in this situation.

In vivo Effects
The left shift of the in vivo PO 2 /SO 2 pairs relative to the in vitro standard ODCs is on an average modest (approx. 2 mmHg) but 17 differences amount to more than 4 mmHg. Differences in ODCs as well as BCs between fresh blood immediately after sampling and blood after equilibration in tonometers have occasionally been observed (e. g. [38,60]). Concentration changes of BPG, ATP, Cl 2 , nitrocompounds or glutathione are possible causes. The means of [BPG] and [ATP] increase slightly but not significantly after equilibration compared to fresh venous samples explaining only 0.7 mmHg of the difference at 50% SO 2 . Intraerythrocytic [Cl 2 ] changes are larger for a given DpH in vivo than in vitro resulting from exchange with the interstitial fluid [59]. Because of the opposite effects of SNO-Hb and Hb[FENO] on oxygen affinity [61] NO usually exerts no measurable influence on the ODC neither in vitro nor in vivo if no methemoglobin is formed [62][63][64][65]. In our experiments MetHb was stable. For an allosteric effect intraerythrocytic [NO] is by far too low even in Tibetans who present very high values [66]. In contrast to NO glutathione is present in millimolar concentrations in the red cells [67] and binds to oxy-Hb thus shifting the curve to the left and reducing the Bohr effect [26,68]. But a marked deficiency of extra-and intracellular glutathione possibly including erythrocytes in CF patients has been suggested [69] which might be related to the disturbed function of CFTR as glutathione transporter [52]. Very recently also an effect of glutamate on P50 was observed [70]; this substance binds to Ca ++ channel proteins in the cell membrane and may be interchanged with muscle fibres. Thus at the moment a clear cause for the in vivo -in vitro difference of PO 2 remains unknown, but apparently it rises with SO 2 . This produces a left shift of the in vivo ODC between 50 and 90% SO 2, which is an advantage for oxygen loading. Recently similar results were found at 3600 m of altitude [13,14]. Interestingly at low saturations ''standardized'' in vivo PO 2 /SO 2 pairs tend to lie right of the in vitro curve, especially in venous blood returning from exercising muscles [36][37][38]55,60]. The result of this opposite changes is a markedly steepened complete in vivo oxygen dissociation curve probably in healthy subjects as well as in CF patients. Such a property is favorable for both loading and unloading of O 2 in lungs and consuming tissues.

Conclusions
The majority of the patients with cystic fibrosis in our study react to the problem of pulmonary oxygen uptake like man at altitude with a small right shift of the in vitro ODC caused by increased organic phosphate concentrations in the red cells. This improves oxygen diffusion into the consuming tissues, but is a drawback for arterialization. Healthy subjects can compensate this by hyperventilation thus reducing arterial PCO 2 with resulting left shift of the ODC during oxygenation. This is not possible for CF patients especially when CO 2 production is increased during exercise. A probably more appropriate left shift by reduction of [BPG] was observed in one patient with the G551D mutation. Also in other papers occasional left shifts can be detected. Whether this is a genetic effect, remains an intriguing question. The slope of the in vitro ODC and the Bohr coefficients were not markedly affected by the disease. Under in vivo conditions, there is a tendency for a left shift of the upper part of the ODC in both healthy controls and patients pointing to unknown affinity modifying factors which improve oxygen loading in the lungs.