Effect of binding to hemoglobin and albumin on pyridoxal transport and metabolism.

Scatchard plot analysis indicated that pyridoxal binds to hemoglobin more than twice as tightly as it does to serum albumin. Comparison of the formation constants for hemoglobin and albumin, using standard competitive binding equations, indicated that the distribution ratio for pyridoxal between erythrocytes and plasma should be 6.5:1. This distribution was approximately the same as that observed when pyridoxal was incubated with whole human blood, suggesting that these two proteins are the primary determinants of the pyridoxal distribution in whole blood. With in situ perfused rat liver the uptake of [3H] pyridoxal from the perfusate was reduced by the inclusion of erythrocytes in the perfusate. This was reflected in the decreased production of 4-pyridoxic acid by the perfused liver from 3.8% to 1.2% of the dose by the addition of erythrocytes to the perfusate. The major labeled metabolites found in the liver were pyridoxal phosphate, pyridoxamine phosphate, and 4-pyridoxic acid for both types of perfusion. In intact animals, reduction of the erythrocytes concentrations to hematocrits of 30-40% increased the recovery in the urine of 3H from administered [3H] pyridoxal from control values of 27-35% to 40-50% of the dose within 48 h. Half of the label in urinary metabolites was in 4-pyridoxic acid.

phosphatase were purchased from Sigma. Pyridoxine phosphate was obtained from ICN Nutritional Biochemicals.
[3H]Pyridoxamine was prepared by the Amersham Corp. using an exchange procedure on pyridoxamine (free base). The synthesis and purification of [3H]PL starting with [3H]pyridoxamine were accomplished as previously described (7). The [3H]PL was more than 90% pure as judged by high voltage paper electrophoresis. Calbiochem-Behring was the source of polyethylene glycol. Spectrapor dialysis membrane tubing was purchased from Fisher Scientific Co. For high performance liquid chromatography, the reverse phase columns used were Rainin Instruments Microsorb C18 and Altex Ultrasphere ODS column. Guard column packing material, Pellicular C18, was obtained from Alltech Associates, Deerfield, IL.
Methods-Fresh human whole blood, collected over heparin, was centrifuged and the plasma and buffy layer were removed. The red cells were then mixed with isotonic saline and washed by centrifugation several times. The erythrocytes were lysed with an equal volume of distilled water and were kept cold for 30 min. The red cell membrane was removed by centrifugation at 15,000 X g for 25 min. The resulting supernatant hemoglobin solution was then concentrated using an Amicon filtration cell with a PM 10 filter until the desired concentration of hemoglobin was attained. The concentration of hemoglobin was determined using molar extinction coefficients for both hemoglobin and methemoglobin at two different wavelengths (8). The methemoglobin content of the freshly prepared oxyhemoglobin was always less than 2%. The dialysis experiments were done using 6-mm spectrapor dialysis tubing (M, cutoff = 14,000) and culture tubes (13 X 100 mm) with screw caps. One-ml solutions of hemoglobin were placed in the dialysis bag and dialyzed against 4 ml of isotonic saline containing various concentrations of pyridoxal and sufficient polyethylene glycol to osmotically balance the hemoglobin. The experiments were done at 4 "C in the dark, allowing 20 h for the system to reach equilibrium. The tubes were rocked throughout the dialysis. The experiments were done initially by adding the pyridoxal to both the inside and outside compartments of the dialysis bag to determine if the equilibration period was sufficient and to provide evidence that reversible binding of pyridoxal to protein was occurring. Following the incubation period, the dialysis bag was removed from the culture tube and the hemoglobin concentration again determined. The distribution of [3H]pyridoxal was then determined by taking 20pl aliquots of the solution both in and out of the bag. Forty pl of a 10% sodium tungstate solution, 40 pl of a 0.667 N sulfuric acid solution, and 100 rl of water were added to each 20-@1 aliquot. After thorough mixing and standing for 10 min, the sample was centrifuged to remove the protein and a 100-pl aliquot of the supernatant was counted.
The methodology of Hems et al. (9) was utilized for in situ liver perfusions with some modification which has been described previously (10). The experiment was started by replacing the recirculating perfusate with 60 ml of fresh perfusate containing [3H]pyridoxal. After 15 min this perfusate was removed and replaced with 70 ml of erythrocyte-free "wash" perfusate which was passed through the liver only once (7 min) to remove residual 13H]pyridoxal which had not yet left the circulation. Finally 60 ml of fresh perfusate were recirculated through the system for 15 min to collect 3H-metabolites which were released by the liver. At this time the liver was removed from the animal and boiled for 5 min. In the experiments where the perfusate contained erythrocytes, they were separated and treated with 3 volumes of water, 2 volumes of 10% sodium tungstate, and 2 volumes of 0.667 N sulfuric acid. The precipitated proteins were removed by centrifugation and the supernatant volume was reduced for high voltage electrophoretic separation of the Bs vitamers.
The high voltage electrophoretic methodology as well as the liver and perfusate treatment in preparation for high voltage electrophoretic analysis have been described (7). High performance liquid chromatographic separations were done to distinguish between pyridoxic acid and pyridoxine phosphate which are not separated by electrophoresis. A Beckman 112 solvent delivery module served as the pump for the reverse phase columns (4.6 X 250 mm) used in the high performance liquid chromatography system. The mobile phase consisted of a 33 mM potassium phosphate solution, pH 2.2, with 0.2 mM hexane sulfonic acid used as an ion-pairing agent.
The animals were fed Ralston-Purina rat chow No. 1104 throughout the experimental period. Some were made anemic by bleeding by heart puncture. The rats were then allowed 48 h to regenerate plasma proteins before they were injected intraperitoneally with 400 pl of [3H]PL. Urine was collected under toluene and analyzed for 3Hmetabolites using the high voltage electrophoretic technique described earlier. Hematocrits were determined using microhematocrit tubes and plasma protein levels were determined using the Lowry method (11).

RESULTS
Binding of Pyridoxal to Hemoglobin-Hemoglobin solutions of various concentrations were placed inside a dialysis bag for dialysis against a saline medium containing appropriate concentrations of PL and polyethylene glycol to provide the same osmolarity as the inside contents. Fifteen hours was found to be sufficient time for the system to reach equilibrium whether the PL was originally inside or outside of the dialysis bag, SO 20 h was used for all dialysis experiments.
The Scatchard plot of PL binding to hemoglobin is shown in Fig. 1, where the line was fitted by method of least squares. The slope of the line which corresponds to the formation constant (K,) for the PL-hemoglobin complex was 730 M-'. The x axis intercept indicated that there are about 2 PL binding sites/molecule of hemoglobin. This is in agreement with previous results (2) which indicated that PL binds to the NH2-terminal valine of the a-chain of hemoglobin, thus pro- The binding of PL to hemoglobin could play a role in the transport of the vitamin from the blood to various tissues. Compounds which are found in the blood may affect the binding of PL to hemoglobin and this may influence the transport process. Table I shows the results of dialysis exper-iments where various compounds were placed with PL outside the dialysis bag, while hemoglobin was inside. Although a more complete analysis of the effect of a given competitor on PL binding would require Scatchard plot construction, a general idea of competitor influence on PL binding can be obtained by examining the relative levels of PL accumulation in the dialysis bag. The presence of plasma or serum albumin outside the dialysis bag reduced the accumulation of [3H]PL in the bag substantially (Table I). Acetaldehyde at concentrations much higher than physiological levels decreased [3H]PL accumulation somewhat while PLP, glucose, and galactose reduced accumulation of [3H]PL only moderately. It should be pointed out that when the fraction of protein which has a ligand bound reaches low values, as when PL concentration was about 10 M m and hemoglobin concentration was greater than 1 mM, the concentration ratio of [3H]PL inside:outside the dialysis bag, when no competitor was present, was lower than would be predicted from the formation constant obtained from the Scatchard plot. This may be due to the presence of traces of impurities which compete with PL for the binding site on hemoglobin.
As a result of the effects found for serum albumin and plasma on [3H]PL accumulation in these dialysis experiments, the binding of PL to albumin was examined via the same dialysis system used for hemoglobin. A Scatchard plot of the data (Fig. 2) showed that the formation constant for PL binding to albumin was 340 M-'. The n axis intercept was 2.4 indicating that there may be two major binding sites, with participation by a weaker third site at higher PL concentrations.
Bovine serum albumin forms three types of complexes with PL phosphate (12). Two are formed by binding at unique sites with high formation constants while the third results from additional binding sites with lower affinity and not identified. All seem to bind by reaction with the t-amino group of lysine. PL accumulation in a dialysis bag containing albumin was examined in the presence of PLP. The accumulation of PL in the dialysis bag was reduced only when the concentration of PLP was 100-fold greater than PL indicating that these two vitamers are not competing for the same binding sites on albumin.
The relative binding affinities of PL for albumin and hemoglobin should determine the distribution of this Be vitamer in an equilibrium dialysis situation and could determine the distribution of [3H]pyridoxal in whole blood. Therefore, standard competitive binding equations (13,14) were used to predict what the distribution of 13H]PL would be in an equilibrium dialysis system where hemoglobin is inside a dialysis bag and albumin is outside the bag. The results of calculations using these binding equations are compared with experimental results in Table 11. The experimentally determined concentrations of PL inside and outside the dialysis bag agreed quite well with the values calculated from the binding equations. These equations were then used to obtain an expected distribution of PL in whole blood by substituting the appropriate physiological concentrations of hemoglobin and albumin into the equations. The predicted concentration ratio of PL inside:outside was 6.51. The distribution of t3H]PL found in whole blood has been experimentally determined by incubating [3H]PL with whole blood for 60 min followed by centrifugation and washing of packed red cells at 4 "C. The concentration ratio of [3HJPL in the red cell relative to outside was found to be 4 4 1 .
Additional experiments were done to investigate the possible influence of plasma protein fractions other than albumin on the distribution of PL in whole blood. Equilibrium dialysis was done as before using concentrations of the various globulin fractions which were about twice those of plasma. The results are presented in Table 111. None of the plasma protein fractions studied bound PL as evidenced by concentration ratios of 1.0 or less.
Liver Perfusions-The binding to hemoglobin inside the erythrocyte may modify the movement of P L between the blood and an organ or tissue. If this transport is influenced by binding then subsequent metabolism may also be influenced since the PL pool is small and a small change in this pool may affect the fate of the vitamin (15). Therefore, in situ liver perfusions were done to determine the uptake and metabolism of PL by the liver. When no erythrocytes were included in the perfusate, 19  "These calculations were done utilizing the following binding equations: System: 2 proteins and 1 ligand NIHb + PL $ HbPL NzA1+ PL $ AlPL Where Nl = binding sites/molecule of hemoglobin and N2 = binding sites/molecule of albumin.
where iPLl = moles of hemoglobin-pyridoxal complex/total moles of hemoglobin and VPk = moles of albumin-pyridoxal complex/total moles of albumin. From the Scatchard plot results: NI = 2; N2 = 2.4; Based on whole blood results with 47% hematocrit and 68.5% of packed cells representing intracellular volume and 15.8 g of hemoglobin/100 ml of whole blood.    Table IV. The metabolites of PL were determined in the initial 15-min perfusate, in the 7-min ''wash'' perfusate which followed, in the final 15-min "release" perfusate, and in the liver. These data show that even in the initial 15-min perfusate metabolites of PL had been released by the liver following uptake and metabolism of PL. The "wash" perfusate contained a smaller percentage of PL and a greater percentage of the other metabolites when compared to the initial 15-min perfusate. The final 15 min of the perfusion was done with fresh perfusate to study the distribution of 3H in compounds released by the liver following uptake of [3H]PL. As shown in Table IV, 49.9% and 14.8% of the label in this release perfusate was PL and 4-PA, respectively. The major 3H-labeled compounds found in the liver were PL, pyridoxamine phosphate, 4-PA, pyridoxine phosphate, and PLP representing After 15 min, the perfusate was removed and the perfusion system was washed (single pass) with 70 ml of fresh erythrocyte-free perfusate. Then a 15-min perfusion was done with 60 ml of fresh perfusate, designated release perfusate. The red blood cells in the wash perfusate are from the initial erythrocyte-containing perfusate. 18.4%, 26.7%, 11.8%, 16.9%, and 17.2%, respectively, of the isotope found in the liver. The distributions of 3H-labeled compounds in the red cells from the initial perfusate and the wash perfusate were similar to those of the media in which they were suspended. One noteworthy exception was PLP, where the percentage of 3H recovered in the erythrocytes was approximately twice that in the corresponding suspending medium. The distribution of 3H-metabolites found in the perfusions done without erythrocytes was very similar to those in perfusions with added erythrocytes. The major 3H-compounds found in the liver were PL, pyridoxamine phosphate, 4-PA, and PLP and they accounted for 22.1%, 24.2%, 23.5%, and 15.7%, respectively, of the total isotope found in the liver after 32 min. The percentage of the original perfused dose of PL (0.8 nmol) which was oxidized to 4-PA by the liver was calculated from the results for isotope accumulation by the liver and metabolite distribution data. Almost 4% of the total isotope perfused was converted to 4-PA by the liver when no red cells were present compared to 1.2% when the hematocrit of the perfusate was 23%.

Urinary Excretion of PL and Its Metabolites in Anemic
Rats-The results with perfused rat livers indicated that red cells may reduce the percentage uptake of PL by the liver from the portal circulation. The consequence of this in the whole animal may be increased delivery of PL to peripheral tissues when erythrocytes accumulate this compound. Conversely, more PL may enter the liver to be metabolized to compounds such as 4-PA in those animals with compromised ability to bind PL in the blood. To check this hypothesis, rats were made anemic by bleeding. They were then injected intraperitoneally with [3H]PL. The amount of 3H excreted in the urine in the ensuing 48 h was measured and compared to urinary excretion levels of rats with normal red cell numbers.
Eight rats were divided into two groups of four each. Group I was bled initially and 2 days later both groups were injected intraperitoneally with [3H]PL. Fourteen days after Group I had been bled, Group I1 was bled and 2 days after that both groups were again injected with t3H]PL. The urinary excretion results for this switchover experiment are shown in Table   V, Group I had hematocrits of about 33% for the first injection and excreted in the urine in 48 h approximately 48% of the ' H injected as PL (Table V). Group I1 had hematocrits which averaged approximately 50% for the initial injection and they excreted 27.2% of the injected dose. Group II was made anemic prior to the second injection. The hematocrits of Group 1 rats

Urinary excretion of 3H during the 48 h following intraperitoneal injections of CaH/PL into rats with normal a d reduced hemoglobin in switchover design
Averages were analyzed by Student's t test, **, significantly lower than * ( p < 0.04) and **** ( p < 0.04); +**, significantIy lower than **** (L) < 0.09) and * (L) < 0.06). acid accounted for about one-half of the isotope in the urine and PLP for another 20%. Plasma protein levels were determined in both groups before they were made anemic and at the time of urine collection when they were anemic. Both groups of rats regenerated plasma proteins to normal concentrations during the 5-day pre-experimental recovery period.

DISCUSSION
The equilibrium dialysis experiments clearly demonstrated a relationship between concentration of hemoglobin in the However, the competition of 2,3-diphosphoglycerate and other compounds for the @-chain site where PLP binds, reduces the concentration of PLP in the erythrocyte below that predicted from the formation constant. Competition with PL for the a-chain NH, terminus was investigated using compounds known or suspected to bind to this subunit (17). Glucose and galactose had little or no effect on the distribution of PL in the dialysis system. PLP or acetaldehyde caused a slightly lower in:out concentration ratio for PL than was predicted from the models, but the values were not lower than controls with the same low concentrations of hemoglobin. The results with glucose and PLP are in agreement with previous conclusions (2) based upon experiments with intact erythrocytes.
Plasma proteins may bind PL and thereby reduce the accumulation of PL in the red cells. The presence of albumin in the dialysate reduced the accumulation of PL in the hemoglobin compartment as predicted from the formation constants of each protein for PL. The calculated distribution of PL between the erythrocytes and plasma in whole blood, based upon the competitive binding equations of Steinhardt and Reynolds (13) using the experimentally determined binding constants, was 6.5:l. The ratio determined in whole human blood varied from 4 to 5 in several experiments. This deviation from the calculated value cannot be explained by the effect of other plasma proteins, since albumin alone gave evidence of PL binding in separate equilibrium dialysis experiments. Two other explanations can be offered for the reduced concentration ratio for whole blood. The washing of the pelleted red cells with isotonic saline at 4 "C as was done in these studies may remove some of the intracellular PL. The temperature sensitivity (1) of the transport of PL through erythrocyte membranes would suggest that this is not a major cause for leakage. The second factor might be the water of solvation. Not all of the water space in cells is available for equilibration of solutes (18). When the concentration of PL in the erythrocyte is expressed as a function of the volume of intracellular water of solvation, the concentration ratio for whole blood would be corrected upward by a factor of 1.3. This correction would bring the ratio to a value in good agreement with the ratio generated from the binding model. Therefore, these data suggest that hemoglobin and plasma albumin are competing binders of PL and that these proteins alone determine the distribution of PL between red cells and plasma.
The liver perfusion experiments demonstrated that erythrocytes in the perfusate reduce the uptake of PL by the liver.
This was presumably caused by the binding to hemoglobin inside the erythrocytes, thus effectively reducing the concentration of P L in the plasma. The major change in the pattern of labeled metabolites in the liver caused by the lack of erythrocytes was the increse in 4-PA from 1.2 to 4% of the dose. The liver has the capacity to oxidize PL to 4-PA via a nonspecific aldehyde oxidase and is the organ most responsible for this metabolic inactivation of vitamin B6. The presence of erythrocytes in the perfusate may be reducing the PL available for metabolic inactivation in the liver.
The observations recorded in Table V are consistent with the vitamin B6 conserving effect of erythrocytes. More severe anemia might cause a greater elevation of urinary excretion of 4-PA, but added complications might also be expected. The size of the dose of PL injected (0.4 mg) in these studies is nearly three times the daily requirement for rats of the body weight used (19). At a dose of pyridoxine approximating the daily requirement, Contractor and Shane (20) reported that pregnant rats excreted 43% in the urine about one-half as 4-PA. At high dosage pyridoxine is excreted as pyridoxine, PL, and 4-PA, and PL as 4-PA (21) by the human. Likewise at a dose of 1.5 mg of pyridoxine 50% of intake was excreted as 4-PA (22). The 4-PA-forming enzymes aldehyde dehydrogenase and aldehyde oxidase are widely distributed in animal tissues including erythrocytes (23) so an increase in the freely diffusible pool of PL in the blood should result in increased metabolism to 4-PA. This mechanism would lead to the incresed 4-PA excretion observed in those rats with reduced erythrocyte volumes.