Main

The sheep is a common animal model for investigation of fetal and newborn cardiovascular, pulmonary, and hematologic development and response to transfusion therapy(1–6). Because recombinant human erythropoietin is being evaluated as therapy for anemia of prematurity, the mechanisms and regulation of erythropoiesis and red cell kinetics in the ovine fetus are areas of intense study(7–15). Because RCV and RCS are important measures of erythropoiesis and red cell kinetics, a practical and accurate method for determination of both RCV and RCS in fetal and newborn lambs and in ewes would be useful. However,51 Cr is not suitable for determining RCS in sheep because the label leaves the red cell too rapidly(16).

In the course of investigating [14C]cyanate as a permanent labeling agent for the determination of RCV and RCS in sheep, we observed that the blood concentration of 14C covalently bound to intracellular protein increased after infusion of the labeled red cells. This study describes the development and validation of the labeling and assay methods, observation of an in vitro increase sufficient to introduce important artifacts in vivo, investigation of the mechanism of the increase, and suggestions to avoid artifacts.

METHODS

Chemicals. Chemicals and their sources were: potassium cyanate,14 C (DuPont NEN, Boston, MA); KH2PO4 (monobasic), K2HPO4 (dibasic), NaCl, glucose (Sigma Chemical Co., St. Louis, MO); 3a70B liquid scintillation cocktail (Research Products International Corp., Mount Prospect, IL).

Supplies. Supplies and their sources were: Nalgene disposable filter ware, 0.2 μM (Nalge Company, Rochester, NY); Millex-GV syringe filter, 0.22 μM (Millipore, Bedford, MA); 50-mL conical tubes (Corning, Ithaca, NY); Econo-Pac 10DG chromatography columns (Bio-Rad, Hercules, CA).

Buffers. Buffers were: PBS-glucose, pH 7.4: 6 mM potassium phosphate, 0.155 M sodium chloride, 22 mM glucose. CPDA-1(citrate-phosphate-dextrose-adenine): 2 mM adenine, 89.6 mM sodium citrate, 15.6 mM citric acid, 160 mM dextrose, 16 mM sodium phosphate (monobasic).

Labeling of red cells. Labeling of sheep red cells with[14C]cyanate was accomplished according to the method of Eschbach(17) with modifications as follows. To increase labeling efficiency, sheep red cells were washed before labeling instead of simple packing for plasma removal. All labeling studies were done using blood obtained by venepuncture on the same day as the study. Blood was anticoagulated with heparin (10 U/mL of blood) under sterile conditions. After transfer into a sterile 50-mL conical tube, red cells were sedimented by centrifugation for 15 min at 1500 × g in a swinging bucket rotor at 15°C. The plasma was removed by vacuum aspiration and saved. The cells were washed three times by resuspension at a volume ratio of 4 mL of PBS-glucose to 1 mL of packed red cells and sedimentation. [14C]Cyanate[370 kBq/mL (10 μCi/mL) of packed red cells] was mixed with the washed red cells and incubated for 1 h at 37°C in a shaking water bath. The red cells were sedimented as above; and the supernatant was removed and discarded. The red cells were then washed four times as described above. The final supernatant was removed, and the volume was adjusted to a hematocrit approximately equal to the original blood by addition of the saved plasma.

Sample preparation for assay. The objectives of assays for the14 C label are to measure the radioactivity covalently bound to intracellular protein; Hb is the primary intracellular protein in red cells and the principal protein carbamylated(18). We developed two new assays for 14C-labeled protein which share a common dilution scheme as follows. Labeled blood and blood samples from the simulation of in vivo circulation described below were assayed in quadruplicate as described below. Volumes of 0.05 and 0.1 mL of labeled blood or 0.5 mL of blood samples from simulation studies were weighed to four significant digits on an electronic balance in 20 mL of glass scintillation vials. Sample volumes less than 0.5 mL were brought to 0.5 mL by adding unlabeled sheep blood. Hematocrits of all blood samples were determined by a microcapillary method(1).

Acid/acetone assay. This assay separates the heme from the globin (and other proteins) using acidification and acetone before scintillation counting. This acid/acetone method detects only 14C covalently bound to proteins large enough to precipitate in acid/acetone.

Blood was pipetted into 20-mL scintillation vials and disrupted by addition of 1 mL of 0.5 N HCl with swirling. Acetone (15 mL) was added, and the mixture was shaken for 30 s. The flocculate (precipitated globin) was sedimented by centrifugation at 750 × g for 10 min. The acetone was decanted and discarded; the pellet was disaggregated by mechanical agitation and resolubilized by addition of 1 mL water with vigorous shaking. Proteins were again precipitated, sedimented, and resolubilized in water. Ten milliliters of liquid scintillation fluid were added. Bound 14C was determined using liquid scintillation counting with quench correction.

Peroxide assay. This assay uses the same general principle as the benzoyl peroxide assay of Landaw(19) but somewhat different reagents. In the assay described here, the heme in Hb is bleached using hydrogen peroxide to oxidize the ferrous iron. Free [14C]cyanate is hydrolyzed to yield 14CO2 which escapes to the atmosphere. The remaining 14C bound is quantitated by scintillation counting.

Red cells were disrupted by adding 1 mL of 10% sodium lauryl sulfate to each vial and swirling. Vials were placed in a shaking 37°C water bath in a fume hood, and 30% hydrogen peroxide (H2O2) was added stepwise in 50-μL increments until the red of the heme samples was bleached to a light yellow color. This procedure should be done in a fume hood because some14 C may be volatilized as 14CO2 (see“Results”). Residual H2O2 was reduced by stepwise addition of β-mercaptoethanol in 50-μL aliquots until the solution did not bubble vigorously with each addition. Ten milliliters of liquid scintillation cocktail were added, and bound 14C was determined using liquid scintillation with quench correction.

Radiation release. To assess inadvertent volatilization of14 C radioactivity during assay, 100 μL of [14C]cyanate solution (37 kBq/μL, 1 μCi/μL) were added to a 5-mL aliquot of whole blood. In the reference sample, no [14C]cyanate was added. The two samples were incubated at 37°C for 20 min to allow labeling. Then, 0.5-mL aliquots were peroxidized and neutralized as described for the peroxide assay. After neutralization, 100 μL of the [14C]cyanate solution were added to the second sample. Radioactivity was determined as described above. A comparison of the two allowed calculation of the loss of radioactivity during labeling and assay by difference.

In vitro simulation of in vivo conditions. 14C-Labeled red cells were diluted in unlabeled autologous red cells at a ratio of about 1 to 100, simulating the dilution that occurs in vivo with RCV and RCS studies. Unlabeled blood was anticoagulated using CPDA-1. The mixture of unlabeled and labeled blood was incubated in a shaking water bath at 37°C; 95% oxygen and 5% CO2 gas was infused into the head space of the incubation vessel at approximately 5 mL/min. Light microscopy of wet mounts and dry mounts using Leishman's stain both revealed normal red cell morphology with minimal hemolysis during the 72 h of the study.

For a typical in vivo determination of RCV, venous blood would be sampled four times in approximately 5-min intervals after adequate time for mixing (5 min). To determine PTR24, samples would be obtained at 24 h. To assess longer term survival, samples would be obtained at intervals varying from 1 d to several weeks. We simulated the sampling regime of an in vivo study by determining the concentration of 14C bound to protein in red cells at 5, 10, and 15 min and also at 24, 48, and 72 h after mixing.

To assess the effect of temperature on label stability, labeled red cells were diluted as described above and incubated without constant shaking for 24 h at 4 or 37°C as described above. Samples were well mixed by inversion before sampling.

Cellular distribution of label. Ultrafiltration with a 30 000-D pore size and molecular sieve chromatography with a 10 000-D pore size were used to separate 14C-bound macromolecules with a size greater than the pore size from free [14C]cyanate and any 14C bound to molecules smaller than the pore size. Hereafter this smaller pore size fraction will be referred to simply as “free.” Dextran-metrizoic acid density gradient was used to separate red cells from plasma. This is a complimentary technique to cell washing.

Specific details of these techniques are as follows.Ultrafiltration. A 0.5-mL volume of blood was lysed by addition of 0.5 mL of water, and the solution was pipetted onto a ultrafiltration device with 30 000-dalton cutoff (Amicon Centrifree units Amicon, Danvers, MA). The Centrifree unit was centrifuged at 1500 × g until approximately 50% of the solution was ultrafiltrated (1 h). A volume of the retentate and ultrafiltrate was counted by liquid scintillation.

Dextran-metrizoic acid density gradient separation. Separation solution was prepared by mixing 1 volume of 32.8% sodium metrizoate solution and 5 volumes of 6% dextran in PBS-glucose buffer(20). A 0.5-mL volume of blood was diluted to 1 mL with PBS-glucose and layered onto 3 mL of the separation solution. The layered gradient was centrifuged at 1000× g for 5 min in a swinging bucket rotor. The plasma layer and dextran layer were removed by vacuum aspiration. The plasma layer and the red cell pellet were used as described in “Results.”

Molecular sieve chromatography. A 0.5-mL volume of blood lysed by addition of 0.5 mL of water, the red cell pellet (e.g. from a dextran-metrizoic acid separation) lysed by addition of 1 mL of water, or the plasma layer from a dextranmetrizoic acid separation was applied to a molecular sieve column: (Bio-Rad Econo-Pac 10DG) and eluted with PBS-glucose. Heme-containing fractions were bleached by peroxide as described above. Radioactivity in eluate fractions was quantitated by liquid scintillation as described above.

Statistical analysis. Statistical analyses were performed with StatView software (Abacus Concepts, Berkeley, CA). Significance of differences between concentrations of 14C at time zero and 24 h were tested using the unpaired, two-tailed t test. For experiments with more than two time points, the significance of trends was tested using one-way analysis of variance; if statistically significant, post hoc testing was performed using Fisher's multiple range test. For all tests, significance was set at p < 0.05. If measurements made between 5 and 30 min were not significantly different by analysis of variance, values were pooled and are referred to as zero time.

Error bars depict ±1 SD. Error bars are omitted when smaller than the symbol diameter. For the populations of percent changes (e.g. between time zero and 24 h), standard deviations were conservatively estimated as the sum of the standard deviations of the replicates at zero and 24 h.

RESULTS

Validation of acid/acetone and peroxide assays. To be accurate, assays for the 14C label must quantitatively extract the 14C covalently bound to intracellular protein and accurately correct for the considerable scintillation quench arising from the residual material in the globin-containing fractions. For both assay methods, we quantitated both the efficiency of recovery of globin from red cells and the accuracy of the quench correction using heme free 14C-labeled globin prepared as described in“Methods.” Globin was extracted by the acid/acetone method as described and was resolubilized in water. For each assay method, the recovery of added reference 14C-globin from unlabeled blood was compared with normal saline (no quench) and to 14C-globin added to the extract after extraction (quench match).

For the acid/acetone assay, total extraction and quench correction efficiency was 98 ± 3%. Quench correction accuracy was 95 ± 3%. Detectable radioactivity was linear with added radioactivity. The slope of the linear regression was 0.98, and the regression correlation coefficient was 0.9998.

For the peroxide assay, total extraction and quench correction efficiency was 87 ± 2%. Quench correction accuracy was 77 ± 3%. Detectable radioactivity was linear with added radioactivity. The slope of the linear regression was 0.86, and the regression correlation coefficient was 0.9997.

Radiation release. Using whole blood, 9.3% of total radioactivity was lost as volatile products for the combination of labeling and acid/acetone assay. For the combination of labeling and peroxide assay, 7.1% of total radioactivity was lost as volatile products.

To assess whether radioactivity was volatilized during the incubation process, [14C]cyanate was incubated with PBS glucose as usual for labeling. The radioactivity was then quantitated directly without peroxide assay; >99% of the counts were recovered, providing evidence that little volatilization of 14C occurs during the red cell-labeling process. In further experiments, the radioactivity released by the assay was determined by subtracting 14C bound to globin and detectable in the precipitate washes from the total radioactivity in the red cell pellet (which equals the difference between total radioactivity added to the labeling process and that recovered in the washes of the labeled red cells). This experiment revealed that 41% of total red cell radioactivity was volatilized by the acid/acetone assay. In an analogous experiment using the peroxide assay, 27% of total radioactivity was volatilized. These studies provide evidence that substantial amounts of radioactivity were released by the assay process. Additional experiments conducted with plasma alone and PBS-glucose buffer suggest that the radioactivity is volatilized at the time of acidification of the solutions. More radioactivity is lost from buffer (>85%) than from plasma(55%), consistent with the hypothesis that free cyanate is the source of the released radioactivity.

Labeling efficiency. Based on acid/acetone assay data, the14 C radioactivity incorporated into red cells was 32 ± 14% of the total radioactivity added to the incubation (n = 7). Labeling efficiency ranged from 14 to 47%. This labeling efficiency exceeds those of methods that directly label whole blood and is similar to the efficiency of another published method that also labels packed red cells(Table 1)(17).

Table 1 Efficiency of [14C]cyanate labeling methods
Full size table

In vitro simulation of in vivo conditions. In some sheep experiments, blood concentrations of 14C bound actually increased from immediately after infusion (“zero time”) to 24 h. Such increases were too great to be explained by experimental error, even if PTR24 was assumed to be 100%. Moreover, in experiments using a permanent biotin label on the same red cells as the [14C]cyanate label, we observed 5-30% greater PTR24 of the [14C]cyanate label than of the biotin-labeled cells. This observation is consistent with continued labeling in vivo. To investigate this [14C]cyanate phenomenon, we initially evaluated several potential sources of increased blood concentration of label.

To evaluate a mechanism based on exchange with circulatory epithelia or transient in vivo sequestration and release of labeled cells by the reticuloendothelial system, we determined whether the phenomenon of increased labeling could be reproduced in vitro under conditions that mimicked in vivo circulation. When 14C-labeled red cells were diluted in normal cells and incubated for 3 d at 37°C, 14C bound increased by d 1 to 130% of the initial value and stabilized thereafter(Fig. 1, sheep 1). In a second experiment using cells from a different sheep, the bound 14C increased by d 1 to 127% and by d 3 to 154% (Fig. 1, sheep 2). These observations suggested that the presence of a circulatory epithelium and in vivo sequestration and release were not required for the observed phenomenon.

Figure 1
figure 1

Time course of bound 14C. ‡, * denotes p < 0.0001 vs pool of 5-30-min values. Values from 5 to 30 min are offset by ±.1 min for clarity. Labeling increases substantially by 1 d.

Full size image

To evaluate the possibility that the 14C label was redistributing between intracellular proteins that were and were not precipitated by acid/acetone and to monitor for an artifact due to a change in the efficiency of the acid/acetone precipitation or extraction, we characterized the nature of the 14C label associated with the red cells. The acid/acetone method detected an increase of 26 ± 3% (n = 4). The peroxide method detected an increase of 25 ± 1% (n = 4). For both methods, the increases were significant at p < 0.0001.

To investigate the temperature dependence of the increase, labeled cells were incubated for 24 h at three temperatures. At 4°C, 14C bound did not increase. In contrast, 14C bound increased by 19 ± 3% at 22°C, and by 35 ± 4% at 37°C (p < 0.02versus d 0 values). These observations provide evidence for a temperature-dependent mechanism (e.g. chemical reaction).

Cellular distribution of label. We investigated the possibility of continued intracellular labeling by measuring the distribution of free and bound 14C within labeled red cells using three techniques: 1) the acid/acetone assay method, which detects total 14C bound;2) ultrafiltration of lysed blood followed by scintillation counting of the ultrafiltrate, which detects free 14C, and of the retentate, which detects 14C bound by subtraction of free from total; and 3) molecular sieve chromatography of lysed blood, which separates bound 14C in the exclusion peak from free [14C]cyanate in the inclusion peak.

The free and bound 14C were determined immediately after labeling and after incubation for 24 h at 37°C by each of the physical techniques. As depicted in Figure 2A, the 14C bound increased as judged by each of the four techniques. Given the different physical nature of the four techniques, this observation argues strongly against analytic artifact as an explanation for the increase. Free 14C decreased, as judged by the two techniques capable of measuring free label. The extent of the decreases was similar and large enough to account for the increases in14 C bound as detected by extraction, chromatography, and density gradient. These data strongly suggested that unreacted [14C]cyanate was the source of the temperature-dependent increase in 14C bound.

Figure 2
figure 2

Effect of time on the intracellular distribution of14 C-label: free decreases as bound increases. (A) In one sheep, results of four methods two of which allow measurement of free label;(B) in three sheep, the results of the combined gradient/column method that allows measurement of bound and intracellular free label;(C) in three sheep, results of four methods, one of which allows measurement of total free label.

Full size image

In a final study using blood from three sheep, the initial gradient separation of red cells from plasma was followed by column chromatography of the lysed red cells. This method detected a decrease in intracellular14 C free that accounted for the increase in intracellular 14C bound (Fig. 2B). Each of the four independent methods detected about the same increase in the total 14C bound (Fig. 2C).

DISCUSSION

Interest in the interaction between cyanate and red cells arose from early attempts to prevent sickling of red cells in patients with sickle cell anemia. Cerami and Manning(21) hypothesized that the urea effect was probably not due to urea per se because the effect persisted long after urea was removed from circulation. These investigators discovered that cyanate forms a covalent bond with the amino group of the terminal valine of Hb. Equation

This carbamylation reaction occurs both when cyanate is infused i.v. in vivo(18), and when cyanate is incubated with red cells in vitro(21). To track the intracellular location of cyanate, Cerami and coworkers used14 C-cyanate(21, 22). These studies ultimately led to the use of [14C]cyanate as a permanent red cell label(23).

Our interest in a permanent red cell label in sheep arose from two sources. First, we are developing a nonradioactive method for determining RCV and RCS in the human fetus and infant; we needed a permanent, radioactive red cell label to use as a reference method for validation of the nonradioactive method in ovine model studies. Second, because recombinant human erythropoietin is available for therapy of anemia of prematurity, the mechanisms and regulation of erythropoiesis and red cell kinetics are being intensively studied in the ovine fetus and newborn lamb. Those mechanistic studies would be greatly aided by a method that allows determination of both circulating RCV and RCS. However, no suitable method has been published and validated to date.51 Cr is not suitable for determining RCS in sheep because the label leaves the red cell surface too rapidly(16).

To measure circulating RCV and RCS in sheep, we validated a labeling method and two assay methods for the 14C label bound to red cell protein as described here. The labeling method was adapted from the method of Eschbach et al.(17) and is about as efficient as that method (Table 1). For the species studied to date, the efficiencies of the in vitro labeling methods are consistently greater than those of the in vivo methods. We speculate that the lesser efficiency of labeling whole blood in vivo reflects unproductive reactions of the cyanate with plasma and tissue proteins, a rapid removal of the cyanate from circulation in vivo, or a combination. Consistent with the first mechanism is the observation that labeling of washed red cells (as used here) is more efficient than labeling of whole blood as reported by Landaw(19).

The studies reported here provide evidence that the acid/acetone assay specifically detects 14C covalently bound to intracellular protein with>95% efficiency. The overall detection efficiency of the peroxide method is variable, probably due to inaccurate quench correction. An acid/acetone extract of blood was used to construct the quench curves for both assay methods. Thus, the observed superiority of the quench correction for the acid/acetone assay compared with the peroxide assay may result from the better quench match. To avoid differences in the absolute concentrations of radioactivity between the acid/acetone assay and the peroxide assay, quench curves should be constructed for each assay. Notwithstanding this quench problem, the peroxide assay can be used to make valid determinations of RCV and RCS. For the RCV dilution calculations, the quench error will enter as a factor in both the numerator and denominator and will cancel as long as the same quench standard is used for all samples. Similarly, for measurements of label concentration over time, red cell kinetic parameters will not be affected by the constant quench error.

The problem with increasing 14C bound is an important one because the magnitude of the increase (about 25% in these studies) would cause an important error in the calculation of the PTR24. For most studies that would use [14C]cyanate as an RCS marker, accurate results can be obtained by referencing survival to d 1 rather than d 0. This is probably a prudent procedure and is commonly used to avoid the artifactual effects of red cell loss due to damage from labeling, handling, and storage. If an accurate PTR24 is needed, our studies indicate that incubation at 37°C for 24 h exhausts almost all the free, unreacted [14C]cyanate. Of course, samples for RCV studies must be assayed soon after collection (or stored at 4°C until assay) rather than being allowed to remain at room temperature.

Care should be taken to conduct in vitro labeling and assays with proper handling of evolved 14CO2. Such precautions might be overlooked, based on published methods(17, 19), but the release of 14CO2 with in vivo labeling has been recognized previously(18). Cyanate decomposes in solution to NH3 and CO2(21). When[14C]cyanate was administered to a mouse as a single intraperitoneal injection, only 7.5% of administered 14C was recovered from the red cells, but 72.2% was found in expired 14CO2(18). The study described here provides evidence that14 C is released by the acid/acetone assay and by the peroxide assay.