Distribution of plasma membrane Ca²⁺ pump activity in normal human red blood cells

The plasma membrane Ca²⁺ pump (PMCA), in its various isoforms, is expressed in all animal cells, where it plays a major role in the control of resting intracellular-free Ca²⁺ levels ([Ca²⁺]_i).^1-4  The PMCA was originally discovered and extensively studied in human red blood cells (RBCs) in which it is the only active Ca²⁺ extrusion transporter. The maximal Ca²⁺ transport capacity (V_max) of the PMCA in human RBCs (approximately 10 mmol [340 g hemoglobin {Hb}]^-1h^-1) is high compared with the normal pump-leak turnover rate of Ca²⁺ (approximately 50 μmol [340 g Hb]^-1h^-1).³  Although early work indicated that Ca²⁺ pump activity may vary from cell to cell over a much broader range than any other known hematologic parameter,⁵  available measurements of PMCA V_max in intact RBCs always reported mean values in the RBC population⁶ ; the actual distribution of PMCA V_max activity among RBCs was unknown. The aim of the present work was to characterize the distribution of PMCA V_max activity in normal human RBCs.

It is important to ascertain the extent of PMCA V_max variation among RBCs because differences in pump-leak [Ca²⁺]_i levels may determine the degree of vulnerability of individual cells to certain physiologic and pathologic stresses. Human RBCs express in their plasma membrane a Ca²⁺-sensitive K⁺ channel (the “Gardos” channel^7-9 ). Hoffman et al¹⁰  recently showed that the isoform hSK4 represents the small conductance Gardos channel of human RBCs. In conditions with elevated [Ca²⁺]_i, Gardos channel activation causes rapid net loss of KCl and water, leading to irreversible cell dehydration. Elevated [Ca²⁺]_i has also been implicated in the activation of cell proteases and in the cross-linking of cytoskeletal proteins, with cumulative effects on cell aging and viability.^11-15 

The first indirect evidence of substantial cell-to-cell variations in PMCA V_max arose when Tiffert et al¹⁶  examined the dehydration response of RBCs uniformly permeabilized to Ca²⁺ with the ionophore A23187. As Ca²⁺ influx into RBCs was progressively increased by increasing the ionophore concentration in a stepwise manner, RBCs suspended in plasmalike media dehydrated following an all-or-none rather than a graded pattern: Instead of all cells dehydrating at progressively faster rates—as expected with a uniform response—only a fraction of RBCs dehydrated and this fraction increased with increasing Ca²⁺ influx. Further experiments showed that Gardos channel activity among the RBCs was fairly uniform and could not have accounted for this all-or-none pattern.¹⁷  These results suggested that uniform Ca²⁺ permeabilization of RBCs did not generate uniformly elevated [Ca²⁺]_i among the cells and that the differences arose from variations in Ca²⁺ pump activity.

Further evidence for marked variations in Ca²⁺ pump activity among RBCs from single donors was provided by Garciáa-Sancho and Lew^18,19  who detected large differences in the total Ca²⁺ content of RBCs ([Ca_T]_i) uniformly permeabilized to Ca²⁺ with A23187.²⁰  As Ca²⁺ influx increased so did the fraction of high-Ca²⁺-containing cells, indicating a gradual variation of some property among the RBCs rather than a sharp distinction between subpopulations. This unexpected behavior suggested that the measured total Ca²⁺ content of Ca²⁺-permeabilized RBCs did not represent a uniform Ca²⁺ content but rather the mean of extremely heterogeneous distributions. To interpret this phenomenon, Garciáa-Sancho and Lew¹⁹  proposed a relatively wide distribution of PMCA V_max activities among RBCs: the fraction of cells with V_max below the set Ca²⁺ influx would gain Ca²⁺ continuously, while their pumps ran at the Ca²⁺-saturated V_max rate consuming cell adenosine triphosphate (ATP).^6,19  In those cells in which the V_max of the pump exceeded the glycolytic capacity, ATP depletion would eventually inhibit the pump and [Ca²⁺]_i would approach the Ca²⁺ equilibrium distribution determined by the ionophore with very high [Ca_T]_i. On the other hand, in those RBCs with a pump V_max higher than the set Ca²⁺ influx and with a glycolytic capacity to match ATP consumption by the pump, pump-leak balance would maintain [Ca²⁺]_i at levels below pump saturation in the micromolar or submicromolar range. More recent work provided additional evidence for V_max differences among RBCs.^21,22  However, the actual distribution of PMCA V_max among RBCs has never been directly investigated. We report here an investigation of the distribution of PMCA V_max in normal RBCs using a new method developed ad hoc. The results exposed a broad unimodal variation of Ca²⁺ pump activity among RBCs from single donors, with a large coefficient of variation and a right-skewed V_max distribution. The significance of such a distribution is discussed.

Finding and characterizing RBCs that differ in the speed at which they can pump out an ionophore-induced uniform Ca²⁺ load²⁰  requires a procedure to detect cells with different Ca²⁺ contents during the Ca²⁺-extrusion process. We designed a method based on the property of RBCs to become rapidly dehydrated when their [Ca²⁺]_i is elevated. The RBC components that participate in Ca²⁺-induced dehydration are illustrated in Figure 1 (modified from Tiffert et al²²  with permission from Elsevier). The PMCA V_max distribution method was based on the “Ca²⁺ load-extrusion” protocol designed by Dagher and Lew⁶  to measure the mean PMCA V_max in RBCs. This is illustrated in the computer simulations of Figure 2A.

To detect V_max heterogeneity and characterize its distribution, the load-extrusion protocol in Figure 2A was used as follows: At the points A to E on the model-generated Ca²⁺ load-extrusion curve, RBCs are sampled and delivered into a low-K⁺ medium (LK medium; see “Solutions”) consisting of a large volume of ice-cold, iso-osmotic buffer containing EGTA (ethylene glycol-bis(2-aminoethylether)-N,N,N′,N′,-tetraacetic acid) and no added Ca²⁺, to prevent any Ca²⁺ gains; 1 mM vanadate to ensure complete pump inhibition; low-K⁺; and 10 mM SCN^- to allow rapid dehydration of the cells with elevated Ca²⁺. This medium instantly arrests Ca²⁺ fluxes mediated by ionophore and pump so that each RBC retains the Ca²⁺ it contained at the time of sampling, which determines whether or not it will dehydrate.

When Gardos channels are fully activated at nonlimiting anion permeabilities, full dehydration proceeds without profile changes in the hemolysis curves indicating that the distribution of Gardos channel-mediated dehydration capacity is fairly uniform among RBCs.¹⁷  Thus, any alterations in the profile of hemolysis curves from cells actively extruding a Ca²⁺ load could be attributed to differences in Ca²⁺ pump extrusion capacity among the cells. Let us assume provisionally that maximal Gardos channel activation is attained when [Ca_T]_i exceeds 20 μmol (340 g Hb)^-1 as indicated by the dashed horizontal line in Figure 2A. Then all RBCs with [Ca_T]_i more than 20 μmol (340 g Hb)^-1 will dehydrate maximally, whereas those with lower [Ca_T]_i will dehydrate partially or not at all. If we induce Ca²⁺ loads much higher than the Gardos channel deactivation threshold, the time required to pump out the subthreshold Ca²⁺ becomes negligible in relation to the time required to pump out the full Ca²⁺ load, justifying the assumption that nondehydrating cells have practically emptied their Ca²⁺ loads. After a brief incubation at 4°C in LK medium to allow the cells containing suprathreshold [Ca_T]_i to dehydrate maximally, the tubes are spun, most of the supernatant is discarded, and the cell pellets are resuspended in the remaining medium for hemolysis curve sampling.

The predicted results were obtained in preliminary experiments whose real hemolysis curves, from samples taken at the times A to E in Figure 2A, are shown in Figure 2B. With samples A and E we see normal hemolysis curves since all the cells would have been essentially Ca²⁺ free (in sample A because they never experienced a Ca²⁺ load and in sample E because the Ca²⁺ load had been fully extruded, even from the slowest pumping cells). The identity of the 2 curves rules out any significant effects of the Ca²⁺ loads implemented here on the critical hemolytic volume of normal RBC populations, as might have been expected from Ca²⁺-induced membrane area losses.³⁷  Sample B gives the hemolysis curve corresponding to full dehydration of all the cells: because RBCs dehydrated by net salt loss need to gain much more water than normal-volume cells to swell to their critical hemolytic volume, the curve is deeply left-shifted toward low relative tonicities. In samples C and D, variable fractions of high-V_max cells did not dehydrate since they had pumped out nearly all their Ca²⁺ at the time of sampling, whereas the slower pumping cells retaining more than 20 μmol (340 g Hb)^-1 of [Ca_T]_i became fully dehydrated; the resulting lysis curves C and D are intermediate (between curves A and B) with an inflected quasi-plateau region, indicating the development of a marked bimodality in the distribution of hydration states in the RBC population. As progressively more cells pump out their Ca²⁺ loads, the mixed lysis curves with the plateau pattern would migrate upwards toward the fully recovered pattern of sample E. The fraction of cells that emptied their Ca²⁺ loads at each sampling time may be estimated from the Y-intercept values read near the midpoint of the quasi-plateau region of the hemolysis curves. An example is shown in Figure 2B by the dashed vertical line at a relative tonicity of 0.3. The important consideration in the choice of relative tonicity is that the analysis to be applied should be largely independent of the value chosen to represent the plateau. In the experiments to be reported below, the results agreed within a margin of ± 3% when the choice of relative tonicity for the plateau Y-intercepts was within the range 0.25 to 0.40. To estimate the percent cells with fully extruded Ca²⁺ loads at each sampling time (f_X) we must first define the 100% value. We use the ordinate values (Y) at the intersection of curves A (or E) and B with the dashed vertical line to define Y_A and Y_B. The full lysis curve migration interval (100%) is Y_A - Y_B. The percent cells with emptied Ca²⁺ loads at the time of sample C (f_C) may be estimated from the ordinate value of curve C at a relative tonicity of 0.3 (Y_C) as follows: f_C = 100 (Y_C - Y_B)/(Y_A - Y_B). And for D, f_D = 100 (Y_D - Y_B)/(Y_A - Y_B).

Rapid sampling for lysis curves during the Ca²⁺ extrusion phase of the load-extrusion protocol could thus provide a detailed characterization of f_X as a function of the time interval between Co²⁺ addition and emptying of the Ca²⁺ load, Δt_X. This function is the integral of the distribution of PMCA V_max values in the RBC population since for each f_X the corresponding V_max is computed from (V_max)_X = (Ca²⁺ load)/Δt_X, where Δt_X is the time interval between Co²⁺ addition and sampling for X.

The 2 basic solutions used (to which specific additions were made, as noted) were (1) high-K⁺ (HK)-containing (80 mM KCl; 70 mM NaCl; 10 mM HEPES-Na [N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid-Na], pH 7.5, at 37°C; and 0.15 mM MgCl₂) and (2) low-K⁺ (LK) (140 mM NaCl; 10 mM NaSCN; 10 mM HEPES-Na, pH 7.5, at 37°C; and 0.15 mM MgCl₂). Unless specified otherwise, the final concentrations of the most frequently added solutes was (in mM) EGTA, 0.1; CaCl₂, 0.14; inosine, 5; sodium orthovanadate, 1; CoCl₂, 0.4 in HK medium and 0.1 in LK medium; ionophore A23187 (from 2 mM stock in ethanol), 0.01. When radioactive ⁴⁵Ca was used, the specific activity was set between 10⁷ and 2 × 10⁸ counts per minute (cpm)/μmol. All additions were done from stock solutions at least 100-fold more concentrated than in the final solutions or cell suspensions.

Venous blood was drawn from healthy volunteers into heparinized syringes after obtaining informed, written consent. Approval was obtained from the Institutional Review Boards of the Physiological Laboratory, University of Cambridge, United Kingdom; and from the Department of Medicine, Albert Einstein College of Medicine, Bronx, NY, for these studies. RBCs were washed twice by centrifugation and resuspension in large volumes of ice-cold HK solution with added EGTA and thrice more with solution HK alone. The buffy coat containing platelets and white cells was removed after each wash. After the last wash, the cells were suspended at 10% hematocrit (Hct) in medium HK supplemented with inosine and ⁴⁵Ca(Ca²⁺) or Ca²⁺, as described in Figures 3 and 4, and incubated for about 10 minutes at 37°C for temperature equilibration before applying the load-extrusion protocol with the specific modalities described in the figure legends.

Two sets of experiments were performed with slightly different protocols. The first set of experiments was designed to test the assumptions implicit in the design of the V_max distribution method by comparing the release of Hb and ⁴⁵Ca induced by hypotonic lysis (Figure 3; Tables 1, 2). The second set of experiments aimed to characterize the distribution of PMCA V_max values among RBCs from healthy subjects.

For the first set of experiments (Figure 3 protocol 1) washed RBCs were suspended at 10% Hct in medium HK, supplemented with inosine and high-specific activity ⁴⁵Ca(Ca²⁺) (2 × 10⁸ cpm/μmol) and incubated for about 10 minutes at 37°C for temperature equilibration. The ionophore A23187 was added at t = 0 followed 2 minutes later by CoCl₂. Samples (0.5 mL) were drawn once before ionophore addition, twice or thrice after ionophore, and just before cobalt addition to estimate the size of the Ca²⁺ load. After cobalt, samples drawn every 30 seconds were delivered into centrifuge tubes containing 12 mL of ice-cold LK solution with added EGTA and vanadate to inhibit the Ca²⁺ pump (preliminary experiments had shown that exposure of the RBCs to 1 mM vanadate at 0°C to 4°C produced irreversible and near-complete Ca²⁺ pump inhibition). In this medium the Ca²⁺ content of the RBCs at the time of sampling was maintained, since low temperature, vanadate, and intracellular cobalt prevented Ca²⁺ loss through the ionophore or extrusion by the pump. After about 30 minutes of this “postincubation” in the ice bath to allow maximal dehydration by the Ca²⁺-containing cells, the RBCs were washed once by centrifugation and resuspension in 10 to 12 mL of LK+EGTA medium, and the cell pellet was finally resuspended in 1 mL LK+EGTA. The wash and dilution reduced external ⁴⁵Ca to near background and rendered about 1 mL of a 5% Hct suspension for each sample, whose intracellular ⁴⁵Ca distribution was maintained from the time of sampling. Of this, 0.1 mL was centrifuged for 15 seconds in 1.5-mL microfuge tubes, and after discarding the supernatant the cell pellet was resuspended in 0.6 mL of 6% trichloroacetic acid (TCA) and centrifuged again. From the clear TCA supernatant, 0.5 mL was taken for scintillation counting of ⁴⁵Ca to estimate [Ca_T]_i (Figure 3A). The remaining 0.9 mL of each sample was used for osmotic lysis curves (see next paragraph), after which the Hb and ⁴⁵Ca²⁺ released by hypotonic lysis was determined for each of the 12 different microwells.

A solution containing 155 mM NaCl and 2 mM HEPES-Na, pH 7.5, at room temperature, iso-osmotic with the HK and LK solutions, was diluted with a medium containing only 2 mM HEPES-Na to render 12 “lysis” solutions with the relative tonicities (RT) indicated in the figures. Two hundred fifty microliters of each of these lysis solutions were delivered into each of the 8 wells of columns 1 to 12 of 96 × round-bottom microwell plates. Between 0.5 and 0.7 mL of the 0.9-mL 5% Hct sample was placed in a grooved container and, using a multichannel pipette, 10 μL was delivered and mixed into each of the 12 wells of the assigned row so that each sample provided a complete lysis curve. The round-bottom plates were centrifuged at 1020g × 5 minutes, and 150 μL of the cell-free supernatant from each well was transferred to corresponding wells on 96 × flat-bottom microwell plates for measurement of Hb concentrations by light absorbance at 415 nm (Soret band) using a microplate reader (Molecular Devices, Sunnyvale, CA).⁵  The absorbance in the 2 mM HEPES-Na buffer (relative tonicity = “0”) was taken as 100% lysis, and the readings at the higher tonicities were compared to it. After the Hb measurement, 130 μL of 11% TCA were added to each well, the plate was centrifuged to sediment denatured protein, and 200 μL of the clear TCA supernatant from each well was transferred to scintillation vials to measure the ⁴⁵Ca released by lysis. These measurements provided a direct comparison of the Hb and Ca²⁺ released from the RBCs at each relative tonicity and at each sampling time.

The second set of experiments (Figure 4 protocol 2) differed from the first in that the original cell suspension was divided in 2, one for measuring the mean V_max of the PMCA by the Dagher-Lew method⁶  using low specific activity ⁴⁵Ca(Ca²⁺) (Figure 4A), the other for measuring lysis curve migration with high time resolution in tracer-free conditions (Figure 4B). Except for the presence of tracer ⁴⁵Ca in the aliquot for measurement of the mean V_max, the load extrusion protocol was the same for both suspensions. Samples for lysis curves (0.5 mL) from the tracer-free suspension were drawn at 10-second intervals after cobalt addition and processed as described in the preceding paragraph, except for the TCA treatment.

We first tested the following assumptions made for the experimental design shown in Figure 2: (1) starting from a high, saturating, and uniform Ca²⁺ load, cell-to-cell differences in Ca²⁺ pumping rates would generate a progressive heterogeneity of Ca²⁺ contents among the RBCs; (2) deactivation of the Gardos channels would occur when [Ca_T]_i fell below approximately 20 μmol (340 g Hb)^-1 (Figure 2A dashed line); and (3) the vertical migration of the quasi-plateau region of hemolysis curves (Figure 2B) reports the progressive fraction of RBCs that have fully extruded the imposed Ca²⁺ load.

A typical result is shown in Figure 3 and Tables 1 and 2. During the load-extrusion protocol parallel samples were taken at the indicated times to estimate the mean PMCA V_max from the initial slope of the Ca²⁺ extrusion curve after cobalt addition (approximately 10 mmol [340 g Hb]^-1h^-1; Figure 3A). Apparent Ca²⁺ desaturation of the pump began after 4.5 minutes when the mean [Ca_T]_i was below 73 μmol (340 g Hb)^-1.

The losses of Hb and Ca²⁺ induced by hypotonic lysis (Table 1; Figure 3B) were compared for 4 selected samples taken between 4 and 5.5 minutes of ionophore addition (2 and 3.5 minutes after addition of cobalt). With the 4-minutes sample (as with earlier samples taken after cobalt; not shown) the Hb and Ca²⁺ release curves had very similar patterns (Figure 3B ○ and •, respectively). But the following 3 samples showed a progressive divergence between the Ca²⁺ release curves and the Hb release curves. The change in Hb release pattern shows that the proportion of dehydrated cells decreased progressively from more than 80% in the 4-minutes sample to less than 20% in the 5.5-minutes sample. In contrast, the ⁴⁵Ca release pattern remained essentially unchanged; more than 80% of the Ca²⁺ was released at relative tonicities below 0.20. Because dehydrated cells lyse only at these low relative tonicities, the Ca²⁺ release pattern indicates that in all samples, more than 80% of the mean [Ca_T]_i was always contained within cells that dehydrated during the postincubation period. Thus, the increasing fraction of RBCs that failed to dehydrate during the postincubation period in LK medium always contained less than 20% of the RBCs' mean total Ca²⁺ content.

These findings indicate that during PMCA-mediated Ca²⁺ extrusion of a uniform Ca²⁺ load, a large and progressive heterogeneity of Ca²⁺ contents is generated among the RBCs, which must be attributed to cell-to-cell differences in PMCA V_max.

In Figure 3C, the fractions of Hb and Ca²⁺ released at each relative tonicity are compared directly. The progressive departures from single linearity expose developing heterogeneities of Ca²⁺ content. At 4.5 minutes and later, with the emergence of low-[Ca_T]_i cells, the curves show 2 distinct slopes reflecting the progressive dissociation between lysis-induced Hb and Ca²⁺ release. It can be seen that the increase in fractional Hb loss with time (steep slope component) is contributed by RBCs that contain less than 20% of the mean [Ca_T]_i in the cell population.

Two additional samples in Figure 3C (dashed lines) were from cells postincubated in HK media so that they retained their original volume distribution from before the experiment. In these samples, as with all HK postincubated RBCs, regardless of the sampling time during the load-extrusion protocol, the points fell within 5% of the unit-slope line, indicating that when the original volume distribution of RBCs was retained during the load-extrusion protocol and subsequent sample processing, cells with different PMCA V_max and [Ca_T]_i remained evenly distributed in the cell population and cannot be distinguished by any of the properties that determine the normal lysis curve distribution such as cell volume, area, Hb, and total osmolyte content.^5,38  A slope of 1 was also obtained from samples drawn just before cobalt addition. Although not unexpected, since the ⁴⁵Ca distribution was uniform at that stage, the result is important because it confirms an important assumption that hypotonic swelling of cells to prelytic states, as must take place within all the microwells with hypotonic media, does not increase their Ca²⁺ permeability.¹⁸ 

The [Ca_T]_i level at which Gardos channels deactivate was estimated from the analysis in Table 2 (from selected data in Table 1). Note that for each of the time points of partial Ca²⁺ extrusion in Table 2 when the mean RBC [Ca_T]_i was 73, 39, and 17 μmol (340 g Hb)^-1 and that of the dehydrated cells was 114, 95, and 78, respectively, the [Ca_T]_i within the cells that did not dehydrate was 15, 6.2, and 2.7 μmol (340 g Hb)^-1. Thus, Gardos channel deactivation always occurred when [Ca_T]_i fell below this chosen boundary of 20 μmol (340 g Hb)^-1. The results also document the large differences of Ca²⁺ contents generated shortly after the initiation of net Ca²⁺ extrusion that result from cell-to-cell variations in PMCA V_max.

The detailed distribution of PMCA V_max was investigated in RBCs from 4 donors in 5 experiments that rendered very similar patterns (compiled in Table 3). Figure 4 shows a typical experiment in which the mean PMCA V_max, measured by linear regression through the first 6 time points after Co²⁺ addition, was 12.8 mmol (340 g Hb)^-1h^-1 (Figure 4A). After the mean RBC [Ca_T]_i fell below approximately 100 μmol (340 g Hb)^-1, the Ca²⁺ extrusion rate dropped progressively. During Ca²⁺ extrusion, frequent sampling for hemolysis curves gave the patterns shown in Figure 4B. Beginning about one minute after Co²⁺ addition, there was a progressive increase in the fraction of RBCs that failed to dehydrate during the postincubation period in low-K⁺ media, indicating that they had pumped out their initial Ca²⁺ loads to levels below those that activate the Gardos channels. The increase in the fraction of such cells was assessed by the upward progression of the quasi-plateau portions of the lysis curves (see “Materials and methods”). The result is shown in Figure 4C where the percent calcium-emptied cells, estimated from the readings at a relative tonicity of 0.3, was plotted as a function of time after Co²⁺ addition. The experimental points were well-fitted by a sigmoid, saturation-like curve (Figure 4C legend). About 90% of the RBCs emptied their Ca²⁺ loads within 3 minutes of Co²⁺ addition, while approximately 10% took longer to complete Ca²⁺ extrusion.

Figure 4D (solid line) shows the derivative of the curve fit in Figure 4C estimated by dividing the Ca²⁺ load by the time interval between Co²⁺ addition and Ca²⁺ emptying, plotted as a function of V_max. This represents the distribution of PMCA V_max activities among the RBCs.

The results show a broad distribution of PMCA V_max values with a right-skewed appearance. A symmetrical Gaussian distribution curve (Figure 4D dotted line) was superimposed on the rising branch of the V_max distribution to highlight its skewed nature. V_max values varied about 7-fold in this experiment, between 5 and 35 mmol (340 g Hb)^-1h^-1 with mean and standard deviation of 12.9 and 5.9 mmol (340 g Hb)^-1h^-1, respectively, and a positive skew of 1.55. The mean V_max was higher than the mode, as expected from the right skew of the distribution. In all 5 experiments of this series the V_max distribution was unimodal, broad, and positively skewed with an identical pattern and a range of V_max variation similar to that shown for the experiment of Figure 4 (6- to 9-fold).

The results presented here provide the first detailed description of the distribution of PMCA V_max activity in normal human RBCs. This distribution characterizes the differences in Ca²⁺ extrusion capacity among RBCs but carries no information on whether these differences result from disparities in the number of pumps, in the fraction of active pumps, in the factors controlling the turnover rate of the pumps in each cell, in a differential response of the 2 pump isoforms, or in a combination of these.

Earlier findings had indicated the existence of differences in Ca²⁺ pumping rate among RBC subpopulations, but the actual pattern and extent of variation was unknown. The present results show that there is a large variation in PMCA V_max among RBCs from single donors and that this variation follows a unimodal, broad distribution pattern with a consistent positive skewness toward higher V_max values, suggesting an excess of cells with V_max higher than the mean value. The apparent range of V_max variation measured in the 5 experiments of this series was 6- to 9-fold. However, it should be noted that the method can detect the early upward deflections in hemolysis curve profiles with high sensitivity (Figure 4B-C), whereas the resolution in the late periods, when nearly all cells have emptied their Ca²⁺ loads, is less precise. Thus, the possibility cannot be ruled out that there are some RBCs with V_max values much lower than those apparent from the indicated range.

It is important to rule out a potential artifactual origin of the skewed distribution. Differences in PMCA V_max among RBCs may result in uneven ionophore-induced Ca²⁺ loads. The cells with the highest V_max would be expected to sustain a pump-leak balance with [Ca_T]_i below the measured mean load. The early emptying of their Ca²⁺ load could result not only from high V_max values but also from lower Ca²⁺ loads. Using the mean Ca²⁺ load to calculate their V_max could then overestimate their true V_max values, thus creating a false right-shift in the V_max distribution. However, Tiffert and Lew^27(Fig8) showed that at the high ionophore and [Ca²⁺]_o concentrations used in the present experiments, the difference between measured mean Ca²⁺ loads in the presence and absence of 1 mM vanadate, was negligible (this finding was repeatedly confirmed in other experiments, not shown). Vanadate at this concentration inhibited more than 99.5% of PMCA-mediated Ca²⁺ fluxes. Therefore, the differences in Ca²⁺ load among RBCs must have been minimal and cannot account for the large positive skew observed here.

The 3 main properties of the PMCA V_max distribution documented here are unimodality, wide-spread, and right-shifted skew. Unimodality indicates that the variations in PMCA V_max in normal human RBCs are smooth and continuous, thus ruling out distinct cell subpopulations with large differences in V_max. The 2 isoforms of the PMCA known to be expressed in human RBCs³⁹  may contribute to the observed variation in Ca²⁺ extrusion capacity among the cells. The unimodal pattern imposes constraints on this potential contribution but does not provide any information about possible functional differences between the 2 isoforms.

The question arises of how the large spread and positive skew of the PMCA V_max distribution compares with those of other hematologic and transport parameters of RBCs. Thus far, only a few constitutive and functional RBC properties have been examined for population variations. RBC volume, osmotic fragility, Hb content, and Hb concentration all follow symmetric Gaussian distributions with coefficients of variation (in percent) of 13.7, 5.4, 12.5, and 7.0, respectively.⁵  RBC membrane transporters, whether highly expressed (anion exchanger 1.1 × 10⁶ per cell^40,41 ) or weakly expressed (Gardos channels, 100-200 per cell^42-44 ), also appear to have parameter distributions with coefficients of variation of less than 7%,⁴¹  indicating relative uniformity. On the other hand, prostaglandin E₂ was reported to induce Ca²⁺-dependent Gardos channel activation but only in about 15% of the RBCs, suggesting that receptor expression occurs in selected RBC subpopulations.⁴⁵  Against this background, the PMCA V_max distribution shows an unprecedented spread with a coefficient of variation of about 50% (range of 45% to 53% in the 5 experiments [Table 3], 46% in the experiment of Figure 4) and a positive skew, unique among the distributions investigated.

What might be the origin of this unusual distribution? Does it represent a static, constitutive feature of mature RBC populations generated during normal “production line” erythropoiesis or is it a dynamic, age-related property with an age-declining pattern? In other words, are individual RBCs “born” with the V_max activity they will sustain throughout their approximate 120 days in the circulation, or does PMCA V_max decline with age? At present, there is no conclusive evidence to distinguish between these 2 alternatives, and results of earlier experiments testing variations in Ca²⁺ pump V_max in density-fractionated RBCs were conflicting.^21,46,47 

If PMCA V_max does decline with RBC age the pattern would appear to differ from that of most other RBC transporters. In the transition from reticulocyte to mature RBC, the ion transport mediated by the sodium pump and the KCl cotransport (KCC) symport become substantially reduced within the first 2 to 3 days in the circulation.^48-52  On the other hand, there is no evidence of early decline in PMCA activity from reticulocyte to mature RBC.⁵³  In the context of the age-decline hypothesis, the broad PMCA V_max variation pattern points toward a progressive fall in pump activity throughout the life of the mature cell. The possible mechanism of V_max declines with cell age, such as progressive protease-induced falls in viable pump units per cell,⁵⁴  metabolically-linked,^15,47  progressive glycation,⁵⁵  or other, remains speculative. The method applied here to study the distribution of PMCA V_max, by enabling the segregation of RBCs with different V_max values, may provide a way to investigate the possible age-V_max relation more directly in the future.

A final consideration arising from the present results concerns the reduction in Ca²⁺ extrusion rate at relatively high mean [Ca_T]_i levels seen during the Ca²⁺ extrusion phase of the load-extrusion protocol (Figures 3A,4A). This pattern has been interpreted in the past as reflecting desaturation of Ca²⁺ by the PMCA and was used to derive kinetic parameters of the pump.^21,56  As explained above in the analysis of Figure 2A, the wide PMCA V_max distribution documented here (Figure 4D) fully accounts for the observed apparent Ca²⁺ desaturation pattern that starts at high mean [Ca_T]_i levels. Thus, apparent desaturation at high [Ca_T]_i is an indicator of the presence of low-V_max RBCs and carries no information on the kinetics of PMCA activation by [Ca²⁺]_i.

We are grateful to the Wellcome Trust, United Kingdom; and to the National Institutes of Health, United States, for funds.