0 1994 by The American Society for Biochemistry and Molecular Biology, Inc. THE JOLIRNAL OF BIOUXICAL CHEMISTRY Vol. 269, No. 4, Issue of January 28. pp. 2399-2404, 1994

Printed in U S A .

Exposure of Phosphatidylserine in the Outer Leaflet of Human Red Blood Cells RELATIONSHIP TO CELL DENSITY, CELL AGE, AND CLEARANCE BY MONONUCLEAR CELLS*

(Received for publication, May 17, 1993, and in revised form, September 28, 1993)

Jerome Connor, Charles C. Pak, and Alan J. SchroitS From the Department of Cell Biology, University of E m s M. D. Anderson Cancer Center, Houston, nxas 77030

Human red blood cells (RBC) were separated by den- sity on self-forming Percoll gradients into five distinct populations. The transbilayer movement and equilib- rium distribution of l-oleoyl-2-(N-(7-nitrobenz-2-oxa- 1,S-diazol-4-yl)aminocaproyl)phosphatidylserine (NBD- PS) was slower in dense cells and equilibrated in the inner leaflet of these cells to a lesser degree than in light cells. Conversely, the outward movement of the lipid was slower in light cells. Assessment of endog- enous PS in the cells’ outer leaflet by the prothrombi- nase activity of externalized PS revealed an increase in its presence at the cell surface with increasing cell den- sity. The presence of PS on the cell surface directly cor- related with the propensity of the RBC to be bound by autologous monocytes. To determine whether increased cell density is associated with increased cell age, the in vivo clearance of density-separated murine RBC was monitored in syngeneic mice. The Tm of circulation of light cells was about twice that of dense cells. The ma- jority of the cleared cells localized in the spleen. Stud- ies carried out in antibody-deficient SCID mice indi- cated that RBC were cleared via a mechanism that was independent of antibody. These data suggest that cell age is related to cell density and that cells of increasing age and density display higher amounts of PS in their outer leaflet.

~~ ~

Normal tissue function is dependent on the disposal of non- functional senescent cells. This is particularly evident in the circulatory system where the human red blood cell (RBC)l survives in the peripheral circulation for about 120 days. Al- though definitive proof is lacking, senescent red cell clearance is probably the result of a series of progressive events (Bar- tosz, 1991; Piomelli and Seaman, 1993) that lead to the cells’ deterioration and subsequent removal from the circulation (Bartosz, 1990). Typically, density-separated aged cells exhibit decreased cell volume (Nash and Wyard, 1980; Linderkamp and Meiselman, 19821, cell size (van Oss, 1982) and deform- ability (Clark et al., 1983; Nash and Meiselman, 19831, as well as increased osmotic fragility (Riflcind et al., 1983) due to loss

* This work was supported in part by National Institutes of Health Grant DK 41714. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Box 173, University of Texas M. D. Anderson Cancer Center, 1515 $ To whom correspondence should be addressed: Dept. of Cell Biology,

Holcombe Blvd., Houston, TX 77030. The abbreviations used are: RBC, human red blood cell(s);

NBD-, l-oleoyl-2-~N-~7-nitrobenz-2-oxa-1,3-diazol-4-yl)~nocaproyl)-;

tidylcholine; RES, reticuloendothelial system; BSA, bovine serum albu- PE; phosphatidylethanolamine; PS, phosphatidylserine; PC, phospha-

min; PBS, phosphate-buffered saline.

~~

of electrolytes and microvesiculation (Dumaswala and Green- Walt, 1984; Greenwalt and Dumaswala, 1988). Membrane-re- lated alterations include decreased surface charge density and lipid content due to loss of sialic acid residues (Bartosz et a l . , 1984) and microvesiculation (Bartosz, 1981), respectively. In addition, senescent RBC exhibit increased adhesiveness to en- dothelial and reticuloendothelial cells (Dhermy et al . , 1987), changes in membrane cation transport (Hentschel et al . , 1986), decreased enzymatic activities (Bartosz, 19901, and ac- cumulation of lipid peroxidation products (Jain, 1988). Al- though these alterations clearly indicate that RBC undergo time-dependent changes, it is not known which process, or combination of processes, results in an “aberrant” senescent cell that will be removed from the circulation (Lutz, 1990). It is also not known which of the cellular changes results in the selective recognition of old cells by mononuclear cells. None- theless, several non-mutually exclusive mechanisms have been proposed. These include fragmentation of the membrane and desialylation (Aminoff, 1988; Schlepper-Schafer et al., 1983), exposure of cryptic, senescent epitopes (Kay, 1978; Kay et al., 1984), oxidative damage (Jain et al . , 1983), aggregation of cell surface proteins, and the presence of increased PS on the cell surface (Schroit et al., 1985; Allen et al . , 1988). Con- ceivably, these membrane changes can be either directly de- tected by mononuclear phagocytes or mediated through the binding of autologous antibody (Kay, 1978; Galili et al., 1986) and complement (Lutz et al., 1987, 1988).

Since PS is predominantly localized in the inner membrane leaflet of normal RBC (Verkleij et al., 1973; Op den Kamp, 19791, changes in endogenous membrane organization could lead to its presence in the outer leaflet. Indeed, it has been shown that certain pathologic (Schwartz et al., 1985; Connor et al . , 1989; Utsugi et al., 1991) and apoptotic (Fadok et al., 1992a, 199213) cells express PS on the outer surface. It is possible that alterations in the system responsible for maintaining lipid asymmetry, i.e. the aminophospholipid transporter, generates a shift in the normal equilibrium distribution of PS (Connor et al . , 1992). Indeed, the initial transport velocity of exogenously supplied aminophospholipid analogs across the membrane bi- layer in sickled RBC and in artificially and in vivo aged RBC was reduced when compared to normal (Blumenfeld et al., 1991) and young cells (Herrmann and Devaux, 19901, respec- tively. Interestingly, cells that express PS in their outer mem- brane surface are bound by macrophages in vitro (Connor et al., 1989; Utsugi et al., 1991; Fadok et al., 1992a, 199213) and are cleared in vivo from the peripheral circulation to the spleen and liver (Schroit et al., 1985). Thus, as a consequence of alterations in membrane lipid organization, RBC are recognized by the RES and removed from the vasculature, a process which may be similar to the normal physiological clearance of RBC.

In this study we show that RBC of increasing density display differences in the transmembrane distribution of PS that are probably associated with the cells age. Relatively dense, old

2399

2400 Phosphatidylserine in Red Blood Cells

RBC transport an exogenously supplied PS analog from the outer to the inner leaflet more slowly, and contain higher amounts of endogenous PS in their outer leaflet than do young, less dense cells. This age-dependent change in lipid asymmetry correlates with the cells' propensity to be cleared from the pe- ripheral circulation and bind to autologous mononuclear cells in vitro.

EXPERIMENTAL PROCEDURES Materials and Routine Procedures-NBD-PC was purchased from

Avanti Polar Lipids (Birmingham, AL). NBD-PS was prepared from NBD-PC by phospholipase D-catalyzed base exchange in the presence of L-serine as previously described (Comfurius et al., 1990). Pyridyldithio- ethylamine was synthesized as previously described (Connor and Schroit, 1988). Factor V and factor X were isolated from bovine plasma and activated as previously described (Connor et al., 1989). Prothrom- bin was from Sigma, and the thrombin-sensitive chromogen S2238 was purchased from Kabi Laboratories (Franklin, OH). Percoll and Ficoll- Hypaque were obtained from Pharmacia LKB Biotechnology Inc. Buffy coats were obtained from the Gulf Coast Regional Blood Bank (Houston, TX). Human RBC were obtained from healthy volunteers by venipunc- ture into heparinized syringes. The blood was diluted with 1 volume of PBS (111.4 m~ NaCl, 10 m~ sodiudpotassium phosphates, 0.5 m~ EDTA, 28 m~ glucose, pH 7.4) and washed twice. The buffy coat was removed after the first centrifugation. Murine RBC were obtained from the tail vein of mice and treated as described for the human cells. Percoll was prepared by mixing 213.5 g of Percoll with 25 ml of 10 x concentrated PBS and water to 250 ml. The pH was adjusted to 7.4 and the osmolarity to 310-320 mosmkg. Steady-state fluorescence of NBD- lipids was measured at a wavelength of 535 nm (excitation 468 nm) with a Farrand MKII spectrophotometer at room temperature using 10-nm slit widths. Radiation was monitored by scintillation counting using a Packard AutoGamma spectrometer. ATP content of RBC was deter- mined using the Sigma Bioluminescent (luciferaselluciferin) assay system.

Density Separation of RBC-RBC were separated by density on self- forming Percoll gradients essentially as described by Lutz et al. (1992). Briefly, 0.5 volume of packed RBC was mixed with 10 volumes of Percoll a t room temperature. The suspension was then centrifuged at 38,700 x g for 20 min. RBC were harvested from the gradients by puncturing the bottom of the tube. Each fraction was further purified by centrifugation in a second Percoll gradient. To determine the density of the various fractions, marker beads of defined density (Pharmacia) were run in a parallel gradient. The cells were washed free of Percoll and diluted to a working concentration of 2 x lo8 celldml in PBS.

Inward Danslocation of NBD-lipids-RBC (2 x 108/ml) were rapidly mixed with NBD-labeled lipid analogs in ethanol (1 pg of lipid10 pl of EtOWml of RBC) and incubated at 37 "C. At the indicated intervals, aliquots were removed and centrifuged through 1% BSA in PBS (back- exchange) to remove lipid not transported to the cell's inner leaflet.' The pellets were solubilized in 1% Triton X-100, and the amount of lipid transported from the outer to inner leaflet was determined by compar- ing the amounts of lipid remaining in the cells before and after back- exchange. The data points were fitted using Slidewrite Plus (Advanced Graphics Software, Carlsbad, CA) exponential curve-fitting routines. Equilibrium distributions at t = m were simulated by calculating the percentage of lipid in the cell's inner leaflet a t t = 10,000 min. Initial rates were estimated from the values calculated at t = 2 min.

Outward Danslocation of NBD-lipids-RBC (2 x 108/ml) were incu- bated with the various lipid analogs (1 pg of lipid10 pl of EtOWml of RBC) for 2 h at 37 "C. The cells were washed once with PBS and incubated with 1% BSA for 5 min to remove any NBD-lipid remaining in the outer leaflet.2 The cells were then extensively washed in PBS to remove all of the BSA. The washed RBC were resuspended to 2 x lo8 celldml and incubated at 37 "C. Aliquots of cells were removed at the indicated intervals, and the fraction of lipid remaining associated with the cells was determined by back-exchange through BSA as described above. The fitted curves were obtained as described above using data points up to and including 5 h. The equilibrium distributions at t = m were simulated at t = 10,000 h.

' The differences observed in the transport of NBD-PS between cells of different density were not due to the presence of ethanol used during lipid loading. Duplicate experiments performed in cells loaded with NBD-PS by lipid exchange from donor NBD-PS vesicles in aqueous buffer produced similar results.

Prothrombinase Complex Assay-RBC (100 pL; 2 x lo8 celldml) were added to 0.5 ml of prothrombinase assay buffer (136 m~ NaC1, 2.7 m~ KCl, 2 m~ MgCl,, 3 m~ CaCI', 10 m~ HEPES, 0.5 mglml human serum albumin, pH 7.4). 50 pl each of factor Va (60 m) and factor Xa (30 m) was then added. The suspension was incubated at 37 "C for 1 min, after which 50 pl of prothrombin (2.0 p ~ ) was added for an additional 10 min. A 100-pl aliquot of the suspension was transferred to a cuvette contain- ing 1 ml of TridEDTA buffer (175 m~ NaCl, 50 m~ Tris, 2 m~ EDTA, pH 7.9) to stop the production of thrombin. The thrombin-dependent chro- mogen S2238 was added to the cuvettes (to 0.4 m ~ ) , and the rate of chromophore formation was monitored at 405 nm with a Gilford Re- sponse Spectrophotometer employing appropriate kinetic software. The initial rate of thrombin-dependent chromophore production, which is directly proportional to the amount of PS present on the cell surface (Connor et al., 1989), was determined from the slope of the absorbance curve. To quantify the amount of endogenous PS exposed on the surface of these cells, a standard curve of the rates of thrombin production/ chromophore formation was generated using pyridyldithioethylamine- treated RBC that contained known amounts of NBD-labeled PS exposed in their outer leaflet (Connor et al., 1989; Utsugi et al., 1991). Thrombin- dependent chromogen production was directly proportional to the amount of PS present in the cell's outer leaflet and linear within the range of interest.

In vivo Clearance of Density-separated Murine RBC-Mouse RBC were incubated with 51Cr for 1 h at 37 "C and washed. Packed cells (0.5 ml) were mixed with 10 ml of Percoll and centrifuged as described above. Fractions were collected and washed extensively, after which lo8 cells (in 200 pl) were injected into the tail vein of C57BU6 mice. The mice were bled 1 h after injection and at the indicated time points. The fraction of Wr-labeled RBC remaining in the peripheral circulation was calculated using the 1-h bleed as the 100% reference point. To determine the organ distribution of the cleared cells, radiation in the liver and spleen was also determined. To examine RBC clearance in antibody-deficient mice, RBC from SCID and BALBIc mice were labeled with 51Cr and injected into both strains as described above.

Preparation of Monocyte Cultures-Buffy coats were diluted with 3 volumes of PBS, layered onto a gradient of Ficoll-Hypaque (1.077 glml), and centrifuged at 730 x g for 25 min. The interface containing mono- nuclear cells was collected, washed, and resuspended to 2 x lo7 celldml in minimal essential medium containing 10% fetal calf serum. Cells (6 x lo7) were plated on 35-mm tissue culture plates (Falcon Plastics, Oxnard, CA) and incubated for 4 h at 37 "C. The plates were then vigorously washed, and lo7 51Cr-labeled autologous Percoll-separated RBC were added for 2 h at 37 "C. The plates were washed to remove unbound RBC. The remaining cells were lysed with 1% Triton X-100, and the number of bound cells was determined by scintillation counting.

RESULTS

Density Fractionation of RBC-Fractionation of human RBC on self-forming Percoll gradients yielded five distinct popula- tions that maintained their density upon recentrifugation through a second gradient (Fig. 1). Table I shows the densities of the different populations, the percentage of total cells in each fraction, and their ATP levels. The densest fractions (Fractions I and 11) represent the smallest population of the total prepa- ration (4.5 and 8.5%, respectively), while fractions of interme- diate densities (Fractions I11 and IV) comprised the majority of the cells, representing about 70% of the total. Table I also shows that, except for the densest fraction, RBC of different densities had similar ATP levels. Mouse cells separated on identical gradients yielded only three populations with densi- ties intermediate to those obtained with human cells (Table I). The lightest fraction, Fraction 111, consisted of 270% of the total RBC preparation.

Pansbilayer Movement of NBD-Lipids-Fig. 2 shows the transport of NBD-PS and NBD-PC from the outer to the inner leaflet in the various cell populations. It can be seen that the rates of NBD-PS transport and the final distribution of the lipid between the inner and outer leaflet were dependent on cell density. The data shown in Table I1 indicate that a direct rela- tionship between cell density, the initial rates of lipid transport, and the final equilibrium distribution of the lipid between membrane leaflets exists. For example, the densest cells (Frac-

Phosphatidylserine in Red Blood Cells 2401 TABLE I

Promrties of red blood cells fractionated from self-forminn

U

>r 1.063.

V) 1.073. .- E Q)

n I .090-

I .098-

1.118-

1 .I 38.

1.07 ' .O.W 4.88 Q.58 .0.32 .o 20

. "

STD I II Ill

Fraction

IV V

FIG. 1. Percoll gradient separation of RBC. 1.5 ml of packed human RBC were mixed with 30 ml of Percoll and centrifuged at 38,700 x g for 20 min at 20 "C. Cells from the five fractions were recentrifuged in individual 10-ml Percoll gradients under the same conditions. A, beads of known density were run in a parallel gradient and their RF values were plotted against their densities (M). B, distribution of RBC populations in the first (30 ml) gradient. The gradient is mounted hori- zontally and the density of the fractions are plotted inA (0). C , fractions from the first gradient were washed and recentrifuged in individual 10-ml gradients together with a standard containing density marker beads (STD).

tion I) transported PS at an initial rate of 1.8 ng of lipid2 x lo7 celldmin, whereas the lightest cells (Fraction V) transported the lipid about twice as fast. The differences in transport rates were consistent with the attained equilibrium distributions where 75 and 90% of the NBD-PS in Fraction I and Fraction V cells was transported to the cell's inner leaflet. Like NBD-PS, the relative movement of NBD-PC also seemed to be dependent on cell density (Fig. 2b). About 8% of the lipid was transported by the densest fraction, whereas the two lightest fractions transported about 20% of the lipid.

Contrary to the rapid inward movement of NBD-PS, the outward movement was -10-fold slower. The attained equilib- rium distribution was, however, like the inward movement, dependent on cell density (Fig. 3u). Table I1 shows that approxi- mately 34,28,26, 25, and 22% of the NBD-PS initially located in the inner leaflet of Fraction I, 11,111, IV, and V cells, respec- tively, moved to the cells'outer leaflet upon incubation at 37 "C.

Percol gradients . . -

Fraction Density T o t a l " ATP

glml 90 nmol lliP cells Human RBC

I 1.112 4.5 7.5 I1 1.105 8.5 8.7 I11 1.100 39.0 Iv

9.8 1.090

V 33.0 9.7

1.081 15.0 9.7

I I1

1.109 6.8 1.100 20.2

111 1.087 73.0

Mouse RBC

cells in each population following the second density separation.

mal preference for the cell's outer leaflet (-80% outside/20% inside) within 5-6 h in all RBC fractions (Fig. 3b).

Since the transbilayer movement of PS is ATP-dependent, it is possible that the different transport activities of the various fractions could be due to different ATP levels (see Table I). The ability of RBC containing decreasing amounts of ATP to trans- port NBD-PS was therefore examined. RBC with different ATP levels were generated by incubating cells at 37 "C in the ab- sence of glucose for increasing times. At various intervals, their ATP levels were measured and their ability to transport NBD-PS was assessed. The results presented in Fig. 4 show that NBD-PS transport was unaltered with a loss of up to -75% of normal ATP levels. These data make it unlikely that the cells' ability to transport lipid was related to the small differences in ATP levels ( ~ 2 5 % ) observed between the various cell populations.

Presence of Endogenous Phosphatidylserine in the Outer Leaflet of RBC and Binding by Autologous Monocytes-Current data support the hypothesis that aged cells display higher amounts of PS in their outer leaflet than young RBC (Her- rmann and Devaw, 1990). Assuming that the density-sepa- rated RBC populations represent cells of increasing age (see below), RBC of increasing density should display more PS. Us- ing an appropriately generated standard curve (see "Experi- mental Procedures"), the amount of PS in the outer leaflet of density-separated RBC was determined (Table 111). It can be seen that, similar to the equilibrium distribution established with NBD-labeled PS, Percoll separated RBC displayed in- creasing amounts of endogenous PS that was dependent on the cells' density. For example, the densest cells (Fraction I) displayed 104 ng of PS/107 cells, whereas the lightest cells (Fraction V) expressed about half as much PS (55 ng of PS/107 RBC).

To determine whether the differences in the presence of PS among the various Percoll-separated cell fractions correlated with monocyte binding, RBC from each Percoll fraction were incubated with autologous monocytes for 2 h at 37 "C. The results presented in Table I11 show a nonlinear relationship between the amount of PS present on the cell's surface (and cell density) and the ability of the cells to become associated with monocytes. These results suggest that a threshold amount of PS must be present for this interaction to occur (see "Discus- sion").

In vivo Clearance of Density-separated Mouse RBC--To esti- mate the relative age of the various RBC populations obtained by Percoll gradient centrifugation, the in vivo lifespan of gra- dient-separated RBC was modeled by examining the in vivo clearance of 51Cr-labeled, density-separated mouse RBC (Fig. 5). The results show that the densest cells (cells from Fraction I) were cleared from the peripheral circulation with a half-time

The percentage of total was determined by measuring the volume of

This effect was not seen with NBD-PC, which attained its nor- of approximately 3 days, whereas lighter cells obtained from

2402 Phosphatidylserine in Red Blood Cells

n

E.

t a

' I b 75

L

50 -

t

0 15 30 45 60

INCUBATION TIME (min)

rated RBC. Density-fractionated RBC were incubated with NBD-la- FIG. 2. Inward translocation of NBD-lipids in deneity-sepa-

beled lipid at 37 "C. The movement of lipid from the outer to inner leaflet was measured as described under "Experimental Procedures." a , NBD-PS; b, NBD-PC. The RBC fractions are indicated as follows; A, Fraction I; 0, Fraction 11; V, Fraction 111; , Fraction W, ., Fraction V.

TABLE I1 Initial transport rates and equilibrium distribution of NBD-PS in

density-separated RBC

Fraction Initial transport rate

~

distribution" Equilibrium

ngl2 x lO'RBClmin W inner leaflet lipid Flip I I1 I11 Iv V

1.8 74.9 2.2 77.6 2.5 83.0 2.9 85.9 3.4 90.3

Flop I I1 I11 Iv V

0.33 0.26 0.25 0.25 0.26

65.9 71.8 73.9 75.4 77.9

Equilibrium distributions were calculated at 10,000 min and 10,000 h for flip and flop, respectively.

Fraction I11 had significantly longer clearance times (Tu2 = 6 days). It therefore appears that RBC density is directly propor- tional to its in vivo clearance rate, suggesting a correlation with cell age. Assuming that the clearance of mouse RBC occurs within 60 days (Momson et al., 1983; Mueller et al., 1987), the Tm of clearance of a normal RBC population with a mean age of 30 days should be about 2 weeks. While the experimentally

a

b

A 70 s I- ! 60 - IA

- v A

A

4 15 100 -I

1

z z 5 80

e3 E - -I 60

40

20

0 0 1 2 3 4 5 6

INCUBATION TIME (hr) FIG. 3. Outward translocation of NBD-lipids in density-differ-

entiated RBC. Density-fractionated RBC were preloaded with NBD- labeled lipid and back-exchanged with 1% BSA to remove lipid not transported to the cell's inner leaflet. Lipid movement from the inner to outer leaflet was assessed as described under "Experimental Proce- dures." a, NBD-PS; b, NBD-PC. The RBC fractions are indicated as follows: A, Fraction I; 0, Fraction 11; V, Fraction 111; , Fraction W, B, Fraction V.

determined half-times are somewhat shorter than theoretically expected, the 2-fold difference in the rates of clearance between Fraction I and I11 cells is a consistent finding.

To examine the organ distribution of the cleared cells, radia- tion in the spleens and livers of mice injected with 51Cr-labeled Fraction I and Fraction I11 RBC was measured (Table IV). On a weight basis, mice injected with dense (old) cells accumulated more W r in their spleens than did mice injected with the lightest (young) cells. The spleediver ratio for mice injected with dense cells ranged from about 7:l at 1 h to 26:l at 5 days. In contrast, the spleediver ratio of mice injected with light cells was only 3:l at 1 h and 5:l at 5 days.

Taken together, these data suggest that cells that express relatively high levels of PS (Fraction I cells) represent a popu- lation of aged cells that are recognized and cleared from the peripheral circulation. Moreover, since the spleen is the pre- dominant organ of clearance of Fraction I cells, it would appear that clearance is mediated by an antibody-independent mecha- nism (Schroit et al., 1985). To verify that in vivo clearance of RBC can indeed be mediated by an antibody-independent mechanism, the kinetics of RBC clearance was determined in immune-deficient SCID mice, which lack both mature T and B cells and have no circulating antibodies (Bosma et al., 1983).

Phosphatidylserine in Red Blood Cells 2403

80

70

60

50

40

30 Incubation Tlme(hn): 0.0 0.5 1.0 1.5 2.0 2.5

20 ATP (nmobllo’cella): 8.1 9.0 8.8 5.9 4.4 2.4

Inltlal Rate (nplmln): 3.8 4.3 3.8 3.4 3.0 3.0

K Inner leaflet (I-): 118 87 108 95 102 85 10

0 15 30 45 60

INCUBATION TIME (min)

FIG. 4. Effect of cellular ATP levels on NBD-PS transport. RBC with different levels of ATP were generated by incubating a normal mixed red cell population at 37 “C in the absence of glucose for 0 (A), 0.5 (+), 1.0 (O), 1.5 (H), 2.0 (+ ), and 2.5 h (V). The cells were then assessed for their ATP levels and ability to transport NBD-PS as described.

TABLE I11 Presence of PS in the outer leaflet of density-separated RBC and

binding to autologous monocytes

Fraction PS” RBC bound

I I1 I11 IV V

~ 1 1 0 7 x IO” 104.0 15.6 81.5 64.9 61.1 54.6

7.2 4.0 2.1 0.9

lished with known amounts of NBD-PS using the prothrombinase com- a The amounts of PS were calculated from a calibration curve estab-

plex assay as described under “Experimental Procedures.”

100

10

1 0 S 16 24 32 40

DAYS

rine RBC, prelabeled with Y k , were separated into three populations FIG. 5. In vivo clearance of density-separated murine RBC. Mu-

on Percoll gradients. The cells were washed and injected into the tail vein of recipient mice. At various times the amount of RBC remaining in the peripheral circulation was determined by counting an aliquot of blood. A, Fraction I cells; 0, Fraction I1 cells; H, Fraction I11 cells.

51Cr-labeled RBC from both SCID and control Balblc strains were injected into the tail vein of mice from each strain. At appropriate intervals the amount of 61Cr-labeled RBC remain- ing in circulation was determined as described in Fig. 5 . Clear- ance rates of SCID RBC injected into SCID or BALB/c mice and BALBlc RBC injected into SCID or BALBlc mice were identical (data not shown), suggesting that antibody-dependent mecha-

TABLE IV Organ distribution of cleared RBC

Time“ Fraction I Fraction 111

Spleen Liver Spleen Liver

61Crlg tissue (cpm) 1 39,270 5,886 28,020 9,015 24 120

40,899 124,400

3,812 4,674 38,696

18,109 8,070 5,499

Percent of inputb 1 0.51 0.84 0.29 1.14 24 1.06 120

1.46 0.21 0.72 3.15 0.93 1.25 0.88

a Post-injection. Mice were injected with lo6 to 2 x lo6 cpm of SICr-labeled RBC.

nisms are not required for the recognition and removal of se- nescent RBC.

DISCUSSION It is well established that the aminophospholipids are asym-

metrically distributed in the plasma membrane of RBC, with most of the PE, and practically all of the PS, residing in the cell’s inner leaflet. This asymmetry is maintained by the activ- ity of the aminophospholipid translocase that specifically transports PS and PE from the outer to inner leaflet of the plasma membrane. Since this inward movement counteracts any outward movement, a major function of the translocase may be to prevent the exposure of PS in the cell’s outer leaflet. This suggests that the maintenance of phospholipid asymmetry is an important component of hemostasis. Indeed, defects in the normal asymmetric distribution of PS result in the expression of altered membrane surface properties that are of physiologi- cal significance. A prominent example of this is the role of PS in hemostasis and thrombosis of activated platelets (Bevers et al., 1983; Zwaal et al., 1989).

Besides the important procoagulant role of PS in platelet activation and in the initiation of thrombi, perturbations in the normal asymmetric distribution of PS have been implicated in the recognition and clearance of PS-expressing cells by reticu- loendothelial cells. It has been shown, for example, that deoxy- genation of sickle RBC leads to increased exposure of PS (Allan et al., 1982; Franck e t al., 19851, which, while providing a procoagulant catalytic surface, also coincides with their bind- ing and subsequent endocytosis by macrophagelmonocytes (Schwartz et al., 1985). Other studies have shown that undif- ferentiated erythroleukemic cells express relatively high levels of PS that disappears from the cell surface upon their differ- entiation to a normal, non-tumorigenic phenotype, a phenom- enon that is directly related to their binding by macrophages in vitro (Connor et al., 1989). Similar experiments revealed that human tumorigenic cell lines that are bound by macrophages displayed 3-7 times more PS in their outer leaflet than did normal human epidermal keratinocytes that are not bound by macrophages (Utsugi et al., 1991). Collectively, these data sug- gest that the expression of PS on the outer leaflet of cells, due to alterations in normal membrane lipid asymmetry, plays a role in the recognition and subsequent removal of RBC by the RES (Allen et al., 1988).

While the relationship between red cell age and density is still controversial (Morrison et al., 1983; Dale and Norenberg, 19901, the view has been upheld that the aging red cell is removed from the circulation as a result of a series of progres- sive events accompanied by a concomitant increase in cell den- sity (Piomelli and Seaman, 1993). Since the relative rates of bidirectional lipid movement across the membrane contributes to the equilibrium distribution of individual phospholipid spe- cies between bilayer leaflets, we examined both the presence of

2404 Phosphatidylserine in Red Blood Cells

endogenous PS in the cells’outer leaflet and the rates of inward and outward movement of an exogenously supplied NBD-la- beled analog of PS. RBC of the different density showed distinct differences in the presence of PS in the cells’ outer leaflet and in the rates of NBD-labeled PS transport. The densest cells displayed decreased inward movement of NBD-PS, in conjunc- tion with increased rates of outward movement, which resulted in the increased presence of NBD-PS in the cells’outer leaflet at equilibrium. Similarly, measurements of endogenous PS re- vealed increased exposure of PS on the surface membrane of denser cells. These data suggest that a direct relationship ex- ists between cell age and the expression of PS on the outer membrane leaflet.

It has previously been shown that the presence of PS in the outer leaflet of cells results in their endocytosis by phagocytes (Schroit et al., 1985; Allen et al., 1988; Fadok et al., 1992a, 1992b). These results led to the suggestion that recognition of senescent RBC by the RES might involve an age-dependent alteration in membrane lipid asymmetry. The data presented here show a relationship between the amount of PS present on the surface of density-separated RBC and their recognition by autologous monocytes. This can be seen from the data pre- sented in Table 111, where a 2-fold increase in the amount of PS present on Fraction V cells (55 ng) and Fraction I cells (104 ng) was accompanied by a 17-fold increase in binding. This appar- ent exponential increase suggests that a critical amount of PS must be present on the cell surface for this interaction to occur. This observation is consistent with the results using mouse RBC as a model, where age/density-related appearance of PS on the cell surface was accompanied by decreased circulation time and accumulation of the cells in the spleen (Fig. 5 and Table IV) and our previous results on the inability of RBC containing subthreshold amounts of PS to be cleared in vivo (Schroit et al., 1985).

Although antibody-dependent mechanisms of senescent cell clearance have been described (Kay, 1978; Kay et al., 19841, several lines of evidence suggest that the clearance of denselold cells can be mediated by antibody-independent mechanisms. First, our data show clearance principally to the spleen, whereas the liver is the primary organ for the removal of op- sonized RBC (Schroit et al., 1985). Second, the rate of dense RBC clearance in antibody-deficient SCID mice was no differ- ent than the clearance rate in control BALB/c mice. These results suggest that while antibody-mediated clearance of RBC does occur, other mechanisms of clearance also participate in the removal of senescent RBC.

Taken together, the data presented here suggest that a den- sity/age-dependent rearrangement in membrane lipid asymme- try occurs in RBC. Although the mechanism by which mem- brane lipid rearrangement occurs is not known, the appearance of PS on the cells outer leaflet could be the result of: 1) de- creased aminophospholipid translocase-mediated inward movement or 2) increased active or passive PS outward move- ment with a fully active translocase. Either of these mecha- nisms could prevent the translocase from establishing its “nor- mal” equilibrium.

Acknowledgments-We thank Karen Gillum, Barbara Hunt, and Mary Lloyd for technical assistance and Dr. J. Killion for many stimu- lating discussions and thorough review of the manuscript.

REFERENCES

Allan, D., Limbrick, A. R., Thomas, P., and Westerman, M. P. (1982) Nature 296,

Allen, T., Williamson, P., and Schlegel, R. A. (1988) Proc. Natl. Acad. Sci. U. S. A.

Aminoff, D. (1988) Blood Cells 14,229-247 Bartosz, G. (1981) Biochim. Biophys. Acta 644,69-73 Bartosz, G. (1990) Blood Cell Biochemistry: Erythroid Cells (Harris, J. R., ed) Vol.

Bartosz, G. (1991) Gerontology 37,3347 1, pp. 45-79, Plenum Press, New York

Bartosz, G., Grzelinska, E., and Bartkowiak, A,, (1984) Mech. Ageing Deu. 24,

Bevers, E. M., Comfurius, P., and Zwaal, R. F, A. (1983) Biochim. Biophys. Acta

Blumenfeld, N., Zachowski, A,, Galacteros, F., Beuzard, Y., and Devaux, P. F. (1991)

Bosma, G. C., Custer, R. C., and Bosma, M. J. (1983) Nature 301,527-530 Clark, M. R., Mohandas, N., and Shohet S . (1983) Blood 61,899-910 Comfurius, P., Bevers, E. M., and Zwaal, R. F. A. (1990) J. Lipid Res. 31, 1719-

Connor, J., and Schroit, A. J. (1988) Biochemistry 27, 848451 Connor, J., Bucana, C. D., Fidler, I. J., and Schroit, A. J. (1989) Proc. Natl. Acad.

Connor, J., Pak, C. H., Zwaal, R. F. A., and Schmit, A. J. (1992) J. Biol. Chem. 267,

Dale, G. L., and Norenberg, S . L. (1990) Biochim. Biophys. Acta 1036, 183-187 Dhermy, D., Simeon, J., Wautier, M. P., Boivin, P., and Wautier, J. L. (1987)

Dumaswala, U. J., and Greenwalt, T. J. (1984) Dunsfusion 24, 490-492 Fadok, V. A,, Voelker, D. R., Campbell, P. A,, Cohen, J. J., Bratton, D. L., and

Fadok, V. A,, Savill, J. S. , Haslett, C., Bratton, D. L., Doherty, D. E., Campbell, P. Henson, P. M. (1992a) J. Zmmunol. 148,2207-2216

Franck, P. F. H., Bevers, E. M., Lubin, B. H., Comfurius, P., Chiu, D. T.-Y., Op den A,, and Henson, P. M. (1992b) J. Zmmunol. 149,4029-4035

Kamp, J. A. F., Zwaal, R. F. A., van Deenen, L. L. M., and Roelofsen, B. (1985) J. Clin. Znuest. 75, 183-190

612-613

85,80674071

1-7

736,5746

Blood 77,849-854

1721

Sci U. S. A. 86, 31843188

19412-19417

Biochim. Biophys. Acta 904,201-206

Galili, U., Clark, M. R., and Shohet, S . B. (1986) J. Clin. Znuest. 77, 27-33 Greenwalt, T. J., and Dumaswala, U. J. (1988) BE J. Haematol. 68,465-467 Hentschel, W. M., Wu, L. L., Tbbin, G. O., Anstall, H. B., Smith, J. B., Williams, R.

Hemnann, A,, and Devaux, P. F. (1990) Biochim. Biophys. Acta 1027,4143 Jain, S . K. (1988) Biochim. Biophys. Acta 937, 205-210 Jain, S . K., Mohandas, N., Clark, M. R., and Shohet, S . B. (1983) BE J. Huemotol.

Kay, M. M. B. (1978) J. Suprumol. Struct. 9,555-567 Kay, M. M. B., Goodman, S . , Whitfield, C., Wong, P., Zaki, L., and Rudloff, V. (1984)

Linderkamp, O., and Meiselman, H. J., (1982) Blood 59, 1121-1127 Lutz, H. U. (1990) Blood Cell Biochemistry: Erythroid Cells (Hams, J. R., ed) Vol.

1, pp. 81-120, Plenum Press, New York Lutz, H. U., Bussolino, F., Flepp, R., Fasler, S. , Stammler, P., Kazatchkine, M. D.

and Arese, P. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 7368-7372 Lutz, H. U., Fasler, S . , Stammler, P., Bussolino, F., and Arese, P. (1988) Blood Cells

14, 175-195 Lutz, H. U., Stammler, P., Fasler, S. , Ingold, M., and Fehr, J. (1992) Biochim.

Biophys. Acta 1116, 1-10 Momson, M., Jackson, C. W., Mueller, T. J., Huang, T., Dockter, M. E., Walker, W.

S., Singer, J. A,, and Edwards, H. H. (1983) Biomed. Biochim. Acta 42,5107-

Mueller, T. J., Jackson, C. W., Dockter, M. E., and Momson, M. (1987) J. Clin. Slll

Nash, G. B., and Meiselman, H. J. (1983) Biophys. J. 43,63-73 Znuest. 79,492499

Nash, G. B., and Wyard, S . J. (1980) Biorheology 17,479484 Op den Kamp, J. A. F. (1979)Annu. Rev. Biochem. 48,47-71 Piomelli, S. , and Seaman, C. (1993) Am. J. Hematol. 42, 4G52 Rifkind, M., Araki, K., and Hadley, E. C. (1983) Arch. Biochem. Biophys. 222,

Schlepper-Schafer, J., Kolb-Bachofen, V., and Kolb, H. (1983) Biochem. Biophys.

Schroit, A. J., Madsen, J. W., and Tanaka, Y. (1985) J. Biol. Chem. 260,5131-5138 Schwartz, R. S., Tanaka, Y., Fidler, I. J., Chui, D. T.-Y., Lubin, B., and Schroit. A.

Utsugi, T., Schmit, A. J., Connor, J., Bucana, C. D., and Fidler, I. J. (1991) Cancer

van Oss, C. J. (1982) Biorheology 19, 725 Verkleij, A. J., Zwaal, R. F. A,, Roelofsen, B., Op den Kamp, J. A. F. and van Deenen,

Zwaal, R. F. A,, Bevers, E. M., Comfurius, P., Rosing, J., Tilly, R. H. J., and

R., and Ash, K. 0. (1986) Clin. Chin. Acta 157,3344

53,247-254

Pm. Natl. Acad. Sci. U. S. A. 80, 1631-1635

582589

Res. Commun. 115,551-559

J. (1985) J. Clin. Znuest. 75, 1965-1972

Res. 51,3062-3066

L. L. M. (1973) Biochim. Biophys. Acta 323, 178-193

Verhallen, P. F. J. (1989) Mol. Cell. Biochem. 91, 23-31