From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
Detection of Altered Membrane Phospholipid Asymmetry in Subpopulations
of Human Red Blood Cells Using Fluorescently Labeled Annexin V
By Frans A. Kuypers, Rachel A. Lewis, Mandy Hua, Mary Ann Schott, Dennis Discher, Joel D. Ernst,
and Bertram H. Lubin
The phospholipids of the human red cell
are distributed
asymmetrically in the bilayer of the red cell membrane. In
certain pathologic states, such as sickle cell anemia, phospholipid asymmetry is altered. Although several methods
can be used t o measure phospholipid organization, small
organizational changes have been very difficult t o assess.
Moreover, these methods fail t o identify subpopulations of
cells that have lost their normal phospholipid asymmetry.
V inflow cytometry and
Using fluorescently labeled annexin
fluorescent microscopy, we were able t o identify andquantify red cells that had lost their phospholipid asymmetry in
populations as small as 1 million cells. Moreover, loss of
phospholipid organization in subpopulations as small as
0.1% of the total population could identified,
be
and individual cells could be studied by fluorescent microscopy. An excellentcorrelation was found between fluorescence-activated cell sorter (FACS) analysis results using annexin V t o
detect red cells with phosphatidylserine (PS) on theirsurface
and a PS-requiring prothrombinase assay using similar red
cells. Cells that bound fluorescein isothiocyanate (FITC)-lausing
beled annexinV could be isolated from the population
magnetic beads covered with an anti-FITC antibody. Evaluation of blood samples from patients with sickle cell anemia
under oxygenated conditions demonstratedthe presence of
subpopulations of cells that had lost
phospholipid asymmetry. While only a few redcells were labeled in normal control
samples (0.21% f 0.12%. n = 81, significantly increased (P
c .001) annexin V labeling was observed in samples from
patients with sickle cell anemia (2.18% f 1.2190, n = 13). We
conclude that loss of phospholipid asymmetry may occur
in small subpopulations of red cells and that fluorescently
labeled annexinV can be used t o quantify and isolate these
cells.
0 1996 by The American Societyof Hematology.
T
where the aminophospholipid translocase is severely impaired, such as during storage of redblood cells (RBC):
endogenous PS molecular species moveonlyvery slowly
from the inner to the outer leaflet, and noequilibration across
the bilayer is observed. Loss of asymmetry of endogenous
PS becomes apparent onlywhen the aminophospholipid
translocase activity is impaired and the membrane phospholipid organization is scrambled by other means as well, as
is seen when RBCs are incubated with calcium and ionophore.”
Loss of normal phospholipid asymmetry has been implicated in red cell pathology in diseases such as sickle cell
anemia.” Impaired aminophospholipid translocase activity
can be measured in sickle cells using exogenous phospholipid probe molecules,’2 and in vitro manipulations of these
cells indicate loss of endogenous phospholipid asymmetry.”
However, it has been difficult to unambiguously show the
loss of phospholipid asymmetry in vivo. This difficulty could
be in part due to the methods used to measure endogenous
phospholipid organization. With the exception of a PS-in-
HE ASYMMETRIC distribution of phospholipids
in membranes was postulated 20 years ago by
Bretscher.l>*Phospholipid asymmetry has since been demonstrated in a wide variety of cell types. Techniques used to
distinguish between the two membrane leaflets and between
headgroups of the different membrane phospholipid molecular species include the use of chemical reagents or phospholipases to specifically alter the endogenous phospholipids of
the outer monolayer, strategies to monitor the redistribution
of exogenously added phospholipid probe molecules across
the bilayer, and the activation of hemostatic processes such
as the phosphatidylserine (PS)-mediated activation of prothrombin to thrombin by prothr~mbinase.~.~
Application of these techniques to normal erythrocyte
membranes shows that the choline-containing phospholipids
phosphatidylcholine (PC) and sphingomyelin (SM) are localized mainly in the outer monolayer of the erythrocyte membrane, whereas the aminophospholipids phosphatidyl-ethanolamine (PE) and PS are mainly PE or exclusively PS found
in the inner monolayer. Although phospholipid asymmetry
is essentially conserved throughout the life span of the cell,
phospholipids are actually in a state of dynamic equilibrium
between the two monolayers. The net equilibrium is a balance of two mechanisms. First, all phospholipids diffuse
passively across the bilayer at a relatively slow rate. Second,
aminophospholipids are actively transported from the outer
monolayer to the inner monolayer by a Mg2+-adenosinetriphosphate (ATP)-dependent aminophospholipid translocase,
and their distribution at equilibrium affects that of the other
phospholipids.’
The rate of active transport of aminophospholipids from
outer to inner monolayer (flip) is significantly higher than
the rate from inner to outer monolayer (flop). It appears, in
particular, that PS present in the outer monolayer is rapidly
transported to the inner monolayer by the aminophospholipid
translocase. Although an inactivated aminophospholipid
translocase could be expected to lead to a loss of asymmetry
due to passive diffusion, driven by a concentration gradient,
this is apparently a slow process. Even under conditions
Blood, Vol 87,No 3 (February I), 1996:pp 1179-1187
From the Children’s Hospital Oakland Research Institute, Oakland; the University of California San Francisco, SFGH, San Francisco; and the Lawrence Berkeley Laboratory, Berkeley, CA.
Submitted November 17, 1994; accepted September 1 l , 1995.
Supported by National Institutes of Health Grants No. DK32094,
HL20985, HL55213, and HL27059 and a grant from the American
Heart Association (California afiliate). This work was pet$ormed
during the tenure of an established investigatorship of the American
Heart Association (J.D.E.).
Address reprint requests to Frans A. Kuypers, PhD, Children’s
Hospital Oakland Research Institute, 747 52nd St, Oakland, CA
94609.
The publication costs of this article were defrayed in part by page
charge payment. This article must therefore be hereby marked
“advertisement” in accordance with 18 U.S.C.section 1734 solely to
indicate this fact.
0 I996 by The American Society of Hematology.
OOO6-4971/96/8703-0003$3.00/0
1179
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
1180
KUYPERS ET AL
duced prothrombinase activity assay or the use of radioactively labeled annexin V,'? current assays to measure phospholipid sidednessdo
not permitassessment
of small
changes in asymmetry.The use of most traditionaltechniques formeasurement of PS asymmetryisalso limited
by hemolysis. Even limited levels of hemolysis will greatly
interfere with the interpretation of the data, because the resulting exposure of the inner monolayer to the probes will
affect the data collected. Furthermore, as all of these measurementsdescribetheaverage
phospholipidorganization
in a population of red cells, noneof themexamines the
phospholipid organization ina population of red cells, none
of them examines thephospholipid asymmetry in small subpopulations of a red cell sample. Even a complete loss of
asymmetry in small subpopulations of a sample might not
be detected using the methods described to date. This
would,
however, be feasible with a probe that can assess phospholipid organization in individual cells.
Annexins are a family of proteins that bind to acidicphospholipids,
particularly
These proteins are
found
in
many organisms, and although extensive
in vitro studies have
described their calcium-dependent binding to phospholipids,
the exact biologic function of the annexins isnotyet
known.'' Fluorescently labeled annexins have been used to
identify PS on the surface
of activated platelets under physiologic conditions in vitro as well as on the surfaceof vesicles
released from activated platelets'' and apoptotic cells." Radioactively labeled annexin V was used toidentify PS exposure in red cells." However, bindingof radiolabeled annexin
V only indicates the loss of phospholipid asymmetry in the
average cell anddoes notallow analysis of the redcell
population. In the present study, we show that fluorescently
labeled annexin V used in flow cytometry and fluorescent
microscopy offers an excellent method for the determination
of loss of normalphospholipid asymmetryin subpopulations
or individual redcells. Using this technique,we have demonstrated that under certain in vitro conditions known to scramblemembrane phospholipids, or in pathologic states such
as sicklecellanemia,loss
of phospholipidasymmetryis
heterogeneous and may beconfined to subpopulations of red
cells.
MATERIALS AND METHODS
Erythrocytes. Human erythrocyte suspensions were prepared
from fresh human venous blood collected in heparin or EDTA after
informed consent was obtained from laboratory volunteers or patients with sickle cell anemia. Erythrocytes were pelleted by centrifugation, washed twice with 0.9% NaCl and once with incubation
buffer, and finally diluted in incubation buffer to the appropriate
hematocrit. Either Hanks' buffered salt solution, pH 7.4 (HBSS;
Sigma, St Louis, MO), or 10 mmol/L Tris/HCI buffered saline, pH
7.4 (TBS), was used as buffer throughout the experiments; similar
results were found with both buffers. Additional ingredients, all of
reagent quality, such as CaCI2 and N-ethyl maleimide (NEM), were
added as indicated.
NEM treatment of RBCs. NEM inhibits the aminophospholipid
translocase8 by complexing a sulfhydryl group necessary for its activity. RBCs at 30% hematocrit were incubated in buffer containing
10 mmoVL NEM (Sigma) for 30 minutes at room temperature and
were subsequently washed in buffer without NEM.
Calcium and ionophore treatment of RBCs. Calcium and ionophore treatment will induce membrane lipid scrambling.1x~21
RBCs
at a 16% hematocrit were equilibrated in incubation buffer with
1 mmol/L calcium for3 minutes at37°C. Subsequently, calcium
ionophore A23187 was added to the RBC suspension to a final
concentration of 4 pmol/L,unless otherwise noted. This final mixture
was incubated for 1 hour. Because incubation of RBCs with calcium
and ionophore will result in a number of cellular changes, including
ATP depletion and vesiculation,22,z3cells will ultimately hemolyse
under these conditions. The process was stopped by a washwith
2.5 mmol/L EDTA to remove calcium. Subsequently, the cells were
washed three times in buffer containing 1 % bovine serum albumin
(BSA) to remove the ionophore and were resuspended in buffer
without BSA to prevent calcium uptake during annexin V-labeling
of the cells (see below). This removal of ionophore was crucial, as
omission and subsequent incubation in the 1.2 mmol/L calcium
buffer used for annexin V labeling led to massive hemolysis. In
some experiments, aliquots of the suspension were removed at designated time points and treated with EDTA andBSA as indicated
above.
Annexin V purijication and juorescein isothiocyanate (FITC) labeling. Annexin V was initially purified from fresh human placentaZ4using a modification of the method of Haigler et al." Currently, recombinant annexin V is purified from an Escherichia coli
expression system by phospholipid affinity chromatography.2hBoth
preparations give similar results.
Purified annexin V is incubated in 50 mmolL Borate buffer, pH
9.0, at 4°C for 16 hours in the dark at a final concentration of 1 mg
annexin V per milliliter in the presence of 20 molar equivalents of
FITC (Molecular Probes, Eugene, OR). Subsequently, unreacted
FITC is removed by incubation with 1.0 m o m Tris/HCl, pH 8.0, and
filtration on a PD-IO Sephadex G-25 column (Pharmacia, Uppsala,
Sweden). The heterogeneously labeled annexin V species are separated by FPLC, and the brightly labeled fractions are ~ollected.~'
The preparation used in our studies had an average of three FITC
molecules per protein molecule. The total fluorescence in solutions
containing FITC-annexin V was measured in a Perkin Elmer LS5B
Luminescence spectrometer (Perkin Elmer, Beaconsfield, Buckinghamshire, UK).
Annexin V labeling of RBCs. RBCs were suspended in buffer
to a final concentration of 4 X 10' mL. Unless noted otherwise, 4
pL of a 500 pmoVL FITC-labeled annexin V solution was added to
0.5 mL of this suspension in the presence of 1.2 mmolL Ca2+.The
samples were incubated for 30 minutes at room temperature and
subsequently washed with buffer to remove unbound annexin V.
The labeled cells were resuspended to approximately 10' cells per
250 pL in buffer with 1.2 mmolL calcium for flow cytometric and
microscopic analysis.
Magnetic cell separation.
Magnetic beads (average size, 15 nm)
covered with an anti-FITC antibody were supplied by Miltenyi Biotec Inc (Auburn, CA). The stock solution of beads was fivefold
diluted in annexin V labeling buffer. Red cells labeled with FITCannexin V were washed, and 6 X IO' cells were taken up in 80 pL
of buffer. To the cell suspension, 20 pL of the diluted beads was
added. After a 10-minute incubation at room temperature, the cells
were separated in a magnetic separation setup (Minimac; Miltenyi
Biotec Inc) according to the standard protocol supplied by the manufacturer.
Flow cytometly. Samples were analyzed on a Becton Dickinson
FACScan flow cytometer (Becton Dickinson, San Jose, CA). Acquisition and data analysis were performed using LYSYS 11 software
(Becton Dickinson). Ten thousand events per sample were acquired
to ensure adequate mean fluorescence levels. The light scatter and
fluorescence channels were set at a logarithmic gain. The forward
angle light scatter setting was E-l with a threshold of 36. The red
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
RBC PS ASYMMETRYANNEXlN
AND FLUORESCENT
V
1181
1
10
1w
1wo
10
'
1
10
100
1wo
10'
100
1wo
10
'
l
10
1w
1wo
10
350
-
250
Fig 1. Flowcytometricanalysis of (AI RBCs incubated with fluorescentlylabeledanti-glycophorin A,
zw
(B) RBCs incubated with FITC-labeled annexinV, (Cl
RBCs incubated with FITC-labeledannexin V after
E
treatment with NEM,and (D) RBCs incubated with
1w
FITC-labeled annexin V after treatment with NEM
for 1 hour in 1 mmollL
5o
andsubsequentincubation
calciumin the presence of 4 pmollL ionophore
A23187. RBCs were selected by their light scattering
properties. The number of cells with fluorescence
1
Fluoresmnce
above background was
defined by gateM1.
8
cell population was defined by size in forward and side scatter plots.
Events that correlated with intact RBCs were analyzed for fluorescence intensity. Mean fluorescence intensities were expressed in
linear mode, and positive fluorescence was defined by comparison
with unlabeled control samples.
The treatment of RBCs with NEM followed by incubation with
calcium and ionophore slightly changed the forward scatter characteristics of the population in the flow cytometer as compared with
control RBCs (not shown). Gated regions were adjusted accordingly.
This shift in characteristics can be expected because this treatment
will result in a number of changes in the cell, including the severe
alterations of morphology and loss of membrane material by vesiculati~n.'~
As expected, RBCs do not show any appreciable fluorescence in fluorescence-activated cell sorter (FACS) analysis in the
absence of an added fluorophore (not shown). Incubation of normal
cells with fluorescently labeled anti-glycophorin A antibody (Becton
Dickinson) showed a clear labeling of the population (Fig IA). In
contrast, these cells were not labeled by FITC-annexin V inthe
presence of 1.2 mmol/L calcium (Fig IB). Treatment of RBCs with
NEM followed by incubation with calcium and ionophore resulted
in the binding of FITC-labeled annexin V to the cells (Fig ID). A
negative control, such as that shown in Fig IB, was used to define
background fluorescence for each assay. The population of cells
significantly labeled with annexin V above background is indicated
by marker MI (Fig 1). For example, 98% of the cells as shown in
Fig ID meet this criterion.
Fluorescence microscopy. Cells were examined under 300 X
optical magnification using a liquid N2-cooled CCD camera as described.28This system allowed accurate generation of digital images
of cells at low fluorescence levels with a minimum of background
noise. The microscope objective is used with a fluorescein filter set
20 nm). Bright fields
(excitation, 480 ? 15 nm; emission, 535
(413 ? 5 nm) were exposed for 10 msec; fluorescent fields, for 600
msec.
Prorhrornbinase assay. The activation of prothrombin to throm-
10
'
Fluotsscence
bin is a measure of the amount of PS available to the prothrombinase
complex.6 To measure the availability of PS in the outer monolayer
of the membrane, RBCs were suspended in buffer to a final concentration of 107/mLin Tris buffer (50 mmoYL Tris-HCI, 120 mmol/L
NaCl, 6 mmovL CaCI,, pH 7.4). The mixture was preincubated at
37T, and 0.33 U/mL of bovine factor V and bovine factor Xa
(Enzyme Research Laboratories, South Lafayette, IN) were added,
followed by 0.13 mg ofbovine prothrombin (Enzyme Research Laboratories). EDTA (to a final concentration of 15 mmoVL) was added
to stop the reaction after 4 minutes. The RBCs were pelleted by
centrifugation, and the amount of thrombin formed was determined
by adding an aliquot (150 FL) of the supernatant to 2 mL of a
100 pmollL solution of the chromogenic thrombin substrate S-2238
(Chromogenix, Molndal, Sweden). The increase in absorbance at
405 nm, 37°C. was determined in a Spectronic 3000 Diode Array
Spectrometer (SLM Instruments, Rochester, NY) equipped with enzyme kinetics software. The conversion rate of the chromogenic
thrombin substrate was used to determine the amount of thrombin
formed.
RESULTS
A recent study demonstrated that radioactive annexin V
can be used as a probe to detect red cells that have lost
phospholipid a~ymmetry.'~
However, neither this study nor
any of the other methods used to date have been able to
determine if loss of phospholipid asymmetry occurs in all
cells or if it is localized to subpopulations of cells. Given
the observation that activated platelets that contain PS on
their surface can be identified and quantitated in vivo using
fluorescently labeled annexin V,'6 we determined if this
method could also be used to identify subpopulations of red
cells that had lost phospholipid asymmetry by flow cytometry (FACS). The red cell population was defined by size in
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
KUYPERS ET AL
1182
As elevated intracellular calcium can inhibit the aminophospholipidtranslocase
as well asscramblemembrane
phospholipids,
we
investigated
the effectsof calcium loading
,/'
on
phospholipid
asymmetry
over
a l-hour time period (Fig
/*'
2). The scrambling effectof calcium and ionophoreon phospholipid asymmetry is a gradual process. At 1 hour, a plateau
seems to be reached, as only a few more cells were labeled.
Also, the average fluorescence of the cells did not significantly increase (notshown). Prolonged incubation, however,
led to substantialhemolysis. Datafromexperimentsdescribed above with NEM-treated cells were obtained after l
hour with calcium and ionophore. A substantial decrease of
this time resulted in a lower level of annexin V binding.
/
To ensure optimal labeling of the cells with annexin V,
0
1
"
.
1
"
.
1
~
'
~
1
.
.
~
0
15
30
45
60
we incubated the cells with different amounts of annexin V
and calcium. The same sample of scrambled cells showed
incubation Time (minutes)
an increase in labeling dependent on the annexin V concenFig 2. The binding of annexin V to cells incubated with calcium
tration (Fig 3). At
the concentrations routinely used,a plateau
and ionophoreA23187. RBCs were incubated in1 mmollL calcium in
is reachedin the percentage of fluorescentlylabeled cells
the presence of 4 pmol/L ionophore A23187. The percentage of cells
(Fig3A). Moreover, the average fluorescenceper cell did
that bind FITC-labeled annexin V is plotted against time.
not increase,indicatingthatindividual
cellsdonot label
more heavily with annexin V at higher concentrations. Maximum labeling was reached within 10 minutes after the addiforward and side scatter plots. Events that correlated with
tion of annexin V. Prolonged incubation with an excess of
intact RBCs were analyzed for fluorescence intensity. RBCs
fluorescently labeled annexin V for up to 12 hours at 4°C
do not show any appreciable fluorescence in FACS analysis
did not change the labeling pattern. For practical reasons, a
in the absence of an added fluorophore (not shown). Incuba30-minute incubation time was chosen to standardize labeltion of normal red cells with fluorescently labeled anti-glycoing conditions and ensure optimal labeling. The binding of
phorin A antibody (Becton Dickinson) showeda clear labelFITC-annexin V to scrambled cells was further monitored
ing of the population (Fig IA). In contrast, these cells were
by measuring the fluorescence in the supernatant before and
not labeled by FTTC-annexin V in the presence of 1.2 mmol/
after
the labeling incubation. Under routine conditions at 2
L calcium (Fig 1 B). We hypothesized that loss of the asympmol/L FITC-annexin V, more than 95% of the label can
metric distribution of RBC membrane phospholipids would
of 2 X IO6 red
be recovered in themediumafterlabeling
lead to the binding of FITC-annexin V to these RBCs. As a
cells (Fig 3B), indicating an excess of label. Only when the
first step, to characterize the binding properties of annexin
concentration of thecells is substantiallyincreasedcana
V to red cellmembranes that have losttheir asymmetric
decrease in free, unbound annexin V be observed. Based on
phospholipiddistribution,
we combinedinhibition of the
the number of cells in the labelingmixture andthe concentraaminophospholipidtranslocasewithacceleration
of the
movement of aminophospholipids from inner to outer mono- tion of annexin V in the supernatant after labeling, an estimate can be made for the number of annexin V molecules
layer. Treatment of RBCs with I O mmol/LNEM, which
blocks the activity of the aminophospholipid translocase, did
not lead to labeling of the cells with annexin V (Fig 1C).
Moreover, when NEM-treated cells were incubated for 20
hours at 37"C, only 2%of the cellslabeled above background
(not shown). Incubation of NEM-treated RBCs with ionophore in the absence of calcium also did not result in any
annexin V binding (notshown). In contrast,treatment of
RBCs with NEM followed by incubation with calcium and
ionophore resulted in the binding of FITC-labeled annexin
V to the cells(Fig ID). These conditions were chosen
as they
had previously been reported to cause loss of phospholipid
asymmetry. The frequencydistribution curves ofthe flow
10'1 0 '
10'
10.1
10.
10'
2' 10'
1.6'108
cytometric data contain information on the number of cells
Annexin
FITC
(PM)
perCells
250 v1
labeled above background and average fluorescence per cell.
/'
cells.
in
the
250 p L in
presence of 1 pmollL annexin V.
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
1183
RBC PS ASYMMETRY AND FLUORESCENT ANNEXIN V
350
100
300
250
r
0
U
150
i.!¶
l00
8
200
.k
6o
40
c
2
20
"
50
0
1
10
101000
100
'
Fluorescence
0
20
40
60
80
100
Scrambled cells in population ("h)
Fig 4. Flow cytometric analysis of FITC-labeled annexin
V binding to a mixture of cells inwhich part of the population exhibited a scrambled
membrane induced by treatment with NEM and incubation for 1 hour in 1 mmol/L calcium in the presence of 4 pmol/L ionophore A23187.
[A) A typical analysis of such apopulationin which 107- of the cells showed fluorescence above background, as
determined by the percentage
of cells gated by M l . (B) Relationship between the percentage of the population treated with calcium and ionophore andthe percentage of
red cells that bind FITC-labeled annexinV (01,as determined by flow cytometry as indicated in panel A, or the prothrombin-converting activity
(0)
of the red cells expressed as percentageof thrombin formed per minute relative to scrambled cells incubatedat 37°C in buffer containing
calcium, purifiedbovine prothrombin, and factors
V and Xa. Both data indicate
the average and standard deviation of four separate experiments.
bound per cell. For the average cell after ionophore treatment, approximately 300,000 molecules of FWC-annexin V
are bound per cell.
The concentration of calcium in the labeling buffer affected the binding of annexin V. Labeling was tested on
fresh control RBCs as well as a cell population in which
94% of the cells could be labeled with annexin V under
conditions as indicated in Fig 3 in the presence of 1.2 mmoV
L calcium. At calcium concentrations below 0.5 mmol/L,
the number of cells labeled with annexin V significantly
decreased below 90% (not shown). From 1 to 2 mmoVL of
calcium, the number of cells or the average fluorescence per
cell did not increase. Also, up to 2 mmol/L calcium, no
increase in the number of control RBCs that were labeled
with annexin V was found.
To show that annexin V labeling could be used to selectively identify those RBCs in a population that have lost
their normal phospholipid asymmetry, wemixednormal
RBCs with NEM and calciudionophore-treated RBCs before the addition of fluorescent annexin V. Figure 4A shows
a typical FACScan of a population in which 10% of the cells
have a scrambled membrane phospholipid organization. A
linear relationship was found between the fraction of scrambled cells in the mixture and the percentage identified as
such by FACScan (Fig 4B). We found that subpopulations
of abnormal cells as small as 0.1% of the total population
were detectable with annexin V labeling.
To determine if annexin V binding results correlated with
another measure of loss of phospholipid asymmetry, we
compared the annexin V labeling of these cell mixtures with
the activation of the prothrombinase complex, a well-known
assay for the presence of PS in a membrane surface.6 In this
case, PS was provided by the red cell membrane. More
thrombin was formed as a function of the number of scrambled cells in the mixture (Fig 4B). This increase in prothrombinase activity indicates, independently of the annexin V
labeling, the presence of PS in the outer monolayer of the
red ce11.6An excellent correlation between the number of
cells detected by annexin V in the mixture and the procoagulant activity of the mixture was observed. To illustrate the
power of the annexin-based assay, the interpretation of the
results using the procoagulant measurement would be that a
10% loss of asymmetry in the red cell mixture was noted.
However, using fluorescently labeled annexin V, one can
determine that 10% of the cells have completely lost asymmetry, while the rest are normal.
To confirm the labeling of red cells by annexin and demonstrate that labeling occurred on the surface of the cells,
we analyzed the fluorescently labeled cells by fluorescence
microscopy. A typical example is shown in Fig 5. In this
case, a population in which 64% of the cells tested positive
as the result of calcium and ionophore treatment (Fig 5A)
showed a heterogeneous labeling by fluorescent microscopy
(Fig 5B and C). A field was chosen in which an RBC ghost
was present (indicated by arrow): the bright labeling of the
ghost indicates that the inner monolayer was also available
for labeling with annexin V. The remainder of the cells
that bind annexin V appear to be labeled at their membrane
surface. The number of cells detected by fluorescent microscopy is identical to that measured by flow cytometry.
Wenext determined ifthered cells that had lost their
phospholipid asymmetry could be isolated from the cells that
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
KUYPERS ET AL
1184
1
l0
100
loo0
Fig 5. Analysis byflow cytometry and fluorescence microscopy of RBCs incubated with
FITC-labeled annexin V after
treatment withNEM and subsequent incubation for l hour in 1
mmol/L calcium in the presence
of 4 pmol/L ionophore A23187.
(A) Analysis showed 64% of the
cells labeled above background.
A representative field of this
population as observed in (B)
bright field andIC) fluorescence
is also shown.Notethe
RBC
ghost in the population (arrows).
10'
Fluorescence
were collectedandanalyzed
by flow cytometry(Fig 69).
All these steps wereperformed in our annexinVlabeling
buffer containing 1.2 mmol/L calcium. Whenthe purified
subpopulation was washed in buffer without calcium, FITCannexinVtogether with the beads wasremoved from the
cells (Fig 6C). Subsequent incubation of cells in buffer containing 1.2 mmol/L calcium andFITC-labeledannexinV
resulted in renewed labeling of the cells (Fig 6D).
Loss of phospholipid asymmetry has been implicated in
sickle cell anemia. Although changes in membrane phospholipid organization can be induced in RBCs in vitro by deoxygenation. it has been difficult to show loss of phospholipid
asymmetry in RBCs in vivo.Hence. we applied our new
method to peripheralbloodcollectedfrompatients
with
sicklecelldisease and compared the results with normal
control cells. In contrast with normal cells from eight volunteers. in which 0.21% of the cells labclcd with annexin (Fig
had not undergone this rearrangement, using a preparation
of small magnetic beads covered with an FlTC antibody. A
typical example is shown in Fig 6. Approximately IO% of
the cells in this sample exhibited an abnormal phospholipid
organization. Figure 6A shows the flow cytometric analysis
of this sample after annexinV labeling and subsequent incubation for I O minutes with FlTC antibody-coveredbeads.
The presence of these very small beads onthe surface of
the cellsdid not affectthe flow cytometric analysis. The
population was separated on a column placed in a magnetic
field, according tostandard protocols suppliedwith the beads
(MiltenyiBiotecInc).
Theapproximately 10% fluorescent
cells in the population labeled with annexin V and magnetic
beadswere retained on the column. A simple wash with
buffer removed the rest (unlabeled portion) of the cells. Subsequently, the column was removed from the magnetic field
andthe annexinV-labeled cells,nownolonger
retained.
1
IO
loo
1000
@-
-
10
'
d.
m
J
c
Fig 6. Magnetic separation of annexin V-labeled
cells. (A) Cell population labeled with FITC-annexin
V andsubsequently incubated with magnetic beads
covered with FlTC antibody. (B) The cells fromthe
population labeled with FITC-annexinV, selected by
magnetic separation. (C) The sample as shown in
panel
B
after
wash
a
in buffer
without
calcium. (D)
The sample as shown in (C) after renewed
labeling
with FITC-annexinV.
5E,
250:
-3
150
1
loo
50
:
1
1
10
too
..."
I
1000
Fluorescence
io'
l
l0
100
lmo
Fluoreacence
10
'
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
1185
RBC PS ASYMMETRY AND FLUORESCENT ANNEXIN V
P
1
Q)
S
.-C
g
2.18 f 1.21
E
H
8
.x
0.21 f 0.12
l
nn
Fig 7. Annexin V binding to normal and sickle cells. Fresh normal
(AA, n = 8 ) or sickle red cells (SS, n = 13) were labeled with FITC-
annexin V. The average and standard deviation is indicated. Despite
the wide spread in data, the average sickle cell sample was significantly increased relative to normal ( P .001).
7), fresh RBCs from 13 different patients with sickle cell
anemia showed a wide range of labeling with annexin V and
a significant increase as compared with normal cells (2.18%
2 1.21%, P < .001). These data support the hypothesis that
small subpopulations of sickle cells in the circulation can be
labeled by annexin V due to their loss of normal phospholipid organization under oxygenated conditions. They also
suggest that variations in the numbers of these cells exist
between patients. Because the flow cytometer enables the
distinction between RBCs and other blood cells, we compared results of experiments in which either washed RBCs or
whole blood were labeled with annexin V. Virtually identical
results were found in the RBC labeling of cells and whole
blood of five patients with sickle cell anemia (not shown).
DISCUSSION
The presence ofPS on the membrane surface has been
proposed to be the main reason for annexin V binding in
platelets,16apoptotic cells,17 and red cells.13Annexin V only
binds to very low numbers of normal red cells in flow cytometric analysis (Figs 1B and 7) in the presence of 1.2 mmoV
L calcium. This indicates that the outer monolayer of normal
RBCs, which lacks the presence of PS, does not bind annexin
V. Research on the binding specificities of annexin V in
vesicle systems has shown that SM and PC do notbind
annexin V in the presence of calcium.29Thus, the presence
of SM and PC on the outer monolayer is not expected to
contribute to annexin V binding. However, the outer monolayer also contains PE and phosphatidylinositiol (PI), phospholipids that do bind annexin V in a vesicle system in the
presence of calcium.29In the normal red cell, approximately
20% of all PE can be found in the outer monolayer, which
would account for approximately 11 mol% of all phospholipid in this monolayer. Of PI, 20% is also present in the
outer m~nolayer,~"
but the relative amount of this minor
phospholipid as compared with the total phospholipid pool
in the outer monolayer is less than 1 mol %. The lack of
fluorescent labeling of normal cells indicates that normal
levels (in the outer monolayer) of PE and PI do not act to
bind annexin V in the presence of relatively low concentrations (1.2 mmoVL) of calcium used in our experiments.
With loss of asymmetric distribution of RBC membrane
phospholipids, there is a significant increase in the exposure
of PS on the outer monolayer, and binding of annexin V to
these RBCs can be expected. Incubation of RBCs with NEM
results in the inhibition of the aminophospholipid translocase,8 and additional treatment of RBCs with calcium and
ionophore will lead to a scrambling of the normal membrane
phospholipid organization by effectively accelerating the
flop of aminophospholipids from inner to outer monolayer.'"
When these cells were incubated with FITC-labeled annexin
V, a very significant binding of annexin V to virtually all
cells was observed (Fig 1D). Although these results do not
prove that annexin V binds specifically to PS in human red
cells, the excellent correlation noted between the binding of
fluorescently labeled annexin V and the procoagulant properties of red cells in which phospholipid asymmetry was disrupted (Fig 4) supports an important role for PSinthis
process.
Our results clearly indicated that both impairment of the
aminophospholipid translocase as well as an additional
scrambling of the phospholipid organization was important
for high levels of labeling with annexin V. The inhibition
of the aminophospholipid translocase by NEM, without subsequent incubation in calcium and ionophore, did not result
in labeling of RBC with annexin V (Fig 1C). Even prolonged
incubations resulted in only low levels of annexin V binding.
While virtually all cells incubated with calcium and ionophore are labeled after NEM pretreatment (Fig lD), only
60% of the cells are labeled when this treatment was omitted
(Fig 2). Aminophospholipid translocase activity is inhibited
by increasing cytosolic calcium to 1 mmol/L,".72conditions
that exist during calcium and ionophore treatment. Unlike
NEM treatment, however, calcium-induced inhibition of the
translocase seems to be at least partially reversible: when
calcium and ionophore are removed by a wash with EDTA
and BSA, restored translocase activity can be expected to
reverse some or all of the scrambling that occurred during
incubation, resulting in a (partial) restoration of the phospholipid organization. NEM treatment effectively prevents any
reversal of membrane scrambling. Collectively, these data
indicate that for maximal labeling of cells with annexin V,
both an inhibited flip (deactivated aminophospholipid translocase) and increased flop are required.
The binding of approximately 300,000 molecules of FITC
annexin V to cells, as deducted from the removal of FITC
from the medium, compares favorably with similar data obtained using radioactively labeled annexin V.I3 However,
this number has to be treated cautiously, given the heterogeneity of the labeling, and is merely a rough approximation
of the number of binding sites per cell.
Experiments in which red cells with PS on their surface
were mixed with normal red
cells illustrated the power of
FITC-labeled annexin V and flow cytometry to detect small,
subpopulations of red cells with abnormal phospholipid
asymmetry (Fig 4). This approach not only facilitates analy-
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
KUYPERS ET AL
1186
sis of very small red
cell samples ( 1 0‘ cells or 0.1 pL of
packed RBCs) but is capable of detecting loss ofnormal
phospholipid organization in less than 1 % of a red cell mixture.Individual cells thatbindannexin
V can be studied
using fluorescence microscopy (Fig 5 ) and can be separated
from the population using magnetic beads (Fig 6). The simple purification from the population combined with the rethecell surface by
moval of annexin V andbeadsfrom
removing calcium from the buffer is a very gentle way to
prepare these cells for further analysis.
For an accurate determination of membrane phospholipid
organization, it is important
that the probe used is able to
discriminate between the outer and inner monolayer of the
bilayer. Either membrane permeability to the probe of the
intact cell or access to the inner monolayer after hemolysis
will interfere withaccurateassessment
ofphospholipid
asymmetry. It was assumed in previous studies that annexin
V, a 35-kD protein, only has access tothe outer monolayerof
intact cells,”.” even after calcium and ionophore
treatment.”
The lack of calcium entry and hemoglobin loss indirectly
seems to confirm this assumption, as it is very unlikely that
annexin V would be able to enter thecell while calcium can
not. More convincingly, however, we showthat all cells that
are fluorescentlylabeledwithannexin
V can be removed
from thepopulation with 15-nmmagnetic beads covered
with anti-FITC antibody. The remainingpopulation does not
show anyfluorescence abovebackground.Moreover,the
positivepopulation(Fig
6B) washedwith buffer without
calcium loses all fluorescence (Fig 6C). We believe it
is very
unlikely that these 15-nm beads could enter the cells tobind
to FITC-annexinpossiblypresentontheinnermonolayer
and that these large complexes composed of annexin V and
beads would reside on the inner monolayer and beremoved
from the cell by a wash in calcium-free buffer. Hence, we
conclude that FITC-annexin labels the outer monolayer of
the RBCs. Collectively, our data indicate that even after the
relatively harsh treatment of
the cell with NEM, calcium.
and ionophore, annexin V cannot enter the cell and label the
inner monolayer. Hemolysis during annexin V labeling was
very low, excluding thebinding of annexin V to the exposed
inner monolayer of RBC ghosts.Moreover,even
the few
ghosts formed werenot analyzed by flow cytometry because
they didnotappear in the selected RBC window. and in
fluorescent microscopy, they could easily be identified (Fig
5).
Loss of normal phospholipid asymmetry has been implicated in red cell pathology in diseases such as sickle cell
anemia.However, it has beendifficult to unambiguously
showthe loss of phospholipidasymmetry in vivo.Using
radioactive annexin V, 20,000 annexin V binding sites were
reported in sickle cell samples, as compared with 300,000
lost their
annexin V bindingsites in thesecellsthathad
phospholipid asymmetry due to calcium and ionophore
treatment.” This could be interpreted asan approximate 3% loss
of phospholipid asymmetry in the average cell. Importantly,
these results can also beinterpreted such that 3%of the cells
have lost their asymmetry. Radioactivity measurements do
not allow a distinction between these two possibilities. However, the population can be studied in detail using fluorescent
annexin V. Our study shows that in the population of sickle
RBCs in vivo, subpopulations exist that bind annexin V (Fig
7). These data indicate that while some cells in the sickle
cell population have lost their normal phospholipid asymmetry, othercellsare normal in that respect. Furthermore, a
variation between patients in the numbers of these cells was
noted. Hence, these results show that the use of fluorescent
annexin V is distinct from the use of radiolabeled annexin
V or other techniques that assess phospholipid asymmetry
of the average cell in the population, and allows the identification in blood of patients of small numbers of cells that
circulate with an abnormal phospholipid asymmetry, possibly of physiologic importance.
In conclusion,ourdatashow
thatfluorescentlylabeled
annexin V binds to RBCs with an altered phospholipid organization. The use of flow cytometry and fluorescent microscopy allows the detection of altered membrane lipid organization in subpopulations or individual cells. This approach
allows for the measurement of very small samples ( 1 O6 cells
or 0.1 pL of packed RBCs) and is capable of measuring the
loss of normal phospholipid organization in less than 0.1%
of that population, and FITC antibody-containing magnetic
beads allow the purification of these cells.
V should be
The use of fluorescentlylabeledannexin
apowerfulprobe
for identification of the loss of normal
phospholipid organization and will be applicable to assessment of this important characteristic in normal and pathologic erythrocytes. The ability to define loss of normal phospholipidorganization
in subpopulations of cells by flow
cytometry, or even individual cells by fluorescence microscopy, should be very useful in studies of red cell pathology,
particularly because loss of normal phospholipid organization in a small subpopulation of pathologic RBCs may be
of significant physiologic importance. The selection of these
cellsfrom thepopulationbased
on theirability to bind
annexin V will facilitate further studies of the mechanisms
that underlie the loss of phospholipid asymmetry in these
cells, as well as the potential pathophysiologic properties of
these cells.
REFERENCES
I . Bretscher M: Asymmetrical lipid bilayer structure for biological membranes. Nature 236:11, 1972
2. Bretscher M: Phosphatidylethanolamine: Differential labeling
in intact cells and cell ghostsof human erythrocytes by a membrane
impermeable reagent. J Mol Biol 71523, 1972
3 . Devaux PF: Static and dynamic lipid asymmetry in cell membranes. Biochemistry 30:1163, 1991
4.Etemadi AH: Membraneasymmetry. A surveyandcritical
appraisal of the methodology. 11. Methods for assessing the unequal
distribution of lipids. Biochim Biophys Acta 604:423, 1980
5. Op den Kamp J: Lipid asymmetry in membranes. Annu Rev
Biochem 48:47, 1979
6. Schroit AJ, Zwaal RFA: Transbilayer movement of phospholipids in red celland platelet membranes.BiochimBiophys Acta
1071:313, 1991
7. Roelofsen B: Phospholipases as tools to study the localization
of phospholipids in biological membranes: A critical review. J Toxicol 1:87, 1982
8. Zachowski A: Phospholipids in animal eukaryotic membranes:
Transverse asymmetry and movement. Biochem 294: I , 1993
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
RBC PS ASYMMETRY AND FLUORESCENT ANNEXIN V
9. Geldwerth D, Kuypers FA, Butikofer P, Allary M, Lubin BH,
Devaux P F Transbilayer mobility and distribution of red cell phospholipids during storage. J Clin Invest 92:308, 1993
10. Sulpice JC, Zachowski A, Devaux PF, Giraud F Requirement
for phosphatidylinositol 4,5-bisphosphate inthe
Ca2+-induced
phospholipid redistribution in the human erythrocyte membrane. J
Biol Chem 269:6347, 1994
11. Kuypers FA, van den Berg JJM, Lubin BH: Phospholipid
asymmetry and diffusion in the membrane of normal and sickle
erythrocytes, in Onishi S. (ed): Membrane Linked Diseases. Boca
Raton, F L , CRC, 1994, p 21
12. Blumenfeld N, Zachowski A, Galacteros F, Beuzard Y, Devaux P Transmembrane mobility of phospholipids in sickle erythrocytes: Effect of deoxygenation on diffusion and asymmetry. Blood
775349,1991
13. Tait IF, Gibson D: Measurement of membrane phospholipid
asymmetry in normal and sickle cell erythrocytes by means of
annexin binding. J Lab Clin Med 123:741, 1994
14. Ernst JD: Epitope mapping of annexin I. Antibodies that compete with phospholipids and calcium recognize amino acids 42-99.
Biochem 289539, 1993
15. Raynal P, Pollard HB: Annexins: The problem of assessing
the biological role for a gene family of multifunctional calciumand phospholipid-binding proteins. Biochim Biophys Acta 1197:63,
1994
16. Dachary-F’rigent J, Freyssinet JM, Pasquet JM, Carron JC,
Nurden A T Annexin V as a probe of aminophospholipid exposure
and platelet membrane vesiculation: A flow cytometry study showing
a role for free sulfhydryl groups. Blood 81:2554, 1993
17. Koopman G, Reutelingsperger CPM, Kuijten GAM, Keehnen
RMJ, Pals ST, van Oers MHJ: Annexin V for flow cytometric detection of phosphatidylserine expression on B cells undergoing
apoptosis. Blood 841415, 1994
18. Comfurius P, Senden JM, Tilly RH, Schroit AJ, Bevers EM,
Zwaal RF: Loss of membrane phospholipid asymmetry in platelets
and red cells may be associated with calcium-induced shedding of
plasma membrane and inhibition of aminophospholipid translocase.
Biochim Biophys Acta 1026:153, 1 9 9 0
19. Baldwin JM, O’Reilly R, Whitney M, Lucy JA: Surface exposure of phosphatidylserine is associated with the swelling and osmotically-induced fusion of human erythrocytes in the presence of
Ca2+. Biochim Biophys Acta 1028:14, 1990
20. Connor J, Gillum K, Schroit AJ: Maintenance of lipid asymmetry in red blood cells and ghosts: Effect of divalent cations and
serum albumin on the transbilayer distribution of phosphatidylserine.
Biocbim Biophys Acta 102532, 1990
21. Williamson P, Kulick A, Zachowski A, Schlegel RA, Devaux
P F Ca2+ induces transbilayer redistribution of all major phospholipids in human erythrocytes. Biochemistry 31:6355, 1992
22. Butikofer P, Kuypers FA, Xu CM, Chiu DT, Lubin B: Enrichment of two glycosyl-phosphatidylinositol-anchoredproteins, acetylcholinesterase and decay accelerating factor, in vesicles released
from human red blood cells. Blood 74:1481, 1989
23. Wagner GM, Chiu DTY, Yee M, Lubin BH: Red cell vesiculation-A common membrane physiologic event. J Lab Clin Med
108:315, 1986
24. Emst JD, Hoye E, Blackwood RA: Use of a novel strategy
for the preparation and characterization of an antipeptide antibody
capable of recognizing members of the annexin family. Biochem
Biophys Res Commun 161:959, 1989
25. Haigler HT, Schlaepfer DD, Burgess WH: Characterization
of lipocortin I and an immunologically unrelated 33 kDa protein as
epidermal growth factor receptorkinase substrates and phospholipase A2 inhibitors. J Biol Chem 262:6921, 1987
26. Ernst JD, Blackwood RA, Hoye E, Mok TL: Identification
of a domain that mediates vesicle aggregation reveals functional
diversity of annexin repeats. J Biol Chem 266:6670, 1991
27. Emst JD, Mall A, Chew G: Annexins possess functionally
distinguishable Caz+ and phospholipid binding domains. Biochem
Biophys Res Commun 200:867, 1994
28. Discher DE, Lenng A, Mohandas N, Evans EA: Molecular
maps of red cell deformation: Hidden elasticity and in situ connectivity. Science 266:1032, 1994
29. Blackwood R A , Ernst JD: Characterization of Ca2+-dependent phospholipid binding, vesicle aggregation and membrane fusion
by annexins. Biochem J 266:195, 1990
30. Butikofer P,Lin Z
W
, Chiu DT, Lubin B, Kuypers FA: Transbilayer distribution and mobility of phosphatidylinositol in human
red blood cells. J Biol Chem 265:16035, 1990
31. Zachowski A, Favre E, Cribier S,Herve P, Devaux P F Outside-inside translocation of aminophospholipids in the human erythrocyte membrane is mediated by a specific enzyme [published erratum appears in Biochemistry 257788, 19861. Biochemistry 25:2585,
1986
32. Bitbol M, Fellmann P, Zachowski A, Devaux PF: Ion regulation of phosphatidylserine and phosphatidylethanolamine outsideinside translocation in human erythrocytes. Biochim Biophys Acta
904:268, 1987
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
1996 87: 1179-1187
Detection of altered membrane phospholipid asymmetry in
subpopulations of human red blood cells using fluorescently labeled
annexin V
FA Kuypers, RA Lewis, M Hua, MA Schott, D Discher, JD Ernst and BH Lubin
Updated information and services can be found at:
http://www.bloodjournal.org/content/87/3/1179.full.html
Articles on similar topics can be found in the following Blood collections
Information about reproducing this article in parts or in its entirety may be found online at:
http://www.bloodjournal.org/site/misc/rights.xhtml#repub_requests
Information about ordering reprints may be found online at:
http://www.bloodjournal.org/site/misc/rights.xhtml#reprints
Information about subscriptions and ASH membership may be found online at:
http://www.bloodjournal.org/site/subscriptions/index.xhtml
Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published weekly by the American
Society of Hematology, 2021 L St, NW, Suite 900, Washington DC 20036.
Copyright 2011 by The American Society of Hematology; all rights reserved.