From www.bloodjournal.org by guest on June 16, 2017. For personal use only.
Denaturing Interaction Between Sickle Hemoglobin
and Phosphatidylserine Liposomes
By Esther Marva and Robert P. Hebbel
It is hypothesized that abnormal interactionbetween
sickle hemoglobin (HbS) and erythrocyte membrane lipid
might promote deposition of denatured hemoglobin (hemichrome) on the membrane. We compared the interaction
of HbS and normal HbA with large unilamellar phosphatidylserine (PS) liposomes under low salt/pH conditions.
Admixture of oxyHb and dioleoyl-PS resulted in loss of absorbance at 412 nm, the apparent first order rate constant
for which was.25 & 0.02 hour” for HbA and .85 & 0.1 8
hour“ for HbS. This was ascribable largely to formation of
metHb and hemichromes and was accompanied by some
actual transfer of heme from hemoglobin t o lipid phase. By
comparison, admixture of oxyHb with liposomes made
from bovine brain PS having unsaturated acyl chains promoted even faster absorbanceloss if the
starting liposomal
material contained detectable peroxidation by-product.
In such cases, actual heme destruction developed with
accompanying liberation of free iron and promotion of lipidperoxidation. Fluorescence quenching experiments indicate that hemoglobin/lipid interaction is characterized by
very rapid initial electrostatic interaction, followed by development of irreversible changes. Similar changes still occur under conditions of physiologic salt/pH, but they develop much more slowly. The 3.4-fold faster oxidation of
HbS versus HbA on lipid observed here represents an additional augmentation of the disparity in oxidation rates for
hemoglobins in solution (1.7-fold faster for HbS than for
HbA) observed previously. The accelerated promotion of
Hb denaturation resulting from lipid contact may help explain deposits of hemichrome on sickle red blood cell membranes, particularly because these cells are in double jeopardy by virtue having
of
both the mutant
HbS and abnormal
amounts of peroxidized membrane lipid.
6 1994 by The American Societyof Hematology.
T
some portion ofthe RBC’s lipid bilayer surface is sufficiently
exposed to allow interaction with the highly concentrated
cytoplasmic Hb.‘ For these reasons, we hypothesized that
hemichrome formation in sickle RBC might result from abnormal interaction between the mutant Hb and the membrane’s lipid bilayer. To test this. we compared the interaction of HbA and HbS with large unilamellar PS liposomes
of two types: dioleoyl PS (dPS) that has low potential for
oxidation because it has only one double bond per acyl
chain, and bovine brain PS (bPS) having higher potential
for oxidation due to the multiple double bonds present in
natural aminophospholipid.’ Our results indicate that
HbS does undergo accelerated oxidative denaturation to
hemichrome-like material upon contact with PS liposomes,
and additional findings strengthen the possible significance
of this thesickle RBC.
HE REMARKABLY pleiotropic effectof the sickle
gene is most evident in the myriad abnormalities characteristic of the sickle erythrocyte (RBC) membrane.’It has
been suggested that a number of these defects derive from
theabnormal deposits ofdenatured hemoglobin (Hb), in the
form of hemichromes, found on thesickle membrane.’ For
example, denatured Hb potentially could target oxidative
damage to adjacent membrane components,’ and it can
promote clustering of membrane protein band 3 and accumulation of harmful Ig.’ Despite the probable importance
of hemichromes, the biochemical basis for their deposition
on sickle membranes has not been defined. As products of
oxidative Hb denaturation, hemichromes can appear after
either spontaneous or induced formation of methemoglobin (metHb).’ However, it is not known whether they form
directly on the membraneor form in the cytoplasm and subsequently migrate to themembrane.
Three earlier observations are of particular interest in this
regard. First, sickle Hb (HbS)is somewhat unstable. This is
evident, for example, in its tendency to precipitate during
mechanical a g i t a t i ~ n ,an
~ unusual property that results
from the mutant Hb’s surface hydrophobicity4 and consequent tendency to exhibit surface denaturation behavior.
Second, when normal HbA interacts in vitro with phosphatidylserine (PS), it tends to denature and even transfer its
heme from globin to lipid phase.5 Third, it is likely that
From t h e Division of Hematology, (he Departmcwt of Medicine
University ofMinnesota Medical School, Minneapolis.
Submitted March 4,1993: accep.predAugus&31, 1993.
Supporled h?]National Institutes ofHealth Grant No. HL30160.
Addre.ss wprint requests IO Roherf P. Hehbel, MD. BOK 480
UMHC, Harvard St at E Riwr Rd,Minneapolis, M N 55455.
The publication costs ofthis article were defrayed in partby page
charge payment. This article must therefore be hereb.y marked
“advertisement” in accordance with 18 U.S.C.section 1734 sole!,’ to
indicate this.fuct.
0 1994 hv The American Society qfHematolog~‘.
000~-4971/94~~301-0016$3.00/0
242
MATERIALSANDMETHODS
Hh Prcparufion
Fresh heparinized whole blood was drawn from volunteer donors
with sickle cell anemia (HbSS) or from normal donors. Hb was isolated from RBC lysates by ion-exchange chromatography using
DEAE-Sepharose CL-6B (Pharmacia, Piscataway, NJ).8 Eluates
were dialyzed against “assay buffer” ( 5 mmol/L KCI, I O mmol/L
Tris, pH 6.5). concentrated by Amicon filtration (Amicon, Beverly,
MA) and assessed for purity using isoelectric focusing. We previously found that Hbs prepared in this manner are devoid of detectable superoxide dismutase or catalase activity.8 All preparative
steps were performed at 4’C. If storage was required, the eluates
were frozen drop-wise in liquid nitrogen and stored at -80°C. At
the time of use, amount of contaminating metHb was measured’
and found to be lessthan 2%. For deliberate preparation of metHb,
we added I .5 molar excess ofpotassium femcyanide to HbS or HbA
and used gel filtration for desalting.’ Bovine metHb also was used.
Liposome P r ~ ~ a r u t i o n
We prepared liposomes from each of three types of phospholipid:
dioleoyl ( I 8: I ) phosphatidylserine (dPS) or bovine brain phosphatidylserine (bps) (Avanti Polar Lipids, Alabaster. AL) or dioleyl
Blood, Vol83, No l (January l), 1994:pp 242-249
From www.bloodjournal.org by guest on June 16, 2017. For personal use only.
243
HbS INTERACTION WITH LIPOSOMES
phosphatidylcholine (PC) (Sigma Chemical CO, St Louis, MO).
Two milliliters ofthe lipid dissolved in chloroform ( I O mg/mL) was
evaporated under nitrogen and desiccated overnight. Assay buffer
(1.5 mL) was added to the lipids and large (100 nm) unilamellar
liposomes were made by repeated extrusion through polycarbonate
filters (Avestin Inc, Ottawa, Canada)." During ourpreliminary experiments, the expected liposome size was kindly confirmed (97.3
f 29.0 nm) by Avestin using quasielastic light scattering. The pHof
the final liposomes in assay buffer was adjusted to pH6.5, and then
the phospholipid content was determined by measurement of lipid
phosphorus."
0.9
0.8
0.7
0.6
HbNdPS
'\
0.5
HbS/dPS
HbNbPS
HbS/bPS
0.4
Hb/Liposome Reaction Mixture
To 1,OOO pL of Hb, 100 pL of liposomes was added to achieve
final concentrations of 6.8 pmol/L heme and0.25 mg/mL lipid in
assay buffer (5 mmol/L KCI, 10 mmol/L Tris, pH 6.5). In some
experiments, conditions of higher pH (7.2) and/or higher salt concentration (100 mmol/L KCI) were used. All incubations were performed at room temperature. The following parameters were monitored.
Soret absorbance. After admixture of Hb andliposomes, we recorded absorbancein the Soretregion (4I 2 nm) as our basic method
for monitoring rate of change in the state of Hb. We used these
values to estimate the apparentfirst order rate constant for the initial phase of the interaction during which the Soret absorbance
change was linear with time on semi-log plot.
Hb oxidation. Concentrations of oxyHb, metHb,and hemichrome were calculated as described by Winterbourn9 after measurement ofabsorbance at560,577,630, and 700 nm.
Heme destruction. The actual total hemecontent ofthemixture
was determined by removing a 100-pL aliquot, adding itto 1 mL of
concentrated formic acid, and measuring absorbance at 398 nm,"
a method that quantitates
heme regardless of specific Hb species or
whether it is associated with globin.
Thiobarbituric acid-reactive substance (TBARS). To monitor
lipid peroxidation, we used theTBARS assay described previously.13At zero-time and after 1 hour of incubation, 1 mL from
the mixture was transferred to theTBARS assay reagents that contained 0.1 mmol/L deferoxamine plus 0.1 mmol/L butylated hydroxytoluene. The amountof TBARS was determined by measuring absorbance at 532 and 453 nm and correcting for presence of
any non-TBARS chromogen.
Tryptophanfluorescence. Changes in the fluorescence intensity
of Hb tryptophan in the mixture was monitored using a PerkinElmer Model LS-5B spectrofluorometer (Norwalk, CT) with excitation at 280 nm.
Hb contactwith fluorescent liposomes. Fluorescent liposomes
were made by incorporating the fluorescent probe 9( 1,2-anthroyl)
stearic acid (Sigma) at 0.1%by weight during liposome preparation.
We defined the initial fluorescence intensity ofthe labeled liposome
before any additions as 100%. Hb and liposomes in assay buffer
(having 5 mmol/L KCI) were admixed to the same final concentrations used for the other experiments (6.8 pmol/L heme, 0.25 mg/
mL lipid), and fluorescence signal was monitored using excitation
at 360 nm andemission at 480 nm. In several experiments, the ionic
strength of the mixture was abruptly increased (to 100 mmol/L
KCI) at I , 10, or 25 minutes afterthe initial admixture step.
Measurement offiee iron. To detect liberation of free iron, we
monitored formation ofthe colored complexbetween free iron and
ferrozine in the presence of a denaturant (sodium dodecyl sulfate)
and a reducing agent (ascorbate). Using the procedure previously
described," we added 0.1 mL of the mixture to 0.9 mL of the ferrozine assay reagents and recorded absorbance at 562 nm after 30
0
0
.
20
3
40
1
60
'
80
'
7
160
minutes
Fig 1. Loss of Hb absorbance inthe presence of liposomes. Absorbance at 412 nm was monitored after admixture of oxyHbA(triangles)oroxyHbS (squares) with dPS (closed symbols) or bPS (open
symbols) liposomes under the following conditions: pH 6.5, salt 5
mmol/L KCI, room temperature, final concentrations of6.8 pmol/L
heme and 0.25 mg/mL lipid.A representative experiment is shown.
minutes. Because of the detection requirements of this assay, this
experiment was performed with 35 pmol/L heme and 1.25 mg/mL
of lipid, but preserving the same Hb/lipid ratio as in other experiments.
Absorption spectra. The absorption spectra of Hb/liposome
mixtures were recorded from 350 to 700 nm at room temperature
in a DU-70 spectrophotometer (Beckman Instruments, Palo Alto,
CA) and using the assay buffer as blank, because the liposomes
made a trivial contribution to absorbance. Comparison was made
with the spectra of oxyHb and metHb and of the hemichrome in
sickle RBC ghost membranes. The latter were obtained as thewhitest possible ghosts using the method of Kuross et al.'' To record
the absorption spectrum of sickle ghosts, we used the comparable
spectrum of normal RBC ghosts prepared in the same manner as
a
blank.
Heme transfer to lipid phase. To test for movement of heme
from Hb tolipid phase, we took aliquots of the incubation mixture
at zero-time and after I hour andadded sufficient salt to bring KC1
concentration to 100 mmol/L in the sample to slow down, if not
terminate, any further interactionbetween Hb andlipid.5 The mixture was then passed through a Sephadex G- 100 column (1.5 X 75
cm), andthe peak fractions containing separatedliposomes and Hb
were identified by screening for heme content (absorbance at 412
nm) andlipid phosphorus. The tryptophan fluorescence of the Hb
fraction separated in this manner was compared with that ofcontrol
Hb (gel filtered without liposomes present) after first adjusting the
two solutions to equivalent absorbance at 280 nm. In addition, we
compared the ratio of heme absorbance (at 4 12 nm) toprotein absorbance (at 280 nm).
RESULTS
In these experiments, we compared the changes in oxyHbA versus oxyHbS after their admixture with large unilamellar liposomes made from either dPS or bPS, with most
experiments being performed at low pH (6.5) and low salt
concentration ( 5 mmol/L KCl).
Addition of either oxyHb to dPS liposomes resulted in a
From www.bloodjournal.org by guest on June 16, 2017. For personal use only.
MARVA AND HEBBEL
244
loss of absorbance at 412 nm, the rate of which was faster
for HbS than for HbA (Fig l). The apparent first order rate
constant for the initial stages of this process was 0.85 f 0.18
hour" for HbS and 0.25 k 0.02 hour" for HbA (n = 5. P <
.005). In contrast, the rate of absorbance changes was dramatically faster using bPS liposomes, although the relative
rate remained considerably faster for HbS than for HbA (Fig
l). A rate constant could not be estimated for experiments
using bPS liposomes. Inclusion of 10 pmol/L deferoxamine
to the Hb/liposome mixture had no effect whatsoever on
rate of these absorbance changes (data not shown). Notably,
under these conditions no absorbance change resulted from
addition of either oxyHbto liposomes made from dioleoylPC (data not shown).
In these experiments, absorbance deteriorationreached a
plateau after which little further change was noted (Fig l).
This was not explained by the system having an excess of
Hb, leaving some without access to the lipid surface; rather
it wascaused by rapid completion ofthose reactions making
the greatest contribution to loss of absorbance at 4 12 nm.
Evidence for this includes the fact that the total Hb present
in this system was sufficient to cover only about 10%)of the
total liposome surface area.* Furthermore, additionof a further aliquot of liposomes to the reaction mixture after attainment oftheabsorbance plateau caused no additional absorbance change (data not shown). As shown below, the
reduction in absorbance at 4 I2 nm resulting from Hb/lipid
interaction results from multiple factors, some common to
both dPS and bPS liposomes and some uniquely stemming
from the presence of unsaturated acyl chains in bPS. In fact,
the difference in rate of absorbance change shown for dPS
and bPS (Fig 1) is accounted for mainly, if not completely,
by presence of peroxidation by-product in the bPS material
(see below).
f l h Osidution
Part of the loss in absorbance at 412 nm was caused by
conversion of oxyHb to metHb and hemichrome, and inspection of Fig 2 suggests that this is a complex process with
ongoing evolution of Hb species over time. Notably, these
changes evolved much more rapidly for HbS than for HbA,
and they were exaggerated for contact with bPS compared
with dPS (Fig 2).
Consistent findings were derived from monitoring ab-
* This estimateis derived from the following values. Preparations
of large unilamellar liposomes like those we used (100 nm in diameter) contain -8 X IO4 phospholipids per vesicle and -7.6 X I O i 3
vesicles per pmol phospholipid, and they have a ratio of phospholipid in inner versus outer monolayers of 0.8 I .7 Our system having
0.25 mg/mL of lipid with an average molecular weight 8 10 daltons
18.7 X
phospholipid molecules.
would containanestimated
The presence of 6.8 pmol/L heme provides I .9 X I O l 5 molecules
of Hb tetramers. Assuming that the surface ofa single phospholipid
and the surface area of Hb tetramer
head group occupies 0.6
is 26 I I ~ ' .this
' ~ provides an estimate of sufficient Hb present to
cover only 10% of the available liposome surface in these experiments.
-
'1
HbNdPS
*OxyHb
"tMetHb
"Hemichrome
HbSIdPS
"
f
t
0
'1
" 4
E
'
20
40
60
80
HbNbPS
6
4
2
0
0
'1
20
40
60
80
HbShPS
0
0
2600
40
80
minutes
Fig 2. Conversion
of
oxyHb
to metHb and
hemichrome.
metHb (A),and hemichrome(m) were calcuAmounts of oxyHb(0).
lated from measurementof absorbance at multiple wavelengths after admixture of oxyHbA or oxyHbS with dPS (top two panels) or
bPS (bottom two panels) liposomes. Conditionswere as for Fig 1.
Representative experiments are shown.
From www.bloodjournal.org by guest on June 16, 2017. For personal use only.
HbSINTERACTION WITH LIPOSOMES
I
1
l
245
I
,
,
,
I
,
,
,
I
450430410390370
1
1
,
I
450430410390370
Heme Destruction, Liberation of Iron, and Lipid
Peroxidation
Whereas no destruction of heme was observed when either oxyHb was incubated with dPS, their incubation with
bPS resulted in destruction of 26% and 14%of the heme in
HbS and HbA, respectively, within 1 hour (Fig 4, top). The
heme in bovine metHb met this same fate with destruction
stimulated by bPS but not by dPS (datanot shown). Thus,
C
I
somewhat less well-defined peak at about 410 nm. These
shifts in absorbance are unlikely to be explained by development of a simple mixture of oxyHb and metHb because
incubation of metHb with dPS liposomes resulted in a shift
in peak absorbance fromthe starting value of 405 nm to the
higher value of 410 nm (Fig 3C), consistent with formation
of hemichrome-like material from metHb. With bPS liposomes, metHb led to an ill-defined spectrum with a peak
absorbance at a somewhat lower value (Fig 3C). Notably,
however, if the oxyHb/bPS mixtureswere allowed to incubate for longer time periods (or if the bPS contained higher
amounts ofperoxidation by-products) they also evolved to
the same ill-defined spectrum shown for metHb/bPS (data
not shown).
Thus, although metHb undoubtedly was formed after
contact of oxyHb with liposomes, metHb appears not to
have accumulated inthese experiments, but rather was rapidly converted to hemichrome. Notably, the reduction in
absorbance at 4 12 nm for metHbS and metHbA added to
dPS was extremely fast and completed within 6 minutes for
both (data not shown). The rapidity of this change for
metHb compared with oxyHb indicatesthat metHbformation is the likely rate-limiting step in the interaction of Hb
with liposomes.
I
4
,
I
I
450 430 410 390 370
Fig 3. Absorption spectra of Hblliposome mixtures. Absorption
spectra were obtained after admixture of Hb and liposomes under
conditions as for Fig 1. Representativeexperiments are shown. (A)
OxyHbS plus dPS after 1 minute (1) and after 1 hour (2); for comparison, absorption spectrum of exhaustively washed sickle ghosts
(B) OxyHbS plus bPS after 1 minute (1) and after 1
is shown (3).
hour (2). (C) Bovine metHb plus dPS for 1 minute (1) or 1 hour (3);
and bovine metHb plus bPS for 1 minute (2)or for 1 hour (4).
0
sorption spectra of the Hb/liposome mixtures. For oxyHbS
incubated with dPS (Fig 3A), the overall absorbance diminished, and peak absorbance shifted from its startingvalue of
4 14 nm (oxyHb) to 4 10 nm which is characteristic of
hemi~hrome.'~*'~ absorption
The
spectrum for sickle RBC
ghost membranes containing hemichrome has peak absorbance at 410 nm (Fig 3A). The interaction of oxyHbS with
bPS (Fig 3B) shows similar results but with evolution of a
20
40
60
80
160
minutes
Fig 4. Destruction of heme during incubation of Hb with bPS.
Occurrence of actual heme destruction after admixture of Hb and
liposomes (under conditions as for Fig 1) was detected as loss of
heme absorbance in concentrated formic acid (see Materials and
Methods). Representativeexperiments are shown for admixture of
oxyHbA (triangles) or oxyHbS (squares)with dPS (closed symbols)
or bPS (open
symbols).
From www.bloodjournal.org by guest on June 16, 2017. For personal use only.
246
MARVA AND HEBBEL
100
iments, we only useddPS liposomes to avoid the confounding influence of heme destruction. The rate of the observed
fluorescencechange was consistent with the previously
noted rate of Hb oxidation in the presence of dPS (Fig 2).
In confirmation, we also used gel filtration to physically
separate liposomes from Hb after l hour of coincubation.
Supporting the occurrence of heme transfer to lipid, we observed substantially greater tryptophan fluorescence and
much lower ratio of Soret to protein absorbance (412 nm/
280 nm) from the Hb that had been incontact with the lipid
compared to the control hemoglobin (data not shown).
-
1
1
6o
40
Mechanism of Hb/Lipid Interaction
To examine the early stages ofthis interaction, we added
Hb to dPS liposomes (to preclude confounding influences
from heme destruction) and monitored the signal intensity
from an incorporated fluorescent probe molecule (Fig 6).
This results first inan instantaneous, stable reduction signal
intensity due to simple dilution, so the remaining fluorescence (about 70%of the original) can be considered to be
the true zero-time or 100%’’signal level for the actual experiments. This dilutional effect is shown by the curve labeled “control,” which was generatedby addition of HbS to
fluorescent control PC liposomes (with which Hb did not
react).
The other curves shown in Fig 6 resulted from addition
of HbA or HbS to fluorescent dPS IiDosomes. They show a
further component of rapid initial signal quenching that is
I
1
1
1
“
20 ‘i
300
320
340
360
380
400
Fig 5 . Changes in Hb tryptophan fluorescence intensity. After
the
admixture of Hb with liposomes (under conditions as for Fig l),
fluorescencefrom globin tryptophan was monitored by excitation at
280 nm and recording of emission spectra. Increasingfluorescence
reflects departure of heme from globin. Illustrated curves are for
30 minutes (3).
and 60 minutes
oxyHbS plus dPSat 30seconds (l),
(5); and for oxyHbA at 30 minutes (2) and 60 minutes (4).A representative experiment is shown.
2
0
Heme Transfer to Lipid Phase
Separation of heme from Hb, accompanied by transfer of
heme to the lipid phase,was detected by observation of globin tryptophan fluorescence, which increasedmore rapidly
for HbS than HbA as the fluorescence-quenching heme
moiety lostproximity to the globin (Fig5). For these exper-
0 control
W
?I
4 iI
G-
’c
actual heme destruction was one component of the diminishing absorbance at 4 12 nm (Fig 1).
Destruction of heme was accompanied by other events.
Upon addition of HbS to bps, heme destruction of 3.5 +0.67 nmol/ml after 3 hours was accompanied by liberation
of 1.51 rt 0.16 nmol/mL of detectable free iron. This was
associated withformation of new TBARS in the amountof
3.41 k 0.56 nmol/pmol lipid within 1 hour. In contrast,
upon addition of HbS to dPS, no free iron liberation or
TBARS formation could be detected.
70
30
HbA
0
5
10
15
minutes
20
25
Fig 6 . Time-dependent changes after addition of Hb to fluorescent liposomes. The signal intensity of fluorescent-labeled liposomes was monitored after addition of Hb under conditions as for
Fig 1. Signal remaining after addition of Hb to dPC liposomes(control) indicates the maximal fluorescenceinthe experimentalsystem
at zero-time signal. Data points indicate changes in signal intensity
or oxyHbA (closedsquares)
after addition of oxyHbS (open squares)
to fluorescent dPS liposomes.Arrows indicatetime-points at which
salt concentration was abruptly increasedto 100m d / L KC1in one
of the HbA samples and one of the HbS samples.To avoid overlap
of symbols, the resulting jumps in fluorescenceare shown offset by
1 minute (inreality, the increased signal was recorded 10 seconds
after addition of higher salt) and with correction for the effect of
simple dilution. See text forinterpretation. A representative experiment isshown.
From www.bloodjournal.org by guest on June 16, 2017. For personal use only.
247
HbS INTERACTION WITH LIPOSOMES
of Hb and bPS liposomes was roughly proportionate to the
amount of peroxidized lipid inthe starting liposomal material (data notshown).
1
0.9
HbNbPSl
0.8
0.7
0.6
HbA/bPS2
H bS/bPS 1
HbS/bPS2
0.5
0.4
!
I
0
20
I
I
40
60
minutes
Fig 7. Rate of Hb absorbance change is influenced by presence
of peroxidation by-products. Absorbance
at 4 1 2 nm was monitored
to bPS liafter addition of oxyHbA (triangles) or oxyHbS (squares)
posomesunderconditionsasforFig
1. Forcurvesindicatedby
"bPS1" (open symbols)the starting liposomes contained only
0.23
nmol TBARS per pmol lipid.
For curves indicated by
"bPS2" (closed
symbols) the starting liposomes contained 1.2 nmol TBARS per
pmol lipid. A representative experiment is shown.
greater for HbS than for HbA. This initial phase was followed bya slower, continuing phase of even greater quenching development, the rate of which wassimilar for HbS and
HbA. The arrowsin Fig 6 indicate times at whichwe
abruptly elevated the salt concentration (to 100 mmol/L
KCl) of some samples to displace any dissociable Hb from
the liposome. However,the resulting restoration of fluorescence signal was never
complete, and the degree of recovery
clearly was lower at later time-points. This experiment is
interpreted in the Discussion below.
Physiologic pH and Salt Concentration
Whereas most ofour experiments were performed at low
salt (5 mmol/L) and low pH (6.5), we confirmed that the
abnormal interaction of HbS with bPS still occurs under
more physiologic conditions (pH 7.2, 100 mmol/L KCl),
although it takes place over hours rather than minutes (Fig
8). Interaction with the less physiologiclipid, dPS, could not
be detected under these conditions over this time frame.
DISCUSSION
Interaction of oxyHb with PS liposomes in vitro leads to
complex changes that may be informative with regard to
sickle disease pathophysiology.
In these studies, we used PS
liposomes made either of dioleoyl-PS(and having low potential for oxidation because of a single double-bond per
acyl chain) or of bovine brain PS (having great potential for
oxidation because of the significant degree unsaturation).
of
The dPS liposomes allow observation of changes inthe absence of destruction of either heme or lipid, but the bPS liposomes comprise a better approximation of the RBC
membrane's inner monolayer lipids." We used large (100
nm) unilamellar PS liposomes for these experiments because interpretation is facilitated by the slower rate of
changes compared with
using
small,
sonicated lipos o m e ~ ~and
* ~because
'~
the gently curved surface of a large
liposome is a better approximation of the membrane's lipid
bilayer than is the highly curved and stressed bilayer of a
small liposome.'
S
1.21
I
1
-
1
1
1.0
HbS/dPS
c
Hb Interaction With Peroxidized Lipids
The bPS contains some unsaturated lipid and nearly alwaysis accompanied by some peroxidation by-products
contaminating the starting material. After admixture with
liposomes, we found that the rate of Hb absorbance change
for bPS liposomes starting with 1.2 nmol TBARSlpmol
lipid was four times faster than that for bPS liposomes starting withjust 0.23 nmol TBARS/pmol lipid. This accelerating effect was evident for both HbA and HbS (Fig 7). Because initial Hb contact with lipid is virtually
instantaneous
relative to the overall time course of these experiments (Fig
6), it is likely that this acceleration in presence of peroxidized lipid is accounted for by the subsequent Hb denaturation phase. In fact, inthe occasional experiment for which
zero TBARScould be detected in the starting bPS liposome
preparation, the rate of absorbance change at 4 12 nm of the
Hb/bPS mixture was significantly slowerand very similar to
that seen for Hb/dPS mixtures (data not shown). The magnitude of the TBARS generation response after admixture
c
(d
,
2
0 . 8 h
0.0
j m l , l a , , l , l
0
200
400
600
800 1 0 0 0
minutes
Fig 8. Interaction between HbS andliposomesunderphysiologic conditions. OxyHbS and PS liposomes were admixed at the
all previous experimentsbut under
same final concentrations as in
more physiologicpH (7.2)and salt concentration (100 mmol/L KCI).
No absorbance changeat 4 1 2 nm is observed using dPS liposomes
(closed symbols). but using bPS liposomes they dooccur(open
symbols). A representative experiment is shown.
From www.bloodjournal.org by guest on June 16, 2017. For personal use only.
MARVA AND HEBBEL
248
We find that admixtureof Hb with dPS liposomes leads
to Hb oxidation, with formation of metHband hemichrome, and even loss of some heme from globin to lipid
phase, thus confirming changes observed previously upon
incubation of HbA with l i p o s ~ m e s ~ ~or
' ' ~fatty
' ~ acids.15
However, we further find that such changes are significantly
accelerated for oxyHbS compared with oxyHbA. As calculated from initial rate of change in absorbance at 412 nm
after Hb/dPS admixture, the apparent first order rate constant for oxidation was about three times greater for HbS
than for HbA (0.85 0.18 hour" versus 0.25 t- 0.02 hour-',
respectively; P < .005). In contrast to these limited changes
resulting from Hb interaction with dPS, using bPS liposomes we observed the additional, subsequent processes of
heme destruction, liberationof free iron, and stimulation of
peroxidation in the bilayer membrane. Again, these developed faster for HbS than HbA.
A tentativedescription ofthe mechanism ofthe basic Hb/
lipid interaction (proximate to heme destruction) derives
from our experiments monitoring quenching of lipid fluorescence by added Hb (Fig 6). The immediate signal
quenching is largelyreversed by abruptly increasing the
mixture salt concentration, a maneuver that displaces Hb
from the lipid surface,53"thus arguing that it reflects electrostaticinteraction between Hb and PS.5,'4,20However,
even this initial quenching is about 1.5 times greater for HbS
than for HbA, and it is not completely reversed by salt. This
presumably reflects the earliest stages of Hb oxidationidenaturation. Indeed, this first stage is followed by a slower
phase of fluorescence quenching. Because we calculate that
the total amount of Hb in our reaction mixture was sufficient to cover only about 10%of the available lipid surface
area (*refer to earlier footnote), thisslower phase should not
be accountedfor by late developmentof newHb/lipid initial
contacts. Rather, it must reflect evolution of the later stages
of Hb interaction with lipid, such as loss of freeheme to lipid
phase or formation of denatured hemichromesthat perhaps
remain associated with lipid. Consistent with thisinterpretation, abrupt elevation of the Hb/liposome mixturesalt
concentration allowed lesserrecoveryof
fluorescence at
longer time points.
Our interpretation that Hb/PS interaction involves an
initial and reversible electrostatic interaction, followed by a
more complex irreversible interaction, is perfectly analogous to theseminal observations of Shaklai et aI2' regarding
Hb interaction with RBC ghost membranes in vitro. Our
data further suggest that, subsequent to the initial contact
step, formation of metHb is the next and rate-limiting step
and thatthis develops faster for HbS than HbA. This accelerated formation of metHb predictably would lead to enhanced likelihood for Hb denaturation (hemichrome formation)and heme loss.2.s,22Differences in experimental
conditions (eg, temperature) preclude direct comparison of
the present rate constants for Hb oxidation on lipid with
those we obtained previously for these Hbs in solution.*
However, it is valid to note that previous studies found that
Hb auto-oxidation in solution proceeds 1.7-fold faster for
HbS compared with HbA,' whereas the present data indi-
cate that Hb oxidation on lipid proceeds 3.4-fold faster for
HbS versus HbA. Thus, it is evident that lipid contact exaggerates the inherent instability of HbS and hastens its oxidative denaturation.
Indeed, as a result of Hb/PS interaction, we here observed
formation of the same hemichrome-like spectral characteristics thatare seen in sickle RBC rnembranes.l6 Significantly, this process is accelerated further and is accompanied by heme destruction, peroxidation of lipid, and
liberation of free iron if the lipid bilayer contains peroxidation by-products. In this regard, it is likely that the physiologically relevant by-product is lipid hydroperoxide,* and
sickle membranes containabnormal accumulations of such
material.23Thus, the sickle RBC is in greater jeopardy than
the normal RBC not only because it containsHbS but also
because of the peroxidation by-products in its membrane.
Because it appears that HbS can make contact with membrane lipid in the intact sickle red blood
this may be
the underlying explanation for deposition of hemichrome
and other iron formson the sickle membrane, a process believed to be proximate to development of various membrane defects.'.'
REFERENCES
I . Hebbel RP: Beyond hemoglobin polymerization: The red
blood cell membrane andsickle disease pathophysiology. Blood 77:
214. 1991
2. Hebbel RP: The sickle erythrocyte in doublejeopardy: Autoxidation and iron decompartmentalization. Semin Hematol 2 7 3 I ,
1990
3. Asakura T, Ohnishi T, Friedman S, Schwartz E: Abnormal
precipitation of oxyhemoglobin S by mechanical shaking. Proc Natl
Acad Sci USA7 I :1594, 1974
4. Adachi K, Kim J, Travitz R, Harano T, Asakura T: Effect of
amino acid at the 06position on surface hydrophobicity, stability,
solubility, and thekinetics of polymerization ofhemoglobin. Comparisons amongHb A (GLUD6),Hb C (LYSo6), Hb Machida
(GLNo6),and HbS (VALu6).J Biol Chem 262: 12920, 1987
5. Shviro Y , Zilber I, Shaklai N: The interaction of hemoglobin
with phosphatidylserine vesicles. Biochim Biophys Acta 687:63,
I982
6. Liu SC, Derick LH, Palek J: Surface areadensity ofmembrane
skeleton (MS) in normal red cells and severe hereditary spherocytosis (HS): Role in lipid bilayer destabilization. Blood 72:31a, 1988
(abstr, suppl)
7. Cullis PR, Hope MJ: Physical properties and functional roles
oflipids in membranes, in Vance DE, Vance JE (eds): Biochemistry
of Lipids and Membranes. Menlo Park, CA, Benjamen/Commings,
1985, p 25
8. Hebbel RP, Morgan WT, Eaton JW, Hedlund BE: Accelerated autoxidation and heme loss due toinstability of sickle hemoglobin. Proc Natl Acad Sci USA 85:237, 1988
9. Winterbourn CC: Reactions of superoxide with hemoglobin,
in Greenwald RA (ed): Handbook of Methods for Oxygen Radical
Research. Boca Raton, FL, CRC, 1988, p 137
IO. MacDonald RC, MacDonald RI, Menco BP, Takeshita K,
Subbarao NK, Hu L Small-volume extrusion apparatus for preparation of large, unilamellar vesicles. Biochim Biophys Acta 1061:
297, 1991
I I . Bartlett GR: Phosphorus assay in column chromatography.
J Biol Chem 234:466, 1959
From www.bloodjournal.org by guest on June 16, 2017. For personal use only.
HbS INTERACTION WITH LIPOSOMES
12. Kuross SA, Rank BH, Hebbel R P Excess heme in sickle
erythrocyte inside-out membranes: Possible rolein thiol oxidation.
Blood 71:876, 1988
13. Kuross SA, Hebbel R P Nonheme iron in sickle erythrocyte
membranes: Association withphospholipids and potential role in
lipid peroxidation. Blood 72:1278, 1988
14. Kimelberg H K Protein-liposome interactions and their relevance to structure and function of cell membranes. Mol Cell Biochem 10:171,1976
15. Akhrem AA, Andreyuk GM, Kisel MA, Kiselev PA: Hemoglobin conversionto hemichrome under the influence of fatty acids.
Biochim Biophys Acta
992: I9 1,1989
16. Asakura T, Minakata K, Adachi K, Russell MO, SchwartzE:
Denatured hemoglobin in sickle erythrocytes.J Clin Invest59:633,
1977
17. van Deenen LLM,Gier DJ: Lipidsof the red cell membrane,
in Surgenor DM (ed):The Red Blood Cell. New York, W ,Academic, 1974, p 147
18. Hebbel RP, Foker W: Accelerated autoxidation of hemoglo-
249
bin S after interaction with phospholipids predicts the excess heme
accumulation insickleredcell membranes. Blood 68:62a, 1986
(abstr, suppl)
19. LaBrake CC, Fung LWM: Phospholipid vesicles promote
human hemoglobin oxidation. J Biol Chem 267:16703, 1992
20. Szundi I, Szelenyi JG, Breuer JH, Berczi A: Interactions of
haemoglobin with erythrocyte membrane phospholipids in monomolecular lipid layers. Biochim Biophys Acta
595:41, I980
21. Shaklai N, Sharma VS, Ranney HM: Interaction of sickle
cell hemoglobin with erythrocyte membranes. Proc Natl Acad Sci
USA 78:65, 1981
22. Bunn HF, Jandl JH: Exchange ofheme among hemoglobins
and between hemoglobinand albumin. J Biol Chem 243:465, 1968
23. Sugihara T, Hebbel R P Exaggerated cation leak from oxygenated sickle red blood cells during deformation: Evidence for a
unique leak pathway. Blood80:2374, 1992
24. Sugihara T, Repka T, Hebbel R P Detection, characterization, and bioavailability of membrane-associated
iron in the intact
sickle red cell.J Clin Invest90:2327, 1992
From www.bloodjournal.org by guest on June 16, 2017. For personal use only.
1994 83: 242-249
Denaturing interaction between sickle hemoglobin and
phosphatidylserine liposomes
E Marva and RP Hebbel
Updated information and services can be found at:
http://www.bloodjournal.org/content/83/1/242.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.