Erythrocyte echinocytosis in liver disease. Role
of abnormal plasma high density lipoproteins.
J S Owen, … , H Isenberg, W B Gratzer
J Clin Invest.
1985;
76(6)
:2275-2285.
https://doi.org/10.1172/JCI112237
.
Echinocytes were frequently found in patients with liver disease when their blood was
examined in wet films, but rarely detected in dried, stained smears. When normal
erythrocytes (discocytes) were incubated with physiologic concentrations of the abnormal
high density lipoproteins (HDL) from some jaundiced patients, echinocytosis developed
within seconds. Other plasma fractions were not echinocytogenic. There was a close
correlation between the number of echinocytes found in vivo and the ability of the
corresponding HDL to induce discocyte-echinocyte transformation. On incubation with
normal HDL, echinocytes generated in vitro rapidly reverted to a normal shape, and
echinocytes from patients showed a similar trend. Echinocytosis occurred without change in
membrane cholesterol content, as did its reversal, and was not caused by membrane
uptake of lysolecithin or bile acids. Abnormal, echinocytogenic HDL showed saturable
binding to approximately 5,000 sites per normal erythrocyte with an association constant of
10(8) M-1. Nonechinocytogenic patient HDL and normal HDL showed only nonsaturable
binding. Several minor components of electrophoretically separated erythrocyte membrane
proteins bound the abnormal HDL; pretreatment of the cells with trypsin or pronase reduced
or eliminated binding. Echinocytosis by abnormal HDL required receptor occupancy, rather
than transfer of constituents to or from the membrane, because cells reversibly prefixed in
the discoid shape by wheat germ agglutinin, and then exposed to abnormal HDL, did not
become echinocytes when […]
Research Article
Find the latest version:
Erythrocyte Echinocytosis
in Liver Disease
Role ofAbnormal Plasma High Density Lipoproteins
JamesS. Owen, David J.C.Brown, David S. Harry, andNeil Mcintyre
AcademicDepartmentofMedicine,RoyalFreeHospital School of Medicine(University ofLondon), London NW32PF Gilbert H. Beaven,HarveyIsenberg, andWalterB.Gratzer
Department of Biophysics and MedicalResearch Council Cell Biophysics Unit,King's College (University ofLondon),
London WC2R2LS,England
Abstract
Echinocytes were frequently found in patients with liver disease when their blood was examined in wet films, but rarely detected in dried, stained smears.Whennormal erythrocytes (discocytes) were incubated with physiologic concentrations of the abnormal high density lipoproteins (HDL) from some jaundiced patients, echinocytosis developed within seconds. Other plasma fractions werenotechinocytogenic. There was a close correlation between the number ofechinocytes found invivo and the ability of the corresponding HDL to induce discocyte-echinocyte transfor-mation. On incubation with normal HDL, echinocytes generated
invitro rapidly reverted toanormalshape, and echinocytes from patients showed a similar trend. Echinocytosis occurred without changeinmembranecholesterol content, as did its reversal, and was not caused by membrane uptake of lysolecithin or bile acids. Abnormal, echinocytogenic HDL showed saturable binding to -5,000 sites per normal erythrocyte with an association constant of108
M-'.
Nonechinocytogenicpatient HDL and normalHDLshowed only nonsaturable binding. Several minor components ofelectrophoretically separated erythrocyte membrane proteins bound the abnormalHDL;pretreatment of the cells withtrypsin
orpronase reduced oreliminatedbinding.Echinocytosis by
ab-normalHDLrequired receptor occupancy, ratherthantransfer of constituentsto orfrom the membrane, because cellsreversibly prefixedinthe discoidshape bywheatgermagglutinin,andthen exposed toabnormal HDL, didnot becomeechinocytes when
the HDL andlectinweresuccessively removed. Binding didnot cause dephosphorylation of spectrin. We conclude that the
echinocytes of liverdisease are generated from discocytes by
abnormal HDL, andweinfer thattheshape changeis mediated by cell-surface receptors for abnormal HDL molecules.
Introduction
The initialassessmentof erythrocyte morphologyinliver disease is usually based onexamination ofdried, stained blood films.
Portions of this studywerepresentedattheMedical Research Society meeting in London (1981.Clin. Sci.
(Lond.].
60:2P)andattheInter-nationalAssociation fortheStudy of the Liver meeting in HongKong
(1982. Hepatology. 2:150).
Address reprint requeststo Dr. Owen, Academic Department of
Medicine,RoyalFredHospital School of Medicine(Universityof
Lon-don),Rowland HillStreet, LondonNW32PF, England.
Receivedfor publication27April 1983 andinrevisedform24 June 1985.
Unfortunately, drying may cause shape artefacts to which wet preparations and scanning electron microscopy(SEM)'are not subject. "Target" cells are common in stained smears from jaundiced patients, but result from distortion because they are
codocytic(bell-shaped) in wet films. In 1968, Grahn and col-leagues (I) found spiculated erythrocytes in wet blood films from
patientswith liver disease; they were not evident in dried, stained specimens. This observation has largely been disregarded. They
pointedoutthat these cells (whichweshall call
echinocytes-seeMethods)werealso present in "spur cellanemia,"ahemolytic
disorderseenoccasionally in severe alcoholic liver disease and characterized by the presence of acanthocytes2 (1-4). Acantho-cytes are bizarre, irregularly spiculated cells recognizable in stained smears, aswellasinwetfilms.
Grahnetal.(1)foundthat theserumfrom bloodcontaining
echinocytes causedechinocytosisofnormalerythrocytes;serum
from blood that contained acanthocytes also induced formation
ofechinocytes, butnotof acanthocytes. They therefore argued thatan additional factoracting in vivo mightbe necessary to
produce acanthocytes and suggested that the spleen might be involved. They were unable to identify the serum fraction(s)
responsiblefor theformationofechinocytes.
HuiandHarmony (5) produced echinocytes byincubating
normal erythrocytes with isolated normal lowdensity
lipopro-teins (LDL). The active constituentwasapparently
apolipopro-tein B(apoB), inasmuch ascyclohexanedione blockage of the arginineresiduesof apoB reducedechinocytogenicpotency(6).
Thehigh densitylipoprotein(HDL)of patientswith liver disease is often enriched in apoE, which competes with the apoB of normalLDLfor "apoB,E-receptors"onthesurface ofnucleated cells(7).Wehavethereforeinvestigatedwhether abnormalHDL
fromjaundiced patients, like normal LDL, wouldcause
echino-cytosis. We found echinocytes in patients with liver
disease;
morphologicallysimilar cellswereformedrapidlywhen normal erythrocytes were incubated with abnormal HDL from thesepatients. Echinocytosisdidnotresult from additionorextraction
of membrane components; theimportant factor seemed tobe
reversible,saturablebindingof abnormal HDL
particles
tothe erythrocyte surface.Methods
Patients. HDLwere isolated from 18 inpatients withliver disease of varying severity (Table I). Diagnostic methods included liver biopsy,
1. Abbreviations usedinthis paper: apo, apolipoprotein; LPDS,
lipo-protein-deficientserum;SEM,scanningelectron microscopy.
2. Theseirregularyspiculatederythrocyteshavealso beentermed"spur
cells"and"spur-shapedcells"(2,3)or"burrcells"(1);we
prefer
touse the nomenclature of BrecherandBessis(9),asdescribed in Methods, andcallthem"acanthocytes."J.Clin. Invest.
©TheAmericanSociety for Clinical Investigation,Inc.
0021-9738/85/12/2275/11 $1.00
Q%ON 0 " 0 O "1 000 o000w1 0 0 ,Q
) Oen en, O t-- O
zzzz 0O O
Z Z Z Z Z CmkN 0% 0 t 0l "Ft 'I
____t" oo00o00
00o O tO 0 - O
0
kn 0,0 ON -0O OCIA O 000
00 0 ^ 0 N 0 eC 00ONq -N - 0 - N
-ON IDN t 00 -0 (ON-s 0
- O" O M ON
%000 \0 "o N -0 (N -0 00) C0N .0 e 0
eQ0(N0000 ...(n e
r-~ Ne en (N (N (N e^ (N (N e^
Cr
-CrC
- N Cr tN- "o t N 0 000 (N t N t a-M - d 20 0 Rt - N n N n (N C
(N 0 %o Rt -N C -
--ON
o>-00--N>o-£-e~ -- - _-e , - d
C,~~~ ~ ~ ~~~~~~~C
0 CNMCrCo 0%00Se'Ce o(008 o0Cr0%N
(N CDC e~ t'- S-(ON C tn e r
Et°.E2! 2 : E Dt 6 t6U 6
0V
V-C
.0. 20.~~ 00.
.0
~0'~~ 0 . 0
0o0 . 00 .0.O .0
2~~'~20 2 ~~ 2
2a
U
0 C .0 Cu
'u<', 0
<
.5
tn tn tN e
" ew Nn (nN
N
(N 00 00
n
0 (N m 0I
m Cfvn (N 00 m o N ^ o eC (N m eo^ '
R)imXSU~~~~~x3 =2 cvt[; 2z*0iX s 6 2t L
0
ai =.0
0
a
.a
,a=
* r.FF
06
Q
_.
.Q
za
_Z
O:z c
t.
a,
.0 _
.0%
0',
E-1
'A
.m
.~
°,
_o=
( '
'0_c o
-._
2 _
=
2-o *,
cd C.) Cu 0
00
CZ_ r
D
C.-, x
'x 3
0 0 0
.Ctc
Cd-^ =>
~ _
CuU
CZ~
0.0
Cd
0. .
- _) 00
0F _ C..
C.o C.. Cu
u 3;0
0 0.= e Z ° _ o
b.
cC .2°-=
-; 0.l
_C);
2276 Owenetal.
I
2.-1.
-DI0
.I
cholangiography, and/orsurgery.Five patientshadobstructive jaundice
(three intrahepatic, twoextrahepatic): thirteen had parenchymal liver
diseaseand twoofthem (M.F. and M.T.), both with hemolyticanemia, showed acanthocytes in dried blood smears. In the last 13patients
in-vestigated, erythrocytes were examined in wet films and by SEM. Re-versibility of echinocytosis was studied in 30 additional patients, mostly
alcoholiccirrhotics, selected because>10%of theircells wereechinocytes.
HDL were isolatedfrom three patients (K.N.,M.J.,and S.J.) with
abe-talipoproteinemia,andtheirerythrocytemorphologywasstudiedbySEM.
ThesepatientslackapoB-containinglipoproteins and their clinical details have beenreportedpreviously(8).Normalsubjectswerehealthy medical and laboratorystaff.
Reagents. Human serumalbumin (<0.005% fatty acids)and allother
materialswerefromSigmaChemicalCo., Poole, England, except Hanks'
balancedsaltsolution(FlowLaboratories Ltd.,Irvine, Scotland),
Cab-O-Siland Mercksilicagel (BDH ChemicalsLtd., Poole, England), and pronase(Calbiochem-Behring Corp.,SanDiego,CA).
Lipoproteins. Very low density lipoproteins(VLDL, d< 1.006 g/ ml), LDL (1.019-1.063), HDL (1.063-1.21) and lipoprotein-deficient
serum(LPDS, d> 1.25)wereisolated by differential ultracentrifugation
and washed oncewithbuffer ofthe appropriate density(7). The lipo-proteins were usedwithin a day or twoof isolationand were passed through a
0.2-mtm
filter before additionto erythrocytesuspensions;tech-niquesusedforcompositional analysis, labeling with 125I,andchemical modificationhavebeen describedpreviously(7).
Erythrocyte isolation, morphology, and analysis. Erythrocytes were separatedbycentrifugationat40C ofvenousbloodanticoagulated with
10U/mlheparin. Theywere washed threetimes withHanks'solution, broughtto ahematocritof'-50%andfixed for 30min at0C by adding
anequal volume of 1% glutaraldehyde in Hanks' solution. Morphology
was examined by light microscopy of wet films and frequently by SEM
(Philips501). When whole blood was fixed, erythrocyte shape appeared
identicaltothat in washedpreparations.
UseofSEM hasallowedreclassification of erythrocyte shape anda betterinsight intotherelationshipbetweenspiculated forms (9).Inliver disease four"spiculated"erythrocyteshapescanberecognized in addition
tonormalbiconcave discs(discocytes).These arediscswith protuberances around theedge(echinocytes I), flat cells with spiculesoverthe surface (echinocytesII),sphericalorellipsoidalcellswithalargenumber of spic-ules evenly distributedoverthe surface (echinocytes III),andcells with
abulbousbody and a small numberof irregularlyspacedspicules,
gen-erally with knobbly, club-shaped ends (acanthocytes) (Fig. 1). These shapesmaygenerallybedistinguishedbylight microscopy ofwetfilms,
Figure1.Scanningelectronmicrograph oferythrocytesfromapatient
(M.F.,TableI)with alcoholic cirrhosisand associated hemolytic
ane-mia.Examples of acanthocytes (AC),and type I(E-I),type II(E-II),
and type III(E-III) echinocytesarelabeled. (Inset) Discocytes froma
normalsubject. Calibrationbar= 10Aim.
but we suggest that the identification of acanthocytes should rest on SEM. Using light microscopytheproportion of echinocytes IIand III
in a cell population was determined by counting 200 cells in three to four fields; the observer was unaware of the source of material.
Erythrocyte and lipoprotein lipids were extracted with isopropanol-chloroformandchloroform-methanol, respectively, and the concentra-tionsof individual phospholipids measured as described previously (10,
I1). Total bile acids were determined by the technique of Barnes and Chitranukroh (12).
General incubationprocedures.Washederythrocytes (usually 10 !d containing 6X 109 cells/ml of Hanks' solution) were incubated with4
vol oflipoprotein or reagent. Lipoprotein concentrations are expressed
asmilligram of protein per milliliter, andallconcentrations of lipoproteins
orreagents refer to the final incubation mixture. Incubationswerestopped by fixation with I volof1%glutaraldehyde or, if addition ofasecond reagent was required, by diluting with 50 vol of Hanks' solution, followed by immediatecentrifugation and a secondwash.Further detailsaregiven in legends to our tables and figures. Washed erythrocytes were enriched incholesterol using cholesterol-rich phosphatidylcholine dispersions (13). Binding studies. Washed erythrocytes (25 glof a 10%suspension) were added to 250zdofvarying concentrations of1251-labeledlipoprotein with 0-40mg/mlhumanserum albumin and incubated for 40 min at room temperature with periodic agitation. Cells were separated from unbound lipoprotein, by brief centrifugation of the suspension (200JA)
throughabarrier (50
jil)
of dibutylphthalate, andbound1251 was mea-sured.Determinationsweregenerally made in quadruplicateandblanks (without erythrocytes) were subtracted; the correction was 10%.Toidentify binding components, membrane proteins were first frac-tionated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electro-phoresis (14). The protein zones were electrophoretically transferred to
anitrocellulose membraneandnonspecific binding sites blocked by ag-itationfor 24 hin 10 mMTris-HCI, 150mMNaCI, 1 mMEDTA, 2 mM sodium azide, pH 7.5, containing 25 mg/ml bovine serum albumin ( 15). The bufferwasthenreplaced by '25I-labeled patientHDL at
0.1-1.0mgof protein/ml in the same buffer. After2hof gentle agitationat
room temperature the nitrocellulose membrane was rinsed, dried, and leftincontact withpreflashed Kodak X5 film at -70°C for upto 4h. Duplicate lanes of the original SDS gel were stained with Coomassie
blue.
Washed erythrocytes (10% hematocrit) were subjected to proteolysis by incubating for 1 hat37°C with 0.5 mg/ml of trypsin, chymotrypsin,
orpronase in 5 mM sodiumphosphate, 150mMNaCl,pH 7.4. The cellswerewashed, membraneswereprepared, and theirprotein
com-positionswereanalyzed bySDS-gel electrophoresisin0.1MTris-Bicine
N,N-bis(2-hydroxyethyl)glycine,pH 8.1 (15).The"dot-blot"technique was used to establish whether binding of patient HDLstill occurred; membraneproteinwas spottedontowashed nitrocellulose, incubated with radioiodinated HDL and theamountof binding measured.
Other studies. For measurement ofphosphorylationchanges inthe
spectrin of the membrane cytoskeleton, erythrocytes were prelabeled
with 32P (16) and incubated with lipoproteins. Cells were removed at intervals and their membraneproteinswereseparated byelectrophoresis
( 14)andstained withCoomassie Brilliant Blue G. Relativeconcentrations
ofspectrinwereestimated by dye elution with 70% (vol/vol) formic acid and the 32Pcontentby Cerenkov counting.
Erythrocytes(107/ml)weretreatedwith wheat germagglutinin (20
,ug/ml),
incubated withabnormal HDL orotherechinocyticagents forI minat37°C,andwashed threetimes; bound lectinwasdisplaced by
additionof 20 mMN-acetylglucosamine (17). Erythrocyteswerealso treated with the oxidant diamidetointroducedisulphidebonds
predom-inantly between spectrin chains(18).Cross-linkswereremovedby ad-ditionof 15mMdithiothreitol.
Results
echino-cytesIIand/or III inwetfilms (mean 47%,range 10-98%). Fig. 2 A and Bexemplify the morphologic variation at the lower (14%) and upper (83%) ends of the range. Discocytesand
echi-nocytes I, II, and III could be seeninmostpreparations: spec-imens with the highest degree of morphologic abnormality
con-tainedmainlytypeIIechinocytesand those with the least mainly type I. No acanthocytes were seen in this group of 30 patients. Twoofthe 18 patients (M.T. and M.F., Table I), from whose blood HDL wasisolated, were alcoholic cirrhotics with a
he-molytic anemia and "spur cells." Acanthocytes were evident bySEM(Fig. I andFig. 2 C)butcomprised only4-6% ofthe total erythrocyte population;mostcellswereechinocytes II or
III.Erythrocytes in the other 16patients showed variable echi-nocytosis but no acanthocytes inwetfilmsorbySEM; in only two (A.S. and S.F.) were spiculated cells reported on routine
examination of dried,stained smears. Light microscopyof wet
films allowed discrimination of the various types of echinocytes (Fig. 3 A) and acanthocytes(Fig.3C).LikeGrahn et al.(1), we found almostnotypeII orIIIechinocytesin dried, stained films (Fig. 3 B), but acanthocytes were readily recognized (Fig. 3 D).
In all three patients withabetalipoproteinemia small pro-portions of acanthocytes were present, but most cells were
echinocytes IIandIII(Fig. 2Dand E).
Effects
ofliver diseaseHDLonnormalerythrocytes.Normal erythrocyteswereincubated for I minat370C withHDL fromjaundicedpatients ataconcentration similartothat in plasma
(500 ,ug/ml) (19, 20); echinocytes I, II, or III appeared in the
suspension (Table I, Fig. 4). With some patient HDL, only
B
I
echinocytes I and II were formed (Fig. 4 A); with HDL from others(A.S., M.T., and J.R., Table I) complete conversion to type II orIIIechinocytes occurred (Fig. 4 B). The ability ofa
patient's HDLtotransform normalerythrocytes correlated with the proportion of echinocytes in the patient's blood (r= 0.92, P<0.001;TableI). Less than 1% of normal cells became
echi-nocyteswhenincubated with HDL from normal subjects. Echi-nocytosis occurred within the time requiredtostopthe reaction (>5 s);nofurther changeswere seenwith incubation forup to 4h. The degreeof echinocytosiswasdependenton HDL
con-centration butindependent oftemperature,giving thesameresult
at0°C andat 37°C aftera 1-min incubation (Fig. 5).
Severalpatients' VLDL, LDL, HDL, and LPDSwereisolated and incubated with normal erythrocytes for Iminat37°C.Only
the HDLand LPDS fractions were echinocytogenic at
concen-trations equaltothose in the original plasma. LPDSwasmuch less effectivethan HDLbutmaystill have contained lipoprotein
components, derived artefactually by ultracentrifugation (21), becauseit became inactive ontreatmentwithCab-O-Sil. VLDL, present at lowconcentration in severe parenchymalliver disease (20), becameechinocytogenic only when concentrated above1
mg/ml. PatientLDL was inactive except with longer incubations (>1 h) and higher concentrations (>2 mg/ml). Suspending the washed cells in normal plasma or normal LPDS before addition
of abnormal HDL causedonlyminor reduction inthenumber
ofechinocytes produced. When active patientHDL was incu-bated overnight at 4°C with normal LPDS and recovered by untracentrifugation, it retained its echinocytogenic ability;
nor-Figure2.Scanningelectronmicrographsof
erythrocytesfrompatientswith liver diseaseor abetalipoproteinemia,and of cholesterol-en-riched normalerythrocytes.(AandB)Examples
of the various types ofechinocytesseenin liver ~~~
~disease.Theacanthocytes,of liver disease(M.T.,
TableI)andabetalipoproteinemia,(K.N.and ~~ 4~ S.J.,Reference8)areillustrated in C, D, and E,
respectively. (F) Erythrocytesenriched in
choles-terolbyincubation in vitro witha cholesterol-richphospholipid dispersion (cholesterol/phos-pholipidmolar ratio= 1.38comparedwith 0.85
incontm]rolcels.0
iolk
J vLJ-1_.i
Add6
'@
Cow.it
by
"Op
a ~
~~
CHaJ024w
-.
Figure 3.Light microscopyof wet filmsand dried stained blood smearsfrom patientswith liver disease. (A) Echinocytes are clearly
vis-iblewhen wet films are examined from patient P.K. (TableI)but are
mal HDLdid notacquire shape-transforming properties after
incubation with patientLPDS.
Reversibility of the discocyte-echinocyte transformation. Echinocytes formed inseconds retained their abnormal shape
onprolongedincubation(upto 4h) withthepatient HDL, and alsoshowed little reversalafter washing and subsequent rein-cubation in Hanks' solution for up to 8 h. When they were
incubated with normalHDL therewas completereversalto
nor-mal erythrocyte morphology within seconds (Fig. 6). Normal LPDSinducedonlypartial reversaleven athighconcentrations.
Reversibility ofechinocytesfoundin liver disease. Normal
HDL(8mg/ml) wasaddedtosuspensions of erythrocytesfrom
rarely seenin driedsmears(B). Inapatient with "spur cell"anemia (M.T., TableI) acanthocytescan bedistinguished fromechinocytesin
wet films (C) and are also recognized in driedfilms (D).
30 patients with liver disease, whose blood contained at least 10% echinocytes II and III. In each case there was a rapid re-duction in the percentage of type II and IIIechinocytes(Fig. 6 andFig. 7). Complete reversal was rarely achieved, possibly be-cause the cells had adapted to theirnew morphology during
prolongedexposure totheechinocyticagentand reversedonly slowly when the agent was removed (22). Excess normal HDL alsoameliorated the echinocytosisof cells from the two patients
with spurcellanemia but the proportion of acanthocytes was
unchanged,suggesting that acanthocytosiswasirreversible.
Binding studies.Lipoproteinbinding by erythrocytes reached
equilibrium within 30 min, but the small number of sites per
Figure4.Ability ofHDLfrom patients with liver diseasetotransformnormalerythrocyte into echinocytes. Normal, washed
erythrocytes (6X 109cells/mlHanks'buffer)wereincubated for 1 min at
370C
withfreshly isolatedHDL(finalconcentra-tion 500jgofprotein/ml) from patients S.F. (A)andA.S. (B). v!p --
ro-V.S.. 0
4
i15
4.-
%Al
v
a
61
.
9
sb^.I
4 c
f4l04;
-01 -.rl A.
r4f
w
(nb
Mi
0
4rp
la
1oo
1-,
0)4
*-,
0-cn
CU
C-,
LL
be-
75-50
251
.00
-/- - A
-
-0' I;_.-- - - --
-0 25 50 15 100
Echinocytogenic Agent (arbitrary units)
Figure5.Transformation of normal
erythrocytes
toechinocytes
at00C
and370C. Normal,washederythrocytes (6
X109
cells/mlHanks'buffer)
wereincubatedfor 1 minateither0C(opensymbols)
or370C
(solid symbols)
withincreasing
concentrations offreshly
isolatedHDLfrom
patient
M.A-M.(o,.;
100U= 1 mgofprotein/ml),
lyso-lecithin-enriched
plasma
(0,
.;
100U=430 nmollysolecithin/ml
plasma)orplasmaenriched with bile salts(a,,; 100U=2.5jmol
totalbile
acids/ml
plasma).Thecellswerethenfixed and the percent-ageofechinocytes
IIand IIIdetermined inwetfilmsby
differentialcounting
of 200cellsin threetofourfieldsusing
light microscopy.
celllimited
precision
ofmeasurement.Strongly echinocytogenic
HDL
(patient
A.S.,
TableI)
was boundby erythrocytes
in asaturablemannerwhen added in
increasing
concentration(Fig.
8).
Saturation of sites wasachieved atan HDL concentration of~-40
nM(10
,ug/ml),
atwhichpoint
conversiontoerythrocytes
II and III had reached a
plateau
at -85%. Labeled HDLwasdisplaced by
unlabeledHDL,
andcountswerealsoprogressively
lostby
repeated
washing
of the cells. Thesefindings
areconsistent with a reversiblebinding
process, andassuming independent,
noninteracting
sites,
thebinding
curve canbesatisfactorily
fittedby
asingle binding
processwithanassociationconstantof-X 108M-1
(inset
toFig. 8).
Thereappeared
tobe4,000-5,000
sitespercell.Defatted albumin reducedtheamountof abnormal
HDL
bound,
butalarge
partwasnotdisplaced (Fig. 9). Thus,
despite
thefit(within experimental error) by
asingle
associationconstant, theremaybemorethanonetypeof
site;
onepopulation
being
specific,
the othernonspecific
andavailableforcompetition
by
albumin. Such studiesrequired
large
amountsofHDLandonly
oneotherpotentHDLwasexamined(patient
M.T.,
TableI);
this also gaveabinding profile
consistent with the presenceof saturable sites.
By
contrastnormalHDL,
andpatient
HDL withlittleor noechinocytogenic
activity,
didnotappeartobindtothesites.No saturable
binding
wasdetected,
andinstead therewas acontinuous
uptake
oflipoprotein
in themannerofa par-titionprocessoverthewide concentration rangetested(0-6,000
nM,
data notshown).
NormalLDL,
which did induceeryth-rocyte
shape changes
onprolonged incubation,
also showednosaturable
binding.
Thisdisagrees
with the report of Huietal.
(6).
The dot-blot
technique
indicated thatbinding
of radioio-dinatedpatient
HDLcould beaffectedby proteolytic
pretreat-mentof
erythrocytes
andpronasetreatmentlargely
eliminatedbinding. However, trypsin caused only partial reduction of binding and chymotrypsinwasineffective,possibly because these
proteasesdegradeonly certain of theexternally accessible
pro-teins(23)orbecause cleavedfragments remain associated with the residue of the parent molecule(24). When labeledpatient
HDL was incubated with the electrophoretically
separated
membraneproteins several bandsweredetectedby
autoradiog-raphy. Thesewereweak, presumably because the small number of sites hadrelatively modest associationconstants,and didnot
obviouslycorrespondtomajor membraneproteinswhen stained
withCoomassie blueorwithperiodate-Schiffreagent.The pro-nase-sensitive membraneproteins which bound patient HDL
wereapproximatelyin thepositions of bands 3, 4.9 (or 5), and 6 in the Fairbanks numberingsystem(14). This heterogeneity
suggests that thereceptors areglycoproteins,manyof whichare
presentin erythrocyte membranes atlow concentrations(25),
somedifferingonly in theircarbohydrate moieties.
Spectrin andechinocytosis. Phosphorylation ofspectrin in the erythrocyte cytoskeleton was unaffected by lipoproteins
withinthe incubationtimerequired for shape changes viz.
sec-ondsfor abnormal HDL andup to Ih for LDL(Fig. 10). The
result forLDL conflictswith that of Hui and Harmony (26).
Low concentrations of diamide were added toerythrocytesto
causelimited thiol oxidation with the formation ofcross-links,
mainly between spectrin molecules (18). These cells became partiallyresistant topatient HDL;45-50%oferythrocytes treated
with diamide remained discocytic after addition of abnormal
HDLcomparedto <I0%ofuntreated cells. Theprotective effect
of diamide was reversed on reduction of the cross-links with
dithiothreitol.
Apolipoproteins andechinocytosis. The proportion ofapoE
in HDL from our 18patientswashigher(I13.2±7.3%, mean±SD, range 4.0-30.0%) than that in HDL from 10normal subjects (3.9±1.0%, range 2.6-5.4%; P < 0.001),butthere was no
sig-nificantcorrelation withechinocytogenic potency (r=0.26, P
> 0.05; calculated from Table I). Agents that prevent apoE
binding by apoB,E receptorsonnucleatedcells,e.g.,protamine (200 utg/ml)orheparin (10 mg/ml),orcyclohexanedione mod-ification ofapoliprotein arginine residues (7), did not impede
echinocytosis. Moreover, discocytes from patients with homo-zygous familial hypercholesterolaemia, which lack apoB,E
re-ceptors, were readily transformed to echinocytes. HDL from
threepatientswithabetalipoproteinemiaalsocontainedexcess
apoE (10.5±0.8%, range 9.6-11.1%)but were nonechinocyto-genicatconcentrationsupto 1.5mg/ml; presumablyother
fac-torsare moreimportantforechinocytosisinthesepatients, such
asthe increasedsphingomyelincontent(8) in the cell membrane.
Twoadditional,morebasicapolipoproteins, previously described (7), were found in HDL from nine of the patients with liver disease,but theirpresencedid not reflectechinocytogenic activity
(r
=-0.31,
P>0.05).Relationofchangesinerythrocyteshape to erythrocyte
cho-lesterol content. Echinocyteformation by patient HDL, as
ex-pectedfrom the rapidityof theprocess,was notaccompanied
bynetcholesteroluptake (TableII). Furthermore, the lipid
con-tentof theseechinocytes,and those ofpatients, did notchange
onincubation with normal HDL and reversion tonormal shape. PartialdelipidationofpatientHDLremoved 92%of cholesterol,
24% ofphospholipid,andessentially all ofthecholesterylester andtriglyceride,butdidnotaffectechinocytogenic potency. We
found,inagreementwithothers(I13, 27, 28), thatnormal eryth-rocytesbecamebroader and flatteronenrichment with
A
V*m
1,
-wFigure 6.Reversibility of echinocytesproducedby incubation of nor-mal red cells with HDL isolated from jaundiced patients and of
echino-cytes foundinliver disease. Normal erythrocytes weretransformedto
echinocytes IIand IIIby HDLfrompatient A.S. as described in Fig. 4 (A).Afterwashing, these echinocytes were incubated for I minat
brane cholesterol byequilibrationwithcholesterol-richliposomes
(Fig. 2 F); many showed peripheral distortion as in typeI echi-nocytes. Type II echinocytes were rare, and probably resulted from prolonged incubation, in that they also appeared in
con-trols.
Amphipaths andechinocyteformation. Small amounts of amphipathic anions or other ligands causeechinocytosisby
par-titioning mainly into theouterleaflet of the membrane bilayer andexpanding its area relative to the inner leaflet (29-31). We studiedtwophysiologic amphipaths,bile salts and lysolecithin,
todetermine whether they could be the active components of abnormal HDL. Echinocytosis by bile salts orlysolecithin,and its reversal by addition of normal HDL, was rapid and the
echinocytes produced resembled by SEM thosegenerated by
patient HDL. However, HDL-inducedechinosytosiscould not
beexplained by detectable increasesoftheseamphipaths in active HDL;freshly isolated echinocytic HDL contained no more ly-solecithin than normalHDL(1-2% oftotalphospholipid)and had a low totalbile acid content (2-9 nmol/mg protein) when comparedtoinactiveHDLfrom fourpatientswithextrahepatic
obstructive jaundice (1 1-17 nmol/mg protein). Furthermore, we foundthe echinocytogenic effects of bile salts,lysolecithin
andpatientHDL todiffer. Normal plasmawasenriched with eitherlysolecithin (by incubationat37°C)orbile salts(byadding equal amounts of sodium taurocholate and sodium
glycoche-nodeoxycholate)untilit produced 65-85% typeIIandIII echin-ocytes on I minof incubation at 37°C. In contrast topatient
w_-ism
S
370CwithnormalHDL(finalconcentration 8mgofprotein/ml)to
give completereversal todiscocytes (B). Echinocytes fromapatient withalcoholiccirrhosisalsoreturned towards a normal erythrocyte
shapewhensimilarly incubatedwith normalHDL;thepercentageof echinocytesII and IIIpresentwerereduced from 14%(C)to zero(D).
HDL,far less echinocytosis occurred when the incubations were repeated at00C(Fig. 5). Echinocytesformedby exposuretobile salts revertedto discocytesafter three washes with Hanks'
so-lution; those generated bylysolecithin or by patient HDLdid
not revert. Washed echinocytes produced by lysolecithin
ac-quiredamorenormalshapeafter7-8hofincubation, presum-ably becauselysolecithin progressivelyenteredthe inner
mem-braneleaflet(31), but thoseproduced by patientHDLdidnot reverse. VLDL, LDL, HDL, and LPDS were prepared from plasma enriched withlysolecithinorbile salts and washedonce
byultracentrifugation, andtheir volumeswereadjustedtothat of the original plasma sample. Only the LPDS fractions were
foundto beechinocytogenic;the HDL produced <1% type II andIIIechinocytes.
These resultsseem to us toexcludelysolecithinandbilesalts
asthe active constituents ofechinocytogenic HDL. To assess
whetherother, sofarunrecognized,amphipaths mightbe present we(reversibly) fixed a erythrocytesuspensionin thediscocytic shapewith wheat germagglutinin (17). Aliquotswereincubated either with plasma enriched inlysolecithinorinbile salts
(pre-paredasabove) orwith abnormal HDL. In each case
echino-cytosiswasmarkedly inhibited,thenumberoftype II and III
echinocytes produced being reduced from 65-85%to 3-10%. Reagents were removedbywashingthreetimeswith 50 vol of
Hanks'buffer (without albumin)andthen the boundlectinwas
dissociated by addition ofN-acetylglucosamine. Cellsexposed
tolysolecithinbecamepredominantly echinocytic;asexpected,
100r
80
.
60
IV
G)
8
40k
20L
oL
+
Buffer
+ NormalIHDL
Figure 7.Reversibilityofechinocytes found in liver disease by normal HDL.Suspensionsof washed erythrocytes (6XI09 cells/ml Hanks' buffer) containing at least 10% ofechinocytesIIandIII wereobtained from 30 patientswith liver disease, mainly alcoholic cirrhotics. One portion of the cellswasincubated with normalHDL(final
concentra-tion 8 mg ofprotein/ml)for 1 minat370C and another with Hanks' buffer. The cells were thenfixed andthepercentage ofechinocytesII and IIIdetermined in wetfilms by differential counting of 200 cells in threetofour fieldsusinglight microscopy.
0.6
S0.4
0 0
C~~~~~~~~~
0~~~~~~~~~u0.2 -£
-IH~l~e h
0 VI0.40 S8
2
0.1 0.2
HDJ1
4fHL~ree(OM)
Totalconcentration of HDL (nM)
Figure8. Saturablebinding of'25I-labeledliver disease HDL by
nor-malerythrocytes.Cells were incubated for 40 min at room
tempera-turewithincreasingconcentrations ofHDLfromapatient with
lu-poid hepatitis(A.S., Table
I).
Eachpoint isthemeanof four determi-nations; bars areISD. The molecularweight of the HDL wasassumed to be 0.25X 10'andthe curve ( ) is calculated for 4,500
bindingsites per erythrocyte andK.= I X 108MW.Inset:double
re-ciprocalplot ofl/[HDL]boundvs.
l/[HDL]f,.
°5 0.34 % 0
I \
No 0.2- °
C, 0.1'
0 10 20 30 40
Albumin (mg / ml)
Figure9.Inability of human serum albumintopreventbinding of liver disease HDL by erythrocytes. ThepatientHDLand incubation conditions were as described in Fig. 8 except that each tube contained up to40 mg of defattedhumanserum albumin per milliliter, Final concentrations ofpatientHDL wereeither31 ug ofprotein/ml (o)or
14
,ig
ofprotein/ml(-).the membrane of lectin-fixed cells took up theamphipathand
retained it during washing andlectin dissociation. Cellstreated with bile salts remained discocytic, because washing reverses bile salt-inducedechinocytosis (unlikethat inducedby lysolec-ithinorby patient HDL,seeabove).Cellsexposedtoabnormal
HDLremainedlargely discocytic, suggestingthat abnormal HDL does not actby adding (orremoving) lipophilicmembrane
con-stituents;it must occupy its receptors to beechinocytogenic.
Discussion
The literatureonerythrocyte shapein liver diseaseisconfused
byinconsistentterminology.The term"spur cell,"whichshould
be reservedforacanthocytes,has often been used forspiculated cells,seenin vivo orgeneratedinvitro,that are in factechinocytes (as defined by Brecher and Bessis [9]). Spurcell anemia, (i.e., with acanthocytes)is rare, but we havefrequently found echino-cytesinjaundiced patientswhen their blood was examined in
wetfilms; unlike acanthocytes, they were notusually evidenton
routine examinationofstainedsmears. Theproteinor
lipopro-Tome (min
Figure JO.Inability of liver disease HDL,or normal LDLand normal
HDL,todephosphorylatespectrin innormalerythrocytes.Erythrocyte phosphoruswaslabeledbypreincubation with32Piand then incubated for upto 1 hwith buffer (o),HDLfromapatient (M.T., Table I) with alcoholic cirrhosis(0;0.5 mg ofprotein/ml), normalHDL(n; 1.0mg of protein/ml)or normal LDL(-; 1.0mg ofprotein/ml).Membrane proteinswereseparated bySDS-polyacrylamide gel electrophoresis (14) andthe bandscorresponding tospectrin were removed. Their32p
contentwasmeasured by Cerenkovcountingandthevalues were
nor-malizedto aconstantspectrinconcentrationasestimated by dye elu-tionofthestainedbands.
2282 Owenetal.
;400 .
m11 -i
E O' 3(W
c a a
i
0 w .
sw-4.
1.200
I2
I
0 0.100
t
0 .G
CL
0-Table II. Unaltered Erythrocyte Lipid Composition after HDL-induced Changes in Erythrocyte Shape
'5 Total lipidphosphorus
Erythrocyte source Cholesterol Phospholipid LL§ SM Pi PS L PE PA
nmti/h107cells nmol/10'cells % % '5 5 % %
Normal cells 3.55 4.08 1.1 21.1 0.6 13.5 29.1 27.5 2.1
Transformedcells* 3.54 4.17 1.0 26.9 0.5 12.8 29.6 27.4 1.8
Transformed cells reversed with normalHDL* 3.68 4.10 1.1 27.8 0.5 12.9 29.4 26.3 2.0
Patientcells 5.68 4.32 0.8 18.9 0.7 11.8 43.7 22.2 1.9
Patientcellsincubated with normal
HDL*
5.86 4.44 0.9 18.6 0.5 12.4 42.9 22.5 2.2*Washed,normal erythrocytes (6 X 109cells/mlHanks'solution) were incubated for I minat
370C
withfreshly isolated HDL from patient M.T.(finalconcentration 250
,gg
ofprotein/ml).All cells weretransformed to echinocytes II orIII.Afterwashing, one portion of the transformed cells wasincubated for I min at370Cwith normalHDL(8mg/ml)to give complete reversal todiscocytes. Similar results were found in a furtherexperiment usingHDLfrompatient A.S. 4 Washederythrocytes (6 X 109cells/mlHanks'solution) from patient A.S. were incubated for I min
at37TC withnormal HDL(8 mg/ml) toreducethepercentage of typeIIor III echinocytes (from 69% to21% of the cells). The experimentwas
repeated twiceusing erythrocytes from two additional patients (M.T. and M.F.); the same conclusions were drawn (data not shown). §
Abbrevia-tions: LL,lysolecithin;SM,sphingomyelin,PI,phosphatidylinositol;PS, phosphatidylserine; L, lecithin;PE, phosphatidylethanolamine; PA,
phos-phatidic acid.
tein fraction(s) present in liver disease plasma that is responsible forechinocyteand/oracanthocyte formationhas notpreviously
been identified (1, 32, 33) but our results establish that the
echinocytogenicagent canbeisolatedwith abnormal HDL.
Inliver disease, erythrocyte membranes tend to acquire ex-cesscholesterolfrom abnormalcirculatinglipoproteins (34, 35), butthis is unlikely to explain echinocytosis, inasmuch as spic-ulated cells have been found unaccompanied by elevation in membrane cholesterol content(32) andviceversa(36). The shape ofcells loaded with cholesterol inourexperiments,and in others (13, 27, 28), bore little resemblance to the echinocytes
ofjaun-diced patients and could not be reversed by briefincubation with normal HDL. Cholesterol enrichment was evidently un-related toechinocytosis inducedin vitro by abnormal HDL, for thelatter occurred in seconds, without increase in membrane cholesterol, and could also be induced by cholesterol-de-pleted HDL.
Two physiologic amphipaths producing echinocytes, lyso-lecithin and bile salts,wereexcludedastheactive constituents ofpatient HDL; echinocytogenesis by these agents, and the
propertiesof theresultingechinocytes, differed in several ways from those induced by abnormal HDL. Moreover,our
experi-mentswithwheat germ agglutinintendtoexcludeother, sofar
unrecognized,amphipathsasechinocyticcomponents ofpatient
HDL,evenwere there evidence oftheirexistence;in any case, thedifferentpropertiesof the
echinocytes
generatedbyknownamphipaths renders sucha mechanism
unlikely.
Weconclude thatechinocytogenesisinvolves theoccupation
ofbinding
sitesonthecellsurface byabnormal HDL. Howthiscausesthe im-balance ofarea between the membrane leaflets
implied
by echinocytosis(22,
37-39) is uncertain, although cytoskeletalchanges (thoughnotdephosphorylationof
spectrin)
may bein-volved, in thatpartial
cross-linking
ofspectrinmoleculesby di-amide inhibitedechinocytosis. Removalof abnormal HDLbywashing did notreverse the echinocytosis, possibly because a
slowpassiveadaptation by thecytoskeleton (22, 38,
39)
is re-quiredfor reversal in the absence of normal HDL.Ourevidence suggests that abnormal HDLactsvia saturable receptors,numbering -5,000per cell. Because
binding
andrapid
echinocytosisoccurredwithin thesameconcentration range, andbecause otherplasma lipoproteinsthat do notshowsaturable
bindingwere inactive,it seems reasonable to conclude that
re-ceptorbinding and shape change arecausally related. The
re-ceptors involvedareunlikelytobe the 160,000-molwtapoB,E surface-receptors present in nucleated cells inasmuch asthese have not beenfound in mature red cells (40). Moreover, neither agents that preventbinding oflipoproteins by apoB,E receptors, northe use oferythrocytes from apatient withagenetic defi-ciency of apoB,E receptors on their nucleated cells, impeded
echinocytosisby abnormal HDL. Our tentative conclusionthat
the receptors areglycoproteins is basedon their susceptibility
toproteolysis and theirheterogeneity; proofwill require addi-tional experimental evidence. Their function is also obscure, although other cell types have surface receptors whichbind
li-poprotein moleculeswithoutinternalization (41). Theymay,as suggestedfor insulin receptors(42),remain fromanearlier stage oferythroidcelldevelopment.
Wehave not yetidentified the constituent(s) of abnormal
HDLrequiredfor receptorbindingandechinocytosis, although
theretention of echinocytogenic activity after partial delipidation suggestsanapolipoproteinasthe active component. If this isa
minor part of the total protein, thenthe association constant will beconsiderably higher than is apparent from the binding
curves. Liver disease HDLismarkedly heterogeneous(19, 20, 43) and ourpresent aim isto isolate a population of potent, homogeneous HDLmolecules andtoidentify theactive
com-ponent.
What is thepathophysiologic consequence oferythrocyte
shapechangesinliver disease?Erythrocytesurvival is often
di-minished inpatientswith liverdisease,butisnotusuallyamajor
clinical problem. Target cellsarethoughttohaveanormallife span(34), butacanthocytosisinliver disease isassociatedwith
marked hemolysis and predominant
splenic destruction,
al-though anemia isnot aprominentfeature of
abetalipoprotein-emiaorofcertainother
acanthocytic
disorders(44).
Echinocytes
mightbeexpectedtohave ashortened lifeasthe
spleen
removesmany cells of abnormalshape.Powelletal.
(45)
considered thatonlyinpatientswith spurcells didmorphologicalterations lead tothe premature removal of cells
(though
theirscanning
electronhemoglo-bin levelsin ourpatients didnotcorrelatecloselywith the
num-berofechinocytes in their blood. However, hemoglobin con-centration in liverdisease is altered bymany factors(34)and
the degree of hemolysis is bestassessedbydirect measurement. Areechinocytes the precursorsofacanthocytesassuggested byGrahn et al. (1) and Cooperetal.(46)andwhatis the
mech-anism for the conversion?Thepreponderanceofechinocytesin ourpatients with spur cell anemia and with abetalipoproteinemia is consistent with suchaprecursor-product relationship.
How-ever,webelieve the suggestion (13, 46) that,inliver disease,the
conversionoccurs inthespleenas aconsequenceof cholesterol
enrichmentmustbeviewed with caution. Evidence forasplenic
conversion isderived from asingle case(46)whereas,in other
disorders, splenectomy can itself produce acanthocytosis (47). Therelevance of cholesterolenrichmenttoacanthocytogenesis mustalso bequestioned, as inabetalipoproteinemiathe increase in cell cholesterol content is only marginal. It seems, therefore,
that inliver disease the echinocytic shape,inducedbyabnormal
HDL, may be a more critical factor to the conversion than the reductions in membrane fluidity (35) and celldeformability(13)
caused by excess cholesterol. Clearly, additional studies are
needed, both toestablish therelationship between erythrocyte survival in liver disease and cell morphologyand membrane
lipidcomposition, and to understand in greater detail factors regulating erythrocyte shape in general.
Acknowledgments
Wethank ProfessorJune Lloyd, Dr. BeatriceWonke, and Dr. Tom
Warnesfor allowingus tostudypatientsundertheir care. We are grateful toJackie Lewin and LindaBoxerinthe DepartmentofHistopathology for their help and patience in preparing erythrocytesamplesforstudy
by SEM. We thankMr. A.Chitranukrohfor the bile acid measurements. ThePhilips SEM 501waspurchasedwithagrant from the Stanley
Thomas Johnson Foundation. Dr. Beaven was a Medical Research
Council projectgrantholder.Dr.Owen thanks the WellcomeTrustfor
anInterdisciplinary Research Fellowship and the British Heart Foun-dationforaprojectgrant.
References
1.Grahn,E.P.,A. A.Dietz, S. S.Stefani,andW.J.Donnelly. 1968.
Burrcells,hemolytic anemia and cirrhosis.Am.J.Med. 45:78-87. 2.Smith, J. A., E. T. Lonergan, and K.Sterling. 1964. Spur-cell anemia: hemolytic anemia with red cells resembling acanthocytes. N.
Engl.J. Med. 271:396-398.
3.Silber,R., E.Amorosi,J.Lhowe, and H. J. Kayden. 1966. Spur shaped erythrocytes in Laennec's cirrhosis.N.Engl.J.Med. 275:639-643.
4.Douglass, C.C.,M.S.McCall,and E. P.Frenkel. 1968.The
acan-thocyte in cirrhosis withhemolyticanemia.Ann.Intern. Med. 68:390-397.
5.Hui,D.Y., andJ. A. K.Harmony. 1979.Interactionofplasma
lipoproteinswith erythrocytes.I.Alteration of erythrocyte morphology.
Biochim. Biophys.Acta.550:407-424.
6.Hui,D.Y.,J.G.Noel,and J. A. K.Harmony. 1981. Bindingof
plasmalowdensitylipoproteinsto erythrocytes.Biochim.Biophys. Acta.
664:513-526.
7.Owen,J.S.,H.Goodall, P.Mistry,D.S. Harry, R. C. Day,and
N. McIntyre. 1984. Abnormalhigh-densitylipoproteins from patients withliver disease regulate cholesterol metabolism in cultured human
skinfibroblasts. J. LipidRes.25:919-931.
8. Herbert,P.N.,A.M.Gotto, andD.S.Fredrickson. 1978. Familial lipoprotein deficiency (abetalipoproteinemia, hypobetalipoproteinemia,
andTangierdisease).InThe Metabolic Basis of Inherited Disease. J. B.
Stanbury,J. B.Wyngaarden,andD. S.Fredrickson,editors.
McGraw-Hill BookCo., Inc.,New York.544-588.
9. Brecher,G.,andM. Bessis. 1972. Presentstatusofspiculatedred
cells andtheirrelationshiptothediscocyte-echinocytetransformation: acritical review. Blood. 40:333-344.
10.Owen,J.S.,R. A. Hutton,R. C.Day, K. R.Bruckdorfer,and N.McIntyre. 1981. Plateletlipid compositionandplateletaggregation
in humanliver disease. J.LipidRes.22:423-430.
11.Owen,J.S.,V.Ramalho,J. C. M.Costa,and M. P. T. Gillett.
1979.Determination of lecithin: cholesterolacyltransferinmouseplasma
and theinfluence of mercaptoethanol andsulphydryl blockingagents
onitsactivity. Comp.Biochem.Physiol.63B:261-265.
12.Barnes,S.,and A. Chitranukroh. 1977. Asimplifiedprocedure
for the isolation of bileacids fromserum basedon abatch extraction
with thenon-ionic resin Amberlite XAD-7. Ann. Clin. Biochem. 14:
235-239.
13. Cooper,R.A.,E.C. Arner,J. S.Wiley,andS. J. Shattil. 1975.
Modification of red cell membrane structure by cholesterol-rich lipid dispersions.Amodel for theprimary spurcelldefect. J. Clin. Invest. 5 1:
3182-3192.
14.Fairbanks, G.,T. L.Steck,andD.F. H.Wallach. 1971.
Electro-phoreticanalysis ofthemajorpolypeptidesof the humanerythrocyte
membrane.Biochemistry. 10:2606-2617.
15.Kendrick-Jones, J.,E.M.Szentkiralyi,and A.G. Szent-Gyorgyi. 1976.Regulatorylightchainsinmyosins.J.Mol. Biol. 104:747-775.
16. Wolfe, L. C., and S. E. Lux. 1978. Membrane protein
phos-phorylation of intactnormal andhereditary spherocytic erythrocytes.J.
Biol. Chem. 253:3336-3342.
17. Lovrien, R. E., and R. A. Anderson. 1980. Stoichiometry of
wheat germagglutininas amorphology controlling agentand as a
mor-phology protective agent forthe human erythrocyte.J. Cell Biol. 85:
534-548.
18. Haest, C.W. M., D. Kamp, G. Plasa, and B. Deuticke. 1977.
Intra- and intermolecular cross-linking of membrane proteins inintact
erythrocytesandghosts by SH-oxidising agents. Biochim. Biophys.Acta.
469:226-230.
19. Agorastos, J., C. Fox, D. S. Harry, and N.McIntyre. 1978.
Lec-ithin-cholesterol acyltransferase and the lipoprotein abnormalities of
ob-structive jaundice. Clin. Sci. Mol. Med. 54:369-379.
20. Day,R. C., D. S. Harry, J. S.Owen,A.Y. Foo,and N. McIntyre. 1979. Lecithin-cholesterol acyltransferase and the lipoprotein
abnor-malitiesof parenchymal liver disease. Clin. Sci. 56:575-583.
21. Oram, J. F., J. J. Albers, and E. L. Bierman. 1980. Rapid regulation oftheactivity of the low density lipoproteinreceptorof culturedhuman
fibroblasts.J.Biol.Chem. 255:475-485.
22. Alhanaty, E.,and M. P.Sheetz. 1981. Control of the erythrocyte membrane shape: recovery from the effects of crenatingagents. J. Cell
Biol. 91:884-888.
23. Furthmayr,H. 1978.GlycophorinsA,B and C:afamily of
sialo-glycoproteins. Isolationand preliminary characterization oftrypsin
de-rived peptides. J.Supramol. Struct. 9:79-95.
24.Steck, T.L., J.J. Koziarz,M. K. Singh, G. Reddy, and H.Kohler. 1978. Preparationand analysis of seven major,topographically defined fragmentsofband 3, thepredominant transmembrane polypeptide of
human erythrocyte membranes. Biochemistry. 17:1216-1222.
25. Robinson, P. J., F. G. Bull, B. H. Anderton, and I. M. Roitt. 1975. Directautoradiographic visualisationinSDS-gels of lectin-binding
components of the human erythrocyte membrane. FEBS (Fed. Eur.
Biochem. Soc.)Lett. 58:330-333.
26. Hui, D.Y.,and J. A. K.Harmony. 1979. Interaction of plasma
lipoproteins with erythrocytes. II. Modulation ofmembrane-associated enzymes. Biochim. Biophys.Acta. 550:425-434.
27. Hui, S. W., C. M. Stewart, M. P. Carpenter, and T. P. Stewart. 1980. Effects of cholesterol on lipid organization in humanerythrocyte
membrane. J. Cell Biol. 85:283-291.
28. Lange, Y., H. B. Cutler, and T. L. Steck. 1980. The effect of
cholesterol and other intercalated amphipaths on the contour and stability
oftheisolatedred cell membrane. J.Bid.Chem. 255:9331-9337.
29.Conrad, M. J., and S. J. Singer. 1981. The solubility of
amphi-pathicmolecules in biological membranes and lipid bilayers and its
im-plications formembrane structure. Biochemistry. 20:808-818. 30. Lange, Y., and J. M. Slayton. 1982. Interaction of cholesterol andlysophosphatidylcholinein determining red cell shape. J. Lipid Res. 23:1121-1127.
31.Mohandas, N., A. C. Greenquist, and S. B. Shohet. 1978. Bilayer
balanceandregulation ofred cellshape changes. J.Supramol. Struct. 9: 453-458.
32. Cooper, R. A. 1969. Anemia with spur cells: a red cell defect
acquired inserum and modified in the circulation. J. Clin. Invest. 48: 1820-1831.
33.Martinez-Maldonado, M. 1968. Role of lipoproteins in the
for-mation ofspur cellanaemia.J.Clin. Pathol.21:620-625.
34.Cooper,R. A.1980. Hemolyticsyndromes and red cell membrane
abnormalities in liverdisease. Semin. Hematol. 17:103-112.
35. Owen,J.S., K. R. Bruckdorfer, R. C. Day, and N. McIntyre. 1982. Decreased erythrocyte membranefluidityandaltered lipid
com-positionin humanliver disease.J.LipidRes. 23:124-132.
36.Godin,D. V.,G.R.Gray,and J.Frohlich. 1978. Erythrocyte membranealterations in lecithin:cholesterolacyltransferase deficiency.
Scand.J.Clin. Lab.Invest. 38(Suppl. 150): 162-168.
37. Kuypers, F. A., B. Roelofsen, W. Berendsen, J. A. F.Opden Kamp,and L. L. M. van Deenen. 1984. Shape changes in human eryth-rocytesinducedby replacementofthenativephosphatidylcholinewith
species containing various fatty acids.J.Cell Bio. 99:2260-2267. 38. Lange,Y., R. A.Hadesman,and T. L.Steck. 1982. Roleofthe
reticulum inthestabilityand shape of the isolated human erythrocyte membrane. J.Cell Bio. 92:714-721.
39. Lange, Y., A. Gough, and T. L. Steck. 1982. Role of the bilayer in theshapeofthe isolated erythrocyte membrane.J. Membr. Biol.69: 113-123.
40. Goldstein, J. L., and M. S. Brown. 1983. Familial hypercholes-terolaemia.InTheMetabolic Basis of InheritedDisease. J. B.Stanbury,
J. B.Wyngaarden, D. S. Fredrickson, J. L.Goldstein,and M. S. Brown,
editors. McGraw-HillBookCo.,Inc., New York.672-712.
41.Owen, J. S., N. McIntyre, and M. P. T.Gillett. 1984. Lipoproteins,
cell membranes and cellularfunctions. Trends Biochem.Sci. 9:238-242. 42. Dons, R. F., L. M. Corash,and P. Gorden. 1981. Theinsulin
receptorisanage-dependentintegralcomponentofthe human
eryth-rocyte membrane.J. Biol. Chem. 256:2982-2987.
43. Wiedman, S. W., J. B. Ragland, and S. M. Sabesin. 1982. Plasma
lipoprotein composition inalcoholic hepatitis: accumulation of apoli-poprotein E-rich high density liapoli-poproteinandpreferentialreappearance
of"light"-HDL during partialrecovery. J.LipidRes. 23:556-569. 44.Shohet, S. B. 1974."Acanthocytogenesis"-orhow the red cell wonitsspurs. N. Engl.J.Med. 290:1316-1317.
45. Powell, L. W., J. W. Halliday,and B. R. Knowles. 1975. The
relationship ofredcellmembranelipidcontent toredcellmorphology
andsurvival in patients with liver disease.Aust. N. Z. J.Med.
5:101-107.
46.Cooper, R. A., D. B.Kimball, and J. R. Durocher. 1974. Role
ofthespleen inmembraneconditioning and hemolysis ofspur cells in
liverdisease.N.Engl.J. Med. 290:1279-1284.
47. Brecher, G.,J. E. Haley, and R. 0. Wallerstein. 1973.Spiculed erythrocytes after splenectomy. Acanthocytesornon-specific poikilocytes?
In RedCell Shape. Physiology, Pathology, Ultrastructure. M. Bessis,
R.I.Weed, and P. F.Leblond, editors.Springer-Verlag,New York.