Direct correlation of structure changes and thermal events in hydrated lipid established by simultaneous calorimetry and time resolved x ray diffraction

Direct correlation

of structure

changes

and thermal events

in

hydrated lipid established

by simultaneous

calorimetry

and

time-resolved

x-ray diffraction

Hesson

Chung and

Martin

Caffrey

Department ofChemistry,TheOhio State University,Columbus, Ohio 43210 USA

ABSTRACT Inmanylipid systems,

polymorphic

and

mesomorphic

behavior

depends

on

sample

thermal

history.

To establish unequivo-cally the structural

origin

of endothermic and exothermic events in suchsystems, we have

performed

simultaneous

calorimetry

and time-resolved x-raydiffraction

(SCALTRD).

To thisend, aluminum

calorimetry

crucibleswereused tocontain the

hydrated

lipid sample,

andthe calorimeterwasmounted withthebase of the crucible oriented

perpendicular

to a

synchrotron-derived

focused monochromatic x-ray beamfor

SCALTRD

data collection. Measurementswere made with

hydrated

monoelaidin and

1,2-dielaidoyl-sn-glycero-3-phos-phoethanolamine (DEPE) contained inhermetically sealed crucibles. Time-resolved x-ray diffraction

(TRXRD)

datawerecollectedusing anx-rayimage

intensifier/video

systemandastreakcamera

containing

anx-raysensitive

image plate

and/or

film. SCALTRD

analysis

of thelamellar geltolamellarliquid crystalline phase transitioninhydrated monoelaidingives identical progresscurves

by

calorimetry

and TRXRD at a scanrateof1

°C/min.

Atfaster rates,

calorimetry

showsabroader

phase

transition thatstarts atalower and endsata

higher

temperature thanisobserved by TRXRD. Thedisparity arises in part because the x-ray beam usedinTRXRD

interrogates

only

asmall portion ofthesample, whereas the calorimeter respondstothe entire

sample

volume.Because data collection timesare

relatively

long, radiationdamage isanimportant potential problem for

SCALTRD

measurements.

Such

aneffectwasobserved with

DEPE/water

inthat TRXRD showsthe lamellargeltolamellarliquid crystalline phase transition occurringatalower temperature than observed

by

calorime-try.Wespeculate thatthe sample accumulatesimpurities locallyas aresult of radiationdamage that has the effect of

lowering

the

phase

transition temperature at the siteofinterrogation by the x-ray beam. This "methods-in-combination"

SCALTRD

approach facilitates the direct correlation ofstructurerearrangementsand thermaleventsinthesame

sample

under identicalconditions of thermal

history.

The information content of the datasoderived far surpasses that available from either method used in isolation.

INTRODUCTION

An

important

aspect

of lipid function

in

a

variety of

set-tings

derives

from

the

liquid

crystalline

character

of

lip-ids. Phase

transitions between the

various liquid

crystal-line

states

have been the

focus of

much

attention for

several

decades,

and

a

variety of

methods have been used

to

study their

thermodynamic, structural,

and

kinetic

properties.

Among

them,

calorimetry

and x-ray

diffrac-tion

have proven

to

be the

most

frequently

used

(Luz-zati, 1968; McElhaney, 1986; Small, 1986).

Calorimetry

provides direct thermodynamic

information,

such as

transition temperature and

enthalpy change.

Unfortu-nately,

however, this method does not reveal the

identity

of

the

transforming phases.

X-ray

diffraction,

on

the

other

hand,

is

a

method

of

choice

for phase

identification

and

structure

characterization. With

the

advent

of

synchrotron

radiation and the

development of

the

time-resolved x-ray

diffraction

(TRXRD)'

technique, it has

been

possible

to use

x-ray

diffraction

as

a

means

for

studying

the

kinetics

and

mechanism of phase

transi-tions

(see

Caffrey, 1989,

and

references therein).

As

such,

the two

methods,

calorimetry

and x-ray

dif-fraction, neatly complement one another. A problem

'Abbreviations

used in this paper: CHESS, Cornell High Energy Synchrotron Source; DEPE, 1,2-dielaidoyl-sn-glycero-3-phosphoeth-anolamine; 1-20, intensityversusscatteringangle;

L,,,

lamellarliquid

crystalline phase; L,B, lamellar gel phase; NSLS, NationalSynchrotron

Light Source;SCALTRD, simultaneous calorimetry and time-resolved

x-raydiffraction (TRXRD).

arises,

however,

in

that in many

lipid

and lipid/water

systems polymorphic and

mesomorphic behavior

can

depend

markedly

on

sample

thermal

history (Garti

and

Sato,

1988; Slater

et

al.,

1989).

Frequently,

a

number of

thermal

cycles are necessary before

reproducible results

are

obtained. It is

not uncommon to

find in the literature

that

the first

calorimetric

scan

has been

omitted

because

of

this effect.

Metastability

is

another

intriguing

aspect

of

lipid

mesomorphism (Caffrey, 1987; Seddon

et

al., 1983)

that

creates

problems when data collected

on

samples

with

different

thermal

histories

are

compared.

Method

of

sample

preparation

as

well

as

the

commercial

source

of

the

lipid

can

give

rise

to

differences

in

mesomorphism

(Finegold

et

al.,

1985), which complicates comparisons

of data collected

using

different methods under

noniden-tical

conditions. These

limitations

can

be overcome by

performing

calorimetry

and TRXRD measurements

si-multaneously

on

the

same

sample.

That such

measure-ments

in

combination might

be

possible in hydrated

lipid

systems

was

suggested by

simultaneous

in-sample

temperature

and TRXRD measurements made on a

fully hydrated phospholipid (Caffrey,

1985).

In

this

study,

a

thermal halt

was

observed

while the sample

un-derwent

a

chain

order/disorder

transition monitored

si-multaneously

by TRXRD.

Concurrently,

Russell and

Koberstein (1985) reported

simultaneous TRXRD and

calorimetry

on

polymers

using

a

method that has been

implemented

subsequently in other systems (Ungar and

Feijoo, 1990;

Kellens

et

al., 1991 a,

b).

In

the

latter,

simul-438 0006 3496/92/08/438/10 $2.00 Biophys .1 . Riophyeimd

Biophys.6J. BiophysicalSociety

Volume 63 August1992 438-447

taneous

measurements

were

relatively straightforward

because

sample hydration

wasnot an

issue.

Indeed,

in

some

of these

studies,

open

calorimetry crucibles

were

used. Our focus

is

on

hydrated lipid

systems.

In this

com-munication,

we

describe

a means

for

performing

simulta-neous

calorimetry and

TRXRD

(SCALTRD)

on

such

samples.

The systems chosen for

study

include

hydrated

monoelaidin and hydrated

1,2-dielaidoyl-sn-glycero-3-phosphoethanolamine (DEPE), both

of which

undergo

lamellar

(L,B)-to-lamellar

liquid crystal (La) phase

transi-tions

in

an

accessible

temperature

range.

As

atest

of

the

performance

of

the SCALTRD

method,

measurements

were

made

as a

function of

scanrate.

It is shown that

at

lower

scanrates

the

agreement

between the

two

methods

in

SCALTRD

measurements

is

remarkably good.

In

the

case

of

DEPE,

radiation

damage proved

to

be

a

problem

because the

sample

was

exposed

to a

high

flux

x-ray

beam for

a

prolonged period.

Methods for

minimizing

radiation

damage

in

SCALTRD

measurements are

de-scribed.

MATERIALS AND METHODS

Materials

Monoelaidin (Nu Chek Prep, Inc., Elysian, MN) andDEPE(Avanti PolarLipids, Inc.,Birmingham, AL) hadapurity of.99%and

.98%,

respectively,asjudgedbythin-layerchromatographyonsilicagelK5F plates (Alltech Assoc., Inc., Deerfield, IL) of >100

/tg

lipidin three solventsystems(chloroform/methanol/water,65:25:4byvol; hexane/ toluene/acetic acid,60:40:1 by vol;chloroform/acetone/acetic acid/ methanol,72.5:25:0.5:2 by vol) andwereusedwithout further purifica-tion.At roomtemperature,monoelaidin consists of 90-92% 1-mono-elaidin and 8-10% 2-mono1-mono-elaidin(Martin, 1953).Allotherchemicals usedwereofreagentgrade.Waterwaspurified by usinga water purifica-tion system(Milli-Q; Millipore Corp., Bedford, MA).

X-ray

source

Measurementsweremade in the A and Fl stationsatthe CornellHigh Energy SynchrotronSource(CHESS)and in the X 12B stationatthe Brookhaven National Synchrotron Light Source(NSLS). Inthe Al stationatCHESS, experimentswerecarriedoutbyusing

wiggler-en-hanced monochromatic focusedx-raysaspreviouslydescribed

(Caf-frey, 1985,1987). X-ray wavelength (1.56 A, 8.0 keV)wasdetermined by usingaleadnitrate standard andacarefullymeasured

sample-to-filmdistance. Themachinewasoperatingat5.26 GeVand 80mAtotal

of electron beamcurrentinthe seven-bunch mode and with the

six-pole wigglerathalfpower.The incidentx-raybeamwasdefinedusinga

0.5-mm-diametercollimator(Charles Supper Co., Inc., Natick, MA). Sample-to-detectordistancewas-40cm.

The Fl station at CHESS receives hard bend magnetas well as

wiggler radiation froma25-pole 1.2-Tpermanent magnetinsertion

device. Thishybrid wiggler providesabeam withacriticalenergyof 22keV and hasapeak field of12kGatthe normal operatinggapof4 cm.Approximately 1-2mrad ofpositronwhitesynchrotronradiation

passesthrougha0.0 1-inchberylliumfilterandtwo0.01 5-inch

water-cooledprefilters.Themonochromator,located 20mfromthesource,

horizontallyfocuses andtunesthe beam from 6to 14keVwithoneof

twodifferentbentasymmetricallycutSi(11 1) crystalsmiscut by 5.05°.

Averticalfocusing rhodium-coated Si mirror,thesize ofwhich is-4x

18in,isavailable 0.8-mdownstreamofthe monochromator. The fo-cused beam hasa1.0-mrad horizontal anda0.4-mrad vertical

diver-gence.X-ray wavelength (13.6 keV,0.92A)wasdetermined by usinga

videomonior

FIGURE 1 Schematicrepresentation of the experimental arrangement for simultaneouscalorimetry and TRXRD diffractionmeasurements

at asynchrotronsource.Focusedmonochromatic x-raysarecollimated througha0.5-mm-diameter collimator beforestrikingthesample

con-tained inanaluminum crucible within the calorimeter. Transmitted x-raysareintercepted byaremotely controlled nickel beam stop. When the streakcamerais used for diffraction datacollection, x-ray sensitive filmand/oranimage plate inademountable holder is translatedat a

fixedratebehindalong narrow slit. When theimageintensifier based systemis used, the demountablefilm/plateholderis removed and the diffracted x-raysareallowedtostrike the fluorescentscreenofa three-stageintensifier tube used forimageintensification. Theimageis dis-playeddynamically ontheintensifier tube and recorded in live-time usingafiberopticallycoupled charge coupled device (CCD)camera

andasuperVHSvideocassetterecorder.Along with the two-dimen-sional diffracted image, the temperature, experiment number, and elapsedtime in seconds and video frame numberarerecorded simulta-neouslyonvideo tape. Theoperation of the calorimeter and the streak

camera aswellasdatacollection from the calorimeterareunder

com-putercontrol.

lead nitrate standardasabove. The incidentx-raybeamwasdefined

usinga0.3-mm-diameter collimator(Supper Co., Inc.).

Sample-to-de-tectordistancewas--28cm.Datawerecollected with Cornell Electron

Storage Ring (CESR) operatingat5.26 GeV and60 mA total of

posi-tronbeam.

AtNSLS, thex-raybeamwasdoubly focused bya48-cm-longx

10-mm-wide 1:1 defocused rhodium-coated silicon carbide(primary)

mirror andwaspassed through an0.5-mm-diameterapertureblock

situatedatthefocalpointof theprimarymirror. Thex-raybeamwas

monochromatizedbyadoublefixed-exit silicon ( 111) flat crystal

mono-chromator. The beamwasdoublyfocusedontheplaneof thedetector usinga 1:1 defocused rhodium-coatedelectroless nickel-plated bent

aluminumcylinder measuring68cmlong and 3cmwide. Before

strik-ingthesample,the beamwasfurtherdefined byasetof slitsandtwo

setsofx-yguardslitarrays.X-ray wavelength (0.91 A, 13.8 keV)at

NSLSwasdeterminedby usingacholesterol myristatestandardanda

carefully measuredsample-to-detector distance. The optical

arrange-ment atNSLS enabledfocusing of the beamtoapoint -0.5mmwide

and 2mmhighatthe detector.Sample-to-detector distancesusedat

NSLS variedfrom 40to55cm.

Simultaneous

calorimetry

and TRXRD

A calorimeter(model FP84; Mettler Instrument Corp.,Hightstown, NJ), originally designed forsimultaneous calorimetry andpolarized light microscopy,wasused in thisapplication. Fig.Ishowsaschematic

heating

blocks

base

plate

thermocouple

x ray beaml

aluminum foil

modified

pan

with

hydrated lipid

sample

FIGURE2 Schematic of the sample cruciblearrangementin the calo-rimeterfor simultaneouscalorimetryand TRXRD. Thesampleand reference (not shown) pansareheatedbycontactwithheatingblocks that surround them. The temperature difference between thesample andreference is measuredusingfivethermocouplesensorspositioned onthe baseplate and in directcontactwith the bottom of the pan. The samplecrucible has been modifiedtoincrease x-ray transmittanceas

described in the text under Materials and Methods.

of theexperimentalarrangementforSCALTRDmeasurementsatthe

synchrotron. The sample crucibleinthecalorimeterwasmounted

or-thogonaltothe incidentx-raybeamon anopticalbench(Fig. 2).The

calorimeter wasinturn mounted onapairoftranslatingstages for

positioningof thesamplepaninthex-raybeam.Fig.2shows how the

samplepanhasbeen modified for SCALTRDmeasurementsandhow

it ispositionedinthe calorimeter.The FP84 includestwoglass plates

on either side of the sample pan that act as heat filters. In the

SCALTRDapplication, theseplateswereremovedtofacilitatex-ray

transmissionwith little ifanyeffectontheperformanceof the deviceas acalorimeter. Temperaturesbelowroomtemperaturewereachieved by supplyingthe calorimeter with cooled air fromaforced-aircrystal heating/coolingapparatus(models XR-85-lA-0 and AD-80XA; FTS Systems, Inc.,StoneRidge, NY).

Twotypesof detectorswereusedtorecord live-timex-raydiffraction

patternsduringSCALTRD. One of these detectors consists ofanimage

intensifier/video-based system,and time-resolved diffractionpatterns

wererecordedonS-VHS videotape(Panasonic, Japan)asdescribed

previously (Caffrey 1985, 1987; Caffreyetal., 1991a). The other isa

streakcamerathat usesx-raysensitive film (DEF5; Eastman Kodak Co., Rochester, NY) and/or image plates (storage phosphor, Eastman KodakCo.) and that is translatedatarateof4mm/min behindalong vertical3.2-mm-wide slit. Becauseafinite slit width is used, data

col-lected with the streakcamerahave time andtemperatureresolution compromised somewhatinthepresentapplication. For example, with

a3.2-mm-wide slit andafilmtranslating speedof 4mm/min, each spatial elementonthefilmorplate is exposed for <[3.2 mm/(4mm/

min)]=0.8min (48s).Withatemperaturescanrateof5°C/min,the

correspondingtemperaturerangethat eachspatial element samples is <[5°C/minx0.8min] =4°C.The timeresolutionof the videotape was33ms.

Usingthestreakcameradetector, both low-andwide-angle

diffrac-tion from thelipid samplewasrecorded simultaneouslyonthesame

film.With theimage intensifier/videosystem,only the low-angle dif-fraction datawererecorded. In the lattercase, acharactergeneratorwas

usedtoinsertelapsed time, experiment number, andtemperature

out-putfrom the calorimeter directlyontothe videoimage during tapingso

that thethree parametersarerecordedsimultaneously alongwith the

two-dimensionaldiffraction pattern.

The calorimeter and the streakcamera were operated under

com-puter(Macintosh IIfx; Apple Computer, Inc., Cupertino, CA) control using LabVIEWTM (National Instruments, Austin, TX) software. A

LabVIEWprogramwaswrittentoindexsimultaneouslythe

calorime-tryand streakfilmdatawiththebeginningand end ofanexperiment.

Thesameprogram wasusedtorecord andtodisplayinreal-time the

thermogramon aMacIntoshIIfx; Apple Computer, Inc., Cupertino,

CAcomputerastheexperimentwasperformed. When the video sys-temwasusedtorecordlive diffractionpatterns,the calorimeterand the

videocassetterecorder (S-VHS model 7500A; Panasonic, Japan)were

notsynchronized automatically.Inthiscase,temporal alignmentwas

made byusing thetemperaturedata thatwererecordedsimultaneously

onthecomputerandonthevideoimageforcalorimetryandTRXRD, respectively.

Sample

crucibles

The aluminum cannisters (model 27331, Mettler InstrumentCorp.) foruseinthecalorimeterweremodifiedtoincrease theirx-ray

trans-mittance for SCALTRDmeasurements. The cannisterscomeintwo parts:apanandalid. The bottom ofthepanis -0.2mmthick and has

analuminumpositioning pin (-0.8mmlong)atitscenter.The lid is

0.1 mmthick.Thus,anintact sealedpanhasatotal aluminum

thick-nessof1.1 mmalong the flight pathof thex-raybeam,which is suffi-cientto attenuatethex-raybeamtoanunusable levelattheavailable

energies. X-raytransmittance is calculatedasfollows:

I

'0 (APP (1)

whereIandIOarethetransmitted and the incident beamintensities,

respectively, , is the linear absorption coefficient,pisthedensityofthe

material, g/pis themassabsorption coefficient,andxis thethickness

oftheabsorbing layer.Thequantity M/pisacharacteristicofthe

absorb-ingmaterial and varies with incidentx-raywavelength (Klug and

Alex-ander, 1974).Foraluminum,the /pvalue is -49.4at1.55 Aand 10.2

at0.90Aandpis 2.70g/cm3at20°C (Macgillavry and Rieck, 1962). Thesevalueswereusedtocalculatetransmittance, assuming that the

,g/pvaluesabove holdat1.56Aandat0.91 A.Using Eq. 1,thex-ray

transmittance of the intact crucibleatitscenteris 4.3x I0- and 4.8%

at1.55 and 0.90A, respectively. By removingthepositioning pin,

re-ducingthe thickness of thepanbase from 0.2to0.1mm,andreplacing

the lid with

10-,gm-thick

aluminum foil(ReynoldsMetalsCo., Rich-mond, VA), the total thickness of aluminuminthex-raybeamwas

reducedto 0.11 mm, correspondingto atransmittance of 23.1 and

73.9%at 1.55 and 0.90A, respectively. Such transmittance values for

the modified cruciblesmeanthat TRXRDmeasurementscanbe

per-formed with relativeeasegiven thex-rayflux andwavelengths available

atCHESS and NSLS. Tosecurepositioning of the cannister inthe

calorimeter,itwasnecessarytoinclude the lidinthefinalassembly. However, alid witha 1.5-mm-diameter hole in itscenterserved the

purposeand didnotcontributetox-rayattenuation.

Sample preparation

Sampleswerepreparedinexcesswaterby placing - 10mgofpurelipid and25,d ofwaterinamodifiedaluminumpan.Thetopofthepan was

covered with aluminumfoil, andalidwithaholeinitscenterwas

placed on topof the foil and the entire assembly was presssealed

(model 27339, Mettler Instrument Corp.).Extrafoil around the

out-side of thepanwascarefully removed,and thepanwasfinallysealed

around the rimwithaminimumamountofglue (SuperBonder;Loctite

Co., Newington, CT).

Possible adverse effects of thegluewereevaluatedby calorimetry.

When testedintherange4-700C, the gluewasfoundtohavenoeffect

440 BiophysicalJournal Volume63 August1992

onthe shape or transitiontemperatures oflipidthermograms. Without modification, the press-sealed crucibles remain sealed up to 90°C. Beyondthis temperature,wefindconsistentlythatthebaseline of ther-mograms recordedonhydratedlipid samplesincreases,suggestingthat the seal has broken and thatwaterisevaporatingandescapingfrom the crucible. The modificationswe have effected onthe crucibles gives reliable thermograms upto70°C. However, above70°Cthecansbegin to leak.Therefore,beforeperformingSCALTRDmeasurements, each sample inamodified cannister is tested in thecalorimetertodetermine ifit isleak-proof upto70°C. Those that leak below70°Carediscarded. Our successrateistypically70-80% inpreparingmodified cannisters usefulto70°C.

Theinternal volume of the modifiedcrucible is 40,l.To ensurethat thereis sample in the 0.5-mm-diameter x-ray beamasitpassesthrough the center of thecannister, sample volumes of .25 ,ulareused.

Data analysis

The video-recorded data are quantified by digital image processing (Trapix 55/48;RecognitionConcepts,Inc.,CarsonCity, NV) usinga

combination of commercial (RTIPS; TauCorp., Dayton, OH) and home-written software asdescribed previously(Caffrey etal., 1990; Mencke and Caffrey, 1991). Sequential digitized images or parts thereof are "grabbed" and stored in the memory of theimage

proces-sor.Intensityversusscatteringangle(1-20)scansaremadebyaveraging

over a20° arccenteredonthevertical diameter. The dataarethen portedto aMacIntoshIIfx computer wherethey are processed further

toresolvecompositeimagesinto their components. The 001 diffrac-tionpeaks of the La and

La,

phaseswerefitted byanonlinear least-square methodtothe modelfunction,

I

4-i

₀

20

25

30

35

40

Temperature

(°C)

FIGURE3 Heating thermogramsofmonoelaidin inexcesswater atthe

scan ratesindicated. The sample consisted of 8-10 mg monoelaidin in 25 1Iwater.

ffit(X;

Me,

:,

00

9

Eft

Ma

ea+a[(

D)

=

Me( ~)/1 +

ae

[xr)al+

D,

(2)

whereMis peakheight, A is peak radial position, is peak half-widthat

half-height,Disaconstant,and thesubscriptsaand,B refertothe La

and the

L6

phases, respectively.Peakarea,A, iscalculatedas

A =

keM,

(3)

where k is the proportionality constant ir"2 (Caffrey et al., 1990; Mencke andCaffrey, 1991).

Astaticx-raydiffractionpatternoftheL0 phase inhydrated

mono-elaidinat 10.1°Crecorded withaknownsample-to-film distance and

periodicity (001 =60.5A)wasusedtocalibratethetwo-dimensional

videoimage.

Radiation

damage

Tominimizelipid radiationdamage,a newportionofthesamplewas

exposedfor each SCALTRDmeasurement.Thus,noonepartof the

samplewasexposedtothex-raybeam for >20min.Experimentsto test

for radiation damage were carried out before SCALTRD

measure-ments.Weassumethatdamageisnegligible ifthereisnodetectable

changeinthediffractionpatternduringatypicalx-rayexposurewhen

othervariables,suchastemperature, pressure,etc.,areheldconstant.

Inthecaseofmonoelaidininwater,a2-hexposureatNSLS produced

nonoticeablechangeinthediffractionpatternat38°C. However,on

the Al lineatCHESS, damagetohydrated monoelaidinat38°Cwas

apparentafteranexposureof 10min.Damagewasevidenced bythe

appearance ofthe so-called x-phase (Caffrey, 1987)in an La phase

background.Asaresult,weconsistentlyheldexposuresatsingle sitesin thesampleto <10minatCHESSand<20minatNSLStominimize

thedamagingeffects ofx-rays. Radiationdamageof DEPEis treated separatelyunderRESULTS. Amorecomplete analysis of radiation

damagewillbepresentedelsewhere(Cheng, A.,and M.Caffrey,

manu-scriptinpreparation).

RESULTS

Monoelaidin

in

water

The

first set of SCALTRD

measurements

was

made

us-ing

monoelaidin in water with a

focus

on

the

L,-to-L,

transition

that occurs at -23°C. Data were

collected

in

the

10-40°C

range,

and the effect of scan rate from 1

to

10°C/min

on

the

fidelity ofthe SCALTRD data

was

eval-uated. The calorimetric results are shown in

Fig. 3 for

heating rates of 1, 5, and

10°C/min.

The onset

tempera-ture of the

order/disorder

endotherm

increases

and the

transition broadens as the

heating rate increases. The

onset

temperatures,

Ti,

are

22.7, 23.5,

and

24.9°C

at

heating rates of 1, 5, and

10°C/min,

respectively.

The

corresponding transition widths

(width at

half-height) are 0.6, 1.5, and

2.3°C.

The scan rate

depen-dence of

Ti

and

of the

transition width

suggests that the

true

values are less than those

observed at 1

°C/min

be-cause

of the finite response time of

the

calorimeter.

Selected

low-angle

TRXRD

images

from

the

SCALTRD measurements

recorded at 1

°C/min

are

shown in

Fig.

4.

The

original

unprocessed

two-dimen-sional

low-angle

diffraction

images

and how these

change with temperature

through

the

La-to-L,

phase

transition at

a

heating

rate

of 1

°C/min

are

presented in

this

figure.

Included is the

001 lamellar

reflection

from

the

L,

phase (61 A)

at

22°C,

the

001

lamellar

reflection

from

the

L,a

phase

(52

A) at

28°C,

and the two

coexisting

Chung and Caifrey HydratedLipid Establishedby X-ray Diffraction

1

°C/mm

5

0C/mm

0°C/min

i~~~~~~~~~~~~~~~

FIGURE 4 Progressof the

L,,-to-La

phasetransition infullyhydrated monoelaidin monitored by TRXRDusinganimage intensifier/video-based detector system.Low-angle x-raydiffractioncorresponding to

the001 reflectionof the

L,

andLatphaseswererecordedcontinuously in live-timeonvideo tape attherate of 30frames/s. Shown in this figureisacompositeof threeimagesrepresentingtheintegrationof 16 (0.53s) enhanced videoframes.SectorsAandC show the 001 lamellar reflections from the

L,

and La phases at 22 and28°C, respectively. SectorBshows coexistence oftwo001 reflections: onefromthe L, phase and onefrom the La phase recordedat24°Cmidwaythrough the transition. Theheating rate usedwas1°C/min.

L,,

and

La

001

reflections midway through the transition

at

240C.

The

quantitative

low-angle diffraction data obtained

by

image processing and

how

these

change with

tempera-ture

through the

L,-to-La,

phase transition of

hydrated

monoelaidin

are

presented

in

Fig. 5. As outlined under

Data

analysis,

the diffraction data

are

fitted

as a

compos-ite of

1 or

2

Gaussians and

aconstant.

The

parameters

of

each

Gaussian, namely, peak position

(f),

height

(M),

and

half-width

at

half-height (e), along with the peak

area

(A),

are

shown

as a

function of

temperature

from 10

to

40°C

in

Fig. 5 A-E

during

a

SCALTRD

measurement

made

at

5°C/min. The corresponding

calorimetric data

(Fig. 5 F) shows

onset

and

completion

temperatures

of

23.4 and

29.0°C, respectively, with

a

maximum

heat

ca-pacity change

at

26.50C.

By

comparison, all of the

pa-rameters

of

the 001

diffraction peaks show

a

sharper

transition

with

the

bulk of the transition

taking place in

the interval

24.0-26.50C.

Therefore,

ata

heating

rate

of

5°C/min,

the

onsettemperatures

of

the transition

are

in

reasonable

agreement

by TRXRD and

calorimetry.

However, the completion

temperature

discrepancy is

quite

marked

at

this

rate.

The

temperature

range

of

the

La-to-La transition determined by calorimetry and

by

TRXRD in SCALTRD

measurements

is

presented

asa

function of

heating

rate

in

Fig. 6 A. At the slowest

heat-ing

rate

of

1

°C/min, the

two

methods show

essentially

identical phase transition

temperature ranges.

At

the

higher

scan rates

of 5 and

10°C/min, the

discrepancy

between

the

two

methods

grows

with the

biggest

differ-ence

observed in the completion

temperature at

the

highest

scanrate.

DEPE in

water

In

an

effort

to

evaluate the versatility of the SCALTRD

technique,

measurementswere

also

performed

on

fully

hydrated DEPE in the

temperature

range

20-60°C. In

this

case,

the La-to-La transition is centered

at

38°C,

some

14°C

higher than the corresponding

transition

in

monoelaidin. The SCALTRD results

are

summarized in

Fig.

6 B, which shows the

onset

and

completion

tempera-ture

ofthe

order/disorder transition recorded in the

heat-ing direction

as a

function of

heating

rate.

As

observed

with

monoelaidin, the transition determined by

calorim-etry

is

quite sharp

at a scan rate

of

1

C/min

and

broadens severalfold

on

increasing

the

scanrateto

10°C/

min. The

onsettemperature

is

shifted

to

higher

tempera-tures at

the

higher

scanrates.

Unlike

the

results

observed

with

monoelaidin, however,

at

all

heating

rates

used

both the

onset

and

completion

of

the transition occurred

ata

considerably

lower temperature when measured

by

TRXRD. This result

proved

to

be

quite

amystery

until it

was

demonstrated that the sample had succumbed

to

radiation

damage

in

the

course

of

the SCALTRD

mea-surement.

To

test

the radiation

damage hypothesis, the

following experiment

was

performed.

Fully hydrated

DEPE in

an x-ray

capillary

at

35.4

±

0.5°C

was

irra-diated

continuously

with

13.8 keV

x-rays

while

the

dif-fraction from the

sample

was

being recorded in the Fl

station

at

CHESS. At

the

beginning of the

exposure,

the

sample

was

in

the

L,

phase.

However, within

5

min,

the

La

phase emerged and

continued

togrowat

the

expense

of the

L,

phase

for

the

duration of

the experiment (15

min;

Fig. 7).

Translation of the sample by 2

mmtoone

side of the

original exposed site revealed the

presence

of

the

L,,

phase alone,

indicating that loss of

temperature

control

during

the

experiment

wasnota

problem. This

experiment demonstrates

clearly that x-radiation

can

in-duce

damage

to

the

products of which

can

lower

the

onset

temperature

of the

Lj6-to-La

transition

in

the

DEPE/water

system.

A

more

complete and systematic

study

of

radiation

damage

to

DEPE

wasnot

performed

nor

indeed

were

the

products ofradiation damage

identi-fied.

Suffice

it

to say

that

for

x-ray

sensitive

samples,

such

as

DEPE,

a

suitable

exposurestrategy

that

may

in-clude

intermittent

sample

exposure must

be

imple-mented

so asto

minimize

or

obviate the

radiation

dam-age

problem

during

SCALTRD

measurements.

g

40

.p

20

c

20

S

10

'0

DO

E

140

*0

0ca

Pz4 11on

1lzv

60

cO O

50_

*g

80

40

Cd

0

10

20

30

Temperature

(°C)

FIGURE 5 Progressof the

Lff-to-La

phasetransition infully hydrated monoelaidin induced byatemperaturerampat5°C/min during the

courseofasimultaneous calorimetry and TRXRD.Thex-ray diffrac-tion dataareshown in A-E and the calorimetric thermogram is pre-sented inF. The open and closed symbols in A-Ecorrespondto the

Lf,

andLaphases, respectively. The low-angle (001) diffraction peak

maxi-mum intensity is shown in A, the half-width at half-height in B, the radialpositionof the peak maximum in C, the d-spacing of the peak maximum in D, and theintegrated diffracted intensity or peak area in E. A-Eand F eachcontain -360 and300discrete data points,

respec-tively.Theincrease in the spread of the data inB-Din the coexistence

DISCUSSION

Comparison between

calorimetry

and

TRXRD and the nature

of the transition

The

results

obtained

using calorimetry and

TRXRD

in

SCALTRD

measurements on

the

L6-to-La

transition of

fully hydrated monoelaidin

are

compared

in

Fig.

8. The

data

in

Fig.

8 A demonstrate

convincingly

the

agreement

that exists between the

two

methods when

SCALTRD

measurements are

performed

at

lower scan rates. As

the

scan rate

increases,

so

too

does the

disparity

between the

two

techniques, particularly

in

regard

to

transition

mid-point

and

completion

temperatures.

To

what can we

ascribe

the

origin of the mismatch

between

calorimetry

and TRXRD

observed

in

the

SCALTRD

measurement

made

at

higher

scan

rates?

There

are

several

possibilities.

To

begin

with,

the

x-ray

beam used

is

small

(0.5

mm

diameter)

in

comparison

to

the

dimension of

the

sample crucible

(5.5

mm

diameter).

Thus,

only

a

fraction

of

the

sample

('-2%

by vol) located

at

the center

of

the

crucible is

actually

interrogated

by

TRXRD. In

contrast,

the

calorimeter is

responding

to

the heat

of

the

transition absorbed

throughout the bulk

of

the

sample.

The

second

possibility

concerns

gradients

that

exist within

the

sample.

Referring

to

Fig.

1,

the

sam-ple

is heated from

the

left-hand side by

contact

with the

heating

block and base

plate.

It

is also

heated

from the

right-hand

side by

contact

between

the open

lid and the

heating block. Thus, it is

likely

that

sizable thermal

gra-dients

exist within the

sample

at

higher heating

rates

such

that

the

region

interrogated by

TRXRD

will

re-spond in

a

manner

somewhat

differently

to

that

of

the

bulk.

The

shift

to

higher

completion

temperatures

at

the

higher

heating

rates

undoubtedly

reflects

the

fact

that

the

thermocouple

sensors

are

located

on

the base

plate in

contact

with

the

heating block and

not

in the

sample.

As

a

result,

there

is

a

thermal

lag

associated with

communi-cating

heat between the heater and

the bulk

of the

sam-ple.

The

excellent

agreement

between the

data

obtained

by

calorimetry

and

TRXRD at a

heating

rate

of

1

C/

min

in

SCALTRD measurements

indicates the

need

for

slow

scan

rates

if

structure

and

thermodynamic

data are

to

be

correlated

directly.

The

following evidence

suggests that the

La-to-La

phase

change

in

hydrated

monoelaidin

involves only the

L,

and La

phases during the

course

of

the

transition.

To

begin

with,

the

001

reflections from the L,

and La

phases

both coexist

throughout

the

transition.

Second,

within

the

sensitivity limits of

the

method, no

intermediates

were

observed

during

the phase

transition. Third, there is

no

loss

of correlation

in the lamellar structure of either

regionarisesas aresultoferrorinfitting the001 lamellardiffraction peakof thenascentLa anddecayingL, phase wheneither is present in smallamounts.

Chung and Caifrey Hydrated Lipid Established OOc4

OOOO0'

8

0E

.6

I ifw .

34

30

26

~ 22

.

i

40

a)

36

32

A

B

2

4

6

8

10

Scan

Rate

(°C/min)

fraction

of lipid existing in each phase during the

L,6-to-La

transition is

given by

a

AsNAa

Ga- A0 +NAa(4

and

(5)

= As+NAa 1-a,

where A is the integrated intensity of the 001 reflection,

N is

a

normalization factor

corresponding

to

the ratio of

the

integrated intensities ofthe 001

reflection in the

pure

phase region, and the subscripts

a

and

A

refer

to

the

La

and

L,

phases, respectively.

From the data in

Fig.

5

E,

it

is

apparent

that the integrated intensities of the 001

re-flection in the

pure

L,

and

La

phases

are

different.

This

necessitates the

use

of

a

normalization factor

in

Eqs.

4

12

FIGURE 6 Comparison of the

L,-to-La

phasetransition temperature rangesfor (A) monoelaidin and (B) DEPE inexcess waterdetermined by simultaneous calorimetryand TRXRDatdifferent heatingrates.

Thetransition temperature range forcalorimetryandTRXRD are rep-resentedby open andfilled rectangular boxes,respectively.The bottom and topofeach boxcorresponds,respectively,totheonsetand comple-tion temperatureof the transition. The calorimetriconsetand comple-tion temperaturesweredeterminedastheintercept ofthe low andhigh temperaturebaselines with theextrapolated risingandfallingedgeof the endothermic peak. The diffraction basedonset andcompletion temperaturescorrespond, respectively,tothose temperatureswhere the firsttraceof thenascentLa phase and thelasttraceof thedecayingL, phaseis apparent by visualinspectionof thelive-timediffraction pat-terns.

Cl

*_4

0-4

':I

1-q

20

30

40

50

60

Radial Position (pixels)

0

min

5 min

14

min

fresh

part

of

sample

phase

during the transition

as

evidenced

by the fact that

the

shape and width of the 001 reflections of both phases

remains

essentially

constant

throughout the transition.

In

the

following,

weset

about establishing

a means

for

directly comparing the calorimetric and

TRXRD

data

obtained

during SCALTRD. We

begin this analysis by

noting

that

the

area

under the

experimentally measured

diffraction

peak in

an

1-20

scan

is

proportional

to

the

amount

of

material

giving rise

to

the

diffraction

peak

(Klug

and

Alexander, 1974). Thus,

we can

write that

the

FIGURE 7 X-radiation damage to fully hydrated DEPE at 35.4 ±

0.5°Cin the L4phase. Low-angle001diffraction datawererecordedon an image plate usingaremotely controlled shuttered streak camera underconditions ofcontinuous x-ray exposure. Each plot corresponds

to an 1-20scan across thediameterof thediffraction ringsfrom the digitizedimage platedata. Scansareshownsequentiallyfrom topto

bottom eachcorrespondingto anexposureof1-min duration.Inthe early stagesof theexperiment,only the

L0

phaseis present.Asecond peakcorrespondingtotheLaphase emerges after

~-5

minof exposure andgrowscontinuouslyattheexpenseof the

Lg,

phase with continued exposure. The lastscanshowsthe pureL, phase inafresh unexposed partof the samplesome2mmawayfrom theoriginalexposuresite. Scansaredisplaced verticallyontheordinate forclarity.

444 BiophysicalJournal Volume63

I~~~

~ ~ ~~~~~~~~~a

I I I

I I I I I

I I I I I

I

r--"

(4)

1.1

c)

a)

0--0*

o

.r-rcWL

1.1

0.7

0.3

0.0

20 24 28 32 36

Temperature (°C)

FIGURE 8 Comparison of the integrated calorimetric thermograms (solid lines)and thefraction oflipidin theLa phase determined from TRXRD data(filledcircles)recordedduring simultaneouscalorimetry andTRXRD measurements onfully hydratedmonoelaidinat

differ-entheatingrates.

and 5. Such

a

difference

is

attributable,

in

part, to

differ-ences

in

phase

density

of

magnitude,

typically

10%

(Na-gle and Wiener, 1988; Marsh, 1990). The major

differ-ence

in

integrated intensity

most

likely originates from

the

disparate

structure

factors that

apply

to

the 001

re-flection in

each

phase.

For

Eq.

4

and

5

to

hold,

it

is

necessary

that

the corresponding

structure

factors

re-main

constant

during the transition. In the absence of

knowing the

structure

factor

profile for the

two

phases,

the

following is presented

in

support

of

the

requirement

for

a constant structure

factor

through the transition.

Referring

to

Fig.

5

D,

we see

that the

change in the

lamel-lar

repeat

distance

of either the

L,

or

La

phases in the

single phase region

amounts to

-2

A. In

this

same tem-perature

interval, the integrated

intensity (area) of the

001

reflection

in

either

phase remains essentially

con-stant

(Fig. 5 E). This

suggests,

therefore, that in both

cases

the

corresponding

structure

factors

are

close

to

their maximum

values and

are

relatively insensitive

to

changes

in

lattice

parameter.

By comparison with other

hydrated lipid

systems

where the

structure

factor profile

is known,

a

2-A change in d-spacing gives rise

toa

1,

20,

and 100%

change in

structure

factor

at

the

maximum,

half-maximum, and

near-zero

structure

factor values,

re-spectively (Kim

et

al., 1987; Hogan, 1989;

Caffrey

et

al.,

199 1

b). Because the reciprocal spacings ofthe 001

reflec-tions do

not

change

significantly [.(0.5

A)-',

corre-sponding

to a

d-spacing change of <0.5 A] during the

course

of the transition

(see Fig.

5

D),

we

consider that

use

of

Eqs. 4 and 5

in

the

analysis is appropriate.

The value of N used in the analysis that follows is

[(960

±

60)/(470

±

90)]

=

2.0

±

0.4

(see Fig. 5

E).

After this analysis, the

progress curve

for

the

L6-to-La

transition observed by TRXRD is shown

in

Fig. 8, where

Oa

(filled circles) is plotted

as a

function of

temperature

through the transition. The integrated calorimetric

ther-mograms

(solid lines)

recorded

simultaneously

with the

TRXRD

are

included

to

facilitate

comparison. The

agreement

between the

two

methods is

demonstrated

convincingly

at

the slowest

heating

rate.

Radiation damage

The potential for radiation-induced chemical

break-down of

a

sample

during SCALTRD

measurements

is

high because of the lengthy

exposurestoan

intense

x-ray

beam

during

the

course

of the temperature

scan.

This

means

that the local accumulated radiation dose

can

be

high for

a

stationary sample under conditions

ofcontinu-ousexposure.

Furthermore, the

effect

can

be

sample

spe-cific

as

evidenced

in

this

study. Thus,

although

DEPE

proved

x-ray

sensitive under conditions of

measure-ment,

monoelaidin did

not.

Our

original SCALTRD

measurements on

the

L,,-to-La

transition of

fully

hydrated

DEPE

gave

confusing

re-sults.

Consistently,

at

all

heating

rates

used, the

transi-tion

began and

was

completed

at

lower

temperatures

when observed by TRXRD. Temperature gradients in

the

sample could

not

explain this result because

a

lower

scan rate

should

give

a

better

agreement

between

calo-rimetry and TRXRD and should reduce the transition

range

determined by both techniques.

We

speculated that radiation damage

was

the

cause

of

the

discrepancy between the

two

methods and ofthe

dra-matic

lowering of the

onsettemperature

ofthe transition

observed

by TRXRD.

A

TRXRD

measurementon

hy-drated

DEPE

incubated

2.2°C below the

L,6-to-La

transi-tion

temperature

demonstrated

the

appearance

of

the La

phase in

an

L,

background

on

continuous

x-ray

expo-sure

(Fig. 7). The

La

phase

grewat

the

expense

of the

L,6

phase throughout the experiment. This result showed

convincingly that

x-ray exposures on

the time scale of

the

current

SCALTRD

experiments

are

sufficient

to

in-duce

radiation

damage and

to

alter the mesophase

prop-erties of

the

lipid.

Furthermore, although TRXRD is

performed

on

and is

responding

to

the

very same

volume

that is

undergoing radiation damage, the calorimetric

Chung and Caffrey Hydrated LipidEstablishedby X-ray Diffraction

measurement is

responding

to

>98% of the

sample

vol-ume

that has not succumbed to

damage

and is still fresh.

It

is

not

unreasonable,

therefore,

to

expect

a

different

behavior

monitored

by

TRXRD and

calorimetry

in

SCALTRD measurements where radiation

damage

ef-fects

are

significant.

The problem

of

radiation

damage can be

minimized

by

moving

the

sample

continuously

in

the x-ray beam

and/or by

intermittent exposure

to

the

x-ray beam. The

object in

both cases

is

to

ensure

that any one part

of

the

sample

is not

exposed

to

x-rays for

a

period long

enough

to

induce

damage.

Alternatively,

with

an

x-ray

source

of

large

divergence, the optical

arrangement can

be

such

as

to

focus

x-rays not on

the

sample

but

on

the detector.

This

has

the

effect of

distributing the incident flux

over

a

larger area

and

of

reducing

the accumulated

dose

at

any

one

location

in

the

sample. Such

was

the arrangement at

NSLS.

Indeed,

with this

configuration,

the transition

range

measured

by

TRXRD

was

within that observed by

calorimetry

at

a scan

rate

of

5°C/min for DEPE/water

(data

not

shown). However,

we

must note

that

a

direct

comparison

between

NSLS

and

CHESS is

not

appro-priate

because

of

the

differences

in overall

flux and

x-ray

energies used.

Flux

was

lower

at

NSLS

and

the

energies

used were 8.0 and

13.8

keV at

CHESS and

NSLS,

respec-tively.

These results

serve to

highlight the potential problems

associated with radiation damage during

SCALTRD

measurements and the need to

investigate

the

magnitude

of

the

effect

on a

per

sample

basis. The

issue of

synchro-tron

x-radiation damage of lipids is

an

ongoing

concern

in

this laboratory (Caffrey,

1984;

Cheng and

Caffrey,

1991, 1992;

Cheng and

Caffrey,

manuscript in

prepara-tion).

CONCLUSIONS

Simultaneous

calorimetry

and TRXRD

has been

carried

out

successfully

on a

single hydrated

lipid sample. This

"methods

in

combination" approach

provides for

a

di-rect

correlation of thermodynamic

and structural

infor-mation

free of artifacts

that can

arise

as a

result

of

differ-ences

in

sample thermal

history, preparation,

and source

when

the

measurements are

performed

separately. At

low scan rates,

the

results

obtained by calorimetry

and

by

TRXRD

agree

well in SCALTRD measurements.

At

higher

scan

rates, a

disparity arises

as a result

of

differ-ences

in

the sample volumes

interrogated

by the two

methods and

the

existence of

thermal gradients in the

sample. Radiation damage

was

found

to

be an

important

issue

for SCALTRD measurements as a result of fact

that the exposure

times

used are

long,

and a very

intense

x-ray source

is

needed to make the

TRXRD

measure-ments. The

damaging

effect

appears to be sample

spe-cific

and must be evaluated

on

an

individual

case

basis.

Despite the

problems, SCALTRD is

a

powerful method

for

collecting

structural

and

thermal data

on

systems

where separate

measurements under

identical

condi-tions

are

impractical.

We thanktheentire CHESS (National Science Foundation grant DMR 12822), MacCHESS (National Institutes of Health grant RR-0 14646), and NSLS(Department ofEnergy grant KP04-01 and-BO43)staff for their invaluablehelp and support. We also thank S.-T. Lin for revising theleast-square fitting program used in data analysisand J. Briggs, A.-C.Cheng, J. Hogan, A. Mencke, J. Wang, and T. Zhufor critical commentsonthismanuscriptandfor valuablediscussions.

This work was supported byagrantfrom the National Institutesof Health(DK-36849).

Receivedfor

publication 30 January 1992 and

infinalform

10

April 1992.

REFERENCES

Caffrey,M.1984.X-radiationdamageofhydratedlecithin membranes detected by real-time x-ray diffraction using wiggler-enhanced synchrotron radiation astheionizingradiationsource.Nuclear In-strumentsand MethodsinPhysics Research.222:329-338. Caffrey,M.1985.Kinetics and mechanism ofthelamellargel/lamellar

liquid crystal andlamellar/inverted hexagonal phase transition in phosphatidylethanolamine: a real-time x-raydiffractionstudyusing synchrotronradiation.

Biochemistry.

24:4826-4844.

Caffrey,M. 1987.Kinetics and mechanism of transitionsinvolvingthe lamellar, cubic, inverted hexagonal, and fluid isotropicphases of hydratedmonoacylglycerides monitored bytime-resolvedx-ray dif-fraction.Biochemistry. 26:6349-6363.

Caffrey,M. 1989. Thestudyoflipidphasetransitionkinetics by time-resolved x-ray diffraction. Annu. Rev. Biophys. Biophys. Chem.

18: 159-186.

Caffrey,M., J.Hogan, andA.Mencke.199la. Kinetics of the barotro-pic ripple(P,)/lamellar liquidcrystal(La)phasetransition infully hydrated dimyristoylphosphatidylcholine (DMPC) monitored by time-resolved x-ray diffraction.Biophys.J.60:456-466.

Caffrey,M., J. Hogan,andA.S.Rudolph. 199 lb. Diacetylenic lipid microstructures:Structuralcharacterization by x-raydiffractionand comparison with the saturatedphosphatidylcholineanalogue. Bio-chemistry.30:2134-2146.

Caffrey,M., R.L.Magin,B.Hummel, and J. Zhang. 1990.Kinetics of the lamellar andhexagonalphase transitions in phosphatidylethanol-amine:time-resolvedx-raydiffractionstudyusinga microwave-in-duced

temperature jump. Biophys.

J.

58:21-29.

Cheng,A., andM.Caffrey.1991.X-radiation damageofaqueous phos-pholipid dispersions monitored by x-ray diffraction. Biophys. J. 59:634a.(Abstr.)

Cheng,A., and M.Caffrey. 1992.Observationof an x-radiation dam-age-induced

H.1

phase infullyhydrated DHPE.Biophys.J.61:1777. Finegold,L., S.J.Melnick,and M.A.Singer. 1985. The thermal prop-erties ofdilaurylphosphotidylethanolamine liposomes are affected bylipidsource andpreparation. Chem. Phys. Lipids.38:387-390. Garti, N., and K. Sato, editors. 1988. Crystallization and

Polymor-phismof Fats andFattyacids. Marcel Dekker, Inc., New York. 450

PP.

Hogan,J. L. 1989. Thepolymorphism of headgroup methylated phos-phatidylethanolamines. Ph.D.thesis. TheUniversity of Southamp-ton,England. 279 pp.

Kellens, M., W. Meeussen, R. Gehrke, and H. Reynaers. 199 la. Synchrotron radiation investigations of thepolymorphic transitions of saturated monoacid triglycerides. Part 1: Tripalmitin and tris-tearin.Chem.Phys. Lipids. 58:131-144.

Kellens, M., W. Meeussen, A.Hammersley, and H. Reynaers. 199 lb. Synchrotronradiationinvestigationsof thepolymorphictransitions insaturated monoacid triglycerides. Part 2: Polymorphism study ofa 50:50 mixture of tripalmitin and tristearin during crystallization and melting.Chem. Phys.Lipids.58:145-158.

Kim, J. T., J. Mattai, and G. G. Shipley. 1987. Gel phase polymor-phism in ether-linked dihexadecylphosphatidylcholine bilayers. Bio-chemistry. 26:6592-6598.

Klug, H. P.,and L. E.Alexander. 1974. X-RayDiffraction Procedures: ForPolycrystallineandAmorphousMaterials. 2nd ed. John Wiley & Sons,Inc., New York. 966 pp.

Luzzati, V. 1968. X-raydiffraction studies of lipid-water systems. In Biological Membranes. Vol. 1. D. Chapman,editor.Academic Press, NewYork. 71-123.

Macgillavry, C. H., and G. D.Rieck, editors. International Tables for X-Ray Crystallography. 1962. Vol. III. The Kynoch Press, Bir-mingham, England. 362 pp.

Marsh, D.1990. HandbookofLipid Bilayers. CRC Press, Boca Raton, FL.387 pp.

Martin, J. 1953.Preparation of saturated and unsaturated symmetrical monoglycerides.J. Am.Chem.Soc.75:5482-5486.

McElhaney, R. N. 1986. Differential scanning calorimetric studies of lipid-protein interactions in model membrane systems. Biochim. Biophys.Acta.864:361-421.

Mencke,A.P., and M. Caffrey. 1991. Kinetics and mechanism of the pressure-inducedlamellarorder/disorder transition in phosphatidyl-ethanolamine:atime-resolved x-raydiffraction study. Biochemistry. 30:2453-2463.

Nagle, J. F., and M. C. Wiener. 1988. Structure of fully hydrated bi-layerdispersions. Biochim. Biophys. Acta. 942: 1-10.

Russell T. P., and J. T. Koberstein. 1985. Simultaneous differential scanning calorimetryandsmall-angle x-rayscattering.J.Polym. Sci. Polym. Phys. 23:1109-1115.

Seddon, J. M., K. Harlos, and D. Marsh. 1983. Metastability and poly-morphism in the gel andfluid bilayer phases of dilauroylphosphati-dylethanolamine: two crystalline forms in excess water. J. Biol. Chem. 258:3850-3854.

Slater, J. L., C. H. Huang, R. G. Adams, and I. W.Levin. 1989. Poly-morphic phase behavior of lysophosphatidylethanolamine disper-sions. A thermodynamic and spectroscopic characterization. Biophys. J. 56:243-252.

Small, D. M. 1986. Handbook ofLipidResearch. ThePhysical Chemis-tryofLipidsfrom AlkanestoPhospholipids. Vol. 4. Plenum Pub-lishing Corp., New York. 672 pp.