Biochimica et Biophysica Acta, I i 28 (1992) 93-104

~.-~1992 Elsevier Science Publishers B.V. All rights reserved 01105-2760/92/$(~5.00

93

BBALIP 54002

Structure of polymerizable lipid bilayers. V synthesis, bilayer structure and properties of diacetylenic ether and ester lipids David G. Rhodes ~', Zhenchun Xu b and Robert Bittman b a Biomolecuko" Stnwtto'e Analysis Centea; Department oJ"Radi~d,,~,y, Unit'ersity of Con,tectict¢t Heahh Center, Farmington CT (USA) ami t, Department of Chemisto, and Biochemistry. Q,:e,,n.~C,d;t-se of CUN~, Flushhtg, NY (USA)

(Received 16 December 1091) (Revised manuscript received 24 April 1092)

Key words: Diacelylene; Lipid bilayer; Permeability; Polymerizable lipid; Ether lipid; Bilayer structure; X-ray diffr~tcfion Four diacetylenic t~hosphatidylcholines (PC's) have been synthesized and the structures of bilayers of these lipids have been determined at low resolution by low-angle X-ray diffraction. The PC's all have 18-carbon chains but differ with rcspccl to the ether/ester linkage at the sn-I and sn-2 positions and the relative position of the diacctylcnc moiety: diester-PC (1): 1,2-bis(octadeca-4',6'-diynoyl).sn.glycero-3.phosphocholine diester-PC (2): I-(octadeca-4',6'-diynoyl)-2-(octadeca-5',7 '-diynoyl)-sn-glycero-3-phosphocholine diester-PC (3): 1,2-bis(octadeca-8',10'-diynoyi).sn-glycero-3-pimsphocholine diether-PC (4): ~-~-(~ctadeca-4'~6~-diyny~)-2-~-(~ctadeca-5''~7"-diny~)-sn-g~ycer~-3-ph~sph~ch~in~ Only (I) exhibits the typical bilayer profile, whereas (2), (3) and (4) show evidence of interdigitation and/or significant disorder. Only (1) polymerized effectively upon illumination with 254 nm light, turning deep blue in seconds, indicating the formation of long, well-ordered polydiacetylenic structures. Lipo~o,nes of these derivatives were tested for permeability by osmotic swelling. Polymerized liposomes of (1) underwent osmotic swelling with urea, glycerol, and acetamide more rapidly than did liposomcs of stearoyl-oleoyl-PC, but the initial rates of osmotic swelling of polymerizcd liposomes of (1) were 3-10-times lower than those of unpolymerized iiposomes of (1). Blue polymerized multilayer samples of (1) exhibited an irreversible thermochromic transition to red at approx. 40°C. Differential scanning calorimetry, with liposome suspensions of (1) revealed an endotherm at 28.3°C witll a transition enthalpy of 40 J/g. PC (1) is a potentially useful diacetylenic lipid which exhibits facile, complete polymcrizalion and a bilayer thickness comparable to lhat ol biomembranc lipids.

Introduction Work with polymerizable diacetylenic phospholipids has been ongoing for several years [1-3]. Because of the tremendous potential of these materials, there has been a significant effort to develop improved procedures for the preparation of unsymmetrical diynes, especially positional isomers of conjugated diacetylenic

Corresp(~ndence to: D.G. Rhodes. Biomolecular Structure Analysis Center, Department of Radiology, University of Connecticut Health Center, Farmington, CT 06032, USA. i Previous papers in this series are: (1) Chem. Phys. Lipids 49, 39-17 (1989); (2) Biochim. Biophys. Acta 1022, 291-295 (199(I);(3)Che,n. Phys. Lipids 58, 41-54 (1991); (4) Chem. Phys. Lipids 59, 215-224. Abbreviations: PC, phosphatidylcholine; T,,,, thermotropic phase transition temperature (melting); AH, n, thermotropic phase transition enthalpy (melting); Tr, thermotropic phase transition temperature (fusion); AH r, thermotropic phase transition enthalpy (fusion); DMPC, dimyristoyI-PC; DNPC, dinonanoyI-PC.

alcohols and acids [4]. In addition, efficient acylation procedures for introduction of polymerizable chains into phosphatidylcholine (PC) have been developed [5]. Suitable derivatives could find potential applications in materials design, drug delivery, biosensors and electronics, as well as in basic membrane research. Despite this work, the relation of monomer structure to the morphology of lipid assemblies and properties is poorly understood at present [6]. Most investigations have been with symmetric long-chain (C.,.~-C27) PC's [7,8] or fatty acids [9]. Spectroscopic data from long-chain diacetylenic lipid bilayers [10] have indicated that the methylene segments in these specimens are highly ordered. There has been some effort to reconcile the conflicting demands of diacetylene polymerization, i.e,, that the successive diacetylene moieties be aligned in a near crystalline array and that there be adequate flexibility in the lattice to accommodate the conformational change induced by polymer formation. It has been shown that

94 under some conditions the DCm.,,PC compounds polymerize poorly, even at T < Tm, the thermal phase transition temperature [11]. Although the reason for the limitation is not clear, it is known that doping certain short-chain saturated PC lipids into the DC,,,.,PC bilayer enhances polymerization [12] *. It has been suggested, based on low resolution structure data [ 13], that the addition of the so-called 'spacer lipids' improves the ability of the lattice in the bilayer to adapt to the polymerization. The model postulates that the m region of the bilayer becomes more ordered, while the n region of the bilaycr is more flexible. This allows the diaceq, lene packing to be established, but not locked into position. Another interesting aspect of the DC,.,,,PC diacetylenie lipids is the k~eation of the diacetylene in the two acyl chains. Models based on small-angle X-ray scattering suggest that the diacetylene moieties may be at different levels (positions along the bih~yer normal), with the diacetylene on the sn.2 chain being deeper in the bilayer [7], The difficulty with this model is that if all diacetylenes are invol~ed in polymer formation, polymer must exist at two separate depths in the bilayer, one for the sn-I chain diacetylene and one for the sn-2 chain diacetylene. Otherwise, polymer could form at only one level, involving only diacetylene in one chain (sa-I or sn-2) of each lipid in the polymer, it is not yet clear whether one or both acyl chains are involved in polymerization, but if only one is involved it may be possible to improve the extent of polymerization by increasing the length of the proximal acyl chain segment on the sn-I chain. The DC.,.PC (C,~-C,7) acyl chain lipids have lamell~r dimensions i'0-15 A, larger than those of bilayers with composition typical of biomembranes [7,8]. This means that for a membrane spanning protein in an t~-helical conformation in a bilayer of C.,.a-C~7 acyl chains with diacetylenie moieties, 7-11) additional amino acids (2-3 turns) would be required to span the bilayer. Thus, any effort to reconstitute proteins that normally reside in lipid bilayers of biological origin into these systems risks distorting the protein so as to inactivate the functional site(s). This report describes the synthesis and structural investigation of several diacetylenic phospholipids with acyl chains that are more comparable in length to those of naturally occurring phospholipids (Fig. 1). The.~ C~, derivatives were examined in terms of lamel-

0

II O OC(CH2)n(C~ C);~(CH2).,CH3 CH3(CH2)m(C~C)2(CH2)n(~O==,- q H O 1: m=10, n=2 ~ O IPIOCH2CH2N*(CH3)3

I.

0

3:m= 4, n=8

O

II O F----OC(CH~);~(C ---~C)2(CH2hoCH3 CHa(CH2~9(C"C)2(CH~)3 ICIo~ = ~ H

~1

,•

L'-'-O~OCH2CH~N (CH3la 2

o

IC)a(CH2)loCH~ ~ '-HO(CHa)3(C O

CH3(CH2)9(CmcC}z(CH2),O " T "~°" II 4 ~'-'---OI~OCH2CH~N (CH~)3 O

Fig. !. Structures of ~he diester (I, Z, 3) and deither (4) ?C's used in this study. Each has an equal number of carbon atoms in the acyl chain.

lar structure, polymerizability and permeability. With the diacetylene moieties offset in derivatives (2) and (4), one migh~ expect to find diacetylenes from both acyl chains at the same depth in the bilayer, so that polymerization might be improved. Because most of these derivatives involve very short proximal acyl chain segments, the packing (alignment) of the diacetylenes may be more strongly influenced by the glycerol backbone conformation and packing. Similarly, the proximal acyl chain segments themselves may be disordered by constraints of the adjacent regions of the molecule (because they are not long enough to establish extended all-trans linear domains) and could provide needed flexibility in the polymer formation process. FinallY,, the ether-linked derivative (4) could provide a more favorable diacetylene arrangement by virtue of differences in the dipole moments near the backbone or steric differences. Because of the competing requirements for diacetylene polymerization, however, any of these modifications could result in diminished polymerizability. Materials and Methods

*

Symmetric diacetylenic PC's are often referred to as DC,,.,PC, where m and n refer to the number of CH ,'s between the ester and diacetylene and between the diacetylene and the terminal (SH,~, respectively, on each chain. The m segment is also referred to as the "'proximal" aeyl chain segment and the n segment as the "distal" acyl chain segment.

The 4,6- and 5,7-octadecadiynoic acids were synthesized from (Z)-l-methoxybut-l-en-3-yne as described previously [4]. The progress of reactions ~as monitored by TLC on 0.25-mm thick silica-gel GF glass plates (Analtech, Newark, DE). Compounds were detected by

95

C H 3(CH 2),~C'Z--C ~ C(CH 2)4°

~

O(CH 2)3C --~C ~ C(CH 2)mCH I, ('IP(()Me)N( Pr-i )2, El 3N 2, H O ( ("H 2 )2 N + ( C H ~ ) ~ ()Ts-.

telnlzole 3. t-i~uOOH 4. (('H O ~ N

H

t___ O H

~

- O(CH, )3C ~ C-----C(CH. )..Ctl .~

CH.~(CH ~),,C---- C-----~C(CH. hO

~,--!t 0 L O

IPIOCI-I,CH. N + (CH ~)~

I O

Scheme I. Prcparalion of diclhcr-PC 4. spraying with 10% sulfuric acid in ethanol followed by charring. Flash chromatography, NMR, and IR methods were as described previously [14]. Elemental analyses were done by Desert Anatytics (Tuseon, AZ). Solvents were dried as described elsewhere [14].

Synthesis

74.99, 7(1.99, 7(1.91, 66.24, 66.17, 66.05, 64.99, 63.24, 59.32, 59.28, 54.26, 32.95, 31.87, 29.60, 29.49, 29.32, 29.10, 28.92, 28.33, 22.65, 19.15, 14.97 (to-CH.a), 14.08 (to-CH3). Analysis calculated |or C44HTsOIINP. 3H20: C, 63.82; H, 9.49; N, 1.69. Found: C, 63.38, H, 9.10; N, 1.72.

1,2-Di-(4',6'-octadecadiynoyl)-sn-glycero-3-phosphocholine (1). To a suspension of 247 mg (0.5 mmol) of L-a-glycerophosphorylcholine, cadmium chloride complex (dried for 5 h over phosphorus pentoxide under vacuum at 78°C), 122 mg (1.0 mmol) of 4-(dimethylamino)pyridine and 586 mg (2.12 mmol) of 4,6-octadecadiynoic acid in 6 ml of freshly distilled alcoholfree chloroform was added 437 mg (2.12 mmol) of dicyclohexylcarbodiimide. The reaction mixture was stirred for 60 h at room temperature under nitrogen in the dark. Chloroform (15 ml) was added, and the mixture was filtered through a Celite pad, which was washed with 15 ml of chloroform. Removal of the solvents gave a residue that was purified by flash chromatography (elution first with 100% CHCI 3, then with 90: 10, 60:40, 20:80 CHCI3/CH3OH) to yield a white solid. The suspended silica gel was removed by filtering a chloroform solution of the solid through a 0.45-~m Metricel filter (Acrodisc-CR, Gelman Sciences, purchased from Baxter Healthcare) three times. The product was lyophilized with benzene, affording 352 mg (0.45 mmol, 81%) of 1,2-di-(4',6'-octadecadiynoyl)-sn-glycero-3-phosphocholine (1) as a white solid: [ot]~5 + 4.41° (c 1.125, CHCI3/CH3OH 1" 1); JHNMR (300 MHz, CDCI3)8 5.21 (m, 1 H, CHzCH_.CH2), 3.74-4.42 (m, 14 H, POCH2CH2N, CI-12CHCI-I2, (HaO)3), 3.33 (s, 9 H, N(CH3)3), 2.56 (m, 8 H, (OCCHaCH a f.~-C)2), 2.24 (t, 4 H, (CH a C~-C)a), 1.191.55 (m, 36 H, (CH2)Is), 0.87 (t, 6 H, (a~-CH3)2). 13C-NMR (75 MHz, CDCi 3) 8 171.42 (CH2OCO), 171.12 (CH2OCO), 78.42, 78.34, 77.42, 77.00, 76.58,

sn-glycero-3-phosphocholine (2). To a solution of 7(1 mg (0.(190 mmoi) of 1,2-di-(4',6'-octadecadiynyl)-snglycero-3-phosphocholine (1) in 10 ml of 49:1 ether/CH3OH solution were added 2 ml of 100 mM sodium borate buffer (pH 7.4) containing 4 mg/ml of calcium acetate and 300 gl (150 traits) of phospholipase A 2 (Naja naja) in sodium borate buffer (pH 7.4). The reaction mixture was stirred for 4 days at room temperature in the dark. TLC analysis (elution with 95 : 35 : 6 C H C I 3 / C H 3 0 H / H 2 0 ) showed complete conversion of diacyl-PC (I) (Rt.. = 0.48) into the desired iyso-PC (RF = 0.27). Water (10 mD was added and the mixture was extracted with ether (3 × 25 ml). The liberated fatty acid was in the ether layer and the lyso-PC remained in the aqueous layer. The lyso-PC was extracted three times with 25 ml of 2 : 1 CHCla/CH 3 OH solution and then dried over sodium sulfate. After the solvents were removed, iyso-PC was obtained as a white solid. Column chromatography on silica gel was avoided, since lyso-PC can undergo signif icant acyi chain migration. A solution of 24 mg ((I.2 mmol) of 4-(dimethylamino)pyridine and 55 mg ((I.2 mmol) of 5,7-octadecadiynoic acid in 2 ml of t'reshly distilled alcohol-free chloroform was added, followed by 83 mg (0.4 mmol) of dicyclohexyicarbodiimide. The reaction mixture was stirred for 2 days at room temperature under nitrogen in the dark. Removal of the solvents gave a residue that was purified by flash chromatography (elution first with l(lf)°/~ CHCI 3, then with 90: 10, 60:40 and finally with 20:80 CHCi.~/

1-(4',6'-Octadecadiynoyl)-2-(5",7"-octadecadiynoyl)-

96 CHaOH) to yield a light yellow residue. The suspended silica gel was removed by filtering a chloroform solution of the residue through a 0.45-/.tm Metricel filter three times. The product was lyophilized with benzene, affording 38 mg (0.049 retool, 54%) of 1-(4'-6 '-octadecadiynoyl )-2-(5",7 "-oetadecadiynoyl)-snglycero-3-phosphocholine (2) as a white solid; [a]~ + 7.15° (c 1.02, CHCla/CHaOH 1" 1); IH-NMR (300 MHz, CDCI a) 8 5.20 (m, 1 H, CH2CHCH2), 3.59-4.42 (m, 14 H, POCH,CH.,N, CH:CHCH.,, (H20)3), 3.36 (s, 9H, N (CH~)~), 2.21-2.85 (m, 12 H, (CH2CO2) 2, (CH, C-~CH:),) 1.26-1.83 (m, 36 H, (CH,)Is), 0.88 (t, J~6,34 Hz, 6 H, (oJ-CH~),). LaC NMR (75 MHz CDCla) c5 172.34 (CH,OCO), 171.36 (CH.,OC_.O) 104.87, 78.38, 78.13, 77.42, 77,26, 77.21, 77.00, 76.57. 75.96, 74.92, 70.65, 66.36, 66.21, 66.04, 65.05, 64.98 64.89, 63.44, 63.35, 63.29, 54.46, 34.75, 32.93, 32.76. 31.86, 29.58, 29.47, 29.30, 29.19, 29.08, 28.89, 28.52. 28.31, 26.97, 25.84, 23.54, 22.65, 19.15, 18.53, 14.95 (to-CHa), 14.09 (to-CHa). Analysis calculated for C4~Hs0Ol., N P . 4 H , O: C, 62.46; H, 9.53; N, 1.66. Found: C, 61.82; H, 9:15, N, 1.54. 1,2-Bis(octadeca-8',10'-diynoyi)-sn-glycero-3.phos. phocholine (3). This compound was prepared from 10,12-octadecadiynoic acid [4] in 52% yield by using the precedure described for the preparation of 1,2-diat t . • ( ,6 -octadecad,ynoyl)-sn-glycero-3-phosphochohne (1); [a]~ -3.10 o (c !.55, CHCI3/CH3OH I:1; IH NMR (200 MHz, CDCla)/$ 5.17 (m, 1 H. CH,CHCH,), 3.67-4.42 (m, 12 H, POCH,CH,N, CH,CHCH,, (H, O),), 3.35 (s, 9 H, N(CH~h~','2.14-~.~5 (m7"i2 "H~ (CH ;CO:),, (CH ~C--.C)4), 1.30- !.56 (m, 36 H, (CH.,)is), 0.90 (t, J -- 6.97 Hz, 6 H, (co.CH a)a. Analysis calculated for C,~ H ~.,O~NP. El ,O: C, 62.46; H, 9.53. Found: C, 61.88; H, 9.71. ! -0-(4',6 '-Oct adecad iynyl)-2-O-(5",7 "-octadecad iynylkvn-glycero-3-phosphocholine (4, Scheme I). To a solution of 52 mg (0.09 retool) of I-O-(4',6'.octadeca. diynylb2-O-(5",7".octadecadiynylbsn.glycerol ($) and 75 #1 (54.5 rag, 0.54 mmol) of triethylamine in 2 ml of CH,CI: was added with a syringe 22.5 #1 (23 rag, 95% pure, 0. Ii retool, about 20% molar excess) of N,N-di. i,~propyimethylphosphonamidic chloride. The reaction mixture was stirred for 5 min at -5°C and then concentrated to dryness under vacuum. I H-Tetrazole (25 mg, (I,36 mmol) and 74 mg ((I.27 mmol) of choline tosylate (both were dried overnight over phosphorus pentoxide under vacuum at 78°C) were placed in the ~action flask and the mixture was dissolved in 5 ml of 1:1 acetonitrile/THF solution. The reaction mixture was stirred for 3 h at room temperature and then evaporated to dryness under vacuum. After the residue w:~s mostly dissolved in 3 ml of THF, 33 ~1 (0.1 mmol, ~ q c~c~s) of a 3 M solution of anhydrous ten-butyl hydrooert~xidc in 2,2,4-trimethylpentane was added. The mixture was stirred for 3 h at ro~m ~perature.

After addition of 5 ml of ethyl acetate, the resulting solution was washed with triethylammonium hydrogen carbonate buffer (1 M, pH 7.5) to remove the excess of 1H-tetrazole and choline tosylate. The aqueous layer was extracted three times with chloroform. The combined organic layer was concentrated to dryness and the residue was rendered anhydrous by repeated evaporation with dry 2-propanol. Finally, 4 ml of toluene was added and the mixture was transferred to a pressure bottle. Anhydrous trimethylamine (1 ml) was added. The reaction mixture was stirred for 18 h at room temperature. Removal of the excess trimethylamine gave a residue that was purified by flash chromatography (elution with 9 5 : 3 5 : 6 CHCI3/ CFI3OH/H.,O) to yield a light yellow residue. The suspended silica gel was removed by filtering a chloroform solution of the residue through a 0.45-/zm Metrieel filter three times. Tile product was lyophiUzed with benzene, affording 39.6 mg (0.053 mmol, 59%) of !.O. (4',6'-octadecadiynyl).2-O-(5 ",7"-octadecadiynyl):~,n. glycero-3-phosphocholine (4) as a light yellow solid; [a]~ -2.27* (c 0.90, CHCI3/CH3OH !" I); IH NMR (200 MHz, CDCI 3) ~$4.41 (m 2H, CHCH 2OP), 3.86 (m, 4 H, POCH,CH2N), 3.53 (m, 7 H, CH_,OCH2CHCH,), 3.31 is, 9 H, N(CH3)3), 2.20-2.34 (m, 8 H, (CH, C-a=C)4), 1.26-1.91 (m, 40 H, (CH2).,o), (I.88 (t, J = 6.2.~ Hz, 6 H, (oJ-CH~)2). Analysis calculated for C44 HT,O, NP" 2H20: C, 67.57; H, 10.31; N, 1.79, P, 3.96. Found: C. 67.81; H, 10.32; N, 1.80; P, 3.99.

Osmotic swelling of liposomes The effect of polymerization on the initial rates of osmotic swelling was measured by light scattering at 25°C. The desired amounts of phospholipids were withdrawn from stock solutions in chloroform and placed in 20-ml vials. The chloroform was removed under a stream of nitrogen and the lipids were evaporated to dryness under vacuum. The lipid film was suspended in 60 mM aqueous KCI solution by agitation on a vortex mixer for 3 min. One glass bead (3-mm diameter) was added per 2 ml of suspending KCI solution prior to mixing. A 0.l-ml aliquot of liposomes suspended in 0.06 M KCI was mixed rapidly with 0.7 ml of 0.4 M nonelectrolyte (acetamide, glycerol, or urea)• Liposeines contained 4 mol% of dicetyl phosphate. The total lipid concentration in the liposomes was 0.5 raM. The change in liposome volume (AV) was measured as the change in the reciprocal of the absorbance at 400 nm due to light scattering (A(I/A)) as described previously [15]. Liposomes containing 4 mol% dicetyl phosphate were photopolymerized by illumination at 254 nm for 1 h at 4°C with a black light (UVG-11 Mineralight, purchased from Fisher Scientific) positioned 4 cm above the suspension.

97

Diffraction Chloroform was HPLC grade (Aldrich) and was used without further purification. Water was purified with a deionization/ultrafiltration system (Vanguard RGW-5). Liposome preparations were made by drying the lipid from chloroform with a stream of nitrogen and then subjecting the sample to mechanical vacuum for several hours to overnight. The lipid film was rehydrated by adding water to produce a final concentration of 4 mg/ml at 50°C, allowing the sample to hydrate and vortex gently. Multibilayers were formed as described previously [7,8], using a variation of the method of Clark et al. [16]. Briefly, aliquots (50 #l) of concentrated liposome suspension containing 200 ~tg of lipid were added to specially constructed sedimentation cells [17] designed to fit in the buckets of a Beckman SW-28 rotor. The caps of the buckets have a 100 ~m pinhole. With the rotor spinning at 1001) RPM the centrifuge vacuum was applied, removing the water through the pinhole. Within 1 h, most of the water had evaporated, leaving a multibilayer stack on a thin aluminum foil. if samples were polymerized, they were exposed to 254 nm light at this point. In a cold room (approx. 4°C, approx. 95% r.h.), unpolymerized multibilayer samples were placed in a shallow plastic container, the edges of which served as a support for the UV light. The samples were then exposed to 254-nm light (UVG-I 1) for the desired period of time, 30 s for complete polymerization of 1. The light source (rated at 1.8 × 10 I's photons/s/cm-' at 76 mm) was approx. 5 mm from the samples. Samples were then mounted on curved glass and placed in sealed brass canisters with a plastic cup of saturated salt solution to reguhtte relative humidity (98-66%). Most of the diffraction data reported here were obtained at 15°C. Canisters were placed at this temperature and allowed to equilibrate for at least 18 h prior to initial collection of diffraction data and following any change of conditions. Mounted samples were not removed from the canisters during the course of the experiment. Diffraction data were collected as described previously using a GX-!8 rotating anode microfocus generator (Marconi Avionics) and a fixed geometry beamline with a Franks mirror providing line focus at the detection plane. An Ni filter (0.025 mm) was used with a Cu target to provide K(,,) X-rays (A = 1.54/~). A one-dimensional quartz wire electronic detector (Braun, Innovative Technology) was used for collecting iamellar scattering data and film was used with a point focus beamline to obtain equatorial data. Diffraction orders were integrated by summing the X-ray counts between channels judged to be within the background region on either side of the order and subtracting a trapezoidal background area determined by applying a linear fit to the background on either side of the order. Phases were determined using a

swelling analysis [18] by plotting F (the square root of normalized intensity) as a function of h/d (where h is the lamellar order number and d is the unit cell dimension in the bilayer normal direction) for a series of exposures at different humidities. The intensity data were then Lorentz corrected and subjected to a Fourier transform with the appropriate phase to produce an electron density profile structure as described previously [8]. Spacings from films were measured using an Enraf Nonius film reader.

Differential scanning calorimeoy All scans were carried out on a Du Pont DSC 2910 differential scanning calorimeter with a TA2000 PCbased analysis system. Samples for calorimetry were prepared by drying a sample of lipid in CHCi 3 overnight under vacuum and then hydrating the dry film by adding ultrapure water to yield a final concentration of 8 mg/ml. The sample was kept at room temperature and intenmttently vortexed (approx. 10 s at intewals of approx. 10 rain) over a period of approx. 1 h. For DSC of suspensions, 15 gl of the 8 mg/ml suspension was hermetically sealed in aluminum pans. For DSC of multibilayers, 25 /zl was sedimented onto aluminum foils (see above). The sample mass was 120 #g for suspensions and 200 #g for multibilayers. Most scans were performed at a gradient of 2C°/min over a temperature range of 4°C to 50°C. Under these conditions, with this instrument, this represents 2-4-times the mass required for quantitative determination of transition enthalpies and temperatures lor dimyristoyl PC (DMPC). Fi~. 2 shows a heating scan of a 151) #g sample of DMPC (11} mg/ml), in tiffs and some other figures, corrections were implemented for a sloped baseline. Extended scans from -50°C to 90°C were also performed. Glycerol was not added to the solu-

E L_

r-

"13 e-

UJ

i

T

(°C)

Fig. 2. Heating thermogramof DMPC. These data were obt~dned with 150 #g of DMPC in 15 ,u,I. and serve to demonstrate the sensitivityof the instrumentused in thiswork.

98 tion, so an endotherm due to the solid/liquid transition of water obscured the scan near 0°C. To minimize this effect, the scan was paused at 4°C to reequilibrate. No transitions were observed at -50°C < T < approx. 0*C or 50°C < T < 90°, so the extended ranges were not routinely used. Typical scans included heating from 4"C to 50"C, a pause of 5 rain, and cooling from 50°C to 40C. This cycle was normally repeated four times. Further details of the scans are listed in the text with the data. Polymerized liposome suspensions were prepared by exposure of the suspension to 254 nm light (as described above) after adding the sample to the DSC pan, Multibilayer samples were polymerized as described above, and then placed into DSC pans, For some experiments the pol~,merization was carried out for time intervals ranging from 1 s to 61) s,

Si~ctrtx~copy All spectra were obtained using a Perkin Elmer Lambda-3B spectrophotometer. Suspensions of polymerized (blue) 1 liposomes at 3 rag/! were transferred to a quartz cuvette at 4"C, The cuvette was placed in the sample holder, which had been pre-cooled to 5"C. A polyethylene coated thermistor probe was inserted intt~ the solution at the top of the cuvette to monitor the temperature. The temperature was regulated to approx. + 0.5"C. Scans were obtained from 700 nm to 400 nm. Data were collected at increasing temperatures, allowing several minutes beween temperature changes for reequilibration, Results

Liposome permeability Liposomes behave as osmometers, as the change in iiposome volume (41F) is proportional to the change in

TABLE 1

hzitial rates of osmotic swelling of PC liposomes at 25°C See Materials and Methods section for experimental conditions. (A ( l / A ) per rain)

PC in Liposomes SOPC Unpolymerized I Polymerized 1

acetamide

glycerol

urea

4,2 :l:0,3 24.1 -+0.8 2.8 :t:0.2

2.6 =l:0.2 18,3-+ 0,6 4.4 + 0.5

0.9 :i:0.1 9,9 + 0.5 2,4 + 0.1

the reciprocal of the absorbance due to light scattering

(A(I/A)) [15,19]. Table I summarizes the results of osmotic swelling experiments with 1-stearoyl-2-oleoylPC (SOPC) iiposomes and unpolymerized and polymerized liposomes of (1). In the absence of polymerization, (1) forms liposomes that undergo osmotic swelling with hyperosmolar acetamidc, glycerol and urea at much higher initial rates than do liposomes of SOPC. On polymerization, the initial rate of osmotic swelling of liposomes formed from (1) decreased markedly, but were still somewhat higher than those formed with SOPC liposomes. These results are consistent with the previous reports of reduced permeability of liposomes after photopolymerization [20,21].

Polymerizobility Samples of (1), (2), (3) and (4) were tested for polymerizability at 4°C in solution and in the form of muitilayers. PC's (2), (3) and (4) showed no color change upon exposure to the UV lamp, and were therefore assumed not to polymerize. (After 2 years of storage in the dark as a solid at -20°C, samples of (2) did turn red.) PC (1) turned dark blue immediately (< I s) upon exposure to the ultraviolet light in both solution and multilayer forms. This color indicates long,

b

,

"

i "

T (°C)

z

|

'l's'

|

e

'do'

z

z

'2's'

T (°C)

*

~

'do'

|

|

I-

Fig. 3. DSC thermograms of (I). Heating scans (a) and cooling scans (b) of (I) were obtained from 120/~g of (1) in suspension. In (a) thermograms are from (top to bottom) unpolymerized (1) (Tm = 28.6_+0.2°C, A H m = 54+ 1 J/g), (1) polymerized for 16 s (Tin= 28.2+2°C, AH,w -- 35 _+6 J/g), and a ~ / , g multibilayer of (I) polymerized for 15 s. In (b), times of polymerization are indicated, with unpolymerized (1) (Tf =24.0+0.3~C, AHf= _;6-+6 J/g)(top), polymerization times from ! s to 16 s, (Tf= 19.4°C and 23.1 +0.4°C), and polymerization for 60 s (Tf = 21.q°C and 17.7°C) (bottom).

99 well-ordered polymer, but further efforts to characterize the length distribution were not undertaken. Sometimes, the solution would turn only violet, indicating less complete polymerization. On one occasion, exposure of an olde," sample; which had begun to polymerize as the dry powder and had a slight pink hue, resulted in a deep burgundy color. This suggests the possibility that short chains of polymerized lipids existed in some samples which interrupted the normal polymerization process. Multibilayer samples of (2), (3) and (4) exhibited no color change following routine exposure to X-rays (A ffi 1.54 ,~,) but samples of unpolymerized (1) developed a dark blue streak where X-rays had struck the sample. Thus, structures described below were for polymerized (1) and unpolymerized (2), (3) and (4). Once the blue polymer of (1) had formed (at T < 25°C), the color, and therefore the polymeric structures, was stable for long periods (weeks) if tl~e temperature was maintained at or below room temperature. If multibilayer samples were heated, however, the color changed irreversibly to red (approx. 350C) and orange (approx. 50°C). This most likely represents a disordering of the polymer chains, rather than a decreased chain length. It is not clear why the color change occurred over such a wide range of temperature, but this is not inconsistent with the DSC results (below). The thermochromic transition for liposomal suspensions occurred at a lower temperature ( < 30°C). Visible spectra (below) demonstrated that the spectral shifts occurred gradually with increasing temperature. These observations were correlated with calorimetric events (described below) and the temperatures of the transitions determined accurately.

Differential scanning calorimetry Heating scans with suspensions of polymerized (1) showed a single endothermic transition at T,,, = 28.2 + 0.20C with A H m = 35 + 6 J / g (Fig. 3a). There was no discer,ible difference between the first heating scan, in which the irreversible blue/red thermochromic transition occurred and subsequent scans, in which the sample was red above and below Tm. Heating scans with unpolymerized (1) were quite similar, with Tm= 28.6 + 0.20C and A H m = 54 + 1 J/g. Cooling scans (Fig. 3b) of unpolymerized samples exhibited a single exotherm ~Of'4 at Tf = 24.0 + Nt,..~ ~ with AHf = 56 + 6 J/g. Scans of samples polymerized fo~ even very short times (1 s) revealed an additional exotherm at Tf = 19.4°C, and a small shift of the other exotherm to 23.1 + 0.4°C. Increasing UV exposure from 1 s to 16 s resulted in increasing transition enthalpy at 19.4°C and decreasing enthalpy at 23.1°C, until at 16 s, initial cooling scans revealed no enthalpy at 23.1°C. Extended exposure (60 s), however, resulted in disappearance of the 19.4°C

transition and appearance of transitions at 21.9°C and 17.7°C. Although the exotherm at 23.1°C was absent in the initial cooling scans of samples polymerized for !6 s, this should not be interpreted as indicating complete polymerization. It is valid to conclude that the system is a uniform phase at T > 28.6°C, probably L,r In cooling scans of subsequent heating/cooling cycles, the transition at 23.1°C gradually reappeared, representing over 35% of the total enthalpy in the fourth cooling scan. This strongly suggests segregation of domains of unpolymerized (1) among mixed regions of polymerized and unpolymerized (1). Polymerized partially dehydrated multibilayers of (1) revealed an endotherm at T,,, = 35 ° (Fig. 3a). Based o11 data obtained for other lipid systems (Chester, D., personal communication), the transition temperature is expected to increase as the bilayers become partially dehydrated in forming the multibilayer. Because the multibilayer samples were sealed at ambient humidity, the amc, :it of water associated with the samples is not accurately known, and quantitative data for these samples are less significant. The qualitative observation is significant, however, in that approximate T,,, values for multibilayer samples can be correlated with changes in structural parameters determined by X-ray diffraction, as well as with a thermochromic transition from blue to red. Thin-layer chromatography of samples recovered from DSC sample pans showed two spots, one at the origin and one with a mobility corresponding to monomer. The intensity of the monomer spot decreased with increasing polymerization time, but quantitative determinations were not made. The unpolymerized sample did not show significant material at the origin, but for samples that had been exposed to UV

0,8

"

0.6-

c~ 0,4--

0 . ~)"

.~0

.~0

X(nm)

do

6~o

Fig. 4. Absorption spectra of (1). Data shown were obtained with a suspension at 3 mg/I, in a 1 cm cell, at temperatures indicated, beginning at the lowest temperature.

100 light the dark spot at the origin represented most of the material. It is interesting to note that the monomer spots turned pink during exposure to UV light (at room temperature), while viewing.

Spectroscopy Spectra of suspensions of polymerized (blue) (1) liposomes (3 rag/l) were determined at temperatures from 5°C to 35°C over a range of 700-400 nm. Fig. 4 shows that at 5°C, two absorption maxima appear at 567 nm and 603 nm. A hypsochromic shift occurred for both peaks as the temperature increased and the extinction of the higher energy peak increased, while that of the lower energy peak decreased. As temperatures approached T,,,, the spectrum became noisy and it was observed that large red aggregates had formed in the cuvette. Before data could be collected at 30°C, the sample euvette was agitated by repeated inversion. No correction for light scattering was incorporated in any of these scans. Although quantitative results could not be obtained at 35°C due to continued precipitation, a

single broad absorption maximum appeared at 520 nm with a slight shoulder at approx. 550 nm. These data are consistent with the DSC results which show that the thermal indications of the transition occur slowly over a wide temperature range. It is not clear whether this is due to domain size, polymer length distribution or other factors.

Small angle X-ray scattering Compared to the lamellar scattering observed with other diacetylenic lipids [7,8] the scattering from these systems was relatively weak (Fig. 5). For PC's (2), (3) and (4) it was unusual to have more than 4 orders. An extra reflection, which did not index on the main lattice, was often observed at 20 corresponding to a first order of 10.5 A for PC (4). This was a lamellar reflection which appeared to be sharper than those of the main lattice and with with mosaic spread comparable to that of the main lattice. A second order was occasionally observed. For PC (1), 7-9 lamellar reflections were routinely observed with electronic detection.

30000'

1 b

$ 4

~0000

I' 2

I ~oo0o

J

3

' 6

o

0

0

?00

((:honnel~)

800

900

1000

20 (channels)

40O00-

30OOO

~0000.

......I

'1 3

4

0 20 (channels)

40O

600

800

1000

20 (channels)

Fig. 5. Lamellar diffraction patterns from diacetylene lipid multibilayers: (a) (1), (b) (2), (c) (3), (d) (4). Intensity data are in terms of "counts", X-ra~ evems detected with the position sensitive detector, and "channel numbers", a scale proportional to 20. Lamellar orders are numbered. Higher orders are plotted on an expanded scale, (a) 10 ×, (b) 10 ×, (c) 35 ×, (d) 20 ×.

101 For most lipid bilayers with a moderate degree of ordering, the first lamellar order is the strongest and t h e fourth order is second greatest intensity. For interdigitated bilayers, the third order is usually strong, relative to the second and fourth. Based on this tendency, PC (3) bilayers would be expected to be interdigitated. Bilayers of PC (2) and (4) should have either some interdigitated character or disorder resembling interdigitation, such that the minimum of electron density of the bilayer is broad and ill-defined. The intensity distribution of the lamellar reflections for PC (1) suggests a more conventional bilayer structure and increased ordering in the multilayer. From film data obtained with (1), it is appparent that the structure of the bilayers is different above and below T,,,. Fig. 6 shows that the iamellar repeat distance decreases significantly at high temperature and that the lateral spacings increase. Data obtained at

10°C intervals shows that although slight changes (approx. 2%) in lattice constants occur with any change in temperature, the larger change (approx. 10%) occurs incrementally between 30°C and 40°C. It was also determined that tl~e lattice spacings of (1) at T < T,,, are similar for (blue) samples which have not been exposed to elevated temperatures and (red) samples that have been exposed to temperatures higher than T,,. There appears to be more mosaic spread, which may indicate some irreversible internal disordering, but the fundamental packing of the lipids is unchanged. The electron density profile of (1) (Fig. 7a), with a lameilar repe~t distance (d) of approx. 55 A, could be determined to approx. 6 ,~, resolution. The profile is characteristic of a well-ordered lipid bilayer, with an electron density minimum in the bilayer center and extended shoulders on the main headgroup electron density peaks corresponding to the glycerol backbone

Fig. 6. Diffractionimagesof (1). Data were collected with a point focuscamera at ril = q8%. The temperaturesof the sampleswere (a) T = 10°C. blue (top) and red (bottom)and (b) 59°C. Lattice repeat distancesare listed in Table II. The speckledpatterns are from the aluminumsubstrate and canislerwindows.

102 and diacetylene. The peaks at 5:5 ,A are Fourier artifacts, as indicated by their behavior upon including increasing the number of orders included in the Fourier series. Although the intensity distribution suggested the possibility of interdigitation in bilayers of (2) and (4), the iamellar repeate distance (approx. 55/~) was not consistent with an interdigitated bilayer. Neither of the electron density profiles showed a well-defined electron density minimum at the bilayer center (Fig. 7b, d), but a broad minimum could be discerned. The shape of this electron density profile is similar to that observed for bilayers above the thermal phase transition, T,,,. The melted chains result in an ill-defined minimum at the bilayer center and a lamellar repeat somewhat smaller than that of an equivalent gel-phase sample. The profile of (3) was, as expected, characteristic of an interdigitated bilayer, with a small lamellar repeat distance (40=42 ,~) and no trough at the bilayer center (Fig. 7c). In this case, the lameilar repe-:;t distance (approx. 40 ,~) is more in keeping with an interdigitared system. However, for (3), as with (2) and (4), no thermal phase transition was observed. By steric con-

siderations, the most likely interdigitation would be to overlap the diacetylene moieties from either side of the bilayer or to overlap the diacetylene with the opposing distal acyl chain segment. In either case, one would expect some disordering of the proximal acyl chains and a correspondingly smaller lamellar repeat distance. The resolution of the present data does not allow the two models to be distinguished. Discussion For (1), the resolution and spacing were sufficient to draw some conclusions about the structure, The profile structure clearly indicates an electron density minimum in the bilayer center, but the minimum differs from that observed in long.chain diacetylenic PC's [7,8]. The bilayer structure does not involve intcrdigitation. The strong shoulder on the headgroup peak represents a well-ordered diacetylene and the fact that (I) is easily polymcrizable indicates that in unpolymerized samples, the diacetylene orientation should be approx. 45° from the z-axis and approx. 5 A (in the x,y plane) from adjacent diacetylenes [22]. Finally, the positions of the

10'

O0

\

oo ~

~ -05-10

.lo, - 20

-10

0

10

20

Z(A)

10.

0,5"

~

g _=

oo

off

-05"

.10. .L

-aO

-lb

b Z(A)

lo

2b

-20

-10

0

10

20

Z(~,)

Fig. 7. Electron density profiles of diacetylenic lipid bilayers: (a) (1), (b) (2), (c) (3), (d) (4). Profiles were calculated as described in the Methods. Note that the z axis scales are different for each sample.

103 TABLE II Rela'at ~pacings Jbr ! as a fimction

of temperature

Data were collected using the point focus beamline described in Materials and Methods, using {a) a sample polymerized at 4°C and equilibrated to 10°C, (b) the same sample reequilibrated to 50°C and (c) the same sample then reequilibrated back to 10°C. Fihns were measured in orthogonal directions, using distance between nonzero positive reflections for lamellar (z*) data and distance between positive and negative reflections for equatorial data. Although it is impossible to know which of the equatorial reflections observed at 10°C became tile single equatorial reflection observed at 50°C, it is expected that the lattice constant would increase at T > Tm. T (a) (b) (c)

(°C)

10 511 10

(A)

Color

d:

blue red red

54.3 48,9 53.4

d~ (A)

d~ (~)

9.53 12.3 9.54

4.37, 3.89 4.15 4.24, 3.81

electron dense componentso(PO.~) in the headgroup should be at approx. 5:18 A fronl the bilayer center. Although one cannot deduce a three-dimensional structure from one-dimensional data, it is possible to derive a feasible model structure similar to that proposed previously for other diacetylenic lipid systems [7,8] with the above features, which does not involve unfavorable conformations or close contacts. Chappell and Yager [23,24] have recently proposed a compelling model for association of DCs.,~PC and related long-chain compounds to form tubules. In that case, two lateral interaction energies, U, and U±, arising primarily from acyl chain interactions, are involved in the packing of lipids in the polymerization direction and the normal direction, respectively. The diacetylenic lipids known to form tubules, have proximal and distal acyl chain segments of > 4 CH2 units, most often >_8. In the case of (1), which has a minimal proximal acyl chain segment and does not appear to form tubules, the interaction which must exist to stabilize the diacetylene monomer lattice does not originate in the proximal (m) acyl chains, in the case of (1), there must be a specific headgroup/glycerol backbone/diacetylene interaction to produce a stable lattice compatible with polymer formation which may be further stabilized by the distal (n) acyl chains. It may be that the chain-based chiral structures, described by Singh et al. [25] in the context of tubule handedness, require significant interaction energy (UII and U l , [23,24]) from both proximal and distal acyl chain segments to propagate the orthogonal energy components now thought to be required for tubule formation. Polymerized lipid (1) bilayers have some characteristics similar to the mixed lipid systems studied previously [12,13,26]. When a mixture of DCs.,~PC and DNPC or DMPC is polymerized at low temperature, the membranes turn dark blue. If the polymerization is carried out at room temperature, the membranes turn

orange, indicating thai the conjugation length of the polydiacetylene is shorter. If the blue preparation is heated to room temperature, there is an irreversible color shift to red or orange. Similar behavior is observed for (1); when the temperature of multibilayers exceeds 35°C, and irreversible thermochromic transition to red is observed. When solutions are exposed to temperatures of approx. 20°C, a visible change to purple is observed and samples become bright red upon heating to 30°C. All of these thermochromie events are irreversible. Given the steric requirements for diacetylene polymerization [22], the fact that any polymer forms indicate that the monomers were aligned well prior to polymerization in both lipid systems. In the case of DCs.,j PC/DNPC mixtures, it is likely [13] that the spacer lipids form alternating rows with the diacetylenic lipids. In the case ot'(1), the monomers must also align in rows, probably as described previously lor diacetylenic fatty acids [27], but the headgroup and glycerol backbone conformation need not be similar to that of the longer-chain diacetylenic lipids, in both cases, the fact that the thermochromic transition is so broad suggests that lateral diffusion of the lipids in gradually increasing with temperature until the initial structure is finally disrupted at T,,,, distorting the polydiacetylene from iinearity and thus decreasing the effective conjugation length. Conclusions

The ability of diacetylcpe-containing compounds to polymerize depends on a number of factors rehtted to the position and orientation of diacetylene moieties in adjacent molecules relative to each other. In order for polymer formation to occur, the n th diacetylene moiety must be precisely positioned with respect to the n + Itl' and n - l " ' moieties. That is, one must have a wellordered solid. This does not necessarily have to be a crystal in the strict sense, but most polymerizablc diacetylene-containing solids are crystalline. At the same time, however, a shift in the inter-monomer distance occurs upon polymerization and the lattice must be able to accommodate this shift. This may involve some internal flexibility that does not compromise the integrity of the original structure or some concerted overall change in the crystal parameters and the size of the solid itself. The behavior of (I)is significant in that it provides a one component diacetylenic PC with good polymerizability and thus represents a good model system fi)r developing robust lipid bihtyers. These could be useful in biosensor applications or in understanding the more complex behavior of the long-chain diacetylenic l'C's and the mixtures with spacer lipids [13]. A recent report by Hui et al. [28] showed that monolayer films of (1) polymerized only at molecular areas of _