Synthesis and Characterization of a New Gemini Surfactant Derivedfrom 3r,12r-Dihydroxy-5�-cholan-24-amine (Steroid Residue) and

Ethylenediamintetraacetic Acid (Spacer)

Mercedes Alvarez Alcalde,† Aida Jover,† Francisco Meijide,† Luciano Galantini,‡

Nicolae Viorel Pavel,‡ Alvaro Antelo,† and Jose Vazquez Tato*,†

Departamento de Quımica Fısica, Facultad de Ciencias, UniVersidad de Santiago de Compostela, AVenidaAlfonso X El Sabio s/n, 27002 Lugo, Spain, and Dipartimento di Chimica, Research Center

SOFT-INFM-CNR, UniVersita di Roma “La Sapienza”, Piazza le A. Moro 5, 00185 Roma, Italy

ReceiVed NoVember 13, 2007. ReVised Manuscript ReceiVed March 4, 2008

A new gemini steroid surfactant derived from 3R,12R-dihydroxy-5�-cholan-24-amine (steroid residue) andethylenediamintetraacetic acid (spacer) was synthesized and characterized in aqueous solution by surface tension,fluorescence intensity of pyrene, and light scattering (static and dynamic) measurements. These techniques evidencethe existence of a threshold concentration (cac), below which a three layers film is formed at the air-water interface.Above the cac, two types of aggregates;micelles and vesicle-like aggregates;coexist in a metastable state. Filtrationof a solution with a starting concentration of 2.6 mM (buffer 150 mM, pH 10) allows isolation of the micelles, whichhave an average aggregation number of 12, their density being 0.28 g cm-3. Under conditions where only the vesicle-like aggregates are detected by dynamic light scattering, a value of 5.5 × 104 was obtained for their aggregation numberat 30 µM, their density being 6.8 × 10-4 g cm-3. At high concentrations, the intensity ratio of the vibronic peaksof pyrene, I1/I3, () 0.68) is very close to published values for deoxycholate micelles, indicating that the probe is locatedin a region with a very low polarity and far from water. A hypothesis to explain the observed aggregation behavior(small aggregates are favored with increasing gemini concentration) is outlined.

Introduction

During the past few years, an increasing number of papershave been published on the surface and micellar properties ofgemini surfactants.1–3 This is mainly because of their betterefficiency in decreasing both the surface tension of water andthe critical micelle concentration (cmc) in comparison to theircorresponding monomeric analogues. Most of them contain twohydrophobic long alkyl chains and two hydrophilic groups thatare linked through a flexible or rigid spacer.4,5

Although bile salts are very well-known surfactants6–8 andgood solubilizers of hydrophobic compounds (including drugs9

and cholesterol,10), little attention has been paid to their potentialuse as amphiphile residues to design new gemini surfactants.Only a few examples of gemini surfactants formed by two bileacid residues have been published,11–13 the first one beingsynthesized by McKenna et al.11 In these examples, the carboxylicacid group is lost because it has been used to link the two steroidresidues to the spacer. Here we have designed, synthesized, andcharacterized a dicarboxylic gemini steroid surfactant derived

from 3R,12R-dihydroxy-5�-cholan-24-amine (i.e., a 24-aminoderivative of deoxycholic acid) (see Figure 1). The gemini,g-2DC24-EDTA, was prepared by reacting the dianhydridederivative of ethylenediaminetetraacetic acid (EDTA), whichconstitutes the spacer of the gemini, with the steroid amine. Thustwo amide bonds are formed, and two carboxylic groups arereleased (Figure 1).

Experimental SectionSynthesis. Dimethyl ester of EDTA (0.61 g, 1.9 mmol) was

dissolved in a mixture of 5 mL of dried dimethylformamide (DMF)and 15 mL of dried tetrahydrofuran (THF). Diethyl cyanophosphate(DEPC, 0.65 mL, 4.28 mmol) was added to this solution. After 30min, the solution was cooled to 0 °C, and a solution of 3R,12R-dihydroxy-5�-cholan-24-amine14 (1.51 g, 4.00 mmol) and triethy-lamine (0.6 mL, 4.30 mmol) in 20 mL of dried THF was addeddropwise with stirring. After 90 min, the ice bath was removed, andthe reaction was maintained for 6 h at room temperature. The solventwas evaporated in vaccum. Then 200 mL of chloroform were addedand washed twice with water (50 mL) to remove all DMF. Theorganic phase was dried (Na2SO4) and partially evaporated underreduced pressure. Finally the product was purified by column

* Corresponding author.† Universidad de Santiago de Compostela.‡ Universita di Roma “La Sapienza”.(1) Rosen, M. J.; Tracy, D. J. J. Surfactants Deterg. 1998, 1, 547.(2) Zana, R. AdV. Colloid Interface Sci. 2002, 97, 205.(3) Hait, S. K.; Moulik, S. P. Curr. Sci. 2002, 82, 1101.(4) Song, L. D.; Rosen, M. J. Langmuir 1996, 12, 1149.(5) Mathias, J. H.; Rosen, M. J.; Davenport, L. Langmuir 2001, 17, 6148.(6) Carey, M. C.; Small, D. M. Arch. Int. Med. 1972, 130, 506.(7) Coello, A.; Meijide, F.; Rodrıguez Nunez, E.; Vazquez Tato, J. J. Pharm.

Sci. 1996, 85, 9.(8) Mukhopadhyay, S.; Maitra, U. Curr. Sci. 2004, 87, 1666.(9) Wiedmann, T. S.; Kamel, L. J. Pharm. Sci. 2002, 91, 1743.(10) Almgren, M. Biochim. Biophys. Acta 2000, 1508, 146.(11) McKenna, J.; McKenna, J. M.; Thornthwaite, D. W. J. Chem. Soc., Chem.

Commun. 1977, 809.(12) Li, Y.; Dias, J. R. Chem. ReV. 1997, 97, 283.(13) Ronsin, G.; Kirby, A. J.; Rittenhouse, S.; Woodnutt, G.; Camilleri, P.

J. Chem. Soc., Perkin Trans. 2002, 2, 13026.(14) Fini, A.; Fazio, G.; Roda, A.; Bellini, A. M.; Mencini, E.; Guarneri, M.

J. Pharm. Sci. 1992, 81, 726.

Figure 1. Structure of the g-2DC24-EDTA gemini, derived from 3R,12R-dihydroxy-5�-cholan-24-amine and EDTA as the spacer.

6060 Langmuir 2008, 24, 6060-6066

10.1021/la7035218 CCC: $40.75 2008 American Chemical SocietyPublished on Web 05/23/2008

chromatography (silica gel 70-230 mesh; eluent 7:3 ethyl acetate/methanol). Identity of the compound was confirmed by NMR andfast atom bombardment mass spectrometry (FAB-MS). Overall yield45%. To remove the methyl groups of the ester in the spacer, thecompound was refluxed with KOH 1 M in methanol for one hourat 80 °C. The solvent was evaporated, and the solid was redissolvedin water (200 mL) and acidified with HCl (pH ≈ 1). When thesolution was cooled, the compound precipitated in its diacid form.The precipitate was filtered and dried in a vaccum oven. The disodiumsalt was obtained by adding the stoichiometric amount of NaOH.Both the diacid and the disodium salts were repeatedly crystallizedto guarantee the purity (checked by thin-layer chromatography (TLC),high-performance liquid chromatography (HPLC), FAB-MS, andsurface tension measurements; see Supporting Information) of thegemini compound. The melting point was 129.56 °C.

Instrumental Techniques. Light scattering measurements werecarried out in a Brookhaven instrument constituted by a BI2030ATdigital correlator with 136 channels and a BI200SM goniometer.The light source was a Uniphase solid-state laser system model4601 operating at λ ) 532 nm. The samples were placed in the cellfor at least 30 min prior to the measurement to allow for thermalequilibration. Their temperature was kept constant within 0.5 °C bya circulating water bath. Samples were filtered through filters withdifferent pore size (see Results and Discussion).

The basic theories of static and dynamic light scattering techniquesare very well-known and can be found elsewhere.15,16 For eachsample, the excess Rayleigh scattering ratio ∆Rq was estimated asa function of the scattering vector q ) 4πnsin(θ/2)/λ, where n is thesolvent refractive index, θ is the scattering angle, and λ is thewavelength. Whenever a q dependence was observed, an apparentgyration radius (Rg) and a qf 0 extrapolated ∆Rq value (∆Ro) wereestimated by fitting the experimental data with the equation

1∆Rq

) 1∆Ro

(1-q2Rg

3 ) (1)

at qRg , 1. Hence an apparent molecular weight was estimated as

Mapp )∆R0

KC(2)

where C is the solute concentration (g mL-1) and K is a constantgiven by

K) 2π2

λ4NA(n dn

dC)2(3)

where NA is Avogadro’s constant, and dn/dC is the differentialrefractive index increment of the solution. The differential refractiveindex was measured in an Atago model DD7 instrument. When noangular dependence was shown by the measured ∆Rq, the value atθ ) 90° was used in eq 2.

In the DLS experiments, the intensity-intensity autocorrelationfunction was measured at a particular q value, and related to thenormalized electric field autocorrelation function g1(q,τ) by the Siegertrelation. The autocorrelation function was analyzed either byCONTIN or the cumulant expansion.16 In the latter case, the so-called apparent diffusion coefficient Dapp was obtained from the firstcumulant by the relation

Dapp )- 1

q2

dlng1(q, τ)

dτ |τ)0

(4)

From this value, an apparent hydrodynamic radius Rapp was calculatedby the Stokes-Einstein equation, Rapp ) kBT/6πηDapp, where kB isthe Boltzmann constant, T is the absolute temperature, and η is thesolvent viscosity. Whenever a q dependence of Dapp was observed,

the q f 0 limiting value was obtained from the linear Dapp versusq2 plot. This value was used to calculate Rapp.

Surface tension measurements were carried out in a Kruss K10STtensiometer by the Wilhelmy method. Fluorescence measurementswere carried out in a Hitachi model F-3010 spectrofluorimeter at anexcitation wavelength of 336 nm, and excitation and emissionbandwidth of 5 nm. Samples were thermostatted at 25 °C.

Results and Discussion

In Figure 2, surface tension data, γ, are plotted against log C.Below the cac, γ varies linearly with log C, as for many classicaland gemini surfactants, while, above this threshold concentration(4.6 µM, in water), γ smoothly decreases with log C. As usual,the cac can be accepted as a critical aggregation concentration(cac) above which aggregates are formed. This value is 3 ordersof magnitude lower than the cmc () 4.4 ( 2.9 mM, calculatedfrom compiled values by Coello et al.)7 for sodium deoxycholate,indicating that the new compound is a much better surfactantthan the natural bile salt. Such a difference is typical of a geminisurfactant in comparison to its monomeric analogue.1 Surfacetension measurements were also carried out at different pH values(9.33-11.2 range) and different buffer concentrations (15-150mM, carbonate/bicarbonate). All plots are similar to the oneobserved in water (Figure 2) and the average value for the cacobtained from these experiments is 2.7 ( 0.5 µM.

Figure 2 also shows the dependence of the intensity ratio,I1/I3, of the vibronic peaks of pyrene with log C. At very lowsurfactant concentrations, I1/I3 is almost constant and thendecreases gradually with increasing concentration of g-2DC24-EDTA, reaching a plateau at 0.68. This value is close to the onemeasured for pyrene included in sodium deoxycholate aggregates() 0.70 ( 0.01)17 and not far from its value in the very apolarcyclohexane () 0.61).18 The value obtained for I1/I3 at verydiluted concentrations is close to the observed value in water() 1.96).18 At these limiting gemini concentrations, the con-centration of pyrene () 2 µM) is comparable to the surfactantconcentration, meaning that not all pyrene is incorporated intothe aggregates. So, the I1/I3 ratio corresponds to the sum on theI1/I3 ratios for pyrene in water and in the aggregates, taking intoaccount the respective fraction of pyrene in both media. This canpartially explain the gradual decrease of I1/I3 over the broadrange of concentration observed in Figure 2 (much wider than

(15) Wyatt, P. J. Anal. Chim. Acta 1993, 272, 1–40.(16) Schmitz, K. S. An Introduction to Dynamic Light Scattering by

Macromolecules; Academic Press: Boston, 1990.

(17) Jover, A.; Meijide, F.; Rodrıguez Nunez, E.; Vazquez Tato, J.; Mosquera,M.; Rodrıguez Prieto, F. Langmuir 1996, 12, 1789.

(18) Hashimoto, S.; Thomas, J. K. J. Colloid Interface Sci. 1984, 102, 152.

Figure 2. Surface tension and fluorescence intensity ratio I1/I3 of pyrene(as a probe) vs log [g-2DC24-EDTA]/M in water. pH (at the highestconcentration of gemini) ) 8.33. T ) 25.0 ( 0.5 °C.

Synthesis and Characterization of g-2DC24-EDTA Langmuir, Vol. 24, No. 12, 2008 6061

the typical sharp drop found for surfactants with cmc values inthe millimolar range),5 since, at the lowest concentrations, thepyrene in water increases the measured value of the I1/I3 ratio.Under these experimental conditions, steady-state fluorescenceis not a suitable technique to determine some parameters of theaggregates such as their aggregation number, but this qualitativeanalysis supports the formation of aggregates. Surface tensionresults suggest that the gemini forms aggregates with increasingconcentration, while fluorescence results indicate that the probeis finally located in a region of the aggregate with a polaritysimilar to that of sodium deoxycholate aggregates and far fromwater. Vethamuthu et al.19 have determined that, in sodiumdeoxycholate aggregates, only a small fraction of pyrene () 4%)is in contact with water, evidencing a closed packing of thesurfactant molecules around the probe.

Below the cac, the values for the slope (∂γ/∂ log C) of thestraight line do not show any clear dependency on pH (9.3-11.2)or ionic strength, and, consequently, only the average value ()-15 ( 4 mN/m) will be discussed. The low influence of ionicstrength on this critical concentration and on slope (or,equivalently, on the molecular area at the interface, As) is inagreement with observations for several zwitterionic surfac-tants.20–22 Prosser and Franses23 have reviewed the applicationof the Gibbs adsorption isotherm to the surface tension of ionicsurfactants at the air-water interface. For a strong ionic surfactantof ν+ free positive ions and ν- free negative ions of charges z+and z-, respectively, the surfactant surface density, Γj , is givenby

Γ_)- 1

RTm(c, cs)( dγdlnC)cs

(5)

where m(c,cs) is a function of ν+, ν-, the surfactant concentration,C, the concentration of added inert salt, Cs, and the stoichiometrycoefficient of the counterion of the surfactant in the supportingelectrolyte, ν+

s . m(c,cs) is given by

m(c, cs)) ν-+ν+

2

ν++ ν+s

Cs

C

(6)

It can be noticed that m(c,cs) is not a function of the coion valenceof the supporting electrolyte ν-s . In the absence of inorganicelectrolyte,m) (ν-+ν+), and thesurfaceexcessdensity is inverselyproportional to the total number of free ions in solution. However,when the electrolyte concentration is high, the term involving ν+becomes negligible and m(c,cs) ) ν-. For the present gemini, ν-) 1. Therefore, at pH ) 10.1 and concentrations of Cs(Na+) muchhigher than C (45 mM and below 3 µM, respectively), m(c,cs) ≈1.Forhighlysurface-activesurfactants indilute solutions, thesurfaceexcessdensitymaybeapproximatedby theadsorbedsurfacedensity,Γj ≈ Γ. Since Γ ) 1/(AsNA) (NA is Avogadro’s number), a valueof As) 27 ( 6 Å2/molecule is calculated. The low influence on thesurface properties of ionic strength mentioned above (a behaviorshared with zwitterionic surfactants) also supports the use of ν- )1, as for zwitterionic surfactants.24,25

A simple theoretical calculation (performed with the programCHEM3D of ChemOffice) gives a value of 90 Å2 for a singledeoxycholic acid molecule lying flat. This value is in closeagreement with the experimental one for lithocholic acid lyingflat on the air-water interface.26 The calculated length of a fullyextended molecule of g-2DC24-EDTA is 37 Å. This value iscomparable to the experimental one (34.662 Å)27 obtained fromX-ray diffraction of the crystal of a methyl diester of a 3�-cholicamine dimer in which the two steroid residues are linked byamide bonds to a succinyl group. The area occupied for the fullyextended g-2DC24-EDTA molecule with the two steroid residueslying flat on the surface is 220 Å2 (Figure 3).

For an upright orientation of the gemini (ionic carboxylic groupsoriented toward the water and steroid moieties oriented towardthe aerial phase), the area occupied by a molecule depends onthe angle formed by the two branches of the gemini. For amaximum packing of the steroids (minimum angle), the projectedarea on the surface is 100 Å2/molecule. None of these theoreticalvalues is close to the experimental values mentioned above.Previous interpretations for bile salts will help in understandingsuch a difference. Small26 has reviewed the compression isothermsof natural bile acids. Particularly interesting is the behavior oflithocholic acid. At pH 10 and 3 M NaCl, there is only onecollapse point at about 44 Å2/molecule which corresponds to themolecule in an upright position. (It can be noticed that ourcalculated value for the upright orientation of the gemini is notfar from twice the experimental value for lithocholic acid). Similarvalues (between 40 and 46 Å2/molecule) were obtained for otherbile acids or salts.26,28 At pH ) 2, the isotherm is completelydifferent. At about 119 Å2/molecule, there is a rise in pressureto a first collapse point at 81 Å2/molecule. This range (119-81Å2/molecule) of areas corresponds to a monolayer of moleculeslying flat on the surface. At about 27 Å2/molecule a secondincrease in pressure originates a decrease in area of 5 Å2/molecule.Since this last value for area is one-third of the minimum area(81 Å2/molecule) of lithocholic acid lying flat in a monolayer,Ekwall29 suggested that the bile acid forms a single bulk phasemade up of a trilayer of bile acid. Rosen et al.30 have also proposedthe formation of multilayer structures to explain the aberrantbehavior of some surfactants (having a large number of carbonatoms in the alkyl chain) belonging to a family of bis(quaternaryammonium halide) gemini surfactants. A similar aberrant behavioris observed by Tsubone et al.31 who have also accepted theformation of multilayer films. Therefore, the comparison betweentheoretical and experimental results suggests that g-2DC24-EDTAforms a multilayer film at the air-water interface. The lengthof the steroid side chain plus the EDTA bridge (∼11.7 Å), whichis almost twice the length of the steroid nucleus, would allowthe formation of the multilayer without preventing the interactionof the ionic groups of upper layers with water. Comparisonbetween experimental surface area and theoretical calculationssuggests that three layers are the most probable structure of thefilm at the air-water interface in the presence of added salt. Athigh pH values, where only the carboxylate groups are charged,the presence of salt would allow the unfolding of carboxylate

(19) Vethamuthu, M. S.; Almgren, M.; Mukhtar, E.; Bahadur, P. Langmuir1992, 8, 2396.

(20) Chorro, M.; Kamenka, N.; Faucompre, B.; Partyka, S.; Lindheimer, M.;Zana, R. Colloids Surf., A 1996, 110, 249.

(21) Zajac, J.; Chorro, C.; Lindheimer, M.; Partyka, S. Langmuir 1997, 13,1486.

(22) Aydogan, N.; Abbott, N. L. Langmuir 2002, 18, 7826.(23) Prosser, A. J.; Franses, E. I. Colloids Surf., A 2001, 178, 1.(24) Sesta, B.; La Mesa, C. Colloid Polym. Sci. 1989, 267, 748.(25) Chevalier, Y.; Storet, Y.; Pourchet, S.; Le Perchec, P. Langmuir 1991,

7, 848.

(26) Small, D. M. In The Bile Acids, Chemistry, Physiology, and Metabolism;Nair, P. P., Kritchevski, D., Eds.; Plenum Press: New York, 1971; Chapter 8, p249.

(27) Soto Tellini, V. H. Doctoral Thesis, University of Santiago de Compostela,2006.

(28) Swanson-Vethamuthu, M.; Almgren, M.; Hansson, P.; Zhao, J. Langmuir1996, 12, 2186.

(29) Ekwall, P.; Ekholm, R. Proc. Int. Congr. Surf. Act., 2nd, London 1957,1957, 23.

(30) Rosen, M. J.; Mathias, J. H.; Davenport, L. Langmuir 1999, 15, 7340.(31) Tsubone, K.; Ogawa, T.; Mimura, K. J. Surfactants Deterg. 2003, 6, 39.

6062 Langmuir, Vol. 24, No. 12, 2008 Alcalde et al.

chains into water, their mutual interaction being screened byadded salt ions.

Static (SLS) and dynamic (DLS) light scattering measurementswere performed to characterize the nature of the aggregates. Themeasurements were carried out in water and at high pH valuesin carbonate/bicarbonate buffer, at concentrations of the geminiabove the cac. Figure 4 shows the results obtained from a CONTINanalysis of the autocorrelation function g1(q,τ) (measured at 90°)of a 3 mM gemini sample at pH 10 and two buffer concentrations

(15 and 150 mM). In both cases, bimodal distributions ofaggregates are observed with two maxima at hydrodynamic radiivalues of 2.0 and 77 nm (buffer 15 mM) and 2.5 and 82 nm(buffer 150 mM). These results evidence that (i) two types ofaggregates are present in solution and (ii) the ionic strength hasa low influence on either of the two maxima.

At NaCl 150 mM, Mazer et al.32 have obtained the hydro-dynamic radius for taurocholate, taurodeoxycholate, taurour-sodeoxycholate, and taurochenodeoxycholate, ranging from 1.0to1.5 nm for taurocholate and from 1.5 to 2.2 nm for the otherthree bile salts. These results were analyzed in terms of theformation of micelles with low aggregation numbers (rangingfrom 3 to 22, depending on bile salt and temperature). As wediscuss below, from SLS measurements we have obtained a valueof 12 for the aggregation number of the aggregate of smallersize. The agreement between size and aggregation number ofthis aggregate and those results for natural bile salts suggests amicellar structure for it. Reviews on micelle structure for bilesalts have been published elsewhere.7,33–35 Table 1 shows thatthe size of this aggregate is invariant with the total surfactantconcentration and that it slightly grows with the addition of inertsalts.

Table 1 and Figure 5 show that the hydrodynamic radius ofthe aggregate of greater size (obtained from a CONTIN analysisof the autocorrelation funtion) does not significantly change withthe gemini surfactant in the range of concentration 0.10-3.0mM. The average values are 77 ( 5 nm and 78 ( 3 nm at 15mM and 150 mM of buffer concentration, respectively, evidencingthat the ionic strength does not affect the size of the aggregate.As we show below, this is a vesicle-like aggregate. Table 1evidences that the aggregate of greater size is always present inthe whole range of concentrations above 10 µM and that the

(32) Mazer, N. A.; Carey, M. C.; Kwasnick, R. F.; Benedek, G. B. Biochemistry1979, 18, 3064.

(33) Small, D. M. AdV. Chem. Ser. 1968, 84, 31.(34) Kawamura, H.; Murata, Y.; Yamaguchi, T.; Igimi, H.; Tanaka, M.;

Sugihara, G.; Kratohvil, J. P. J. Phys. Chem. 1989, 93, 3321.(35) Campanelli, A. R.; Candeloro de Sanctis, S.; Giglio, E.; Pavel, N. V.;

Quagliata, C. J. Inclusion Phenom. Macrocyclic Chem. 1989, 7, 391.

Figure 3. Representation of the surface configuration of the g-2DC24-EDTA molecule lying in flat (above) and upright (below) positions. Thetheoretical value of 37 Å in the extended structure is very close to the one () 34.662 Å)27 obtained from the resolution of a crystal of a diester ofa 3�-cholic amine dimer (in which the two steroid residues are linked through a succinyl group).

Figure 4. Histogram size distributions of particles calculated from aCONTIN analysis of the autocorrelation function for a sample of 3.00mM of the gemini g-2DC24-EDTA in bicarbonate/carbonate buffer: (a)15 mM, (b) 150 mM at pH 10. Temperature 25.0 ( 0.5 °C. Scatteringangle θ ) 90°.

Synthesis and Characterization of g-2DC24-EDTA Langmuir, Vol. 24, No. 12, 2008 6063

distribution is monomodal with a low polydispersity (σ < 0.2)in the range 30-60 µM (buffer 150 mM). In this interval ofconcentrations, the aggregate of greater size has an averagehydrodynamic radius of R90 ) 61 ( 2 nm (determined at ascattering angle of 90°). Although micelles are not detected, itdoes not mean that they are not present in the solution since theycan be completely masked by the contribution of the vesicle-likeaggregates because of their much higher molar mass (seebelow), 36 i.e., micelles and vesicle-like aggregates coexist onthe whole range of concentration above the cac.

Under these conditions, a fit of the correlation functions of thescattering intensity with a cumulant expansion was performed,and a realistic apparent translational diffusion coefficient, Dapp,at q f 0 was estimated. From these data, an average value ofRapp ) 114 ( 5 nm is obtained for the apparent hydrodynamicradius of the particle.

Figure 6 shows the apparent molecular weight versus log C(buffer 150 mM and pH 10). Above 10 µM, a sudden increaseof Mapp is observed, suggesting the formation of aggregates abovethe cac. From the initial value determined at 20 µM, Mapp smoothlydecreases with concentration, reaching a plateau at relativelyhigh gemini concentration (∼1 mM). Identical results are obtainedindependently of the method used to prepare the samples, asevidenced in Figure 7. Samples were prepared either by directweighing of the gemini and its solubilization in the right amountof solvent or by dilution from a more concentrated sample. Thesecond method allows us to discard possible bias errors due toloss of mass because of the filtration since it is not required.

SLS measurements were carried out at experimental conditionswhere only the vesicle-like aggregates have been detected byDLS experiments. Values of 5.8, 4.6, and 3.6 have been obtainedfor Mapp/107g ·mol-1 at surfactant concentrations of 30, 45, and60 µM, respectively (buffer 150 mM, pH 10). For the calculationof the molar mass, a value of 0.168 mL ·g-1 was used for theincrement of the solution refraction index with gemini concen-tration, dn/dc. As we have already noticed, micelles are probablypresent in this interval of concentration, thus the observedtendency of Mapp would be in agreement with an increasingpercentage of micelles with increasing surfactant concentration.It also suggests that the average molar mass of these vesicle-like

Table 1. Hydrodynamic Radii of the Two Populations Observedfor the Gemini g-2DC24-EDTA Determined from DLS at a

Scattering Angle of θ ) 90°, at Two Buffer Concentrations andpH 10a

buffer 15 mM (pH 10) buffer 150 mM (pH 10)

C/mM Rh90(1)/nm Rh90(2)/nm Rh90(1)/nm Rh90(2)/nm

9.00 2.5 81.55.00 3.0 82.03.00 2.0 77.5 2.5 82.02.50 2.0 79.5 2.0 79.51.50 2.0 75.0 2.5 80.51.00 2.0 72.0 2.5 74.00.750 2.0 79.0 2.5 75.00.650 2.0 72.50.500 2.5 79.30.320 2.0 71.70.250 2.0 76.6 2.5 74.30.150 2.0 76.5 2.5 76.00.100 2.0 89.0 2.5 78.50.060 58.20.050 2.0 75.0 62.30.040 60.90.030 61.10.020 62.40.010 63.0

a Temperature 25.0 ( 0.5°C. All samples were filtered through filterswith a diameter of 0.45 µm. Typical standard error of the hydrodynamicradius for the larger aggregate is lower than 7%, while the one for the smalleraggregate is lower than 1 nm.

Figure 5. Histogram size distributions of particles calculated from aCONTIN analysis of the autocorrelation function of g-2DC24-EDTAsamples of different concentrations in 150 mM bicarbonate/carbonatebuffer at pH 10. Temperature 25.0 ( 0.5 °C. Scattering angle θ ) 90°.Samples were filtered through filters with a diameter of 0.45 µm.

Figure 6. Apparent molecular weight (scattering angles of 40° and 90°)vs log [g-2DC24-EDTA]/M. Bicarbonate/carbonate buffer 150 mM atpH 10. Temperature 25.0( 0.5 °C. Samples were filtered through filterswith a diameter of 0.45 µm.

Figure 7. Rayleigh scattering ratio (θ ) 90°) vs log [g-2DC24-EDT]/M.Bicarbonate/carbonate buffer 150 mM at pH 10. Temperature 25.0 (0.5 °C. (b) Sample prepared in the scattering cell by dilution of a startingsolution of higher concentration. (g) Samples prepared individually byweighing the right amount of sample and solvent. Samples were filteredthrough filters with a diameter of 0.45 µm.

6064 Langmuir, Vol. 24, No. 12, 2008 Alcalde et al.

aggregates is probably higher than the maximum value for Mapp

measured at 30 µM since Mapp is a weigth average molar massof all species in the solution. In this interval of concentrations,an average value of Rg ) 132 ( 15 nm was obtained for theradius of gyration of aggregates. The influence of the temperaturewas studied at the interval 25-60 °C at a gemini concentrationof 30 µM, and no significant changes were observed either inthe intensity of solution or in the size of aggregates. No changeswere observed after sonication of the solution for 1 h at 25 °Cand different ionic strengths (NaCl 25-350 mM).

Transmission electron microscopy (TEM) measurements wereperformed to obtain information about the morphology of theaggregates. Figure 8 shows a typical TEM image obtained fora sample with an initial gemini concentration of 50 µM. Circularstructures (vesicle-like) are evident, having an average radius of227 ( 45 nm, in acceptable agreement with the Rg and Rh valuesfrom light scattering measurements, if effects on the samplepreparation for TEM measurements are considered.

On the other hand, the comparison of the hydrodynamic radiuswith the radius of gyration is usually done by the parameterdefined by F ) Rg/Rh,37 which has distinctly different valuesdepending on the architecture of the particle. For instance, fora homogeneous impenetrable sphere, F ) 0.788, while a valueof F close to unity suggests the presence of spherical shells, i.e.,vesicles.38,39 From the average value of Rg and the value of Rapp,it results in F)1.1 ( 0.2, supporting the formation of vesicle-like aggregates.

Although a gradual decrease of the vesicle-like aggregatecontribution to the CONTIN distribution is observed by increasingthe gemini concentration, it does not disappear completely, evenat the highest surfactant concentrations. However, we observedthat, at these concentrations, it is possible to completely removethese aggregates by filtering the samples. This is probably dueto fact that the large aggregates are in a metastable state, as isobserved in several vesicle systems. The aggregation number ofthe small aggregate was estimated as follows. A sample ofg-2DC24-AEDT 2.6 mM (buffer 150 mM, pH 10) was prepared

and filtered through a filter with a porous diameter of 0.03 µm,which should retain the vesicle-like aggregates. The concentrationof the filtered sample was determined by measuring the amountof sample retained by the filter (∼19%) of the initial mass. Underthese conditions, a monomodal distribution of aggregates insolution is observed, their hydrodynamic radius being 2.6 nm(polydispersity 0.14). Similar results are obtained after 24 h.That value is close to those values obtained from unfilteredsolutions of similar initial concentration resolved by CONTIN(see above). Under these conditions, ∆R90 ) 1.35 × 10-5, andby accepting a negligible angle dependence (an approximationsupported by the low value of Rh) we have calculated a valueof 1.3 × 104 g ·mol-1 for the molar mass of the micelle. Thisgives a value of 12 for the aggregation number, which is closeto those values for natural bile salts determined from DLS.32 Theisolation of vesicles by filtering the solution is reminiscent of theobservations by Schurtenberger et al.40 for a bile salt/lecithinsolution. These authors have observed that is possible to almostcompletely remove by dialysis the bile salt molecules from mixedvesicles prepared by dilution, without changing the vesicle sizeby more than 10%. They presume that such systems are in ametastable state.

From previous parameters and Mapp ) 5.8 × 107g ·mol-1 forthe large aggregate, densities of 0.28 g cm-3 and 6.8 × 10-4 gcm-3 are estimated for micelles and vesicle-like aggregates,respectively. This large difference confirms the different natureof both aggregates, although the vesicle-like aggregate value isprobably underestimated. The last value is comparable topublished values (1.3-1.8 × 10-3 g cm-3) for homopolymersforming large spheres with a hollow cavity.41

Surface tension, fluorescence intensity of pyrene, and lightscattering (static and dynamic) measurements evidence theexistence of a threshold cac. Below the cac, a multilayer film isformed at the air-water interface, while above it vesicle-likeaggregates in equilibrium with micelles are present, i.e., transitionsfrom a multilayer to micelles or vesicle-like aggregates areinvolved. Kumaran42 has discussed the formation of vesicles byweakly charged membranes due to spontaneous curvature, as aconsequence of an asymmetry in the charge density on the twolayers. The decrease in the electrostatic energy when themembrane forms a vesicle would compensate for the increasein the bending energy of the membrane. This could be the drivenforce for the spontaneous formation of vesicle-like aggregatesin this system.

Furthermore, above the cac, vesicle-like aggregates andmicelles are simultaneously present in solution, although micellesare favored at high gemini concentrations, suggesting avesicle-micelle transition. Micelle-vesicle transitions have oftenbeen observed for mixtures of cationic and anionic surfactants43,44

by the addition of inert salts or by varying the anionic/cationicsurfactant ratio. In some cases, the transition can be induced bythe addition of apolar hydrocarbons45 or the addition of divalentmetal cations in a gemini surfactant.46 The ratio of vesicles tomicelles can be actively controlled by the addition of polar

(36) Pedersen, J. S.; Egelhaaf, S. U.; Schurtenberger, P. J. Phys. Chem. 1995,99, 1299.

(37) Burchard, W. EnVironmental Particles; Buffle, J., Van Leeuvwn, H. P.;Environmental Analytical and Physical Chemistry Series; Lewis Publishers: BocaRaton, FL, 1992; Vol. 2, Chapter 4.

(38) Zhou, S.; Burger, C.; Chu, B.; Sawamura, M.; Nagahama, N.; Toganoh,M.; Hackler, U. E.; Isobe, H.; Nakamura, E. Science 2001, 291, 1944.

(39) Savariar, E. N.; Aathimanikandan, S. V.; Thayumanavan, S. J. Am. Chem.Soc. 2006, 128, 16224.

(40) Schurtenberger, P.; Mazer, N. A.; Kanzig, W. J. Phys. Chem. 1985, 89,1042.

(41) Duan, H.; Chen, D.; Jiang, M.; Gan, W.; Li, S.; Wang, M.; Gong, J. J. Am.Chem. Soc. 2001, 123, 12097.

(42) Kumaran, V. J. Chem. Phys. 1993, 99, 5490.(43) Yatcilla, M. T.; Herrington, K. L.; Brasher, L. L.; Kaler, E. W.; Chiruvolu,

S.; Zasadzinski, J. A. J. Phys. Chem. B 1996, 100, 5874.(44) Renoncourt, A.; Vlachy, N.; Bauduin, P.; Drechsler, M.; Touraud, D.;

Verbavatz, J.-M.; Dubois, M.; Kunz, W.; Ninham, B. W. Langmuir 2007, 23,2376.

(45) Yin, H.; Lei, S.; Zhu, S.; Huang, J.; Ye, J. Chem. Eur. J. 2006, 12, 2825.(46) Huang, X.; Cao, M.; Wang, J.; Wang, Y. J. Phys. Chem. B 2006, 110,

19479.

Figure 8. TEM image of the aggregates formed by g-2DC24-AEDT inD2O. Starting gemini concentration was 50 µM.

Synthesis and Characterization of g-2DC24-EDTA Langmuir, Vol. 24, No. 12, 2008 6065

additives.45 In our case, none of the above-mentioned variableswere modified since only the total concentration of the geminihave been changed. This phenomenon is reminiscent of thoseoccurring in systems of mixed micelles of species with differentsolubilities in water. In these systems, the dilution beyond a limitconcentration gives rise to a micellar growth because of aprogressive loss of the more soluble surfactant from micellesand a consequent increase of micellar low soluble surfactantpercentage. Schurtenberger et al.40 observed a micelle-to-vesicletransition in aqueous solutions of bile salts and lecithin upondilution. According to these authors, the phase limit betweenthese species is reached when, upon dilution, the remaining bilesalt molecules are too few to solubilize the lecithin in disk-shaped mixed micelles with a finite size. This is due to the factthat the amount of bile salt in the mixed micelles (lecithin/bilesalt) decreases by diluting the system.

For the present case, the mechanism of the growth process isunclear, and only some considerations can be outlined. pKa valuesin the ranges 3.5-4.4 and 6.7-7.3 have been published for thedissociation of the tertiary amino groups of EDTA diamides.47

It is expected that g-2DC24-AEDT has pKa values in thoseintervals.48–50 Thus, in buffer solution (pH∼10), only a verysmall amount of monoanionic form is present in equilibriumwith the most abundant dianionic form. The calculation ofconcentration of all species in the bulk solution is straightforward,as the system can be simplified because of the large difference(3 pKa units) between the basicity of the two nitrogen atoms inthe EDTA bridge. Thus the ratio [dianion]/[anion] is equal toKa/[H+]. For instance for pKa ) 7.3 and pH ) 9.3, the ratio is100:1. However, hydrogen ions have the tendency to be adsorbedonto bile salt micelles as they form (micellization tends to increasethe pH of the solution), 28 and acidic species in micelles and atthe interface are present in a more appreciable extent than in thebulk aqueous solution.51 Thus micelles of natural bile salts aremixed micelles formed by species with a different degree ofionization in which anionic bile salt solubilizes neutral forms.51,52

Furthermore, protons are adsorbed more weakly onto smalleraggregates than onto larger aggregates, 28 meaning that neutralforms (with a much lower solubility) are better accommodatedin larger aggregates than in smaller ones. In analogy with previousphenomena observed for natural bile salts, we propose thatg-2DC24-AEDT aggregates are formed by a mixture of dianionicand monoanionic species; and that the proportion of dianionic

species/monoanionic species is probably higher for micelles thanfor vesicle-like aggregates. The previous hypothesis means thatspecies with a lower proportion of dianionic species are favoredby dilution since large aggregates are mainly formed at lowsurfactant concentrations. Lowering the concentration of sur-factant reduces the total number of aggregates that will besurrounded by the same concentration of protons, provided thatpH is kept constant. This will increase the probability of protonadsorption for a given aggregate. Additional dianionic specieshave to join the aggregate to maintain the ratio monoanionicspecies/dianionic species under an acceptable (unknown) limitin the aggregate. This would accrue the size of the aggregate,facilitating the adsorption of additional protons, leading to afurther increment of the aggregate size.

It is important to remark, however, that other processes suchas those involving the binding equilibrium of the counterion tothe micelles can affect the growth. These equilibriums couldslightly change by dilution and sensitively bias the structure ofthe EDTA derivative dianionic polar head with relevantconsequences on the gemini aggregation. Further work is inprogress to outline a detailed mechanism. However, the highpuritiy of our gemini surfactant allow us to rule out hydrophobicimpurities as being responsible for the light scattering results.

Conclusion

A new gemini steroid surfactant, g-2DC24-AEDT, derived from3R,12R-dihydroxy-5�-cholan-24-amine (steroid residue) andEDTA (spacer) was synthesized and characterized in aqueoussolution. Surface tension measurements evidence the existenceof a threshold concentration, or cac, that is 3 orders of magnitudelower than the cmc for sodium deoxycholate. Below the cac, athree-layer film is formed at the air-water interface, and abovethe cac two types of aggregates, small micelles and vesicle-likeaggregates, coexist in a metastable state since they can beseparated by filtration. Micelles have a very apolar region witha polarity similar to that of cyclohexane.

Acknowledgment. The authors from USC thank the Min-isterio de Ciencia y Tecnologıa (Project MAT2004-04606)and Xunta de Galicia (PGIDIT05PXIC26201PN) for financialsupport. The authors from “La Sapienza” are thankful for theMUR financial support (PRIN 2006039789.00). We alsoacknowledge the Spain-Italy Integrated Action grant.

Supporting Information Available: Synthesis of the dimethylester of EDTA; synthesis of 24-deoxycholate amine; synthesis schemeof g-2DC24-EDTA; 1H and 13C NMR spectra of previous compounds;FAB-MS spectra of the dimethyl ester and diacid form of g-2DC24-EDTA. This material is available free of charge via the Internet athttp://pubs.acs.org

LA7035218

(47) Danil de Namor, A. F.; Pacheco Tanaka, D. A. J. Chem. Soc., FaradayTrans. 1998, 94, 3105.

(48) Wang, Y.-M.; Wang, Y.-J.; Wu, Y.-L. Polyhedron 1998, 18, 109.(49) Fisher, A. E. O.; Naughton, D. P. Transition Met. Chem. 2004, 29, 315.(50) Platas-Iglesias, C.; Corsi, D. M.; Vander Elst, L.; Muller, R. N.; Imbert,

D.; Buenzli, J.-C. G.; Toth, E.; Maschmeyer, T.; Peters, J. A. Dalton Trans. 2003,727.

(51) Carey, M. C. Hepatology 1984, 4, 66S(52) Igimi, H.; Carey, M. C. J. Lipid Res. 1980, 21, 72.

6066 Langmuir, Vol. 24, No. 12, 2008 Alcalde et al.