Thermodynamic and Structural Behavior of α‐Galactosylceramide and C6‐Functionalized α‐GalCer in 2D Layers at the Air–Liquid Interface

Abstract α‐Galactosylceramide (α‐GalCer; KRN7000) is a ligand for the glycoprotein CD1d that presents lipid antigens to natural killer T cells. Therefore, KRN7000 as well as some modified versions thereof have been widely investigated as part of novel immunotherapies. To examine the impact of structural modification, we investigated KRN7000 and C6‐modified KRN7000 at the air–liquid interface using monolayer isotherms, BAM, IRRAS, GIXD, and TRXF. The amino group has no influence on the highly ordered sub‐gel structures found at lateral pressures relevant for biological membranes. Neither lateral compression nor the protonation state of the amino group has a measurable effect on the lattice structure, which is defined by strong and rigid intermolecular hydrogen bonds. However, the first‐order phase transition found for the C6‐functionalized α‐GalCer is connected with an extraordinary surface‐inhibited nucleation. Our study demonstrates that KRN7000 can be functionalized at C6 without significantly changing the structural properties.


Introduction
In the early 1990s, severalg lycosphingolipids from the marine sponge Agelas mauritianus were isolated. [1,2] Structural elucidation revealed that the glycosphingolipids consisted of ceramides a-glycosidically linked to d-galactose (e.g., 1). Subsequent structure-activity work found the a-galactosylceramide (a-GalCer)K RN7000 1 to be ah ighly potent tumor growth inhibitor in mice ( Figure 1). [3] The therapeutic properties and applications of KRN7000 have been the subjecto fm any reviews. [4][5][6] Glycosphingolipids continuet ob eo fi nterest due to their biological significance and potential applications. [7][8][9][10] KRN7000s erves as ligand fort he glycoprotein CD1d. [11] CD1d is expressed on the surface of antigen presenting cells and is involved in presenting lipids antigens to natural killer T cells. This event leads to the production of cytokines, resulting in aT h1 or Th2 immune response. Due to the immunological significance of NKT cells, KRN7000a nd other analogues have been investigated in the contexto fn ovel immunotherapies. [12] Work towards ynthetic vaccines to protect against av ariety of infectious diseases [13,14] in our laboratory [15] as well as others [16] have illustratedt he utility of KRN7000 to self-adjuvant conjugate vaccines. [17] Conjugation of a-GalCer 2 to the capsular polysaccharide of Streptococcus pneumoniae serotype 4 inducedl ong-term protection against serotype 4i nm urine models. Serotype 4p olysaccharide was conjugated to a-GalCer 2 via an alkyl amino linker at the C6 hydroxy group of galactose. KRN7000 can be functionalized at specific sites. Functionalizationa tt he C6 position, either by the introductiono fs mall molecules [18] or spacerst oa llow for late-stage conjugation, is possible without significantly inhibiting effective CD1d presentation. [19] Biological and crystallographic data of murine CD1d [20] and human CD1d, with and without KRN700 binding support this notion. [21] Monolayers of amphiphilic molecules are meaningful models of one half of biomembranes to understand structure formation in 2D as well as interactions of biologically important molecules, dissolved in the subphase, with the membrane surface. Early work suffered from the lack of sophisticated surface-sensitive tools to investigate liquid interfaces with molecular and microscopicr esolution. Meanwhile, numerous techniques, like Brewster angle microscopy,X -ray and neutron scattering, infrared reflection-absorption spectroscopy (IRRAS), and nonlinear optical spectroscopy,a re available. Langmuir monolayers can be easily manipulatedb ys imple mechanical compression (increasingt he density of moleculesa to therwise identical condi-a-Galactosylceramide( a-GalCer; KRN7000) is al igand for the glycoprotein CD1d that presentsl ipid antigens to natural killer Tcells. Therefore, KRN7000a sw ell as some modified versions thereof have been widely investigated as part of novel immunotherapies. To examinet he impact of structural modification, we investigated KRN7000 and C6-modified KRN7000a tt he airliquid interfaceu sing monolayer isotherms, BAM, IRRAS,G IXD, and TRXF.T he amino group hasn oi nfluence on the highly ordered sub-gel structuresf ound at lateral pressures relevant for biological membranes. Neither lateral compression nor the protonation state of the amino group has am easurablee ffect on the lattice structure, which is defined by strong and rigid intermolecular hydrogen bonds. However,t he first-order phase transition found for the C6-functionalized a-GalCeri sc onnected with an extraordinary surface-inhibited nucleation.O ur study demonstrates that KRN7000 can be functionalized at C6 without significantly changing the structuralproperties.   tions). [22][23][24][25] However,L angmuirm onolayers sufferd rawbacks as the phospholipid systemsa re metastable, and transmembrane processes as wella st he incorporation of proteins cannot be investigated. The first step in Langmuir monolayerr esearch is recording of surfacep ressure (p)-molecular area (A)i sotherms. Phase transitions, triggered by changing the packing density of the molecules, can be easily identified.T he first-order phase transition from the disordered LE phase to any orderedc ondensed phase is characterizedb yaplateau region in which both phases coexist. The hydrophobic chains are in an ordered (alltrans)s tate, and the monolayer structure can be studied on an Angstroms cale by grazing incidence X-ray diffraction (GIXD). [26][27][28][29] Little is knowna bout the influencet hat structural modifications impart on the physical properties of the glycolipid KRN7000 1 and C6-functionalized a-GalCer 2.W ei nvestigated 2D monolayers of these glycolipids at the air-liquid interface using different surface-sensitive methods, and examined the impact of the C6 functionalization on the physical properties and phase behavior of KRN7000.
The p-A isotherms of KRN7000 1 and the C6-functionalized a-GalCer 2 are markedly different ( Figure 2). The isotherm of KRN700 1 is typical for am onolayer which does not form a liquid-expanded (LE) phase ( Figure 1) up to temperatures of 37 8C. Theg as-analogouss tate transforms during compression directly into ac ondensed (LC) state (re-sublimation process) at practically zero lateral pressure. Apparently,t he compressibility at 20 8Ci sh igher than at 37 8C, at emperature at whicht he film is almosti ncompressible. However,t he reason for this apparentd iscrepancy must be the different spreading behavior at the two temperatures. Likely,islands of the LC phaseformed during spreading must be much larger at lower temperature. Therefore, the defectd ensity in the layer is much higher and difficult to be removed during compression.A th igher temperature, the film is more homogeneous.
The p-A isotherms of C6-functionalized a-GalCer 2 have a lift-off point around 120 2 indicating the existence of al iquidexpanded (LE) phase at low lateral pressures. The transition into the LC phase is characterizedb yadrastically hindered nucleationobserved by the huge hump in the isotherms.The sur- face inhibited nucleationn eeds an excessive over-compression (supersaturation of the LE phase) for startingt he nucleation process. The nucleation rate is high and the growth rate low leadingt om any small domains (BAM image in Figure S1). Reducing the compression speed does not change the shape of the isotherms( data not shown). Increasingt emperature shifts the transition to higher lateral pressures but does not change the surprising behavior.T he highest pressure needed for starting the nucleation is always approximately1 1mNm À1 higher than the corresponding transition pressure p t .T he observed difference ( Figure S2 in the Supporting Information) increases only by 0.8 mN m À1 if the temperature is increased by 10 K. Therefore, the compression curvesa re far from equilibrium and cannotb eu sed for the thermodynamic analysisb yt he Clausius-Clapeyrone quation. The pH hasp ractically no influence on this behavior even if the terminal amino group should be protonated and therefore positively charged at low pH values and uncharged at high values (pK a % 10.6). [36] The only influence of pH is observed in the decompression curve at pH 10. Zerop ressure is not achieved even at large molecular areas what could indicate that denser areas of an on-homogeneous filmare still in contact with the Wilhelmy plate.
The decompression curves at low pH values are equilibrium ones and can be used for the thermodynamic analysis. The influence of temperature on the decompression curves at pH 1 is shown in Figure 3.
The phase transition pressure (p t )i sd eterminedf rom the kink in the p-A isotherms, which indicates the onset of the first-order LE-LC phase transition. Evaluating the temperature dependence of the phase transition pressure gives accesst o the transition entropy. The slope dp t /dT of the linear fit to the experimental data amounts to 0.696 mN m À1 K À1 .T he entropy change DS of the phase transition was calculated using the two-dimensional Clausius-Clapeyronequation [Eq. (1)]: with A e as the molecular area at the onset of the phase transition at the surface pressure p t and A c as the area of the corresponding condensed phase at the same lateralp ressure. [37] The temperature dependence of the entropyc hange DS is presented in Figure 3, right. The main phase transition atc ompression of amphiphilic monolayers is an exothermic process, therefore, negative DS values are obtained. Extrapolationo fDS to zero yields the criticalt emperature T c of 327.2 K( 54.0 8C), above which the monolayer cannotb ec ompressed into the condensed state. On the other hand, the temperature T 0 is 279.8 K (6.6 8C). Below this temperature, no LE phase is existing and the gas-analogous phase transformsd irectly into the LC phase (re-sublimation process) as observed for KRN7000 1. IRRAS supports the conclusions drawn from the isotherm experiments. The CH 2 stretching vibrations are sensitivet ot he phase state of the alkyl chains. The wavenumbers n asym obtained in IRRAS experimentsa long the isotherms of KRN7000 1 and the C6-functionalized a-GalCer 2 are presented in Figure 4 (IR spectra are presentedi nF igures S3 and S4). The values between 2926 and 2924 cm À1 indicate substantial gauche conformers. The first-order phase transition in the C6-functionalized a-GalCer 2 monolayers occurs above 20 mN m À1 ,i na greement with the compression isotherm. Please notet hat the IRRAS experiments are performed at constant pressure. Each experiment at fixed pressure takes approximately 10 minutes.  This means that even keeping the monolayer in the over-compressed state for al ongert ime does not inducet he phase transition below 20 mN m À1 .T he LE-LC phaset ransition can also be observed in the intensity of the OH band. The increase in OH-band intensity ( Figure S3) indicates an increasei nt he effective thickness/density of the monolayer.I nc ontrast, the intensity of the OH band along the isotherm of KRN7000 is practically constant ( Figure S4) indicating ac onstant thickness and only marginal changes in the packing density (removing defects upon compression).
In the LC phase, low wavenumbers between 2919 and 2918 cm À1 are typical for at ightly packed condensed state. The differences between the wavenumbers of KRN7000 1 and the C6-functionalized a-GalCer 2 in the condensed state are only marginal.
To determine the protonation state of the amino group in monolayers of the C6-functionalized a-GalCer 2,T RXF measurements were performed on subphases with pH values between 7.0 and 9.4 around the expected pK a value ( Figure 5).
The positively charged amino group attracts negatively chargedc ounter-ions from the subphase in order to form the electricald ouble layer (EDL). In the present case, the anion Br À is used in ac oncentration of 1mm.T he experiments show clearly ad ecreasing intensity of the Br fluorescence lines (Ka at 11.92 keVand Kb at 13.29 keV) with increasing pH validating a pK a value of % 9f or the terminal amino group.T he fluorescence intensity of the Br at pH 9.4 is only double of that of the background (bare subphase). KRN7000 1 as an uncharged layer does not show any fluorescencei ntensity above the corresponding background values.
The pK a value of the amino group of the C6-functionalized a-GalCer 2 can be determined with high precision using as ubphase buffer with only one type of anion to avoid any compe-tition of anions for interactions with the chargedh ead groups. [38] Grazing incidence X-ray diffraction (GIXD) was employed for defining the structuralo rder of the monolayers down to an ångstrçml evel.I nterestingly,b oth compounds, feature very different thermodynamic properties ande xhibit the same structuralp roperties. The structurald ifferences observed on different subphases (salts, pH) are too marginal to be discussed. The GIXD data obtained from both compounds in the condensed phase revealed that the lateral order of the chains is defined by three strong diffractionp eaks ( Figure 6) measured in the wide-angle region.
The diffraction pattern practically does not change upon compression. This indicatesa"frozen" obliquel attice structure. The structure is characterized by strongly tilted alkyl chains and av ery small cross-sectional area A 0 indicating extremely tight packing with no chain rotation. Comparing the two compounds shows the similarity of the condensed phase structures:K RN7000 1 has ac hain tilt of 34.28 and ac ross-sectional chain area A 0 of 18.4 2 whereas the C6-functionalized a-GalCer 2 has as lightly lower chain tilt of 33.98 and as lightly larger A 0 of 18.6 2 leadingt ot he same in-plane molecular area of % 44.5 2 in agreement with the values determined from the area-pressure isotherms.
Interestingly,a dditional Bragg peaks can be seen in the GIXD patterns of both compounds (Figure 7). These additional peaks indicate ah igher degree of order involving whole molecules. The indexing of those additional Bragg peaks is based on al arger supercelli nduced mostp robably by strong intermolecular hydrogen bonds established between theh ead groups,similar to the previously reportedm onolayer structures of other glycolipids and aG PI fragment. [39,40] The supercell of this sub-gel phase mustb ec ommensurate with the hydrocarbon chain lattice.U sing as unit cell parameters a' = 2 a chains = 9.0 , b' = 2 b chains = 10.4 ,a nd g = 108.58,t he following Bragg peaks can be determined and compared with the experimentally observed ones in Ta ble 1.
The calculated area of 88.8 2 corresponds to that of two molecules. The headg roup order based on ar igid hydrogenbond network, which is practically uncompressible, dictates the packingo rder of the chains and is responsible for the high tilt angle.  Selectivei nteractions between ions and charge-neutral glycolipids can have far-reaching consequences in many biological events. Earlier,p referential interactions of ions with unchargedg lycolipid Langmuir monolayers at the air-water interface have been directly quantified. [40] All monolayers featuring head group order (formation of sub-gel phases)e xhibitedp referential interactions with certain ions suggesting that the defined structuralm otifs of these highly ordered surfaces are responsible for the observed ion selectivity.B ecauseg lycolipid KRN7000i sc hargen eutral and forms highly ordered monolayer structures, first TRXF experiments were carriedo ut to quantify preferential interactions of calcium with the KRN7000m onolayer.I nt he above described experiments,i th as been shown that Br À ions are not attracted by the orderedh ead groups. Nevertheless, experiments with calcium ions indicate another behavior.T he excesso fc alcium is deduced from the intensity of its element-characteristic X-ray fluorescence( integrated Ka and Kb lines) normalized to the intensity of the evanescent standing wave F at the locationo ft he ion. [38] The intensity of the standing wave decaysa pproximately exponentially with the depth [Eq. (2)]: with L % 7nma st he decay length for the used beam energy of 15 keV and incident angle of 0.078,a nd d HC as the thickness of the hydrophobic part of the monolayer. I ex = (I-I 0 )/I 0 is the relative change in the measured intensity I with respect to the intensity I 0 expected in the absence of am onolayer. I 0 is obtained by measuring the bare ionc ontaining aqueous subphase, and taking into account the reduction of the intensity at the place where the ion is located in presence of the monolayer (Figure 8).
Ion excess can be determinedo na nabsolutescale (absolute number of excess ions per nm 2 )b yu sing charged amphiphiles and ions with no noticeable mutual complexing ability.For cations, highly charged (charge density of % 0.64 Cm À2 )L angmuir monolayers of long-chain alkylsulfates (behenylsulfate, BS) have been used successfully. [41] The headg roups of these monolayers are completely deprotonated under the used conditions. The maximum packing density,o bserved in pressurearea isotherms,isc lose to 25 2 ·molecule À1 .
The thickness of the hydrophobic part d HC of the monolayer can be estimated by Equation (3): with l max as the maximum length of as tretched alkyl chain with n CH 2 groups l max = (n 1.26 + 1.5) . [42] With the tilt angle of the chains t (see above)d etermined by GIXD, 28 is calculated for KRN7000a nd 23 for BS. The excessi ntensity I ex amountst o0 .604 for KRN7000 and 28.419 for BS. With one Ca 2 + ion per 50 2 (compressedB Sl ayer), the experimentally determined excess for the KRN7000 layer corresponds to an area per adsorbed calcium ion of % 24 nm 2 or one calcium ion

Conclusion
Most surprising is the formation of condensed phasesi nm onolayerso fm olecules with such al arge chain mismatch. Additional experiments,w hich are beyond the scope of this paper, are required to develop ar easonable model. Amino groups linked via an alkyl spacert ot he galactose of KRN7000 have no influence on the highly ordereds ub-gel structuresf ound at lateral pressures relevant for biological membranes (30-35 mN m À1 ). [43] Neitherl ateral compression to higher surface pressures nor the pH of the subphase has am easurable effect on the lattice structure. The extremelys mall cross-sectional chain area indicates very tight packing (herringbone mode) with no rotational freedom.T he supercelli ndicates ordering of entire molecules which is most likely induced by strong and rigid intermolecularh ydrogen bondsb etween the galactosylceramide head groups dictating the packingo ft he chains. The amino group is decoupled from the strongly interacting head groups and can be even positively charged, as provedb y TRXF,w ith no effect on the monolayers tructure. However,t he thermodynamic monolayer properties are clearly influenced. The C6-functionalized a-GalCer 2 forms al iquid-like phase( LE) at low lateral pressures in contrastt op arent compound KRN7000 1.T he first-order phase transition to the highly ordered condensed phase is connected with an extraordinary surfacei nhibited nucleation( supersaturation of the LE phase). Most likely,t he linker with the amino groups bends towardt he interface in the disordered LE phase. Therefore, over-compression of the layer is neededf or the formation of this highly ordered sub-gel structure, and the isothermsa re similar at all subphase pH values. This study demonstrates that KRN7000 can be functionalized at specific sites without significantly changing the structural properties at biologically relevant lateral pressures in accordance with biological data.

Experimental Section
Materials and monolayer experiments: For the monolayer experiments, the glycolipid (e.g., 1 or 2)w as dissolved in chloroform to a concentration of 1mm.T he solutions were spread by am icro-syringe onto different subphases. The p-A isotherms were recorded on ac omputer-interfaced Langmuir trough (R&K, Potsdam, Germany) equipped with aW ilhelmy balance system after a1 0min wait to ensure complete solvent evaporation. The surface tension of water s w ,m easured with af ilter paper Wilhelmy plate, decreases with increasing concentration of amphiphilic molecules at the surface (s F ). The surface pressure p = s w Às F is plotted versus the molecular area A. Ac onstant temperature was adjusted and stabilized using arecirculation cooler.T he compression speed of the film was 5 2 molecule À1 min À1 .A ll isotherms were measured at least twice for reproducibility.
Brewster angle microscopy (BAM): The morphology of the monolayer was imaged with aB rewster angle microscope, model BAM2-plus from NanoFilm Te chnologie (Gçttingen, Germany), equipped with am iniature film balance from NIMA Te chnology (Coventry, UK), both mounted on an antivibration table. The microscope was equipped with af requency-doubled Nd:YAG laser (532 nm, % 50 mW), ap olarizer,a nanalyzer,a nd aC CD camera. When p-polarized light is directed onto the pure air-water interface at the Brewster angle ( % 53.18), zero reflectivity is observed. The presence of am onolayer causes light to be reflected because of the changed refractive index of the surface layer,w hich is then registered by the CCD camera after passing the analyzer.The lateral resolution of the BAM images was % 2 mm. [44,45] Infrared reflection absorption spectroscopy measurements (IRRAS): The experiments were performed using an IFS 66 FTIR spectrometer equipped with al iquid nitrogen cooled mercury cadmium telluride detector attached to an external air-water reflection unit (XA-511, Bruker). The principle of the method and its application to Langmuir films at the air-water interfaces has been previously described. [46][47][48][49] As mall reference trough and the larger sample trough are alternatively moved into the IR beam path by a shuttle system. The reflectance absorbance was calculated using Àlog(R/R 0 ), with R being the reflectance of the sample and R 0 ,t he reflectance of the reference. The resolution and scanner speed in all experiments were 8cm À1 and 20 kHz. The incident IR beam is polarized with aK RS-5 wire grid polarizer.S pectra are co-added over 200 scans for s-polarized light and over 400 scans for p-polarized light before being apodized using Blackman-Harris three-term function and fast Fourier transformed after one level of zero filling.
Grazing incidence X-ray diffraction (GIXD) experiments: The GIXD experiments were performed at the high resolution diffraction beamline P08 (PETRA III, DESY,H amburg, Germany). For the measurements, aL angmuir trough was located in ah ermetically sealed container with Kapton windows transparent for X-rays. The trough was constantly flushed with helium (He) to avoid scattering of molecules from the air and to increase the signal to background ratio considerably.T he synchrotron beam was monochromated by as et of two monochromators (silicon double crystal (Si111)a nd germanium double crystal (Ge311)). The photon energy was adjusted to 15 keV corresponding to aw avelength of 0.827 .A pproximately 2mm 50 mm of the monolayer surface were illuminated. The incident angle was adjusted to 0.078 to be below the critical angle for total external reflection for water.T od ecrease mechanically excited surface waves, ag lass block was present in the subphase beneath the illuminated area of the monolayer.AM ythen (microstrip system for time resolved experiments) detector (DECT-RIS, Baden, Switzerland) was rotated around the sample to detect the intensity of the diffracted beam as af unction of the vertical scattering angle a f and horizontal scattering angle 2q.ASoller collimator (JJ X-ray,D enmark) was located between the sample and the detector to restrict the in-plane divergence of the diffracted beam.
Model peaks taken as Lorentzian in the in-plane (Bragg peaks, Q xy ) and Gaussian in the out-of-plane direction (Bragg rods, Q z )w ere fitted to the integrated data. Subsequently,t he Bragg peak positions, respectively the centers of the Bragg rods are obtained and structure relevant information is gained. The presence of one Bragg peak located at the horizon describes ah exagonal lattice of non-tilted lipid alkyl chains in ar otator phase, two peaks are typical for an orthorhombic and three peaks for an oblique lattice. [40,[50][51][52][53] Total reflection X-ray fluorescence (TRXF) experiments: In recent years, TRXF has been established as an element-specific comple-  [38,41,54] The TRXF measurements were carried out at beamline P08 (PETRA III, DESY,H amburg, Germany) using the above described set-up. The fluorescence signal was detected by an Amptek X-123SDD detector (Amptek, Bedford, USA) placed almost parallel to the liquid surface and perpendicular to the photon beam axis. This detector position was chosen in order to keep the Compton scattering at the given polarization of the photons as low as possible. The footprint center of the incident beam was adjusted to the middle of the trough at the middle of the view angle of the fluorescence detector. [38,40]