PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
J Am Chem Soc. Author manuscript; available in PMC 2010 May 11.
Published in final edited form as:
PMCID: PMC2867465
NIHMSID: NIHMS91351

Label-Free Detection of Drug-Membrane Association using Ultraviolet-Visible Sum-Frequency Generation

Abstract

Drug-membrane interactions play a crucial role in the pharmacology and activity of drugs. The measurement of drug association to lipid membranes has conventionally been measured by fluorescence and other spectroscopic methods. However, a main disadvantage of fluorescence labeling of drugs is that the introduction of fluorophores may change the molecules physical properties, such as charge, hydrophobic or hydrophilic character and structure. To circumvent these problems, Ultraviolet-Visible Sum Frequency Generation (UV-Vis SFG) has been developed as an ultra-sensitive and label-free technique to detect small-molecule drug association to lipid membranes. Four different classes of drugs, a non-steroidal anti-inflammatory drug (ibuprofen), antibiotic (azithromycine), anti-fungal (tolnaftate), and local anesthetic (tetracaine), were examined. Drug association was measured on planar supported lipid bilayers (PSLBs) of 1,2-dioleoyl-sn-glycero-3-phophocholine (DOPC). Equilibrium association constants of the drugs were obtained and correlate well to the partition coefficients of the drugs in a liposome membrane-water system. UV-Vis SFG is a powerful and novel technique to directly measure the association of drugs to a single biological membrane without chemical modification.

Drug-membrane interactions play a crucial role in the pharmacology and activity of drugs.1,2 The equilibrium of a drug molecule between plasma and the cellular membrane has historically been modeled by bulk phase partitioning, usually between water and 1-octanol.3 On a fundamental level, the thermodynamics of a lipid membrane, which is comprised of a bilayer of lipids ~50Å thick, cannot adequately be modeled by the bulk thermodynamic properties of a homogenous liquid phase. In addition, the membrane, composed of various phospholipids and proteins, is best described as a finite interfacial region with a very high surface-to-volume ratio and a limited number of potential binding sites.4 For this reason, solution phase liposome based assays have been developed to more accurately model and study drug-membrane association.5

Spectroscopic methods for detecting drug-membrane interaction such as UV-Vis absorbance spectroscopy,6 fluorescence,7 IR,8 Raman9 and NMR10 have all been utilized to measure drug interactions with solution phase vesicles. Such analyses require milliliter solution volumes, large quantities of analyte (i.e. drug) and lipid, and usually involve a separation step to remove unbound drug from the solution prior to analysis. These requirements are necessary due to the limited sensitivity of the spectroscopic methods employed.

The use of planar supported lipid bilayers (PSLBs) for measuring drug-membrane association has several advantages over solution phase vesicles, including the use of smaller solution volumes, elimination of the separation steps employed in liposome-based assays, and in principle, a more rapid screening of interactions. However, in order to fully exploit the potential benefits of a PSLB assay, what is needed is a detection method with high sensitivity, a low limit of detection and the capability of measuring drug-membrane interactions at the interfacial level without interference from solution phase species.

Several possible methods could be employed to measure drug-membrane interactions on surfaces. Surface enhancement vibrational techniques such as attenuated total reflection (ATR) IR and surface enhanced Raman scattering (SERS) are sensitive spectroscopic methods but they suffer from spectral congestion and the inability to selectively isolate the resonances of the drug molecule from the surrounding lipid matrix. Fluorescence is an extremely sensitive method, capable of single molecule detection limits,11 however, if the drug molecule of interest lacks any intrinsic fluorescence, an extrinsic fluorescent tag must be covalently linked to the molecule for detection. Surface plasmon resonance (SPR) is another potential technique. However, SPR is not suitable for investigating drug-membrane interactions because the change in the local refractive index of a lipid bilayer (n ~1.5) is not altered significantly upon drug binding due to the similar refractive index of the drug and membrane. The inability of SPR to interrogate drug-membrane association is evidenced by the lack of any definitive literature example of a low molecular weight drug interaction with a PSLB being measured by SPR.

In this paper, the use of a novel deep-UV spectroscopic method, ultraviolet-visible sum-frequency generation (UV-Vis SFG), has been implemented to directly detect drug association to lipid membranes without the need for chemical modification. UV-Vis SFG is a surface sensitive technique which involves spatially and temporally overlapping a UV and visible laser source at the sample of interest, yielding photons at the sum of the two input frequencies. The UV-Vis SFG intensity is expressed as:

ISFG=(f˜sumfUVfVisχ(2))2
(1)

where fUV, fVis and fsum are the geometric Fresnel coefficients for the ultraviolet (UV), visible (Vis) and sum-frequency light respectively.12 χ(2) is the second-order nonlinear susceptibility which has a nonresonant (χ NR) and resonant contribution (χ R):

(χijk(2))χNR+Na,b,ca|μi|ca|μj|bb|μk|c(hωSFGEcaiΓca)(hωUVEabiΓab)(hωVisEbciΓbc).
(2)

In equation 2, N is the surface density of molecules, h is Planck’s constant, Eca, Eab, and Ebc, are the energies of the optical transitions in the deep UV, UV and Vis respectively, μ is the Cartesian coordinate dipole operator, and Γ represents the linewidth for the transitions. The indices on χ(2) denote the input (j,k) and output (i) fields which can take on any of the three Cartesian coordinates (x,y,z). Examination of equation 2 shows that an increase in the UV-Vis SFG signal will be observed when the incident or SFG frequencies are resonant with electronic transitions of the molecules comprising the interface. When χR is large compared to χNR, which is the case when the incident and/or SFG wavelengths are in resonance with an optical transition, the nonresonant term can be neglected.13

To demonstrate the ability of UV-Vis SFG to detect drug association to a membrane, four drugs were examined: ibuprofen, azithromycin, tolnaftate and tetracaine. The structure of these compounds and their associated UV-Vis spectra are shown in Figure 1. These drugs where chosen as they belong to four important classes of known membrane associated drug compounds; non-steroidal anti-inflammatory drugs (NSAIDs), antibiotics, anti-fungals and anesthetics, respectively.1417 Ibuprofen, tolnaftate and tetracaine all contain a π-conjugated ring system in their structures, resulting in strong electronic transitions in the UV, see Figure 1. Azithromycin was chosen as it possesses only a single carbonyl bond, giving rise to a small absorbance at 213 nm (ε213 = 988 M−1 cm−1), which makes it a significantly weaker absorber compared to the other drugs examined. Unlike UV-Vis or fluorescence, UV-Vis SFG is not dependent on direct electronic excitation or emission as it is a coherent scattering process. As a result, problems associated with photodegradation or photobleaching are considerably reduced, especially when the electronic transitions being probed are near the sum-frequency wavelength, as direct electronic excitation from the pump lasers is eliminated.

Figure 1
UV-Vis spectra and molecular structures of ibuprofen, azithromycin, tetracaine and tolnaftate.

The UV-Vis SFG experiments were performed by combining the UV (355 nm, mixed-polarization, γ=45°) and Vis (532 nm, γ=45°) outputs of a Nd:YAG laser both spatially and temporally on a 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) PSLB which was deposited on a fused silica trapezoidal prism, Figure 2. Increasing concentrations of a drug in phosphate buffered saline (PBS) pH 7.5 were equilibrated above the membrane surface and the s-polarized UV-Vis SFG intensity was recorded. The resulting adsorption isotherms for the studied drugs are shown in Figure 3. In order to allow for a direct comparison of the system response for the various drug molecules examined, the SFG intensities were corrected for changes in collection efficiency, so that a relative comparison of intensities from experiment to experiment and between drugs is possible (see the Supporting Information).

Figure 2
UV-Vis SFG cell showing the geometric arrangement of the UV, Vis and sum-frequency beams. Insert: representation of the polarization state of the incident and output fields denoted as the angle γ with respect to the propagation direction (k).
Figure 3
UV-Vis SFG adsorption isotherms for: (top) ibuprofen (blue), azithromycin (red), tetracaine (green); (bottom) tolnaftate. The solid lines are the fits to the data using the Frumkin isotherm (tetracaine and azithromycin) and Langmuir isotherm (ibuprofen ...

For these resonant UV-Vis SFG experiments, the surface density of the drug is proportional to the square root of the SFG intensity. To extract the equilibrium drug-membrane association constants from the UV-Vis SFG data, the adsorption isotherms were fit to the Langmuir model or Frumkin model (equation 3) which accounts for repulsive interactions of the ionized drug molecules:18

exp(g·ISFGISFGMAX)·[drug]Ka=ISFGISFGMAXISFG
(3)

where [drug] is the bulk concentration of drug, Ka is the equilibrium association constant, ISFG is the SFG intensity, ISFGMAX is the maximum SFG intensity at surface saturation and g is a constant used to account for repulsive/attractive interactions of the adsorbate. When g = 0, the Frumkin isotherm becomes the Langmuir isotherm. The results from a nonlinear least squares regression of the data in Figure 3, to equation 3, are listed in Table 1.

Table 1
Measured equilibrium association constant Ka, ISFGMAX, and g values for ibuprofen, azithromycin, tetracaine and tolnaftate.

The affinity constants for the drugs were found to increase in the order ibuprofen < tetracaine < azithromycin < tolnaftate. This order agrees well with the partition coefficients of the drugs into liposomes (listed in Table 2).19 The negative value of g for tetracaine and azithromycin illustrates the repulsive interaction between the charged drug molecules. At pH 7.5, tetracaine and azithromycin have one and two positive charges, respectively. Ibuprofen possesses one negative charge; however the Langmuir model provides a better fit to the adsorption data, presumably due to the electrostatic shielding from the supporting electrolyte. It has been reported that ionized ibuprofen (weak acid) locates closer to the membrane-water interface making it more accessible to the solvent while the ionized tetracaine and azithromycin (weak bases) associate strongly into the hydrophobic tails of the lipid membrane.20 In order for the supporting electrolyte to shield the electrostatic interaction between the ionized tetracaine and azithomycin, ions from the bulk have to move deeper into the membrane, which is not an energetically favorable.20 The screening of the charge on ibuprofen has been verified by performing the adsorption experiment using a lower PBS salt concentration (15 mM, see the Supporting Information for results). Under these conditions, the electrostatic screening between ibuprofen molecules is reduced and a negative g value of −2.36±0.67, was obtained. The absence of any electrostatic repulsion is clearly seen for tolnaftate, which is the only neutral drug in this study and is best fit with a Langmuir isotherm (g = 0). For every drug, the validity of using a Frumkin isotherm or Langmuir isotherm was verified by an F-test.

Table 2
Partition constants, normalized surface densities and limits of detection (LOD) for the drugs studied.

In addition to retrieving information on the thermodynamics of drug-membrane association, UV-Vis SFG can also be used to obtain information on the surface excess of the drug (Γmax) in the membrane. As with other spectroscopic methods, UV-Vis SFG needs to be calibrated. This could be achieved by using lipid membranes which have fixed and known concentrations of drug in the membrane and measuring the UV-Vis SFG response. As drug-membrane association is a dynamic process, it is not possible to create such standards easily. However, data does exist, in the form of partition coefficients for the drugs in liposome-membrane systems or octanol-water partitioning data,21 which can be used to calibrate the UV-Vis SFG data. In the linear region of the binding isotherms, at low surface density, the partitioning of the drug in the membrane is identical to that in a solution phase liposome as there is effectively no competition for binding sites.22 Using the linear regions of the isotherms shown in Figure 3, the UV-Vis SFG intensity was calibrated for each drug. The liposome-water partition coefficients were used to determine the membrane concentration from the bulk aqueous concentration of the drug, [membrane] = Pi ˙[aqueous]. The surface concentration in molecules (molc)/cm2 was then determined by assuming an effective thickness of the DOPC bilayer of 50 Å.

The calculated surface excess (in molecules/cm2) is plotted as a function of bulk concentration in Figure 4. Although the affinity constants for the drug are in the order ibuprofen < tetracaine < azithromycin < tolnaftate, the same cannot be said of the saturation concentration of the drugs in the membranes. Azithromycin and tetracaine have very similar surface saturation values (1.58–5.31±0.16×1010 molc/cm2) while ibuprofen and tolnafate are considerably larger (9.22±0.14×1011 and 1.00±0.03×1013 molc/cm2). The large difference in surface converges is a reflection of the repulsive interactions between drug molecules. Azithromycin and tetracaine have the lowest surface converge and largest g values, while ibuprphen and tolnaftate which shown no repulsive term (g=0) have the highest saturation levels.

Figure 4
Surface excesses Γ for: (Top) ibuprofen (blue), azithromycin (red), tetracaine (green); (Bottom) tolnaftate. Solid lines are the fits to the data using the Frumkin isotherm (tetracaine and azithromycin) and Langmuir isotherm (ibuprofen and tolnaftate). ...

The absence of a direct link between Ka and Γmax is not surprising, as the factors determining these two quantities are not necessarily correlated. Ka is not dependent on the absolute number of binding sites, but rather is a reflection of the fraction of surface occupancy at a specific bulk concentration; which is why knowledge of the affinity constant alone is not sufficient to determine the absolute surface density. A quantitative determination of drug saturation in the membrane is possible when both bulk equilibrium measurements and UV-Vis SFG are used in combination.

A quantitative assessment of the limit of detection (LOD) of UV-Vis SFG can also be made, based on the results presented in Table 2. Using the spectroscopic sensitivity determined from the calibration of the UV-Vis SFG intensity and the standard deviation in the measured signal, the LOD for the three drugs examined were calculated and are listed in Table 2. The lowest calculated LOD is 1.3±0.1 pg/cm2 for tetracaine with the highest observed for tolnaftate at 1.31±0.05 ng/cm2. These LOD are far superior to the only other surface specific and label-free method, SPR, and are comparable to fluorescence measurements but without a label being needed and with reduced photodegradation. The impressive LOD and the surface specificity of UV-Vis SFG illustrate the potential application of the method for measuring low-molecular weight drug-membrane interactions.

We have demonstrated here that UV-Vis SFG is an ultrasensitive and powerful technique to directly detect drug-membrane association without chemical modification. The equilibrium association constants of ibuprofen, azithromycin, tolnaftate and tetracaine into a lipid membrane have been measured and have been shown to increase with the drugs’ hydrophobicity. A quantitative assessment of membrane concentrations is achievable using bulk distribution coefficients and UV-Vis SFG measurements. These findings suggest that UV-Vis SFG is a valuable alternative in measuring the association of drugs to the membranes.

Supplementary Material

1_si_001

Acknowledgment

This work was supported by the NIH (R01-GM068120). Mr. Rembert was supported by the NSF (0649039).

Footnotes

Supporting Information Available: Detailed experimental description. This material is available free of charge via the Internet at http://pubs.acs.org.

References

1. Seydel JK. Methods and Principles in Medicinal Chemistry. 2002;15:217–289.
2. Triggle DJ. In: Drug-Membrane Interactions. Analysis, Drug Distribution, Modeling. Methods and Principles in Medicinal Chemistry, Vol.15. Seydel JK, Wiese M, editors. Vol.46 2003.
3. Seydel JK, Wiese M. Drug-Membrane Interactions. Weinheim: Wiley-VCH Verlag GmbH; 2002.
4. Tamm LK, McConnell HM. Biophys. J. 1985;47:105–113. [PubMed]
5. Choi YW, Rogers JA. Pharm. Res. 1990;7:508–512. [PubMed]
6. Custodio JBA, Almeida LM, Madeira VMC. Biochem. Biophys. Res. Commun. 1991;176:1079–1085. [PubMed]
7. Lau WF, Das NP. Experientia. 1995;51:731–737. [PubMed]
8. Cao A, Hantz-Brachet E, Azize B, Taillandier E, Perret G. Chem. Phys. Lipids. 1991;58:225–232.
9. Fox Christopher B, Horton Robert A, Harris Joel M. Anal. Chem. 2006;78:4918–4924. [PubMed]
10. Seydel JK, Wiese M. Drug-Membrane Interactions: Analysis, Drug Distribution, Modeling. 2002 [In: Methods Princ. Med. Chem., 2003; 15]
11. Cornish PV, Ha T. ACS Chem. Biol. 2007;2:53–61. [PubMed]
12. Shen YR. The principles of non linear optics. John Wiley&Sons; 1984.
13. Heinz TF, Tom HWK, Shen YR. Phys. Rev. A: Atomic, Molecular, and Optical Physics. 1983;28:1883–1885.
14. Barbato F, La Rotonda MI, Quaglia F. J. Pharm. Sci. 1997;86:225–229. [PubMed]
15. Tyteca D, Schanck A, Dufrene YF, Deleu M, Courtoy PJ, Tulkens PM, Mingeot-Leclercq MP. J. Membr. Biol. 2003;192:203–215. [PubMed]
16. Auger M, Jarrell HC, Smith IC, Siminovitch DJ, Mantsch HH, Wong PT. Biochemistry. 1988;27:6086–6093. [PubMed]
17. Zhang J, Hadlock T, Gent A, Strichartz GR. Biophys. J. 2007;92:3988–4001. [PubMed]
18. Bockris JOM, Reddy AKN. Modern Electrochemistry. Vol. 2. New York: Plenum Press; 1970.
19. Avdeef A. Absorption and Drug Development: Solubility, Permeability, and Charge State. Hoboken: NJ: Wiley-Interscience; 2003.
20. Avdeef A, Box KJ, Comer JEA, Hibbert C, Tam KY. Pharm. Res. 1998;15:209–215. [PubMed]
21. Seydel JK, W M. Drug-Membrane Interactions. Weinheim: Wiley-VCH Verlag GmbH; 2002.
22. Schreier S, Malheiros SVP, de Paula E. Biochim. Biophys. Acta, Biomembranes. 2000;1508:210–234. [PubMed]