Confocal Raman Microscopy for pH-Gradient Preconcentration and

Article pubs.acs.org/ac

Confocal Raman Microscopy for pH-Gradient Preconcentration and Quantitative Analyte Detection in Optically Trapped Phospholipid Vesicles Chris D. Hardcastle and Joel M. Harris* Department of Chemistry, University of Utah, 315 South 1400 East, Salt Lake City, Utah 84112-0850, United States S Supporting Information *

ABSTRACT: The ability of a vesicle membrane to preserve a pH gradient, while allowing for diffusion of neutral molecules across the phospholipid bilayer, can provide the isolation and preconcentration of ionizable compounds within the vesicle interior. In this work, confocal Raman microscopy is used to observe (in situ) the pHgradient preconcentration of compounds into individual optically trapped vesicles that provide sub-femtoliter collectors for smallvolume samples. The concentration of analyte accumulated in the vesicle interior is determined relative to a perchlorate-ion internal standard, preloaded into the vesicle along with a high-concentration buffer. As a guide to the experiments, a model for the transfer of analyte into the vesicle based on acid−base equilibria is developed to predict the concentration enrichment as a function of source-phase pH and analyte concentration. To test the concept, the accumulation of benzyldimethylamine (BDMA) was measured within individual 1 μm phospholipid vesicles having a stable initial pH that is 7 units lower than the source phase. For low analyte concentrations in the source phase (100 nM), a concentration enrichment into the vesicle interior of (5.2 ± 0.4) × 105 was observed, in agreement with the model predictions. Detection of BDMA from a 25 nM source-phase sample was demonstrated, a noteworthy result for an unenhanced Raman scattering measurement. The developed model accurately predicts the falloff of enrichment (and measurement sensitivity) at higher analyte concentrations, where the transfer of greater amounts of BDMA into the vesicle titrates the internal buffer and decreases the pH gradient. The predictable calibration response over 4 orders of magnitude in source-phase concentration makes it suitable for quantitative analysis of ionizable compounds from small-volume samples. The kinetics of analyte accumulation are relatively fast (∼15 min) and are consistent with the rate of transfer of a polar aromatic molecule across a gel-phase phospholipid membrane.

P

therapeutic compounds into liposomes for applications in drug delivery.13−16 This approach exploits the semipermeable character of phospholipid bilayers, where a small, neutral molecule can permeate a lipid membrane, while its ionized (protonated or deprotonated) counterpart cannot. Phospholipid vesicles are prepared with an interior buffer at a pH that differs significantly from the pH of a solution into which they are transferred, where the external solution contains the compound of interest. Neutral ionizable molecules can then diffuse across the phospholipid bilayer and be protonated or deprotonated within the vesicle and thereby accumulate as their charged forms in the vesicle interior.17,18 In the case of neutral basic compound (e.g., an amine), a low pH in the vesicle interior and a high pH in the surrounding source phase lead to accumulation of protonated amine within the vesicle. This process continues until the activity of the neutral amine on the vesicle interior inside becomes equal to its activity in the source phase; see Figure 1.

reconcentration is a well-known method for enhancing the sensitivity of detecting trace-level analytes in a sample. Common materials used for sample preconcentration include functionalized chromatographic silica,1−4 low-viscosity and porous polymers,5−7 or functionalized magnetic particles.8−10 Typically the collector is isolated from the source phase and then washed to release the concentrated analyte for subsequent ex situ analysis by chromatographic or spectroscopic detection.1−3 With the application of microfluidics and labon-a-chip sample manipulation, there has arisen a need to reduce the scale of preconcentration methods. In order to reduce the large sample volumes required for extraction, one may consider analysis of the contents of a single microscopic particle used as a concentrator phase. Utilizing the small volumes of single particles would be especially useful when quantifying precious or biological compounds when there is a limited volume supply of sample.7,11 The use of a pH gradient in phospholipid vesicles or liposomes is a potentially powerful means of concentrating ionizable analytes into small volumes for detection and quantification.12 The loading of lipid vesicles or liposomes by a pH gradient was first developed as a means of transferring © 2015 American Chemical Society

Received: May 21, 2015 Accepted: July 1, 2015 Published: July 1, 2015 7979

DOI: 10.1021/acs.analchem.5b01905 Anal. Chem. 2015, 87, 7979−7986

Article

Analytical Chemistry

With such large potential factors for analyte enrichment, pHgradient loading of vesicles can be a powerful method for concentrating ionizable molecules into small volumes for subsequent analyte detection. There are several experimental constraints that must be addressed to observe concentration enrichment factors comparable to those predicted by eq 3. First, if the internal solution volume of the vesicles in dispersion exceeds the source-phase volume by more than a factor 1/E, then analyte in the source phase will be depleted by transfer into the vesicles. As an example, for a typical dispersion of 1.0 μm vesicles, hydrolyzed and extruded from a 1.0 mg/mL solution of lipid and diluted 1:100 into a source-phase solution, the vesicle volume would represent only a 4 × 10−4 fraction of the total solution. Nevertheless, this vesicle volume fraction would limit the concentration enrichment to 2,500 or 1/400times smaller than what is achievable based on eq 3 in the preceding example. To achieve the highest concentration enrichment with this approach, one should ideally quantify the contents of a single vesicle, whose volume is negligible compared to the sourcephase solution surrounding it. Optical-trapping confocal Raman microscopy19−22 is capable of characterizing the structure of individual micrometer-sized phospholipid vesicles23−28 and has also been successfully employed to characterize and quantify compounds within them.24,26,29 Confocal Raman microscopy was recently demonstrated as a method capable of detecting the pH-gradient transfer of an ionizable analyte into individual vesicles.12 This study reported the source-phase pH dependence of the concentration of a protonated amine transferred into a single vesicle and demonstrated enrichment factors as large as 2.5 × 104. A significant limitation of this study, however, was that concentration of amine in the source phase was relatively high (>10 μM), which led to the observed enrichment being limited by the capacity of the internal buffer. As neutral amine passes through the membrane and is protonated, the source of protons for this reaction is the internal buffer within the vesicle. At high concentrations of the amine in the source phase, this “titration” within the vesicle lowers the concentration of the acid form of the buffer, which raises the internal pH, decreases the magnitude of the pH gradient driving the process, and limits the internal concentration of the protonated amine to the capacity of the buffer.16 At this point, the concentration of analyte in the vesicle becomes insensitive to changes in the external analyte concentration and the method loses its quantitative response. In the present study, we employ confocal Raman microscopy to investigate the conditions under which pH-gradient preconcentration into lipid vesicles can be used as a quantitative method for sensitive detection of ionizable compounds. Individual phospholipid vesicles are prepared with a low-pH citric acid internal buffer, optically trapped and transferred into a high-pH source phase. The source phase contains low concentrations of a model analyte, N-benzyl-N,N-dimethylamine (BDMA), which incorporates both a basic, tertiaryamine functionality and an aromatic ring that serves as a Raman scattering reporter. BDMA accumulates within the vesicle in its protonated form, the concentration of which is determined by a Raman scattering measurement relative to sodium perchlorate internal standard. Optical trapping allows the same 1.0 μm diameter vesicle to be manipulated and investigated for several hours, during which time the dependence of the accumulated analyte concentration can be measured over a 20,000-fold change in the concentration of the neutral amine in the source-

Figure 1. pH-gradient loading of a weak base into a phospholipid vesicle. The neutral base, B, can permeate the lipid membrane, while the protonated acid form, HB+, cannot. Protonation of B by the lowpH buffer leads to a high concentration of HB+ in the vesicle interior, until the interior activity of B is equal to that of the source phase.

The process of pH-gradient loading is governed by the acid− base equilibria for protonation of the neutral basic compound inside and outside of a vesicle, which are given by the acid dissociation constant of the protonated base, KA: KA =

[H+]out [B]out [H+]in [B]in = [HB+]in [HB+]out

(1)

+

where [B] and [HB ] are the activities of the neutral (basic) and protonated forms, respectively. Because the neutral base, B, is membrane permeable, then, at equilibrium, the activities (neglecting activity coefficients and concentrations) of the neutral form on the inside and outside the vesicle become equal, [B]in = [B]out (see Figure 1). Substituting this relationship into eq 1 and solving for the total concentration of analyte inside the vesicle give total = [HB+]in + [B]in [B]in

=

[H+]in + KA ([HB+]out + [B]out ) [H+]out + KA

=

[H +]in + KA total [B]out [H+]out + KA

(2)

The concentration enrichment of analyte within a vesicle, E, is given by the ratio of the internal analyte concentration to the source-phase concentration, which in terms of the pH of the two phases is given by total total E = [B]in /[B]out =

10 pKA − pH in 1 + 10 pKA − pHout

(3)

For a given internal pHin within the vesicle, the enrichment increases with increasing external solution pHout until the limit pHout ≫ pKA. Under these conditions, essentially all of the basic analyte in the source phase is in the neutral form, [B]total out ≈ [B]out, the denominator of eq 3 approaches unity, and the enrichment approaches its limit, Emax = 10pKA−pHin. Under ideal conditions, the simple model of eq 3 can predict the enrichment of analyte of known KAp from a source-phase solution of known pHout into a vesicle prepared with a known internal pHin. For example, a base having a pKA of 9 in a source phase of pH > 10 could theoretically be concentrated in its protonated form within a vesicle having an internal pHin = 3 by a factor of nearly Emax = 106. 7980

DOI: 10.1021/acs.analchem.5b01905 Anal. Chem. 2015, 87, 7979−7986

Article

Analytical Chemistry

the fraction of lipid present on the outer leaflet of the bilayer, which can be related to the average lamellarity of the vesicles in the dispersion. The results (see Supporting Information) show that upon quenching the fluorescent labels on the outer leaflet, the fraction of fluorescence remaining is 0.52 ± 0.01, where the expected fraction remaining for a population of 1.0 μm diameter unilamellar vesicles is 0.498. This result indicates that the vesicles are nearly all unilamellar, where the difference would correspond to less than a 9% fraction of concentric twobilayer vesicles. To assess the stability of the vesicles and their ability to preserve a pH gradient over time, an extruded batch of vesicles prepared with an internal pH 2.8 citrate buffer was diluted into an osmotically matched, pH 10.0 phosphate/borate buffer. The dispersion was monitored daily for a period of 14 days. The preservation of the pH gradient was tested by measuring their accumulation of BDMA (see later) from a 500 nM sourcephase solution in pH 10.0 phosphate/borate buffer. Confocal Raman Microscopy. The 647.1 nm line from a Kr+ laser (Coherent 90K Innova) was sent through a 4× beam expander mounted on the back of an inverted fluorescence microscope (Nikon TE100). The expanded beam was sent through the rear aperture of the microscope and through a bandpass filter and dichroic beam splitter (Chroma). The 35 mW excitation beam was then directed through a 100×, 1.4 NA oil-immersion objective and tightly focused to a ∼0.56 μm diameter spot inside a flow cell. The focused excitation beam also serves as an optical trap23,28 to immobilize the vesicle in the center of the excitation and collection volume. The scattered light is collected back through the objective and dichroic beam splitter, a high pass filter, and a holographic notch filter (Kaiser) and focused onto the entrance slit of a spectrograph (Bruker 500IS) and detected by a CCD camera (Andor DU401A). The 50 μm entrance slit of the spectrograph defines the confocal spot in the horizontal dimension, and collecting the spectrum from a three-pixel row or 78-μm region defines the confocal spot in the vertical dimension.35 The Raman shift was calibrated daily using a 1:1 acetonitrile/ toluene mixture, and its peaks were assigned according the ASTM standard E1840-96 using a cubic interpolation between peaks. To minimize the fluorescent background of the coverslip glass, Raman scattering was collected from optically trapped vesicles by focusing the objective at distance of ∼15 μm above the coverslip surface. All spectra were then processed and analyzed using MATLAB (The MathWorks, Inc., Natick, MA, USA), where they were dark-background-subtracted, whitelight-ratioed, and then baseline-corrected using a fifth-order polynomial fit of multiple regions along the baseline. To determine the area of a given peak in the Raman spectrum, the wavenumber region of interest for each band was fit to a Gaussian function by nonlinear least-squares method. The bestfit Gaussian function of each band was then integrated to determine its area, which is proportional to the concentration of component of interest. Fabrication of an All-Glass Microscopy Flow Cell. To avoid potential loss of hydrophobic analytes to polymer gasket materials (a process for which we find evidence especially from low-concentration analyte solutions), an all-glass microscopy flow cell was constructed from an optically flat float-glass top plate and coverslip, assembled by cold welding where the clean flat-glass surfaces are likely held together by strong hydrogen bonding interactions. A 0.5 mm wide by 0.2 mm deep flow channel was machined into a 25 mm diameter float-glass top

phase solution. The quantitative results are compared with the predictions of theory developed for the titration of the analyte by the internal buffer within the vesicle.

■

EXPERIMENTAL SECTION Reagents and Materials. Lyophilized samples of 1,2 dipalmitoyl-sn-glycerol-3-phosphocholine (DPPC), 1-stearoyl2-hydroxy-sn-glycero-3-phosphocholine (SPC), ovine-wool cholesterol (Chol), and N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine (NBDPE) were purchased from Avanti Polar Lipids (Alabaster, AL, USA) and used without further purification. Lyophilized lipid powders and lipid in chloroform solutions were stored at −15 °C prior to use. An Avanti mini-extruder equipped with polycarbonate membranes having a 1.0 μm pore size from Nucleopore (Pleasanton, CA, USA) was used for extrusion of vesicle dispersions. N-Benzyl-N,N-dimethylamine (>99%), sodium perchlorate (98%), and sodium dithionite (>85%) were purchased from Sigma-Aldrich (St. Louis, MO, USA) and used without further purification. Solid calcium chloride, potassium phosphate monobasic, sodium hydroxide, sodium chloride, citric acid, and boric acid were acquired from Fisher Scientific (Waltham, MA, USA). Water used in preparing all aqueous solutions was first purified using a Barnstead GenPure UV xCAD ultrapure water system. Vesicle Preparation. A dried film of DPPC:SPC:Chol was prepared by pipetting chloroform solution of each lipid in a 9:1:1 mol ratio into a glass vial. The addition of 10% lysolipid (SPC) produces unilamellar vesicles (see later) without requiring freeze/thaw cycles,30 and cholesterol was added to increase the structural stability of the vesicle membranes.31 The lipid solution was dried under a stream of nitrogen, and then the vial was placed under vacuum (100 mTorr) for at least 2 h to remove chloroform. Sample vials were then sealed and stored at −15 °C until used. To prepare vesicles, lipid films were hydrated with 1 mL of a buffer containing 0.250 mM citric acid, 0.100 M sodium perchlorate (an internal standard), and 1 mM CaCl2; the buffer was titrated to a pH of 2.8 with 5 M sodium hydroxide. Note that calcium ion is added to both the internal and external buffer solutions to help stabilize the vesicle membrane.32,33 The hydrated lipid was heated to 52 °C, well above the DPPC/SPC gel-phase melting transition of 42 °C,30 extruded 11-times through a polycarbonate track etch filter having a pore diameter of 1.0 μm, and diluted into 1.5 mL of a buffer comprising 0.050 M boric acid and 0.050 M phosphoric acid, titrated to a pH of 10.0 with 5 M sodium hydroxide. In order to check the lamellarity of the prepared vesicles, DPPC:SPC:Chol lipid films were prepared with the addition 0.5 mol % fluorescently labeled NBD-PE. The lipid was then hydrated in the internal citrate buffer, as described earlier, excluding sodium perchlorate. Following extrusion, 50 μL of the dispersion was diluted into a 3.5 mL glass cuvette containing 2.95 mL of 100 mM phosphate buffer (pH 7.2) brought to an osmolarity to match that of the internal buffer using NaCl. The fluorescence emission from the dispersion was excited at 470 nm and detected at 540 nm using a Hitachi F7000 fluorescence spectrometer; the change in fluorescence intensity upon addition of 150 μL of 1 M sodium dithionite in 1 M pH 10 Tris buffer was measured at 1 s intervals to determine the lamellarity through reductive quenching of the NBD labels on the outside leaflet of the vesicle membrane.34 The decrease in fluorescence intensity from the reduction of the NBD label on the outer leaflet by the dithionite represents 7981

DOI: 10.1021/acs.analchem.5b01905 Anal. Chem. 2015, 87, 7979−7986

Article

Analytical Chemistry

trigonal ring stretching mode at 1003 cm−1 relative to perchlorate scattering, multiplying by a factor that corrects for differences in the scattering cross-sections determined from a standard solution of BDMA and perchlorate. Uncertainties in the reported concentration are estimated from three replicate measurements. Measurements of BDMA accumulation kinetics were made by selecting and optically trapping a single vesicle in the channel, transferring it into the well filled with 500 nM BDMA solution, and initiating a sequence of 1 min acquisitions for an hour. Evolution of the BDMA scattering intensity ratioed to that of the perchlorate internal standard was measured to determine the rate of transfer of BDMA into the phospholipid vesicle.

plate (Edmund Optics) using with a TC-11EF diamond burr (Teng Yuan Diamond Products, Shanghai, China) operated at 5,000 rpm. The channel spans a path between a 0.8 mm inlet and 6.4 mm well, both drilled through the plate. The top plate thus machined was sonicated for 45 min to remove any glass particles from the surface and channel. After a thorough rinse in methanol followed by deionized water, the top plate and a No.1 glass coverslip were immersed in a 5:1:1 solution of water:H2O2(30%):NH4OH, which was then heated to 70 °C for 10 min. The plate and coverslip were then rinsed with the deionized water, held in contact with an aluminum screw clamp, and heated to 180 °C for at least 3 h. After a gradual lowering to room temperature (4−6 h), the coverslip and top plate were bonded, producing a sealed channel between the inlet and well. More details of the cell fabrication are provided in the Supporting Information. Measurement Procedures. Solutions of BDMA were prepared at concentrations ranging from 0.025 to 500 μM in a 0.05 M potassium phosphate/boric acid buffer solution. These source-phase solutions were titrated to a pH of 10.0 with 5 M sodium hydroxide and balanced to the osmolarity of vesicle internal buffer (0.38 M) with sodium chloride. The vesicle suspension was then injected into the flow cell, partially filling the well. The well was emptied by suction and refilled with the BDMA source phase, repeating a minimum of 10-times. While the vesicles in the well were removed by this procedure, a large number of vesicles remain in the channel for testing. A final 50 μL volume of source phase was then pipetted into the well, and from within the channel, a single vesicle was selected, optically trapped, and transferred into the well by translating the microscope stage. BDMA accumulation within the vesicle was monitored by Raman spectroscopy until the concentration reaches equilibrium (∼30 min.). Final spectra were then acquired using an integration time of 120 s. The well was once again emptied and refilled with the next source-phase solution, repeating 10-times; another vesicle was selected, trapped, and transferred into the well; and the process was repeated. To quantify the concentration of BDMA accumulated in the internal volume of a vesicle, an internal standard is needed to correct for variations in vesicle volume or changes in excitation and collection efficiency of the instrument. Sodium perchlorate was chosen as an internal standard added to the buffer in which the vesicles are hydrated and extruded; perchlorate anion is membrane impermeable23 and exhibits a strong Raman scattering from a symmetric Cl−O stretching mode at ∼932 cm−1.36 Because perchlorate occupies the same volume as BDMA in the vesicle, the concentration of BDMA within the vesicle can be determined from BDMA scattering from the 3

[Na +] =

■

RESULTS AND DISCUSSION Predicting pH-Gradient Enrichment. In order to employ pH-gradient preconcentration as a quantitative method for detection of analytes in a source-phase solution, the enrichment factors predicted by eq 3 must be corrected for the finite capacity of the buffer within the vesicle, because the interior pH may change significantly from its initial value as the accumulated analyte concentration approaches that of the buffer.12,16 To account for the pH change due to accumulation of analyte, the acid−base equilibria of the buffer must be considered. For the citric acid buffer within the vesicle, three acid dissociation equilibria are relevant: K1 =

[H+]in [H 2C−]in [H3C]in

(4a)

K2 =

[H+]in [HC2 −]in [H 2C−]in

(4b)

K3 =

[H+]in [C3 −]in

(4c)

[HC2 −]in

where pK1 = 3.13, pK2 = 4.76, and pK3 = 6.4037,38 and the total buffer concentration is given by total [C]in = [H3C] + [H 2C−] + [HC2 −] + [C3 −]

(5)

The citric acid used to hydrate the vesicles was titrated with sodium hydroxide to set the initial pH of the interior buffer. From the equilibria in eqs 4a−4c, the total concentration of + citrate, [C]total in , and the initial pH giving [H ]0, the sodium ion concentration in the internal buffer can be determined as follows:

2

2

total total total [H+]0 [C]in K1 + 2[H+]0 [C]in K1K 2 + 3[H+]0 [C]in K1K 2K3 + K wβ − [H+]0 β [H+]0 β

where β = [H+]3 + [H+]2K1 + [H+]K1K2 + K1K2K3. As neutral analyte passes through the vesicle membrane and is protonated to form HB+, protons are transferred from the acidic forms of the buffer; no charge passes through the membrane, so the internal charge within the vesicle is conserved. A conservationof-charge constraint includes all of the ions within the vesicle and accounts for the acid−base reactions and is given by the following:

(6)

[HB+]in + [H+]in + [Na +] = [H 2C−] + 2[HC2 −] + 3[C3 −] + [OH−]in

(7)

where [OH−]in = K w /[H+]in

(8)

In total, there are 10 variable concentrations constrained by conservation of charge, five equilibrium constants (three buffer dissociation constants, the analyte KA, and the autoprotolysis 7982

DOI: 10.1021/acs.analchem.5b01905 Anal. Chem. 2015, 87, 7979−7986

Article

Analytical Chemistry

optically trapped vesicle before and after BDMA accumulation from a 100 nM solution is shown in Figure 3. To quantify the

constant of water) and four parameters set by conditions of the experiment (concentration of the citrate buffer, the initial internal pH which establishes the internal sodium ion concentration, the source-phase pH, and concentration of analyte in the source phase). Because accumulation is measured on a single vesicle in the confocal Raman experiment, the vesicle volume fraction is infinitesimal (∼10−11) and depletion of analyte from the source phase can be neglected. Partitioning of the neutral amine into the lipid bilayer as a source of vesicleassociated analyte was expected to be insignificant because of the small (3%) volume fraction of the membrane; this assumption was tested with vesicles not having a pH gradient, and indeed no BDMA could be detected (see the Supporting Information). The preceding system of equations is solved for [H+]in, the proton activity inside the vesicle at equilibrium, and the result is a fifth-order polynomial having one positive real root (see the Supporting Information). The resulting value of [H+]in at equilibrium thus determined can be used to determine the internal concentration of [B]total in from eq 2 and the enrichment factor, E, from eq 3. This model can be used to explore the influence of the analyte concentration on its enrichment within a vesicle, as shown in Figure 2. The predicted source-phase pH

Figure 3. Raman spectra of a single optically trapped vesicle before (black) and after accumulation of BDMA from a 100 nM (red) and a 25 nM (blue) source phase. The trigonal ring breathing mode of BDMA (1003 cm−1, blue arrow) was used for quantification relative to the 930 cm−1 band from the perchlorate internal standard.

BDMA accumulated within the vesicle, interpretation of scattering intensity must account for variations in vesicle volume and changes in excitation and collection efficiency. This challenge is easily met by incorporating an internal standard, 100 mM sodium perchlorate, in the buffer in which the vesicles are hydrated and extruded. Perchlorate anion is membrane impermeable23 and exhibits a strong Raman scattering from a symmetric stretching mode at ∼932 cm−1,36 well separated from BDMA scattering from the trigonal ring stretching mode at 1003 cm−1 (see Figure 3). For calibration, the relative scattering intensities of perchlorate and BDMA were measured in free solution standards (see the Supporting Information). Because perchlorate occupies the same volume as BDMA within a vesicle, the BDMA concentration can be determined from its scattering intensity relative to that of the perchlorate internal standard. In Figure 3, the intensity of the 1003 cm−1 BDMA peak thus calibrated corresponds to a concentration inside the vesicle of 52 mM, a 520,000-fold enrichment compared to the 100 nM source-phase solution. Similar data were acquired with source-phase solutions ranging in concentration from 25 nM to 500 μM, and results are plotted in Figure 4 along with predictions of the equilibrium model developed earlier. The results show excellent agreement with the model, especially for source-phase concentrations in the range of 100 nM to 50 μM. It should be noted that the model is not f itted to the data; instead, it is a prediction based on the acid−base equilibrium constants of the analyte and citrate internal buffer, pH of the source phase, and the concentration and initial pH of the internal buffer. The agreement between this a priori model and the measured results indicates that pHgradient preconcentration into vesicles is capable of providing a predictable response that can be interpreted quantitatively to infer analyte concentrations in the source phase. The concentration enrichment that this method provides, especially for very low concentration samples, is quite

Figure 2. Predicted concentration enrichment of an amine analyte having a pKa = 8.91 within a vesicle containing 250 mM internal citrate buffer at an initial pH = 2.8, for varying source-phase concentrations of the amine.

dependence for enrichment of the model analyte (BDMA) into a vesicle with a 250 mM citrate buffer at an initial pH = 2.8 is plotted for source-phase concentrations [BDMA]total out that vary from 10 nM to 100 μM. The results show that enrichments from source-phase solutions having a pHout ≫ pKA = 8.91 are predicted to be significantly smaller for higher analyte concentration samples, compared to what can be achieved in the limit of very low concentration samples. At sufficiently low concentrations, [BDMA]total out ≤ 10 nM, the enrichment is within 15% of the limit for this system, Emax = 10pKA−pHin = 1.29 × 106 (see Figure 2). In the section that follows, this model is tested for its quantitative predictions of preconcentration by pHgradient loading. Quantitative Analysis of Preconcentration in Individual Vesicles. In order to investigate pH-gradient accumulation of BDMA into a vesicle, a DPPC:SPC:Chol vesicle, prepared in 250 mM pH 2.8 citrate buffer, is optically trapped and dragged from the glass channel into an adjacent well containing BDMA in pH 10.0 buffer. Example confocal Raman spectra from an 7983

DOI: 10.1021/acs.analchem.5b01905 Anal. Chem. 2015, 87, 7979−7986

Article

Analytical Chemistry

concentration. Enrichments can remain relatively high (>40,000) at source-phase concentrations as high 5 μM, but here the local sensitivity has dropped to