Imaging GABAc Receptors with Ligand-Conjugated Quantum Dots

We report a methodology for labeling the GABAc receptor on the surface membrane of intact cells. This work builds upon our earlier work with serotonin-conjugated quantum dots and our studies with PEGylated quantum dots to reduce nonspecific binding. In the current approach, a PEGylated derivative of muscimol was synthesized and attached via an amide linkage to quantum dots coated in an amphiphilic polymer derivative of a modified polyacrylamide. These conjugates were used to image GABAC receptors heterologously expressed in Xenopus laevis oocytes.


INTRODUCTION
Quantum dots (qdots) are nanometer-sized semiconductor crystals that have unique physical properties that differ from bulk material. The fluorescent properties of qdots have been widely described, and numerous applications based upon these fluorescent properties have been reported. In addition, previous studies have reported the properties of many varieties of qdots [1][2][3][4][5][6][7][8]. Of these, the most widely studied are cadmium selenide/zinc sulfide core-shell nanocrystals. These consist of a semiconductor core of cadmium selenide encapsulated in a multilayer shell of zinc sulfide doped with cadmium [9]. The shell passivates the surface of the core, and the band gap is wider than that of the core, enabling quantum confinement of an electron-hole pair generated in the core after photoexcitation. Ultimately, the electron hole pair recombines, resulting in a fluorescent emission of a lowerenergy photon in the visible region of the spectrum [10]. The energy of the emitted photon is determined by the size of the quantum confinement (or the size of the qdot). Smaller qdots emit blue light and larger ones emit red light. Qdots have several advantages over conventional fluorescent dyes; these include increased photostability, increased brightness, quantum yields in excess of 80-90% [1,9,11], and a narrow emission spectrum (less than 30 nm full width at halfmaximum in commercial products) [12][13][14][15]. Furthermore, their multivalent surfaces enable the attachment of more than one type of ligand or multiple copies of a ligand to a single qdot.
Since their introduction into biology as imaging agents in 1998 [16,17], qdots have increasingly found applications as fluorescent probes in biology. To be useful as fluorescent probes in biological systems, qdots must be soluble in water and commonly used buffers. Additionally, they must have colloidal stability and low nonspecific adsorption to cellular membranes. These properties have been achieved using a number of techniques, including encapsulation in micelles [18], silanization [19], encapsulation in amphiphilic polymers [20,21], and encapsulation in proteins such as streptavidin [22]. To further reduce nonspecific adsorption to cellular membranes, a number of techniques may be used to modify the surface chemistry of qdots. For example, we have recently demonstrated that nonspecific binding can be significantly reduced by attaching polyethylene glycol chains (i.e., by PEGylating) qdots coated in an amphiphilic modified polyacrylic acid polymer (AMP) [23]. The length of the PEG chain and the PEG loading were demonstrated to be important in reducing nonspecific adsorption to cellular membranes. When PEGs with short (less than 12) repeat units were conjugated to qdots, a small reduction in nonspecific adsorption to cellular membranes was observed. This reduction increased in magnitude when larger PEGs were used. Figure 1 shows the effects of PEGylation (PEG2000) on nonspecific adsorption to 6 different cell types. These cells were treated with a 50 nM solution of PEGylated AMPcoated qdots or a 50 nM solution of AMP-coated qdots. A significant reduction in nonspecific adsorption to cellular membranes was obtained by the addition of PEG2000. The nonspecific adsorption is cell-type specific, as can be seen in the relatively low nonspecific adsorption of AMP-coated qdots to the surfaces of 3T3 cells compared to the high levels of nonspecific adsorption to HEK cells.
Our research efforts focus on the central nervous system. We are interested in using qdots that have been conjugated with small molecules [55][56][57][58][59][60], antibodies [61], and peptides [34] to image receptors and transporters in cell cultures, oocytes, and, ultimately, neurons. In our early work, we used qdots to image the serotonin transporter (SERT) using PEGylated serotonin ligands [62] attached to the surfaces of qdots via an acid-base interaction (see Figure 2). These conjugates antagonized the serotonin transporter protein (SERT) with an IC 50 of 115 μM in transfected HEK-293 cells. Using these conjugates we were able to image SERT expressed in HEK-293 cells [55].
Numerous biofunctionalization methods for qdots have been reported in recent years. Qdot preparations that con-tain an amphiphilic coating on the qdots surface are commercially available, and a variety of methodologies, including those involving sulfo-SMCC [63] and adaptor proteins [64], have been used to conjugate ligands to the coated qdot. Our current strategy uses commercially available qdots that have either an amphiphilic coating (AMP) on the surface of the dots, or AMP qdots with an additional coating of streptavidin. PEGylated ligands may be attached to the surface of these dots using two different methodologies. Either they may be covalently attached to the AMP coating using 1-[3-(dimethylamino)propyl]-3ethylcarbodiimide hydrochloride (EDC) coupling chemistry, or a biotinylated derivative of the biologically active ligand may be attached to the surface of streptavidin-coated qdots via a streptavidin-biotin interaction. Using the PEGylated ligand approach, we have synthesized a novel qdot conjugate and tested its binding activity to the GABA C receptor, a ligand-gated ion channel that is found in retina and other central nervous system tissue and that is activated in vivo by γ-aminobutyric acid (GABA). Specifically, we have investigated a PEG derivative of muscimol, a known agonist of both GABA C and GABA A receptors (see Figure 3). Multiple copies of this ligand have been conjugated to the surface of AMP-coated qdots and used to image GABA C receptors expressed in Xenopus laevis oocytes [65].

Synthesis of the muscimol ligand
Muscimol was synthesized using the method described by Frey and Jäger [66]. This was then coupled to the PEG linker via an aminohexanoyl NHS ester to give the PEGylated muscimol ligand. The ligand was characterized by matrixassisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectroscopy and conjugated to AMP-coated qdots via an EDC coupling.

Muscimol synthesis
Muscimol was synthesized using the synthetic methodology shown in Scheme 1. Dimethyl acetylenedicarboxylate (3.1 mL) was added dropwise to a solution of N-hydroxy urea (1.9 g, 25 mmols) and DBU (4.19 g, 28 mmols) in methanol (25 mL) at 0 • C. The resulting solution was stirred at 0 • C for 10 minutes, and then evaporated under reduced pressure. Concentrated hydrochloric acid was added until a pH of 1 was obtained. This solution was extracted with diethyl ether, dried over magnesium sulfate, filtered, and then evaporated. The resulting solid was recrystallized from methylene chloride to yield 1.1 g of methyl 3-hydroxy isoxazole-5carboxylate (I) in a 32% yield. This was converted to (II) by stirring 0.84 g of (I) in ammonium hydroxide (3 mL) and methanol (3 mL) for 1 hour, followed by recrystallization from ethanol to give 0.75 g of 3-hydroxyisoxazole-5-carboxamide (II) in 88% yield as the ammonium salt. Muscimol (III) was obtained from 1 g of (II) by reduction with borane dimethyl sulfide in tetrahydrofuran (THF) to give 0.2 g of (III) in a 22% yield after purification by ion exchange chromatography.

MALDI-TOF mass spectroscopy
Compounds (VIII) and (IX) were characterized by MALDI-TOF mass spectroscopy (Applied Biosystems Voyager mass spectrometer equipped with a 337 nm nitrogen laser) using an acceleration voltage of 25 kV, and the spectra were obtained by averaging of 30-64 scans [65]. The samples were prepared using a saturated matrix stock solution, consisting of 2,5-dihydroxybenzoic acid and 0.01 M sodium iodide dissolved in methanol. The PEG derivatives (VIII) and (IX) were prepared (5 mM) in methanol. The sample was added to the matrix by mixing the sample and stock solutions in a 2 : 5 : 2 ratio of sample to matrix to salt (v/v). A 1-μL aliquot of each sample solution was placed on the sample plate. Mass calibration of the instrument employed a PEG standard, and was prepared using the same protocol as that employed for the other samples. Analysis of the resulting spectra indicated that compounds (VIII) and (IX) were polydisperse. Compound (VIII) exhibited masses ranging from 3241 Da to 4188 Da (indicative of muscimol conjugation to PEGs of different lengths), and a primary peak at 3726 Da. The treatment of (VIII) with TFA to yield compound (IX) resulted in a MALDI-TOF spectrum shift of 100 Da (primary peak at 3626 Da), consistent with loss of the BOC protecting group.

Ligand conjugation
The ligand was conjugated to qdots using an EDC coupling in which 1000 equivalents of ligand were mixed with 750 equivalents of NHS and EDC in borate buffer at pH 8.5.
To this was added a solution of AMP-coated qdots (8.4 μM). This mixture was stirred for 1 hour at ambient temperature. Unbound ligand was removed by Sephadex G-50 chromatography. The coupling of amino-terminated PEG2000 to AMPcoated qdots using EDC has been studied in an earlier publication, and the efficiency of coupling has been reported to be ∼20% when 2000 equivalents of methoxy-terminated aminoPEG2000 are reacted with 1 equivalent of AMP-coated qdots [23]. Since the terminating muscimol of the present ligand is attached to PEG3400, the coupling efficiency is likely to be similar. On this basis, we estimate the number of muscimol ligands to be around 150-200 per qdot [65]. The derivatized qdots were characterized by electrophoresis in 1% agarose gel (see Figure 4). The gel demonstrates that the muscimol-conjugated qdots (Lane 3), as well as qdots conjugated with PEG2000 (Lane 4), have a wide distribution in the number of ligands attached to their surface, as they streak on the gel more than unconjugated qdots (Lane 2). It is important to note that mobility in the gel does not depend merely on mass, but rather on mass-to-charge ratio. Thus, despite the substantial difference in mass of the PEG2000 versus the muscimol-terminated PEG3400 ligand, the bands representing the qdots conjugates that contain (numerous copies of) these ligands exhibit similar mobilities (Lanes 3 and 4). The present experimental conditions (1% agarose gel) do not separate protein standards that span a molecular weight range of 10-250 kDa (data not shown).

Oocyte imaging
The oocytes used in this study were obtained from adult female X. laevis toads. The oocytes were stored in physiological saline (Ringer solution; 100 mM NaCl, 2 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 10 mM glucose, and 5 mM HEPES, pH 7.4). Using previously reported procedures, we expressed GABA C receptors (human ρ1 and perch ρ1B) in X. laevis oocytes [67,68]. cRNA (50 nL) for each of the receptor subunits was injected into the oocyte, and the oocytes were assayed after 18-72-hour incubation in Ringer solution containing 0.1 mg/mL gentamycin at 16-19 • C to allow for expression of the GABA C receptors.
Oocyte imaging was carried out in a glass-bottom dish into which GABA C expressing oocytes and oocytes that did not express GABA C were placed. These oocytes were incubated for 5-10 minutes in a drop (∼25 μL) of solution containing either 34 nM AMP-coated qdots conjugated to the muscimol ligand, or 34 nM AMP-coated qdots that lacked conjugated muscimol ligand. The oocytes were then imaged using a confocal microscope (Leica model DM-IRE2 with 20x objective) with excitation at 476 nm, and with detection of fluorescence emission over a wavelength range (580-620 nm) that included the qdot emission peak (605 nm). At the beginning of experiments conducted on a given day, we established microscope settings relevant to excitation illumination and detection of fluorescence emission (gain and offset) with use of either a human ρ1 GABA C -expressing or perch ρ1B GABA C -expressing oocyte incubated with 34 nM muscimol-conjugated AMP-coated qdots. These settings were maintained without change for the entire day's measurements [65]. Figure 5 shows the binding of muscimol-conjugated AMPcoated qdots and unconjugated AMP-coated qdots to oocytes expressing the human ρ1 GABA C receptor, and to nonexpressing control oocytes (see Figure 5 legend). When GABA C -expressing oocytes were incubated with a 34 nM solution of muscimol-conjugated dots for 10 minutes, a fluorescent halo was observed at the oocyte surface membrane 6 Journal of Biomedicine and Biotechnology (panel A). The intensity of this halo exceeded that of the surrounding extracellular medium. The fluorescent image can be compared with the corresponding bright-field image (panel B), which shows the position and focus of the oocyte. By comparison with panel A, no fluorescence halo was observed upon similar incubation of a ρ1 GABA C -expressing oocyte with AMP-coated qdots, that is, with a structure that lacked muscimol (panel C). Halo fluorescence of the oocyte surface membrane was also absent when a nonexpressing oocyte was incubated with 34 nM muscimol-conjugated dots (panel E). These results indicate that the muscimol ligand is necessary for binding of the conjugate to the oocyte surface membrane.

Labeling of GABA C -expressing oocytes with muscimol-conjugated qdots
As noted in Section 2.1.4, the muscimol-conjugated AMP-coated qdot preparation used in the oocyte imaging experiments contained ∼150-200 muscimol-terminated chains per qdot. In some preparations (not illustrated) of these muscimol-conjugated AMP-coated qdots, the extracellular medium surrounding the oocytes exhibited aggregation of the fluorescent particles. In developing the method of preparation of the conjugate, we observed that if 2000 equivalents of the muscimol ligand were reacted with AMPcoated qdots in the presence of 1500 equivalents of EDC and NHS, aggregates formed that subsequently precipitated from solution. It is likely that this aggregation is due to hydrogen bonding between muscimol ligands on adjacent qdots. The size and solubility of these aggregates likely depended on the number of ligands conjugated to the qdots.

Image analysis
To quantify the extent of binding of muscimol-conjugated AMP-coated qdots to the oocytes, we analyzed the surface membrane and extracellular regions of a given fluorescence image [65]. Using MetaMorph software (Offline Version 6.3r0; Universal Imaging Corp., Downington, Pa, USA), we determined the intensities of pixels underlying a multisegmented line that traced the arc-like border of the oocyte (15-25 straight-line segments; 450-750 pixels), and tabulated the resulting pixel values in relation to a 0-255 gray scale. We similarly determined the intensities of pixels that corresponded with an identical multisegment line constructed within the extracellular region of the image; tabulated intensities for this control extracellular region were taken as a measure of background (i.e., surround) fluorescence. For the image shown in Figure 5A, fluorescence intensities determined for the halo (henceforth termed "border") at the oocyte surface membrane and the surrounding extracellular medium (background) were 67.31 ± 36.79 (mean ± SD) and 22.30 ± 21.18, respectively. As reported by Gussin et al. [65], results obtained in experiments similar in design to that described in Figures Figures 5E, 5F), border and background fluorescence intensities were 15.14 ± 22.35 and 16.78±22.17, respectively, [65]. Two-way ANOVA analysis of results obtained with the muscimol-conjugated AMP-coated qdots showed that for both human ρ1 GABA C -expressing and perch ρ1B GABA C -expressing oocytes, the fluorescence intensity of the border differed significantly from that of the background. For nonexpressing oocytes incubated with the conjugate, there was no significant difference between border and background values. In addition, the treatment of GABA C -expressing oocytes with free (i.e., non-qdotconjugated) GABA, muscimol, or PEGylated muscimol significantly reduced binding of the muscimol-qdots conjugate to the oocyte surface membrane (see Gussin et al. [65] for further details).

DISCUSSION
The primary finding of the experiments involving the incubation of muscimol-conjugated AMP-coated qdots with GABA C -expressing oocytes is that these conjugates exhibit specific binding at GABA C receptors. This binding depends on the presence of muscimol in the conjugate, as (unconjugated) AMP-coated qdots show no significant binding to oocytes expressing GABA C receptors. The approach described here builds on our earlier work with PEGylated serotonin attached to qdots in which we found that these conjugates exhibited binding at serotonin transporters expressed in HeLa and HEK cells. These findings indicate that it is possible to specifically label transporter proteins and ligand-gated receptors with qdots that have multiple copies of a membrane receptor or membrane transporter ligand attached through a PEG linker.
The presence of numerous copies of ligand in the muscimol-qdot conjugate described here raises the possibility that these conjugates bind to multiple GABA C receptors in a cross-linking fashion. Indeed, it is reasonable to hypothesize that the multiplicity of the muscimol ligand, as well as the length of the PEG linker that tethers each ligand to the qdot, favors such cross-linking. However, experiments conducted to date, while clearly establishing the ability of this conjugate to bind to cell-surface-expressed GABA C receptors, do not address the extent to which receptor crosslinking affects this binding activity. Other investigators have used fluorescent probes to track the diffusion dynamics of single receptors (single-particle tracking (SPT)), and such an SPT approach could be useful for evaluating the crosslinking activity of the present muscimol-qdot conjugate. For example, Dahan et al. [50] have examined the diffusion dynamics of glycine receptors in neuronal membranes by labeling the receptor with a conjugate consisting of a primary antireceptor antibody, biotinylated secondary antibody, and streptavidin-coated qdots. To test the extent of receptor cross-linking by this conjugate, they investigated, as a comparison system, an Fab fragment of the primary antibody that had been linked to an organic fluorophore (Cy-3). Dahan et al. [50] found that the receptor dynamics determined with the qdot conjugate and the Cy-3-containing molecule were similar, indicating that neither the presence of the SA-qdots nor some other feature of the qdot-containing conjugate promoted significant receptor cross-linking. It should be emphasized that the structure of the presently described muscimol-qdot conjugate (see Figure 2) differs from the qdot-containing conjugate studied by Dahan et al. [50] in several respects, including the presence of a smallmolecule ligand (muscimol) rather than an antibody as the receptor-reactive moiety, a high valency (copy number) of ligands per qdot, and a separation of each ligand from the qdot by a long linking chain (PEG3400). In future experiments, it may be possible to test for cross-linking by the muscimol-qdot conjugate using an approach in which GABA C receptor dynamics determined with the muscimolqdots conjugate are compared with those determined using a structure optimized for SPT, for example, a fluorescent probe attached to a single receptor [26,27].