Single-Molecule Imaging of a Fluorescent Unnatural Amino Acid Incorporated Into Nicotinic Receptors

Abstract

We report on the first, to our knowledge, successful detection of a fluorescent unnatural amino acid (fUAA), Lys(BODIPYFL), incorporated into a membrane protein (the muscle nicotinic acetylcholine receptor, nAChR) in a living cell. Xenopus oocytes were injected with a frameshift-suppressor tRNA, amino-acylated with Lys(BODIPYFL) and nAChR (α/β19′GGGU/γ/δ) mRNAs. We measured fluorescence from oocytes expressing nAChR β19′Lys(BODIPYFL), using time-resolved total internal reflection fluorescence microscopy. Under conditions of relatively low receptor density (<0.1 receptors/μm²), we observed puncta with diffraction-limited profiles that were consistent with the point-spread function of our microscope. Furthermore, diffraction-limited puncta displayed step decreases in fluorescence intensity, consistent with single-molecule photobleaching. The puncta densities agreed with macroscopic ACh-induced current densities, showing that the fUAA was incorporated, and that receptors were functional. Dose-response relations for the nAChR β19′Lys(BODIPYFL) receptors were similar to those for wild-type receptors. We also studied nAChR β19′Lys(BODIPYFL) receptors labeled with α-bungarotoxin monoconjugated with Alexa488 (αBtxAlexa488). The nAChR has two αBtx binding sites, and puncta containing the Lys(BODIPYFL) labeled with αBtxAlexa488 yielded the expected three discrete photobleaching steps. We also performed positive control experiments with a nAChR containing enhanced green fluorescent protein in the γ-subunit M3-M4 loop, which confirmed our nAChR β19′Lys(BODIPYFL) measurements. Thus, we report on the cell-based single-molecule detection of nAChR β19′Lys(BODIPYFL).

Introduction

Fluorescent labeling of proteins is a valuable technique for understanding biological processes at the cellular and subcellular levels. In a widely used method with subcellular resolution, a fluorescent protein (FP) (Fig. 1 A) (1) is genetically fused to a protein of interest (2,3). Such FP fusions are used to measure protein expression levels, colocalization, and trafficking. The FP moieties, however, are typically ∼27 kDa, and therefore may interfere with protein function, folding, and stability. Several methods were developed to overcome these limitations. For example, fluorophores with appropriate chemical groups can react with appropriate amino-acid side chains, but such fluorophores label all proteins in a nonselective manner in living cells (4,5). The selective posttranslational labeling of proteins was achieved by introducing unnatural amino acids that can then be derivatized via bio-orthogonal reactions (6–8). However, these reactions may not reach completion, and the reagents may partition nonspecifically into membranes. Orthogonal enzymatic labeling approaches have expanded the toolkit for protein labeling, but require the insertion of a peptide sequence and subsequent labeling (9,10). Peptide sequences with binding affinity or reactivity toward lanthanide ions (11,12) and synthetic dyes (13) can also be encoded into proteins and synthetic peptides. Nevertheless, the intracellular labeling of proteins remains challenging.

Figure 1.

Structure of molecules studied in these experiments. (A) Crystal structure of GFP, Protein Data Bank accession code 1EMA. (B) Structure of Lys(BODIPYFL): same scale as GFP (bottom), and magnified (top). (C) Top view of Torpedo nAChR cryo-EM structure, Protein Data Bank accession code 2BG9. Green residue with red-dashed circle indicates nAChR β19′ site. (D) Side view of nAChR β-subunit, with 19′ site indicated in green. (E) Structure of BODIPYFL-C3-MTS.

Fluorescent unnatural amino acids (fUAAs) represent an attractive alternative strategy. Previously, fUAAs were incorporated into various types of proteins, using nonsense (14–19) and frameshift (15,18) suppression methodologies. Orthogonal tRNA/aminoacyl-tRNA synthetase pairs that can incorporate fUAAs were also developed (20–22). Several fUAAs were incorporated into functional proteins and detected with conventional cuvette or gel-based fluorescence measurements. Fluorescence moieties that were incorporated include: 5-hydroxytryptophan (21), 7-azatryptophan (16), dansyl (20), coumarin (23), NBD-Dap (14), Aladan (19), and BODIPYFL derivatives (15,18). These fluorophores span a wide spectral range, indicating the generality of nonsense and frameshift-suppression methodologies for the site-specific incorporation of fUAAs into proteins.

For example, fUAAs were previously incorporated into receptors and ion channels heterologously expressed in Xenopus oocytes by nonsense suppression. The NBD-Dap and Aladan were incorporated into the functional neurokinin-2 (NK2) receptor (14) and voltage-gated potassium channel Kir2.1 (19), respectively. Plasma membranes from oocytes were isolated, and the fluorescence of NBD-Dap incorporated into the NK2 receptor was measured with cuvette fluorescence spectrometry. In addition, a NK2 cysteine mutant was labeled with a tetramethylrhodamine (TMR) thiol-reactive fluorescent probe, and Förster (or fluorescence) resonance energy transfer (FRET) was detected between the NBD donor and TMR (14). However, to the best of our knowledge, there are no reports of fluorescence imaging with fUAAs incorporated into membrane proteins in living cells. Live-cell fluorescence imaging is essential in understanding the fundamental function and trafficking of ion-channel proteins. This is especially true of neuronal ion-channel proteins, whose function (24) and expression patterns (25) have subcellular specificity.

To extend fUAA incorporation to living cells and also to the single-molecule level, we chose an ion-channel molecule that is well-characterized, but still poses important scientific questions. We studied the incorporation of Lys(BODIPYFL) (Fig. 1 B) into the mouse muscle nicotinic acetylcholine receptor (nAChR) heterologously expressed in Xenopus oocytes. The nAChRs are cation-selective, ligand-gated ion channels (LGICs) in the pentameric Cys-loop superfamily. Muscle nAChR has a subunit stoichiometry of 2α₁/β₁/δ/γ. Each subunit contains an extracellular ligand-binding domain, a transmembrane region with four α-helices (M1, M2, M3, and M4), and a substantial intracellular region between the M3 and M4 helices (Fig. 1 C–D) (26,27). The ACh binding sites are at the α-γ and α-δ subunit interfaces. The transmembrane M2 α-helices from each subunit line the pore axis (28). The M2 transmembrane regions of all LGICs are spanned by >20 amino acids, whose sequence is often summarized according to a numbering system (29) in which 0′ is the most intracellular, N-terminal position of the M2 helix. Despite extensive structure-function studies on the conformational changes that translate ligand-binding at the ACh binding site into the opening of the gate ∼60 Å away (30–33), the nature and order of structural motions are still undetermined.

This lack of direct conformational measurements arises in part because many residues in the M2 domain and other “buried” domains are at best partially accessible for posttranslational fluorescent labeling using conventional methods such as cysteine reactions. Therefore, we incorporated the fUAA Lys(BODIPYFL) into the nAChR β-subunit M2 domain 19′ position (β19′) (Fig. 1 D), a site that was partially probed by fluorescence labeling (30). The nAChR β19′ site is an important gating residue and a candidate for future single-molecule FRET studies.

We used single-molecule total internal reflection fluorescence (TIRF) microscopy to measure fluorescent nAChRs. We studied the photobleaching of diffraction-limited puncta to assess the incorporation of fUAA. We also compared fluorescence signals from the nAChR β19′Lys(BODIPYFL) labeled with a second fluorophore, α-bungarotoxin monoconjugated with Alexa488 (αBtxAlexa488). We compared signals from nAChR containing enhanced green fluorescent protein in the γ-subunit M3-M4 loop (γ_eGFP) with our nAChR β19′Lys(BODIPYFL) observations. Fluorescence measurements with nAChR β19′C labeled with BODIPYFL-C3-MTS (Fig. 1 E), a thiol-reactive BODIPYFL variant, were not possible because BODIPYFL-C3-MTS does not label the β19′C residue. Thus, our study illustrates the advantages and current limitations of biosynthetically incorporating fUAAs into ion channels at buried residues.

Methods

Reagents and materials

Preparation of α-NVOC-L-Lys(BODIPYFL)-O-dCA

All reagents were purchased from Sigma-Aldrich (St. Louis, Mo.) unless otherwise noted. The high-performance liquid chromatography (HPLC) instrumentation was described elsewhere (35). We prepared α-NVOC-L-Lys(BODIPYFL)-OCH₂CN using the protocol reported for α-NVOC-ɛ-(6-(biotinoyl)amino)hexanoyl-L-lysine cyanomethylester (35). Briefly, ɛ-Boc-L-Lys (100 mg, 0.41 mmol; Bachem, Torrance, CA), NVOC-Cl (168 mg, 0.62 mmol), and Na₂CO₃ (65 mg, 0.62 mmol) were dissolved in 18 MΩ H₂O/dioxane (1:1.4, 85 mL) and stirred at room temperature for 8 h. The presence of product was confirmed with APCI⁻-MS: calculated for C₂₁H₃₀N₃O₁₀⁻ 484.19; found [M − H]⁻: 484.1. Then, 15 mL of 1 M NaHSO₄ were added to the reaction, and α-NVOC-ɛ-Boc-L-Lys was extracted with CH₂Cl₂ (30 mL × 3). The CH₂Cl₂ fractions were combined and removed by rotary evaporation. The sample was not further purified, and α-NVOC-ɛ-Boc-L-Lys (∼19.8 mg, 0.04 mmol) was dissolved in CH₂Cl₂ (3 mL). The TFA (3 mL) was added, and the mixture was stirred at room temperature. The CH₂Cl₂ was added 30 min later, and the solvents were removed with rotary evaporation. The α-NVOC-L-Lys was purified with semipreparative HPLC, using a linear gradient of 100% 25 mM NH₄Ac buffer (pH 4.5) against the organic phase (98%/2% ACN/18 MΩ H₂O) for ∼60 min. Fractions corresponding to the 350-nm absorption peak from the NVOC group were pooled and lyophilized. The NVOC absorption extinction coefficient at 260 nm ɛ₂₆₀ (2140 M⁻¹ cm⁻¹) was used to determine that 12.1 mg (77% yield) of product were obtained. The presence of α-NVOC-L-Lys was confirmed by positive mode atmospheric pressure chemical pressure ionization mass spectrometry: calculated for C₁₆H₂₄N₃O₈⁺ 386.16; found [M + H]⁺: 386.1. Twelve milligrams of α-NVOC-L-Lys and 5 mg of 4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionic acid, succinimidyl ester (BODIPY FL, SE, Invitrogen, Carlsbad, CA), were dissolved in 2 mL of dimethylformamide (DMF) and 4.4 μL of diisopropylethylamine, and mixed for 7 h. An aqueous workup was performed with 25 mL of 1 M NaHSO₄, and the product was extracted with CH₂Cl₂ (30 mL × 3). The CH₂Cl₂ extractions were pooled together and dried over Na₂SO₄ for 1 h. The CH₂Cl₂ was removed with rotary evaporation. The DMF was removed by simultaneously immersing the round-bottomed flask with the sample in a 60°C water bath and attaching it to a high vacuum line with solvent traps. Next, the sample was mixed with 42 μL of ClCH₂CN, 3 μL of triethylamine, and 1 mL of DMF. The reaction was allowed to proceed for ∼7 h before an aqueous workup was performed with 15 mL of NaHCO₃ and CH₂Cl₂ (25 mL × 3). The organic phase was dried over Na₂SO₄ for 30 min, and the CH₂Cl₂ was removed with rotary evaporation. The product was purified with semipreparative HPLC, using a linear gradient of 100% 25 mM NH₄Ac buffer (pH 4.5) against the organic phase (98%/2% ACN/18 MΩ H₂O) for ∼60 min. The fractions that had both absorbance peaks of the NVOC (∼350 nm) and BOPIDYFL (∼504 nm) and identical elution times were collected, pooled, and lyophilized. The product yield was 6.8 mg (76% yield). The sample was identified by positive mode electrospray ionization mass spectrometry: calculated for C₃₂H₃₈BF₂N₆O₉⁺ 699.28; found [M + H]⁺: 699.2, and the solvent was lyophilized. The ¹H NMR (300 MHz, CD₃CN) δ 7.70 (s, 1H), 7.36 (s, 1H), 7.13 (s, 1H), 7.00 (d, 1H), 6.44 (m, 2H), 6.31 (d, 1H), 6.23 (s, 1H), 5.43 (m, 2H), 4.82 (s, 2H), 4.24 (m, 1H), 3.95 (s, 3H), 3.89 (s, 3H), 3.15 (m, 4H), 2.53 (m, 2H), 2.27 (s, 3H), 2.17 (m, 3H), 1.8 (m, 2H), and 1.42 (m, 4H).

The α-NVOC-L-Lys(BODIPYFL)-O-CH₂CN and dCA were dissolved in 1 mL of DMF and stirred under argon (Ar) for ∼24 h (in vacuo). The presence of sample was confirmed with analytical HPLC. Preparative HPLC was used to purify α-NVOC-L-Lys(BODIPYFL)-O-dCA, and negative mode electrospray ionization mass spectrometry: calculated for C₄₉H₅₉BF₂N₁₃O₂₁P₂⁻ 1,276.35; found [M − H]⁻: 1276.4 confirmed the presence of product from the fractions that contained absorption peaks at 350 and 505 nm. The product yield was 36.1 μg (1.6% yield), based on the BODIPYFL absorption 504-nm absorption peak and an extinction coefficient of ɛ₅₀₄ (80,000 M⁻¹cm⁻¹) (36).

Preparation of BODIPYFL-C3-MTS

Twelve milligrams of 3-aminopropyl methanethiosulfonate hydrobromide (Toronto Research Chemicals, Toronto, Ontario, Canada) and 250 μg of 4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionic acid, succinimidyl ester (BODIPY FL, SE, Invitrogen), were dissolved in 1 mL of DMF and 10 μL of diisopropylethylamine and mixed for 1.5 h. The product was purified with semipreparative HPLC, using a linear gradient of aqueous phase (2%/98% ACN/18 MΩ H₂O) against the organic phase (98%/2% ACN/18 MΩ H₂O) for ∼30 min. The fractions that had the BOPIDYFL (∼504 nm) absorbance peak were collected, pooled, and lyophilized. The sample was then identified by positive mode electrospray ionization mass spectrometry: calculated for [M + K]⁺ 482.13; found [M + K]⁺: 482.0, and the solvent was lyophilized. The product yield was determined to be 324 μg (57% yield), based on the BODIPYFL absorption 504-nm absorption peak and extinction coefficient of ɛ₅₀₄ (80,000 M⁻¹cm⁻¹) (36).

Preparation of YFaFS_ACCC-α-NVOC-L-Lys(BODIPYFL)

The YFaFS_ACCC tRNA was prepared by linearization of pUC19 containing the YFaFS_ACCC gene with Fok1, which yields 74mer-tRNA upon in vitro transcription with the MegaShortScript kit (Ambion, Austin, TX) (37). The dCA-α-NVOC-L-Lys(BODIPYFL) was ligated to the 74mer-YFaFS_ACCC with T4 RNA ligase, which yielded YFaFS_ACCC-α-NVOC-L-Lys(BODIPYFL) as previously described (38–40). The product was desalted with a CHROMA SPIN-30 DEPC-H₂O column (Clonetech, Mountain View, CA). The presence of product was confirmed with ultraviolet absorption at 260 nm and matrix assisted laser desorption ionization mass spectrometry. The major peak indicated that YFaFS_ACCC-α-NVOC-L-Lys(BODIPYFL) was the principal product of the ligation reaction.

Construct preparation and mRNA transcription

The mouse muscle nAChR subunits (α, γ, and δ) are in the pAMV vector. The β19′GGGU construct was prepared by QuikChange (Stratagene, La Jolla, CA) mutagenesis on the masked β-subunit in the pAMV vector (37). The mutation was verified by sequencing (California Institute of Technology Sequencing/Structure Analysis Facility). The mRNAs were prepared from NotI linearized plasmid DNA, using a T7 mMessage mMachine kit (Ambion), and were purified with the RNeasy Mini kit (Qiagen, Valencia, CA). Concentration was determined by absorption at 260 nm.

The γ_eGFP construction was prepared as follows. The eGFP insertion in the γ-M3-M4 cytoplasmic domain has no effect on the electrophysiological properties of nicotinic receptors (αβγ_eGFPδ) (41). Therefore, we inserted eGFP after position S380 in the γ-M3-M4 cytoplasmic loop by polymerase chain reaction. The construct was synthesized from three segments by polymerase chain reaction. Primers to amplify the first segment are γ_eGFP - Forward No. 1 (ACC ATG GCA CAA GGG GGC CAG AGA CCT CAT CTC CTC TTG CTG, where ATG is the initiation codon of the γ-subunit); γ_eGFP -Reverse No. 1 (5′- CAG CTC CTC GCC CTT GCT CAC CAT TGA GGA AGA GCC ATT CTG GAG TCG-3′; bold sequences represent the N-terminal of eGFP). The second segment was synthesized by γ_eGFP -Forward No. 2 (5′- CGA CTC CAG AAT GGC TCT TCC TCA ATG GTG AGC AAG GGC GAG GAG CTG -3′; bold sequences represent the N-terminus of the eGFP) and γ_eGFP -Reverse No. 2 (5′- CTC TCG AGC CAT GAT GGG CCA CCC CTT GTA CAG CTC GTC CAT GCC GAG -3′; bold sequences represent the C-terminus of eGFP). The third segment was synthesized by the following γ_eGFP -Forward No. 3 (5′- CTC GGC ATG GAC GAG CTG TAC AAG GGG TGG CCC ATC ATG GCT CGA GAG -3′; bold sequences represent the C-terminus of the eGFP) and γ_eGFP -Reverse No. 3 (5′- GTC GAC TCA GTC TGG CAA AGG CAG GTA GGG GCG GGG GTC TCC -3′; bold TCA represents the termination codon of the γ-subunit). Finally, the entire construct was amplified from all three segments, using γ_eGFP -Forward No. 1 and γ_eGFP -Reverse No. 3, and was subcloned into the pAMV vector. The entire construct was verified by DNA sequencing (Davis Sequencing, Davis, CA).

Oocyte preparation, injection, and αBtxAlexa488 labeling

Stage VI oocytes from Xenopus laevis were isolated and maintained at 16°C in ND96 solution consisting of 96 mM NaCl, 2 mM KCl, 1 mM MgCl₂, 1.8 mM CaCl₂, and 5 mM HEPES (pH 7.5), supplemented with Na pyruvate (2.5 mM), gentamycin (50 μg/mL), theophylline (0.6 mM), and horse serum (5%). Before injection, YFaFS_ACCC-α-NVOC-L-Lys(BODIPYFL) was irradiated at 350 nm to cleave the NVOC group, yielding the deprotected tRNA-fUAA (YFaFS_ACCC-Lys(BODIPYFL)). Then the YFaFS_ACCC-Lys(BODIPYFL) was kept at 4°C until mixing with the mRNA and loaded for oocyte injection. Each oocyte was injected with a 1:1 volume mixture of YFaFS_ACCC-Lys(BODIPYFL) (12.5 or 25 ng) and mRNA (25 ng of total concentration ratio of 2:5:1:1 for α/β19′GGGU/γ/δ subunits) in a 50-nL volume. Wild-type (WT) nAChR α, β, and δ and either WT γ or γ_eGFP mRNA were mixed at a ratio of 2:1:1:1 and diluted to a final concentration of 1 ng/25 nL.

Oocytes were labeled with α-bungarotoxin conjugated to Alexa Fluor 488 (αBtxAlexa488) (Invitrogen), dissolved in ND96. We experimented with various labeling concentrations and incubation times, and determined that incubating oocytes with 20–40 nM αBtxAlexa488 for 8–12 h resulted in optimal labeling. The incubations were terminated by transferring oocytes to a 5 mg/mL bovine serum albumin solution, followed by two ∼30-min washes in ND96 before imaging. The manufacturer furnished αBtxAlexa488 at a nominal concentration of 1 μg/μL. We determined the molar ratio of αBtx to Alexa Fluor 488 more precisely, using the known αBtxAlexa488 MW = 8600 (αBtx MW = 7966.23 Da, Alexa Fluor 488 carboxylic acid = 532.5 Da, and linker MW = unknown, assumed to be 100 Da), the Alexa488 absorption peak at 488 nm, the extinction coefficient of 65,000 M⁻¹cm⁻¹ (36), and the NanoDrop (Thermo Scientific, Wilmington, DE) A₂₈₀ protein assay. The molar ratio of αBtx/Alexa488 was 1:0.74. The attempted labeling of the nAChR β19′C receptor with 5 nM and 5 μM BODIPYFL-C3-MTS was performed as previously reported (30).

Electrophysiology

A two-electrode voltage clamp was used to record currents from oocytes. Recordings were performed 24–48 h after injection with a GeneClamp 500 amplifier (Axon Instruments, Union City, CA). The pipette microelectrodes were filled with 3 M KCl, and had resistances ranging from 0.5–2 MΩ. Oocytes were perfused continuously with a calcium-free solution containing 96 mM NaCl, 2 mM KCl, 1 mM MgCl₂, and 5 mM HEPES (pH 7.5). The ACh-induced currents were recorded in response to a series of ACh applications (1–1000 μM).

Total internal reflection fluorescence microscopy

Xenopus oocytes were prepared for single-molecule TIRF microscopy as follows. The oocytes were transferred to a petri dish filled with hypertonic solution (220 mM Na aspartate, 10 mM EDTA, 2 mM MgCl₂, and 10 mM HEPES, pH 7.4) and incubated for 5–10 min. The vitelline membrane was removed with forceps, under a dissecting microscope. Oocytes were transferred to an imaging chamber mounted on the microscope stage. The imaging chamber bottom had a glass coverslip that was cleaned with one of two protocols. The first protocol consisted of rinsing with 90% EtOH and 18 MΩ water. Alternatively, they were immersed in a solution composed of (v/v) 90% 18 MΩ water/7% H₂SO₄/3% H₂O₂. The oocyte was positioned with the animal pole oriented toward the coverslip. The vegetal pole was not used because of a high autofluorescence background. Experiments were performed at 18–20°C, to minimize sample drift and maximize the amount of time the oocytes remained mechanically stable. Before imaging, we allowed oocytes to incubate in the imaging chamber for ∼10 min, to maximize adherence. Next, transmitted bright-field illumination was used to confirm oocyte adherence, using coarse focus. Oocytes typically adhered over a circular area ≥200 μm in diameter. A region of interest was selected to focus and establish the TIRF condition, using laser excitation (<140 mW/cm²). An adjacent area was used to conduct measurements, so as to limit unintended photobleaching during fine focus. In experiments where macroscopic experiments show adequate expression, nearly all injected oocytes yield TIRF images with clearly isolated puncta. The most informative data are generated from images whose puncta density is in the measurable range, i.e., 0.001–0.06/μm². Anomalously high backgrounds are found in the <5% of oocytes that collapse and tear in the imaging chamber. These oocytes were most likely damaged during vitelline membrane removal, and are excluded from analysis. The TIRF microscope for collecting images consists of a Melles-Griot Ar ion laser (CVI Melles Griot, Carlsbad, CA) coupled to a fiber optic that extended to an Olympus TIRF illuminator adapted to an IX-71 Olympus inverted microscope (Center Valley, PA). The Ar laser emits three wavelengths: at 457 nm (∼2 mW), 488 nm (∼15.6 mW), and 514 nm (∼10 mW). We selected the 488-nm wavelength, required to excite and detect Lys(BODIPYFL), γ_eGFP, and αBtxAlexa488, using the Z488 filter cube (Chroma Technology Corp., Rockingham, VT). The unfiltered maximum power at the objective entrance was 4.4 mW (∼140 mW/cm²). Where appropriate, neutral density filters were used to reduce power by up to threefold during imaging. The integration time was 0.5 s for all images. A 100× 1.45-NA TIRF objective (Olympus) was used. Time-lapse images were captured with a Photometrics Cascade 650 front-illuminated CCD camera (Princeton Instruments, Trenton, NJ). The frame area at ×100 magnification is 48.3 × 36.4 μm = 1728.1 μm². Slidebook (Intelligent Imaging Innovations, Inc., Santa Monica, CA) was used to acquire data. The data were subsequently processed and analyzed with ImageJ (National Institutes of Health, Bethesda, MD) (42), Clampfit 9.2 (Axon Instruments, Foster City, CA), Origin 7.0 (OriginLab, Northampton, MA), and MatLab R2006a (The MathWorks, Inc., Natick, MA).

Image analysis

The images acquired with Slidebook were exported as 16-bit time series stack files in TIFF format. The ImageJ (National Institutes of Health) region of interest (ROI) manager was used to document the puncta in the first frame of time stacks. Rectangular 15 × 15 pixel ROIs, centered on a putative single punctum, were duplicated and batch-processed using MatLab R2006a. The point-spread function (PSF) of individual puncta were fit to a two-dimensional Gaussian function:

$$PSF(x,y,x_c,y_c,S_x,S_y,A,B)=B+A\exp \left[-\frac{{(x_{}-x_c)}^2}{2S_x^2}-\frac{{(y_{}-y_c)}^2}{2S_y^2}\right],$$ (1)

where (x_c, y_c) is the centroid of the punctum, S_x and S_y are the respective standard deviations along the x axis and y axis, A is the amplitude, and B is the baseline. Next, the full width at half-maximum (FWHM_x,y) along the x and y axes were determined, $FWH{M}_i=2\times \left(\sqrt{\ln \left(4\right)}\right)\times S_i$, and compared against the FWHM_theoretical PSF of the microscope, which is approximated by the equation (43,44):

$$FWH{M}_{\text{theoretical}}\approx \frac{0.55\times {\lambda }_{\text{emission}}}{NA},$$ (2)

where λ_emission is the emission peak of the fluorescent molecule, and NA is the numerical aperture of the objective (NA = 1.45). The PSF is deformed by several factors, including spherical aberrations, excitation-field nonuniformity (45,46), pixel nonuniformity, background nonuniformity (45), and others. Demuro and Parker reported that muscle nAChRs undergo lateral motions of ±<50 nm on a time scale of ∼10 s (47). Therefore, puncta with FWHM_x,y = 195 ± 74.5 nm (or 2.91 ± 1 pixels) were classified as diffraction-limited, immobile spots, and were selected for puncta density and photobleaching analysis. Time-lapse images with minimal lateral and vertical drift were selected for photobleaching step analysis. The peak pixel value against time was plotted for ROIs with puncta that fulfilled the stated FWHM_x,y criteria. Time-lapse traces were imported into Clampfit and analyzed.

Results

In the nonsense and frameshift-suppression methodologies for site-specific UAA incorporation, the chemically amino-acylated tRNA is a stoichiometric reagent that limits the number of functional channels that can be expressed (48). This low expression level is well-suited to the sensitivity of electrophysiology. Here we demonstrate that single-molecule TIRF microscopy complements the sensitivity of electrophysiology. TIRF microscopy excites only molecules that are within ∼100 nm of the glass surface, thereby minimizing interference from cytoplasmic autofluorescence, including unincorporated fUAAs (tRNA-fUAA or free fUAA). Thus, TIRF microscopy, integrated with a CCD camera with single-molecule sensitivity, enables the optical detection of ion channels expressed at the plasma membrane in low numbers (< 0.015/μm²).

tRNA-fUAA and heterologous expression in Xenopus oocytes

Based on previous cell-free synthesis applications (15,49,50) and photophysical properties, i.e., favorable absorption coefficient (ɛ, ∼80,000–91,000 M⁻¹ cm⁻¹) and quantum yield approaching 1.0 (15,36), Lys(BODIPYFL) (Fig. 1 B) was selected as the fUAA for incorporation into nAChR. The single amino-acid substitution of Lys(BODIPYFL) (Fig. 1 B) is presumably less perturbing than substituting eGFP (Fig. 1 A) into the transmembrane α-helix (Fig. 1 D). For fUAA incorporation, we chose to use the yeast Phe frameshift-suppressor (YFaFS_ACCC) that recognizes the quadruplet codon GGGU and contains seven mutations in the acceptor stem (denoted by lower-case “a”) to avoid recognition by endogenous aminoacyl-tRNA synthetases (aaRSs). The YFaFS_ACCC is recognized least by the endogenous aaRSs of any suppressor tRNA screened in our laboratories, and therefore should not incorporate natural amino acids at the suppression site (37,51,52). As described in Methods, we first synthesized dCA-Lys(BODIPYFL), the fUAA chemically amino-acylated on the dinucleotide deoxyCA, and then enzymatically ligated dCA-Lys(BODIPYFL) to the frameshift-suppressor tRNA lacking the terminal CA nucleotides (74mer-YFaFS_ACCC), which yielded YFaFS_ACCC-Lys(BODIPYFL), the full-length 76mer-tRNA ligated to the fUAA. The YFaFS_ACCC-Lys(BODIPYFL) was mixed with nAChR α/β19′GGGU/γ/δ mRNA and injected into Xenopus oocytes. After incubation for 1–2 days, the presence of nAChR β19′Lys(BODIPYFL) on the plasma membrane was confirmed with two-electrode voltage-clamp electrophysiology (currents for a series of 5-s ACh applications are shown in Fig. 2 A). The average maximal currents evoked by 1 mM ACh (I_max) were 6.3 ± 0.6 μA (n = 5 oocytes) and 3.5 ± 0.2 μA (n = 4 oocytes) for oocytes injected with 25 and 12.5 ng YFaFS_ACCC-Lys(BODIPYFL), respectively. The ACh dose-response relationships were generated for nAChR β19′Lys(BODIPYFL) and WT nAChR. The measured EC₅₀ values were 38 ± 2 and 60 ± 4 μM ACh, respectively, and the Hill coefficients were 1.1 ± 0.05 and 1.3 ± 0.01, respectively (n = 5 and 9 oocytes, respectively) (Fig. 2 B). Substantially smaller currents were detected from oocytes injected with nAChR β19′GGGU mRNA only (I_max = 0.78 ± 0.5 μA; n = 3 oocytes) or nAChR β19′GGGU mRNA with 74mer-YFaFS_ACCC (I_max = 1.1 ± 0.3 μA; n = 3 oocytes) (Fig. 2 C).

Figure 2.

Electrophysiological properties of oocytes expressing nAChR β19′Lys(BODIPYFL). (A) Representative ACh-induced currents during a dose-response series, in an oocyte expressing nAChR β19′Lys(BODIPYFL) receptor. Each ACh application lasted 5 s. (B) Dose-response curve for nAChR β19′Lys(BODIPYFL) (open circles), compared with WT data (solid circles). Error bars are mean ± SE (n = 5 oocytes). Smooth curves represent a fitted single-component dose-response relationship for nAChR β19′Lys(BODIPYFL) (EC₅₀, 38 ± 2 μM; Hill coefficient, 1.1 ± 0.05) and WT (EC₅₀, 60 ± 4 μM; Hill coefficient, 1.3 ± 0.01). (C) Average current induced by 1- mM ACh application in oocytes injected with 25 ng of α/β19′GGGU/γ/δ mRNA + 25 ng YFaFS_ACCC-Lys(BODIPYFL) (mRNA + tRNA-fUAA), 25 ng α/β19′GGGU/γ/δ mRNA only (mRNA), and 25 ng α/β19′GGGU/γ/δ mRNA + 25 ng 74mer-YFaFS_ACCC (mRNA + 74mer-tRNA).

Single-molecule TIRF microscopy of an incorporated fUAA

Single-molecule TIRF microscopy was used to detect the presence of nAChR β19′Lys(BODIPYFL) in the plasma membrane (Fig. 3, A and B). Sparse fluorescent puncta (<0.10 puncta/μm²) were detected 24 h after coinjection of YFaFS_ACCC-Lys(BODIPYFL) (12.5 ng/oocyte) with nAChR β19′GGGU mRNA (25 ng/oocyte; n = 5 oocytes; Fig. 3 A and Fig. S1 A in the Supplementary Material). Control oocytes from the same batch were always injected with YFaFS_ACCC-Lys(BODIPYFL), (i.e., with tRNA-fUAA, but no mRNA), and sparse puncta were detected at lower levels (< 0.015 puncta/μm²) (Fig. S1 B). The PSF of emission profiles (Fig. 3 B) of fluorescent single molecules were fitted to a two-dimensional Gaussian function (45,53) (Fig. 3 C). We compared the two-dimensional Gaussian FWHM_x,y along the x and y axes against the theoretical FWHM. The emission peak of BODIPYFL is ∼515 nm (36). Therefore, we defined puncta with a FWHM_x,y = 195 ± 74.5 nm (1 pixel) as diffraction-limited spots for further analysis. Twenty percent to 50% of all puncta detected were diffraction-limited regions, consistent with the PSF of the TIRF microscope. Time-series traces with single-molecule photobleaching events were observed from puncta with diffraction-limited regions (Fig. 3 D). Time-series imaging experiments demonstrated that the puncta originated from one (80%), two (17%), or three (3%) molecules (n = 3 oocytes, 151 puncta), as determined by discrete photobleaching events. Thus, although the majority of puncta represented a single molecule, a subset represented more than one molecule per diffraction-limited spot.

Figure 3.

TIRF microscopy images of a representative oocyte expressing nAChR β19′Lys(BODIPYFL). (A) An oocyte injected with nAChR α/β19′GGGU/γ/δ mRNA and YFaFS_ACCC-Lys(BODIPYFL). Scale bar represents 12 μm. The counts calibration bar with linear dynamic range is selected for best display. The square is a representative punctum selected for presentation. (B) Representative punctum in 15 × 15 pixel region corresponding to square in A. Scale bar represents 10 pixels, or 0.74 μm. (C) Two-dimensional Gaussian fit of punctum in B. (D) Single-molecule time-series photobleaching trace from selected punctum.

To verify our observations with nAChR β19′Lys(BODIPYFL), we synthesized a thiol-reactive BODIPYFL derivative, BODIPYFL-C3-MTS (Fig. 1 E). The nAChR β19′C was previously labeled with sulforhodamine-MTS (MTSR), and voltage-clamp fluorometry was performed (30). Thus, we experimented with labeling nAChR β19′C with BODIPYFL-C3-MTS, to compare the two labeling strategies directly. However, the TIRF microscopy images show no difference between uninjected oocytes and those expressing nAChR β19′C labeled with BODIPYFL-C3-MTS. This observation agrees with reports that nAChR β19′C is primarily accessible to positively charged thiol-reactive probes (54).

Ulbrich and Isacoff demonstrated the power of counting ion-channel subunits by using genetically fused fluorescent proteins (55). Therefore, we evaluated an alternative strategy, which consisted of generating a nAChR mutant with genetically fused eGFP. Previous data indicated that function and trafficking are well-preserved when FPs are incorporated into the M3-M4 intracellular loop of muscle (and neuronal α4β2) nicotinic receptors (41,56,57). Thus, a nAChR γ-subunit construct with eGFP inserted in the M3-M4 loop (nAChR γ_eGFP) served as a positive control for our observations with nAChR β19′Lys(BODIPYFL). The nAChR γ_eGFP receptors were heterologously expressed in oocytes at similar levels as in nAChR β19′Lys(BODIPYFL), and diffraction-limited puncta were detected (Fig. 4 A and Fig. S1 C). Puncta from uninjected oocytes were essentially nonexistent (Fig. S1 D). The nAChR γ_eGFP puncta consisted of ∼1 (80%) (Fig. 4 B) and 2 (20%) photobleaching steps (n = 3 oocytes, 179 puncta), i.e., similar to nAChR β19′Lys(BODIPYFL). The amplitudes obtained from the Gaussian fits of individual puncta from nAChR β19'Lys(BODIPYFL) and nAChR γ_eGFP at a similar excitation power density (133 mW/cm²) were compared. The nAChR β19′Lys(BODIPYFL) and nAChR γ_eGFP displayed log-normal distributions (58), with a peak value at 1064 counts and log standard deviation (log SD) of 0.43 (n = 3 oocytes, 135 puncta), and a peak value at 501 counts with a log SD of 0.51 (n = 2 oocytes, 78 puncta), respectively (Fig. 5).

Figure 4.

TIRF microscopy images of representative oocyte expressing nAChR γ_eGFP. (A) An oocyte injected with nAChR γ_eGFP. Scale bar represents 12 μm. The counts calibration bar with linear dynamic range is selected for best display. (B) Representative single-molecule time-series trace from selected punctum.

Figure 5.

Histograms of puncta amplitudes from oocytes expressing either nAChR γ_eGFP (black bars) or nAChR β19′Lys(BODIPYFL) (gray bars), using similar imaging conditions. The nAChR β19′Lys(BODIPYFL) and nAChR γ_eGFP displayed log-normal distributions with a peak value at 1064 counts and a log SD of 0.43 (n = 3 oocytes, 135 puncta), and a peak value at 501 counts and a log SD of 0.51 (n = 2 oocytes, 78 puncta), respectively.

nAChR fluorescent ligand confirms the presence of nAChR β19′Lys(BODIPYFL) and nAChR γ_eGFP

To gain confidence that the puncta originated from nAChR β19′Lys(BODIPYFL) and nAChR γ_eGFP receptors, we compared the signals with those for receptors labeled with α-bungarotoxin monoconjugated to Alexa488 (αBtxAlexa488). Two αBtx molecules bind each muscle nAChR irreversibly (59) on the time scale of interest in this study. Thus, time-series traces with three photobleaching events were expected when we labeled nAChR β19′Lys(BODIPYFL) and nAChR γ_eGFP. First, WT nAChRs were used to determine the αBtxAlexa488 baseline level of labeling and the photobleaching step distribution. The WT nAChR was expressed at similar levels as nAChR β19′Lys(BODIPYFL), as confirmed by ACh-induced currents (I_max = ∼2 μA). The WT nAChR labeled with αBtxAlexa488 displayed puncta (Fig. S2 A) with a log-normal amplitude distribution with a peak value at 773 counts and a log SD of 0.45 (n = 2 oocytes, 149 puncta; 43 mW/cm²), and the peak value increased to 1157 counts with a log SD of 0.43 at a higher power density (n = 1 oocyte; 102 puncta; 69 mW/cm²) (Fig. 6 A). All photobleaching step analysis was restricted to a relatively low power density of 43 mW/cm², to resolve multiple steps before photobleaching. The WT nAChR labeled with αBtxAlexa488 resulted in 1 (48% ± 2%), 2 (43% ± 3%), 3 (6% ± 1%), and 4 (3% ± 1%) detected photobleaching steps (n = 2 oocytes, 144 puncta) (Fig. 6 B). Despite the two αBtx sites per receptor, WT receptors labeled with αBtxAlexa488 displayed a plurality (∼48%) of single bleach steps under all conditions. We ascribe this result to the incomplete labeling of αBtxAlexa488 (∼74% of αBtx are labeled by a fluorescent Alexa488; see Methods) and to the unintentional photobleaching of fluorescent molecules during sample-handling and a brief fine focus before image acquisition. Thus, if (1), ∼20% of puncta contain two receptors and 80% contain one receptor, as based on results from nAChR β19′Lys(BODIPYFL) and nAChR γ_eGFP, and (2), 74% of the αBtx have a fluorescent Alexa488, then the theoretical binomial fluorescent dye distribution per puncta is: 0 (6%), 1 (32%), 2 (48%), 3 (8%), and 4 (6%). Excluding the nonmeasurable nonfluorescent dyes (0 steps), the theoretical photobleaching step distribution becomes 1 (34%), 2 (51%), 3 (9%), and 4 (6%). However, Alexa488 photobleaches during fine focus. The measured photobleaching step distribution is adequately explained by assuming that the original 74% of fluorescent αBtx molecules decreased to 57%: the theoretical observable distribution becomes 1 (49%), 2 (41%), 3 (8%), and 4 (2%). We use the following assumptions throughout our step photobleaching analysis: 20% of diffraction-limited puncta contain two nAChRs, and 57% of the αBtx contain fluorescent Alexa488.

Figure 6.

TIRF microscopy of oocyte expressing WT nAChR labeled with αBtxAlexa488. (A) WT nAChRs labeled with αBtxAlexa488. Representative puncta in boxed squares correspond to one and two photobleaching steps. Scale bar equals 12 μm. (B) Representative traces, with one and two photobleaching steps. Dashed guidelines indicate fluorescence levels.

The nAChR γ_eGFP were labeled with αBtxAlexa488 (Fig. 7 A and Fig. S2 B), and we detected a distribution of 1 (40% ± 3%), 2 (33% ± 2%), 3 (18% ± 3%), and ≥4 (9% ± 3%) photobleaching steps (n = 2 oocytes, 148 puncta) (Fig. 7 B). Interestingly, the puncta density ratio between the nAChR γ_eGFP receptors and the receptors labeled with αBtxAlexa488 was 0.33 (n = 4; two oocyte batches). The apparent underreporting by the γ_eGFP label, compared with αBtxAlexa488, arises from at least two factors. The first is unintended photobleaching during fine focus: eGFP bleaches more quickly than Alexa488. The second is that <100% of eGFP molecules reach the mature fluorescent state (2). Again, if we assume that αBtxAlexa488 is reporting the presence of only 57% of receptors, then the expected binomial distribution for photobleaching steps is: 1 (36%), 2 (41%), 3 (15%), and 4–6 (8%).

Figure 7.

TIRF microscopy of oocyte expressing nAChR γ_eGFP labeled with αBtxAlexa488. (A) nAChR γ_eGFP labeled with αBtxAlexa488. Puncta in boxes are diffraction-limited spots with 1, 2, and 3 photobleaching steps. Scale bar represents 12 μm. The intensity calibration bar with linear dynamic range is selected for best display. (B) Representative traces with 1, 2, and 3 photobleaching steps. Dashed guidelines indicate fluorescence levels.

As expected, a subset of puncta containing the nAChR β19′Lys(BODIPYFL) labeled with αBtxAlexa488 yielded three discrete photobleaching steps, in agreement with the predicted stoichiometry (Fig. 8, A and B, and Fig. S2 C). The oocytes injected with nAChR β19′GGGU mRNA and YFaFS_ACCC-Lys(BODIPYFL), and incubated with αBtxAlexa488, displayed puncta with 1 (44% ± 8%), 2 (36% ± 2%), 3 (18% ± 8%), and 4–6 (2% ± 1%) photobleaching steps (n = 3 oocytes, 194 puncta). The puncta density ratio between nAChR β19′Lys(BODIPYFL) and nAChR β19′Lys(BODIPYFL) labeled with αBtxAlexa488 was 0.47. If the probability of a punctum having one detectable Lys(BODIPYFL) per nAChR is 47%, then the binomial distribution for measurable photobleaching steps is: 1 (31%), 2 (41%), 3 (19%), and 4–6 (9%). Thus, the measured photobleaching-step percentages are within ∼13% of values expected from straightforward molecular labeling (Fig. 9 A).

Figure 8.

TIRF microscopy of an oocyte expressing nAChR β19′Lys(BODIPYFL) labeled with αBtxAlexa488. (A) TIRF image of an oocyte expressing nAChR β19′Lys(BODIPYFL) receptors labeled with αBtxAlexa488. Puncta in boxes are diffraction-limited spots with 1, 2, and 3 photobleaching steps. Scale bar represents 12 μm. The intensity calibration bar with linear dynamic range selected for best display. (B) Representative traces with 1, 2, and 3 photobleaching steps, corresponding to puncta in squares. Dashed guidelines indicate fluorescence levels.

Figure 9.

Summary of photobleaching steps and puncta-density statistics. (A) Percent of puncta versus number of photobleaching steps measured for oocytes with surface-expressed WT nAChRs, nAChR β19′Lys(BODIPYFL), and nAChR γ_eGFP labeled with αBtxAlexa488, and imaged under similar conditions. Error bars report the mean ± SE. (B) Average puncta density for oocytes expressing nAChR β19′Lys(BODIPYFL) (β19′Lys(BODIPYFL)), nAChR γ_eGFP (γ_eGFP), WT nAChR (WT), YFaFS_ACCC-Lys(BODIPYFL) (tRNA-fUAA), and uninjected (UI) oocytes with and without αBtxAlexa488 labeling. The puncta density predicted from current density is plotted for nAChR β19′Lys(BODIPYFL), nAChR γ_eGFP, and WT nAChR oocytes. Dashed line represents correspondence between puncta density of 0.019/μm² and expected current of 1 μA. Error bars report mean ± SE.

Fluorescent puncta density and current density analysis

We observed that the optimal single-molecule TIRF microscopy range was I_max = 1–5 μA for WT nAChRs (subsequently labeled with αBtxAlexa488) and for nAChR γ_eGFP. In oocytes with ACh-induced currents <1 μA, no nAChR γ_eGFP receptors were detected. Thus, we compared puncta density with recorded currents. We made the following assumptions: 1), The nAChRs were distributed homogenously throughout the oocyte plasma membrane. 2), The oocyte plasma membrane uniformly adhered to the glass coverslip during single-molecule TIRF imaging, so that the measured image frames can be extrapolated to the entire oocyte. 3), Stage VI Xenopus oocyte plasma membrane capacitance is 0.25 μF (reported range, 0.19–0.25 μF) (60,61). 4), The specific membrane capacitance is 0.8 μF/cm² (62), implying that the average total oocyte plasma membrane area is 3.13 × 10⁷ μm². 5), Desensitization was negligible during the growth phase of ACh-induced currents, so that I_max measures the simultaneous activation of all receptors. 6), Receptors in microvilli that contribute to capacitance measurements are retained during imaging (63). Additional experimental details were as follows: 7), The image frame area was 1758.1 μm². 8), The oocyte holding potential was −60 mV. 9), Under the ionic conditions of our experiments, the single-channel nAChR conductance was 36 pS (64).

In experiments on WT nAChRs, I_max = 1.4 ± 0.5 μA, implying an expected density of 0.027 receptors/μm². When these oocytes were labeled with αBtxAlexa488, we measured 0.036 ± 0.01 puncta/μm² (range, 0.017–0.058/μm²; n = 3 oocytes) (Fig. 9 B).

In experiments with nAChR γ_eGFP, I_max = 1.5 ± 0.8 μA, leading to an expected receptor density of 0.029/μm². The measured puncta density was 0.020 ± 0.006/μm² (range, 0.013–0.032/μm²; n = 3 oocytes) (Fig. 9 B). Other oocytes from this nAChR γ_eGFP batch were also labeled with αBtxAlexa488. The measured puncta density for nAChR γ_eGFP labeled with αBtxAlexa488 was 0.059 ± 0.03/μm² (range, 0.018–0.126/μm²; n = 3 oocytes) (Fig. 9 B). Puncta with single-molecule profiles were observed at a lower density for relevant control oocytes: the measured puncta density for uninjected and unlabeled oocytes was 0.0003 ± 0.0003 puncta/μm² (range, 0.0–0.001/μm²; n = 4 oocytes) (Fig. S1 D).

The average puncta density for coinjection of nAChR β19′GGGU mRNA (25 ng/oocyte) and 12.5 ng YFaFS_ACCC-Lys(BODIPYFL) was 0.021 ± 0.003/μm² (range, 0.015–0.028; n = 5 oocytes). The expected puncta density was 0.068 puncta/μm² for this batch (I_max = 3.5 ± 0.2 μA, n = 4 oocytes) (Fig. 9 B). When these oocytes expressing nAChR β19′Lys(BODIPYFL) receptors were labeled with αBtxAlexa488 (Fig. S2 C), the puncta density was 0.045 ± 0.009/μm² (range, 0.027–0.07/μm²; n = 6 oocytes) (Fig. 9 B). Thus, as expected from the data regarding labeling and dye photobleaching (Figs. 3 A and 8 A), the di-labeling of nAChRs revealed receptors that went undetected in experiments incorporating Lys(BODIPFYFL). Control oocytes from the same batch were injected (as usual) with YFaFS_ACCC-Lys(BODIPYFL) (12.5 ng) (i.e., with tRNA-fUAA but no mRNA), to assess how unincorporated Lys(BODIPYFL) contributed to the observed puncta. In these control oocytes, puncta were detected at a much lower average density of 0.007 ± 0.002/μm² (range, 0.003–0.014/μm²; n = 5 oocytes) (Fig. 9 B and Fig. S1 B). When other similarly injected control oocytes were labeled with αBtxAlexa488, we measured a puncta density of 0.004 ± 0.002/μm² (range, 0.002–0.008/μm²; n = 3 oocytes) (Fig. 9 B and Fig. S2 D). Another type of control used uninjected oocytes incubated with αBtxAlexa488; the puncta density was 0.003 ± 0.002/μm² (range, 0.0–0.006; n = 3 oocytes) (Fig. 9 B and Fig. S2 E). Thus, the experimental values were all internally consistent within a factor of 3.2, which is acceptable, given the number of assumptions made. Two types of control yielded much lower puncta densities (Fig. 9 B).

Discussion

We performed imaging of Xenopus oocytes heterologously expressing nAChR β19′Lys(BODIPYFL) in the plasma membrane. Background signals were minimized by using the frameshift-suppressor tRNA (YFaFS_ACCC) (37,51,52). The frameshift strategy was crucial, because the more common nonsense-suppression strategy, using the amber suppressor tRNA (THG73), produced unacceptably high background signals (data not shown). The β19′ site is buried in the resting state of the receptor (Fig. 1 C), and the efficient incorporation of Lys(BODIPYFL) demonstrates an advantage over genetic FP insertions or posttranslational labeling, such as BODIPYFL-C3-MTS (Fig. 1 E) labeling of β19′C, which was not possible. Plasma membrane imaging of channels incorporating fUAAs with single-molecule TIRF microscopy may be considered a technique that extends beyond the incorporation of fUAAs in receptors and subsequent purification and reconstitution.

The puncta originating from nAChR β19′Lys(BODIPYFL) were brighter than those puncta from nAChR γ_eGFP, by an average factor of ∼2 (Fig. 5), in agreement with the known photophysics of BODIPYFL and eGFP in solution (2,15,36). This is surprising, considering that Lys(BODIPYFL) is incorporated in the interior of an ion channel (Fig. 1 C) and interacts with many other amino acids. Also, labeling with the highly specific fluorescent ligand, αBtxAlexa488, allowed us to confirm that nAChR molecules produced most of the puncta from oocytes expressing nAChR β19′Lys(BODIPYFL) and nAChR γ_eGFP receptors (Figs. 7 and 8). The αBtxAlexa488 studies suggest that ∼47% of the predicted Lys(BODIPYFL) are detected. Thus, similar to FPs (65,66) and quantum dots (67), there is a nonfluorescent fraction of molecules, but there are partially different reasons for each fluorophore. Dye photobleaching during synthesis, handling, expression, and focusing of a subset of nAChR β19′Lys(BODIPYFL) are the primary reasons for the dark nAChR fraction. Although our data on αBtxAlexa488 labeling might be limited by impurities in the commercial αBtxAlexa488, the resolved multistep photobleaching was consistent with the known 2:1 αBtxAlexa488/nAChR stoichiometry (59).

In general, the single-molecule labeling schemes based on genetically encoding (fUAA and eGFP) yield a ratio, (puncta density)/(receptor density expected from electrophysiology), between 0.5 and 2. The experiments with the tightly bound ligand, αBtxAlexa488, gave a ratio of 3–4. We consider this agreement quite satisfactory, considering the numerous assumptions 1–6 (above in Fluorescent Puncta Density and Current Density Analysis). The measured ratio would be changed in unknown directions by variable adherence. Although the puncta density differed between mono-labeled and di-labeled receptors because of photobleaching, many viable single fluorophores were still present, enabling parallel single-molecule measurements.

An important advantage of small-molecule fluorophores is that they are less likely than fluorescent proteins to perturb protein structure, function, and trafficking (see comparison of FP size in Fig. 1 A compared with Lys(BODIPYFL) in Fig. 1 B). This study describes a methodology to incorporate small-molecule fluorophores directly into nascent proteins, using UAA incorporation technology, and allows for the detection of single-membrane proteins in live cellular membranes. Site-specific and single-step fUAA incorporation will facilitate investigations of many membrane proteins, but are particularly useful in measuring the gating dynamics, colocalization, and trafficking in ion channels, because these channels provide complementary single-molecule measurements of function. The incorporation of fUAAs is particularly important for sites within α-helices (Fig. 1 D) and within other structural motifs that may not tolerate the large perturbation of fluorescent proteins (Fig. 1 A).

Finally, although the focus of our report is live-cell imaging, recent reports describe the super-resolution microscopy of fixed cells (45,68). The localization of fluorescent molecules with ∼2-nm resolution was reported. The resolution of single molecules is dependent on dye brightness (69). In addition, it was proposed that the power of super-resolution microcopy will be significantly enhanced with genetically encoded synthetic dyes into proteins (70). The puncta originating from nAChR β19′Lys(BODIPYFL) were brighter than those puncta from nAChR γ_eGFP, by an average factor of ∼2 (Fig. 5). Thus, Lys(BODIPYFL) is an attractive alternative to FPs for fixed-cell super-resolution microscopy.

Uncited references

(34).

Supplementary Material

Two figures are available at http://www.biophysj.org/biophysj/supplemental/S0006-3495(08)00012-X.