Photostable and Orthogonal Solvatochromic Fluorophores for Simultaneous In Situ Quantification of Multiple Cellular Signaling Molecules

Abstract

Ratiometric fluorescence sensors are powerful tools for direct quantification of diverse biological analytes. To overcome a shortage of solvatochromic fluorophores crucial for in situ ratiometric imaging of biological targets, we prepared and characterized a small library of modular fluorophores with diverse spectral properties. Among them, WCB and WCR showed excellent spectral properties, including high photostability, brightness, and solvatochromism, and are ideally suited for dual ratiometric imaging due to their spectral orthogonality. By conjugating WCB and WCR with protein-based lipid sensors, we were able to achieve robust simultaneous in situ quantitative imaging of two metabolically linked signaling lipids, phosphatidylinositol-4,5-bisphosphate and phosphatidylinositol-3,4,5-trisphosphate in live cells. This study shows that any combination of signaling molecules can be simultaneously quantified in a spatiotemporally resolved manner by ratiometric imaging with finely tuned solvatochromic fluorophores.

Graphical Abstract:

Introduction:

Due to its high sensitivity and non-invasive nature, fluorescence has been a popular choice for biological sensing and imaging^{1, 2}. However, quantification of a biological analyte using a fluorogenic sensor with a single emission band is subject to complications caused by many variables including instrumental parameters, heterogeneous biological environments, and photobleaching of the sensor^{3, 4}. To overcome these limitations, a wide range of ratiometric fluorescent sensors have been developed^4–6. Because of the analyte-induced changes in dual or multiple emission bands displayed by these fluorescent sensors and built-in referencing between emission bands, they allow robust quantification of biological targets devoid of the aforementioned complicating factors.

Ratiometric fluorescent sensors can be classified into two categories; small molecule-based and protein-based. Over the past two decades, a wide variety of small fluorescence molecules^{4, 6} and nanomaterials^{7, 8} have been developed for quantification and imaging of biological molecules. These approaches have been successfully applied to the quantification of small biological analytes^{4, 6}, such as metal ions and protons (i.e., pH); however, they have shown significant limitations in the quantification of large and complex biological targets, such as proteins and lipids, due to the small size-related inherent low specificity. Protein-based ratiometric sensors are generally constructed from a sensing unit and a pair of conjugated fluorescent proteins that undergo a major change in fluorescence resonance energy transfer when the sensing module binds to a target molecule^9–11. Although these genetically encoded biosensors offer high specificity and experimental convenience they often suffer from inherently low dynamic range, which ultimately results in low sensitivity^{12, 13}.

To take advantage of the high specificity of protein-based sensors and the high sensitivity of small molecule fluorescence probes, we^14–17 and others¹⁸ have developed a hybrid approach in which highly specific protein-based sensors are conjugated with a small fluorophore with desired spectral properties. Although a fluorophore can be genetically and thus homogenously introduced to the protein sensor as a part of a non-natural amino acid, this approach still poses major technical challenges due to low availability and low incorporation efficiency of large non-natural amino acids¹⁹. Thus, single-site or orthogonal multi-site chemical labeling remains a mainstay in hybrid sensor preparation¹⁸. Specifically, we have developed highly specific and sensitive ratiometric sensors for cellular lipids by chemically conjugating an environment-sensitive fluorophore with engineered lipid binding protein domains^14–17. These sensors have been successfully used for spatiotemporally resolved in situ quantification of diverse cellular lipids in live cells, providing new insight into the lipid-mediated cell regulation.

Solvatochromic fluorophores containing electron donor and acceptor groups in the same molecule are classical environment-sensitive fluorophores^{20, 21}. These so-called push-pull fluorophores have a highly dipolar excited state, and thus their fluorescence emission is highly sensitive to the polarity of solvent, typically showing a hypsochromic shift in a less polar environment, such as non-polar solvents or lipid membranes. Many of these chromogenic molecules, including 2-(dimethylamino)-6-acryloylnaphthalene (DAN), nitrobenzoxadiazole (NBD), and Nile-Red, show greatly enhanced intensity at the blue-shifted emission band²¹. Due to these favorable properties, solvatochromic push-pull fluorophores (referred to as solvatochromic fluorophores hereafter) are well suited for ratiometric sensing. For example, hybrid protein sensors labeled with DAN and Nile-Red derivatives have been successfully used for in situ quantification of lipids in live cells^14–17. Unfortunately, the number of known solvatochromic fluorophores is still small and many reported solvatochromic fluorophores have major drawbacks. In general, short-wavelength fluorophores, such as DAN, have low photostability whereas long-wavelength fluorophores, such as Nile-Red, suffer from weak solvatochromism and low water solubility²¹. These obstacles have limited the utility of solvatochromic fluorophores in ratiometric sensing.

Herein we report preparation and characterization of a small library of ready-for-protein-labeling solvatochromic fluorophores (Figure 1A), some of which show exceptional spectral properties, including high photostability, brightness, and solvatochromism, over a wide spectral range. Among these new fluorophores, WCB and WCR are equipped with not only favorable spectral properties but also non-overlapping (orthogonal) spectral ranges, which allow for simultaneous in situ quantitative ratiometric imaging of multiple biological targets, most notably two lipid species in live cells (Figure 1B).

Figure 1.

(A) Structures of new solvatochromic fluorophores (F1-F16) synthesized from diverse core scaffolds. The red circle shows fluorophores in the near IR region whose emission bands are shifted from red to orange/yellow regions in non-polar solvents or in hydrophobic lipid membranes. The blue circle includes fluorophores that show a green-to-blue hypsochromic shift in a non-polar environment. Fluorophores in the center (WCY, WCB and WCR) are solvatochromic fluorophores with ideal spectral properties. (B) A basic strategy for quantifying membrane lipids. Our lipid sensors (e.g., WCB-eENTH and WCR-eMyoX-tPH) undergo a hypsochromic shift in emission spectra upon binding to their target lipids, allowing ratiometric quantification of lipids through calibration.

Results and Discussion

Despite their potential utility in quantitative biological sensing and imaging, availability of solvatochromic fluorophores suited for ratiometric imaging is extremely limited. In particular, the DAN and Nile-Red pair is essentially the only spectrally non-overlapping combination that allows simultaneous dual ratiometric imaging but neither fluorophore has ideal spectral properties¹⁵. Ideal solvatochromic fluorophores for ratiometric imaging would possess high photostability and brightness²². They should also exhibit desirable solvatochromism (i.e., they should show both a large spectral shift and a large intensity change) and have narrow emission bands, which would result in spectral orthogonality with other fluorophores, hence allowing for simultaneous ratiometric imaging of multiple targets. For the preparation of protein-based hybrid sensors, they should also have good aqueous solubility for efficient protein labeling and show consistent spectrally properties whether they are free in solution or conjugated with a protein²². Our strategy for improving ratiometric imaging with a new generation of solvatochromic fluorophores hinges on first identifying the most desirable fluorophore motif, followed by building a library of molecules with diverse spectral and chemical properties through modular synthesis. Special emphasis was placed on identifying the structural motifs that can be modified to cover both the blue and red fluorescence ranges and exhibits desirable spectral properties when conjugated with a protein sensor. For protein–fluorophore conjugation, we primarily employed the engineered epsin1 ENTH domain (eENTH) with a single cysteine conjugation site, which has been extensively used for quantitative imaging of a cellular lipid, phosphatidylinositol-4,5-bisphosphate (PI4,5P₂)^{14, 15}. Due to its high specificity PI4,5P₂, eENTH-based ratiometric sensors specifically detect and respond to PI4,5P₂ in model and cell membranes under physiological conditions^{14, 15}. Successful ratiometric PI4,5P₂ sensing by this type of sensor critically relies on the membrane-induced hypsochromism of the conjugated fluorophore as a function of the PI4,5P₂ concentration in the membrane (Figure 1B)^{14, 15}, and thus serves as a rapid and sensitive screening assay for ideal ratiometric fluorophores.

We first built a small library of cysteine-reactive fluorophores by linking a Michael acceptor to the reported solvatochromic fluorophores (Figure 1A), including Nile-Red derivatives (F1–F3), benzochromenone (F4), styrylbenzoindole (F5), coumarin derivatives (F6, F10–F13), BODIBY derivative (F7), TCF derivative (F8), and fluorene derivatives (F9, F14–F16). These molecules showed solvatochromism in solution in response to the change in solvent polarity, which is consistent with the reported spectral properties of their parent molecules²³. When conjugated with eENTH, however, all but one compound showed simple fluorogenic behaviors (Figure S2). Only one fluorene derivative F15 showed a large hypsochromic shift when binding to PI4,5P₂-containing large unilamellar vesicles (LUVs). In general, fluorene derivatives including FR0, developed by Klymchenko and coworkers, have been reported to have favorable spectral properties in terms of photostability, molecular brightness, and solvatochromism^{21, 24}; however, they have not been utilized in quantitative ratiometric imaging in biological systems. We thus prepared a panel of derivates using the core fluorene scaffold of F15 as a template (see Scheme 1). Our primary goal was to generate two spectrally orthogonal fluorene derivates that can be used for simultaneous dual ratiometric imaging of two different lipid species.

Scheme 1.

Synthesis of FR0, WCY, WCB and WCR

First, we synthesized 3 (FR0), the core scaffold of F15, from commercially available 1 via intermediate 2 through a two-step procedure (Scheme 1). To tune the extent of π-conjugation and install a cysteine-reactive Michael acceptor, the formyl group of 3 was specifically transformed to a vinyl ketone in 5 (WCY), to an acrylohydrazone in 7 (WCB), and an acryloyloxime in 10 (WCR). It was found that despite the relatively small structural change from F15, 5 (WCY) showed much improved spectral properties over F15 as a PI4,5P₂ sensor. F15-conjugated eENTH showed a hypsochromic shift upon PI4,5P₂ binding but its shifted emission spectra have two peaks (Figure S2) presumably due to two different types of membrane interaction, which complicates the ratiometric calculation. In contrast, WCY-conjugated eENTH (WCY-eENTH) showed a large hypsochromic shift (i.e., the emission maximum shifted from 570 nm to 480 nm) with a single emission peak upon PI4,5P₂ binding, and a large fluorogenic increase at 480 nm in a PI4,5P₂ concentration-dependent manner. To test the utility of WCY-eENTH for in situ quantification of PI4,5P₂ in live cells, we performed in vitro ratiometric calibration of WCY-eENTH using giant unilamellar vesicles (GUVs) containing varying concentrations of PI4,5P₂ (Figure 2B). We then delivered WCY-eENTH into NIH 3T3 cells by microinjection and quantified cellular PI4,5P₂ in the cross-sectional images of cells using two separate emission channels (Figure 2C) as reported previously^{14, 15}. It is evident that the blue channel signal that represents PI4,5P₂-bound sensor molecules is predominantly localized to the plasma membrane (PM) wherein a large majority of PI4,5P₂ molecules exist²⁵, whereas the yellow channel signal that derives from both free and membrane-bound sensors is distributed both at PM and in the cytosol. Ratiometric conversion of intensity plots yielded a 3-dimensional spatial PI4,5P₂ concentration profile in NIH 3T3 cells (Figure 2D), and the spatially averaged PI4,5P₂ concentration at PM was 1.2 ± 0.1 mol%. The PI4,5P₂ concentration profile can also be determined in real time to yield complete spatiotemporal profiles (see Figure 4B for example).

Figure 2.

(A) Fluorescence emission spectra of WCY-eENTH in response to binding to POPC/POPS/PI4,5P₂ (80–x/20/x: x = 0–2 mol%) LUVs. The spectra were obtained spectrofluorometrically with the excitation wavelength set at 405 nm. The intensity values were normalized using the maximal intensity of the free sensor as the reference. The black line represents the spectrum of the sensor alone and the red line that with 0% PI4,5P₂ (i.e., POPC/POPS (80/20)). (B) The ratiometric calibration curve of WCY-eENTH for PI45P₂ quantification. WCY-eENTH was mixed with POPC/POPS/PI4,5P₂ (80–x/20/x: x = 0–2 mol%) GUVs and the fluorescence emission intensity in two separate channels were measured by a confocal microscope. Nonlinear least-squares analysis of the plot using the equation F_B/F_Y = (F_B/F_Y)_min + (F_B/F_Y)_max / (1 + K_d/[PI4,5P₂]) yielded K_d, (F_B/F_Y)_max, and (F_B/F_Y)_min values and the calibration curve was constructed using these parameters. F_B/F_Y, K_d, (F_B/F_Y)_max, and (F_B/F_Y)_min are the ratio of the fluorescence intensity in the blue channel to that in the yellow channel, equilibrium dissociation constant (in mol%), and the maximal and minimal F_B/F_Y values, respectively. The blue channel depicts the membrane-bound sensor whereas the yellow channel shows membrane-bound plus free sensors. Error bars indicate standard deviations calculated from three independent sets of measurements. (C) Representative blue-channel and yellow channel cross-sectional images of a NIH 3T3 cell microinjected with WCY-eENTH. (D) A spatially resolved PI4,5P₂ concentration profile at PM calculated from Figure 2C. The z-axis scale indicates mol% of PI4,5P₂. A pseudo-coloring scheme with red and blue representing the highest (i.e., 2 mol%) and the lowest (i.e., 0 mol%) concentration respectively is used to illustrate the spatial PI4,5P₂ concentration heterogeneity. Scale bars indicate 5 μm.

Figure 4.

(A) Four-channel images of representative NIH 3T3 cells in response to PDGF treatment. PI4,5P₂ and PIP₃ levels were simultaneously quantified by WCB-eENTH and WCR-eMyoX-tPH, respectively. Blue and orange channels depict membrane-bound sensors whereas green and red channels show membrane-bound plus free sensors. (B) Spatial PI4,5P₂ and PIP₃ concentration profiles calculated from Figure 4A. z-axes indicate PI4,5P₂ and PIP₃ concentrations in mol%. The pseudo-coloring schemes for PI4,5P₂ and PIP₃ are the same as in Figure 2D and Figure 3I, respectively. (C) Real-time changes of spatially averaged PI4,5P₂ (green closed circles) and PIP₃ (red closed squares) concentrations in the PM after PDGF treatment. Controls show PI4,5P₂ (green open circles) and PIP₃ (red open squares) concentrations under the same conditions minus PDGF. Scale bars indicate 5 μm.

Although WCY has excellent solvatochromic properties, its broad emission band covering 400–700 nm (Figure 2A) essentially precludes its use in dual ratiometric imaging. We thus further modified WCY to make its emission bands narrower and shift the emission peak to both lower and upper wavelengths such that orthogonal fluorophores can be obtained for simultaneous dual ratiometric imaging. We first introduced structural modification in WCY to generate WCB1 (see structure in Figure 1A and synthetic details in Scheme S1) by replacing the vinyl ketone moiety with an oxime ether connected to an acrylate via a glycerol linker. We then conjugated eENTH with WCB1 to produce WCB1-eENTH, which showed excellent solvatochromic properties in the shorter wavelength range, similarly to DAN-eENTH (See Figure S3) that has been extensively used for PI4,5P₂ quantification^{14, 15, 17}. When microinjected into NIH 3T3 cells, however, WCB1-eENTH produced a high level of cytosolic puncta (see Figure S4). This high background, which might derive from intracellular aggregation of the sensor, precludes its use in quantitative imaging. To overcome this major obstacle, we prepared a slightly different derivative WCB (see structure in Figure 1A and synthetic details in Scheme 1) by substituting the oxime functionality with a hydrazine moiety. WCB-eENTH showed solvatochromic behaviors comparable to those of WCB1-eENTH (Figure 3A and Table S1). Most importantly, it yielded clear blue and green channel images without any appreciable cytoplasmic background when microinjected into NIH 3T3 cells (Figure 3C). Quantitative analyses of these cross-sectional cell images using the ratiometric calibration curve of WCB-eENTH (Figure 3B) yielded a 3-dimensional spatial PI4,5P₂ concentration profile in NIH 3T3 cells (Figure 3D). The main problem associated with short-wavelength fluorophores is the necessity to excite them with a high energy laser that causes cell damage and rapid photobleaching, the latter being the main weakness of DAN. When continuously irradiated with a 405-nm laser, WCB-eENTH showed much higher photostability than DAN-eENTH (Figure 3E) although both show comparably high brightness at their blue-shifted wavelengths. The fluorescence intensity of DAN-eENTH was reduced below 5% within 5 min while that of WCB-eENTH remains >50% even after 10 min. Also, due to its relatively high fluorescence brightness after the hypsochromic shift, WCB-eENTH could be excited with low-power illumination, thereby preventing extensive cell damage. Collectively, WCB is an excellent new solvatochromic fluorophore for ratiometric imaging in the blue to green fluorescence range.

Figure 3.

(A) Fluorescence emission spectra of WCB-eENTH in response to binding to POPC/POPS/PI4,5P₂ (80–x/20/x: x = 0–2 mol%) LUVs. The spectra were obtained spectrofluorometrically with the excitation wavelength set at 380 nm. The black line represents the spectrum of the sensor alone and the red line that with 0% PI4,5P₂ (i.e., POPC/POPS (80/20)) (B) The ratiometric calibration curve of WCB-eENTH for PI4,5P₂ quantification. The calibration was performed as described for Figure 2B except that the blue channel represents the membrane-bound sensor. (C) Representative blue-channel and green channel cross-sectional images of NIH 3T3 cell containing WCB-eENTH. (D) A spatial PI45P₂ concentration profile calculated from Figure 3C. The z-scale indicates mol% of PI4,5P₂. The pseudo-coloring scheme is the same as in Figure 2D. (E) Relative photostability of WCB-eENTH and DAN-eENTH in NIH 3T3 cells. Total blue-channel fluorescence intensity at the PM of a fixed number (e.g., 15) of cells was counted as a function of time with continuous illumination. (F) Fluorescence emission spectra of WCR-eMyoX-tPH in response to binding to POPC/POPS/PIP₃ (80–x/20/x: x = 0–1 mol%) LUVs. The spectra were measured with the excitation wavelength set at 488 nm. The black line represents the spectrum of the sensor alone and the red line that with 0% PIP₃ (i.e., POPC/POPS (80/20)) (G) The ratiometric calibration curve of WCR-eMyoX-tPH for PIP₃ quantification. The calibration was performed as described for Figure 2B except that the ratio of fluorescence intensity in the orange channel to that in the red channel was used for analysis. (H) Representative orange- and red-channel cross-sectional images of NIH 3T3 cell containing WCR-eMyoX-tPH. (I) A spatial PIP₃ concentration profile calculated from Figure 3H. The z-scale indicates mol% of PIP₃. A pseudo-coloring scheme was used with red and blue representing the highest (i.e., 0.6 mol%) and the lowest (i.e., 0 mol%) concentration of PIP₃, respectively. (J) Relative photostability of WCR-eMyoX-tPH and NR3-eMyoX-tPH in NIH 3T3 cells. Total orange-channel fluorescence intensity at the PM of a fixed number (e.g., 15) of cells was counted as a function of time with continuous illumination. Scale bars indicate 5 μm.

To find a red or near infrared solvatochromic fluorophore that is compatible with WCB for simultaneous dual ratiometric imaging, we further extended the π-conjugation of the fluorene core 3. Multiple rounds of optimization resulted in WCR (see scheme 1 for synthesis), which showed desirable spectral properties. In solution, free WCR behaved like an ideal fluorophore with a large hypsochromic shift and high fluorescence brightness for a red fluorophore (see Figure S1 and Table S1). For protein conjugation with WCR, we employed the engineered tandem PH domain of myosin X (eMyoX-tPH) that has served as a sensor for phosphatidylinositol-3,4,5-trisphosphate (PIP₃)^{15, 17}. Conjugation of WCR with a single cysteine in eMyoX-tPH produced WCR-eMyoX-tPH, which showed a large hypsochromic shift (i.e., the emission maximum shifted from 660 nm to 600 nm) upon binding to PIP₃ in LUVs (Figure 3F). It should be noted that this type of solvatochromism is unprecedented for a red or near infra-red fluorophore, which is evident when WCR-eMyoX-tPH is compared with eMyoX-tPH conjugated with a Nile-Red derivative NR3 (see Figure S3)¹⁵. As was the case with WCB-eENTH, WCR-eMyoX-tPH was ratiometrically calibrated using GUVs containing varying concentrations of PIP₃ (Figure 3G). When microinjected into NIH-3T3 cells, WCR-eMyoX-tPH showed weak but clear PM localization (Figure 3H). No cytoplasmic puncta or background was observed. Ratiometric analysis of cross-sectional cell images in red and orange channels (Figure 3H) yielded a 3-dimensional spatial PIP₃ concentration profile (Figure 3I), with the average PIP₃ concentration 0.16 ± 0.04 mol% at PM. This low basal PIP₃ concentration at the PM accounts for weaker PM intensity of WCR-eMyoX-tPH than that of WCB-eENTH. Most importantly, WCR-eMyoX-tPH allows much more accurate ratiometric quantification of PIP₃ than NR3-eMyoX-tPH due to its significantly higher solvatochromism. Furthermore, when continuously irradiated with a 488-nm laser, WCR-eMyoX-tPH was much more photostable than NR3-eMyoX-tPH (Figure 3J). Collectively, WCR is a superb replacement for Nile-Red derivatives.

Most importantly, WCB and WCR have negligible spectral overlap (Figure 3A and 3F) and should thus be suited for dual ratiometric imaging. To test this notion, we performed simultaneous ratiometric quantification of PI4,5P₂ and PIP₃ in NIH 3T3 cells by WCB-eENTH and WCR-eMyoX-tPH, respectively. PI4,5P₂ and PIP₃ are metabolically linked signaling lipids. In response to specific cellular stimuli, such as growth factors, PI4,5P₂ at PM is phosphorylated by Class I phosphoinositide 3-kinase to PIP₃²⁶, which triggers a wide range of cellular responses, including cell growth and proliferation^{26, 27}. PIP₃ is then converted to PI4,5P₂ by a lipid phosphatase, PTEN^{28, 29}. Before cell stimulation, the basal concentration of PI4,5P₂ in the PM is much higher than that of PIP₃ (Figure 4A and 4B) as reported previously¹⁷. When cells were stimulated with platelet-derived growth factor (PDGF), the concentration of PIP₃ reached a maximum at 4 min and slowly returned to its basal concentration in 20 min (Figure 4C). The increase in the PIP₃ concentration was counterbalanced by the decrease in the PI4,5P₂ concentration throughout the time course, confirming that WCB-eENTH and WCR-eMyoX-tPH faithfully and accurately monitored the conversion between PI4,5P₂ and PIP₃ under physiological conditions. It should be noted that we recently reported similar results using DAN-eENTH and NR3-eMyoX-tPH¹⁵; however, due to low photostability of DAN and low solvatochromism of NR3, the sensors were excited by time-lapse two-photon stimulation and PIP₃ data were subjected to significant computational errors during the ratiometric conversion¹⁵. In contrast, dual quantitative imaging with WCB-eENTH and WCR-eMyoX-tPH allowed continuous one-photon stimulation of sensors for up to 10 to 30 min (depending on the laser power) in a conventional confocal microscope, which led to much more accurate and reproducible ratiometric calculation of lipid concentrations. Collectively, WCB and WCR are orthogonal solvatochromic fluorophores ideally suited for dual ratiometric imaging.

Conclusion

We have developed new environment-sensitive fluorophores ideally suited for ratiometric sensing and imaging. WCY, WCB and WCR, which are derived from the same fluorene core motif, all have highly desirable spectral and chemical properties including brightness, photostability, chromogenic and fluorogenic properties, and aqueous solubility. When chemically conjugated with protein-based lipid sensors, these fluorophores enabled accurate and reproducible in situ ratiometric quantification of membrane lipids. WCB and WCR serve as an excellent pair for dual ratiometric imaging, allowing for accurate simultaneous quantification of two signalling lipids due to their spectral orthogonality. The same approach can be applied to simultaneous in situ quantification of any combination of cell signaling molecules. Modular structures of WCB and WCR also allow for their easy derivatization to create a series of related variants of finely tuned spectral and chemical properties. Collectively, WCB and WCR represent a new generation of environmentally sensitive fluorophores, which should have significant utility in dual ratiometric imaging of diverse biological targets.

Experimental Procedures:

General Information:

All synthetic operations were carried out in oven- or flame-dried glassware under the argon or nitrogen atmosphere unless otherwise noted. Starting materials were purchased from commercial sources (Aldrich, TCI America or Acros) unless otherwise noted. Tetrahydrofuran (THF) was distilled over sodium and benzophenone under the nitrogen atmosphere. The reactions were monitored by thin-layer chromatography (TLC) visualized both by UV (254 and 365 nm) and by a relevant staining agent (KMnO₄ or vanillin) followed by subsequent heating. Flash chromatography was performed using the 60 Å silica gel (32−63 mesh) purchased from Silicycle Inc. Analytical thin layer chromatography (TLC) was performed using pre-coated (0.25 mm) silica gel 60 (particle size: 0.040−0.063 mm) purchased from Silicycle. Yields were calculated for chromatographically and spectroscopically pure compounds unless otherwise stated. ¹H NMR and ¹³C NMR spectra were recorded on a Bruker DRX-500 spectrometer. Multiplicities are indicated by s (singlet), d (doublet), t (triplet), q (quartet), m (multiplet). Coupling constants, J, are reported in Hz (Hertz). Electrospray ionization (ESI) mass spectra were recorded on a Waters Micromass Q-Tof Ultima in the University of Illinois at Urbana-Champaign. 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) and 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoserine (POPS) were from Avanti Polar Lipids. 1,2-Dipalmitoyl derivatives of PI4,5P₂ and PIP₃ were from Cayman Chemical Co.

Sensor Preparation

The sensors for PI45P₂ (eENTH)¹⁴ and PIP₃ (eMyoX-tPH)¹⁷ were expressed as GST-tagged proteins in BL21 RIL codon plus E. coli cells as described previously. Protein expression was induced overnight at room temperature with 0.5 mM (final concentration) isopropyl β-d-1-thiogalactopyranoside when the optical density of the media reached 0.6–0.8. Cells were harvested and pellets were suspended in 20 mM Tris buffer, pH 7.4, with 0.16 M NaCl, 1 mM tris(2-carboxyethyl) phosphine (TCEP) and 1 mM phenylmethylsulphonyl fluoride (PMSF), then lysed by sonication. The supernatant was incubated with Glutathione Resin (GenScript) for 2 h. The resin mixture was then poured into a small column and washed with 20 mM Tris buffer, pH 7.4, containing 0.16 M NaCl. After the resin became clear, the washing buffer was replaced with the labeling buffer (50 mM Tris-HCl, 300 mM NaCl, 1 mM TCEP, 50 mM arginine, 50 mM glutamic acid (pH 8.0)). After adding 100 μL of labelling fluorophore (e.g., WCB or WCR) (10 mg/mL in DMSO), the mixture was gently shaken at room temperature for 2 h. The resin was then washed with 20 mM Tris buffer, pH 7.4, containing 0.16 M NaCl and 5% DMSO until the free fluorophore was completely removed. The resin was then suspended in a digest buffer (20 mM Tris-HCl, 160 mM NaCl, 20 mM CaCl₂, 0.5 mM TCEP, 50 mM arginine, 50 mM glutamic acid (pH 7.4)), and 40 U of bovine α-thrombin (Haematologic Technologies)). After overnight, thrombin digest at 4°C, the sensor was eluted from the column and concentrated by centrifugation at 4°C.

Fluorimetry Measurements

All fluorescence measurements were performed with Horiba Flurolog-3 spectrofluorometer. Lipid sensors (typically 500 nM) were added to LUVs with various lipid compositions and the emission spectra of DAN/WCB1/WCB and NR3/WCR1/WCR were measured with the excitation wavelength set at 390 nm and 520 nm, respectively in a single-photon excitation mode. All solvents for fluorescence experiments were of spectroscopic grade. Absorption spectra were recorded on a Cary 300 Scan UV-Visible spectrophotometer (Varian) using 1 cm quartz cells. Stock solutions of fluorophores were prepared using dioxane (< 5% (v/v) of DMSO was added in case of poor solubility).

Lipid Vesicles Preparation:

LUVs were prepared by extrusion through the 100 nm polycarbonate filter in a Microextruder (Avanti Polar Lipids). Giant unilamellar vesicles (GUVs) were prepared by electroformation. The lipid mixture was spread onto the indiumtin oxide (ITO) electrode surface and the lipid was dried under vacuum to form a uniform lipid film. Vesicles were grown in a sucrose solution (350 mM) while an electric field (3 V, 20 Hz frequency) was applied for 4–5 h at room temperature.

Calibration of Lipid Sensors:

In vitro calibration of PI4,5P₂ or PIP₃ sensors was performed with POPC/POPE/POPS/PI/cholesterol/PI4,5P₂ (or PIP₃) (14–x/40/26/10/10/x: x = 0–2 mol%) giant unilamellar vesicles (GUV) as described previously^{14, 17}. These GUV were mixed with the sensors (0–500 nM) and fluorescence imaging was performed with a custom-modified six-channel FV3000 (Olympus) confocal laser scanning microscope. DAN- or WCB (WCB1)-labelled sensors were excited by the 405 nm laser source and the emission intensity was measured in two separate channels with the spectral detector setting of 425–470 nm (blue channel) and 515–535 nm (green channel), respectively. NR3- or WCR-labeled sensors were excited with the 488 nm laser source and the emission intensity was collected in two separate channels with the spectral detector setting of 540–620 nm (orange channel) and 630–660 nm (red channel). For each PI4,5P₂ (or PIP₃) concentration, at least 10 GUVs were selected for image analysis by Image-Pro Plus (Media Cybernetics, Inc.). All calibration curve fitting was performed by non-linear least-squares analysis with Origin (OriginLab Co.) using an equation: F_B/F_G = (F_B/F_G)_min + (F_B/F_G)_max / (1 + K_d/[PI4,5P₂]) (for DAN/WCB-labeled sensors). F_B/F_G, K_d, (F_B/F_G)_max, and (F_B/F_G)_min are the ratio of the fluorescence intensity in the blue channel to that in the green channel, equilibrium dissociation constant (in mol%), and the maximal and minimal F_B/F_G values, respectively. The same calibration was performed for NR3/WCR-based sensors using fluorescence intensity values in the yellow and red channels.

Cell culture and in situ cellular imaging:

NIH 3T3 cells were seeded into 100 mm round glass–bottom plates (MatTek) and grown at 37 °C in a humidified atmosphere of 5% CO₂ in Dulbecco’s modified Eagle’s medium (DMEM) (Life Technologies) supplemented with 10% (v/v) fatal bovine serum (FBS), 100 U/mL penicillin G, and 100 mg/mL streptomycin sulfate (Life Technologies) and cultured in the plates for about 24 h before lipid quantification. All cell lines were cultured bi-weekly and stocks of cell lines were passaged no more than ten times for use in experiments. PDGF stimulation was performed with 50 ng/mL PDGF-BB (final concentration). Typically, 20–30 femtoliter of 5–10 μM sensor solution was microinjected into the cell to reach the final cellular concentration of about 100 nM. For simultaneous PI4,5P₂ and PIP₃ quantification in NIH 3T3 cells, an equimolar mixture of WCB-eENTH and WCR-eMyoX-tPH was delivered into the cells. All image acquisition and imaging data analysis were performed as described for GUV calibration. Detailed data analysis was performed as described previously^{14, 17}. The three–dimensional display of local lipid concentration profile was obtained using the Surf function in MATLAB.