PI(4,5)P2 regulates various cellular functions, but its nanoscale organization in the cell plasma membrane is poorly understood. By labeling PI(4,5)P2 with a dual-color fluorescent probe fused to the Pleckstrin Homology domain, we describe a novel approach to study the PI(4,5)P2 spatial distribution in the plasma membrane at the nanometer scale.
Phosphoinositides in the cell membrane are signaling lipids with multiple cellular functions. Phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) is a determinant phosphoinositide of the plasma membrane (PM), and it is required to modulate ion channels, actin dynamics, exocytosis, endocytosis, intracellular signaling, and many other cellular processes. However, the spatial organization of PI(4,5)P2 in the PM is controversial, and its nanoscale distribution is poorly understood due to the technical limitations of research approaches. Here by utilizing single molecule localization microscopy and the Pleckstrin Homology (PH) domain based dual color fluorescent probes, we describe a novel method to visualize the nanoscale distribution of PI(4,5)P2 in the PM in fixed membrane sheets as well as live cells.
Phosphoinositides contribute to a small portion of total membrane lipids, but play critical roles in a variety of cellular processes. They include seven members derived from reversible phosphorylation or dephosphorylation of the inositol rings on the 3rd, 4th and 5th positions1. Phosphatidylinositol 4-phosphate (PI4P) and phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) are two major phosphoinositides that function relatively independently as the determinant lipids of cell plasma membrane (PM) 2,3. Phosphatidylinositol (3,4,5)-trisphosphate (PIP3) is much less abundant than PI4P and PI(4,5)P2, but it has unique functions in different cellular processes including cancer4 and diabetes5. These lipids have complex molecular interactions with their effectors and many other proteins. Therefore, it is crucial to understand the spatial organization of these phosphoinositides in the PM at the nanometer scale.
Mounting evidence has shown that protein complexes or molecule clusters in the confined PM areas can serve as signaling hotspots6. For example, syntaxin 1a, a key protein that regulates membrane fusion7-9, displays cluster organization in the PM. Distinct from the consensus view of the cluster organization of syntaxin 1a, the spatial distribution of phosphoinositides in the PM is controversial. PI(4,5)P2 distribution patterns range from uniform10-12, large patches13,14, to dense clusters14-18, depending on cell types and experimental methods used. The spatial organization of PI(4,5)P2 at higher resolution is also inconsistent. A study using stimulated-emission-depletion (STED) microscopy 19 has revealed a large number of dense PI(4,5)P2 nano-clusters (~73 nm in diameter) in the PM sheets of PC-12 cells 20. This result is different from studies using quick freezing electron microscopy (EM)21,22, an approach that preserves the intact PM structure of live cells much better than chemical fixation. The latter showed distinct pools of PI(4,5)P2; relatively concentrated PI(4,5)P2 in caveolae and coated pits as well as uniform distribution on the flat PM region. Moreover, nanoscale PI(4,5)P2 organization in membrane sheets may differ in live and fixed cells. Our recent work investigated this issue in both fixed and live INS-1 cells using single-molecule localization microscopy (SMLM)23.
SMLM is based on stochastically switching on only a small subset of fluorophores at any given time so that individual fluorophores can be localized with high precision. Many super-resolution imaging approaches have been developed using similar principles to surpass the diffraction limit of conventional light microscopy, such as photoactivation localization microscopy (PALM)24, fluorescence photoactivation localization microscopy (FPALM)25, stochastic optical reconstruction microscopy (STORM)26,27 and direct STORM (dSTORM)28. With photo-switchable or photoactivatable fluorophores (dyes or fluorescent proteins), SMLM techniques allow scientists to image biological structures at nanometer resolution24,29,30 with video-rate in living cells31,32.
Using PI(4,5)P2 as an example, we introduced the SMLM approach to study the nanoscale distribution of phosphoinositides on the PM. The PH domain of PLCδ1 (Phospholipase C δ1) that specifically binds to PI(4,5)P2 is a well-established probe for imaging PI(4,5)P2 sub-cellular distribution and dynamics33,34 in the PM. We have genetically tagged this domain with two fluorescent proteins, PAmCherry135 and iRFP36 to produce a dual-color fusion protein (iRFP-PAmCherry1-PHPLCδ1) (Figure 1A-B). PAmCherry1 serves as the photoactivatable probe of PH domain for PALM and iRFP serves as a general indicator to identify transfected cells before PALM acquisition. We apply this dual-color fluorescent probe for SMLM imaging in the fixed membrane sheets. For live PALM imaging, we tagged mEOS337 instead of PAmCherry1 to the PHPLCδ1 domain to generate the mEOS3.1-PHPLCδ1 probe for its better photon efficiency and brightness (Figure 1C-D).
SMLM imaging with these novel probes in the PM of insulin-secreting INS-1 cells38 has uncovered homogeneous labeling of PI(4,5)P2 in the majority of the PM regions as well as concentrated PI(4,5)P2 microdomains that are sparsely intermixed in the flat PM and some filopodia-like structures23. The nanoscale distribution of PI(4,5)P2 provides a structural base for rethinking how it functions in living cells.
1. Membrane Sheet Preparation and Fixation
2. Cell Culture Preparation for Live-cell Imaging with mEOS3.1-PHPLCδ1
3. PALM Image Acquisition of Membrane Sheets and Live Cells
4. SMLM Image Processing and Reconstruction
The localization uncertainty (σ) of our super-resolution system is 14.73 nm23. Direct comparisons between TIRF and PALM images demonstrated a significant improvement of spatial resolution. Figure 3A-B shows the representative PI(4,5)P2 TIRF images labeled with PI(4,5)P2 antibody and iRFP-PAmCherry1-PHPLCδ1 in typical membrane sheets and live cells. The images with conventional TIRF microscopy in membrane sheets are remarkably similar to those in intact live cells labeled with the enhanced green fluorescence protein (EGFP) tagged PH domains (EGFP-PHPLCδ1) (Figure 3C). All samples showed an even distribution of probes. In contrast, non-optimal fixation of the samples resulted in sharp dense PI(4,5)P2 clusters and a decrease in signal intensity (Figure 4). Under optimal fixation conditions, the super-resolution images of PI(4,5)P2 in fixed cells (Figure 5) revealed a homogenous distribution of probes in a significant portion of the PM with only limited concentration gradients. Some membrane patches enriched with PI(4,5)P2 probes were sparsely distributed and had various sizes. Live cell PALM images display a similar spatial distribution as fixed cells (Figure 6). Detailed analysis of PI(4,5)P2 signals over time results in fast dynamics in local areas, without significant changes of their abundance in broad areas.
Figure 1. Scheme for fluorescent probes used in this study. (A-B) iRFP-PAmCherry1-PHPLCδ1 probe used in the fixed membrane sheet experiments. During conventional TIRF imaging (A), a 640 nm laser is used to excite the iRFP (Ex: 692 nm; Em: 713 nm). TIRF image taken in this condition serves as a TIRF reference image for the super-resolution imaging obtained by PALM imaging (B). A 405 nm laser is used to photo-activate the PAmCherry1 fluorophore and a 561 nm laser is used to excite PAmCherry1 (Ex: 564 nm; Em: 595 nm) for PALM imaging. (C-D) mEos3.1-PHPLCδ1 probe used in the live cell experiments. (C) During conventional TIRF imaging, a 488 nm laser is used to excite mEOS3.1 (Ex: 506 nm; Em: 519 nm). (D) After photoconversion by a 405 nm laser, mEos3.1 turns into the red form (Ex: 573 nm; Em: 584 nm) and a 561 nm laser is used for PALM acquisitions. Please click here to view a larger version of this figure.
Figure 2. Scheme for membrane sheet preparation from INS-1 cells. (A) Place the coverslip with cultured cells facing down onto a PDL-coated coverslip and wait for 7 ~ 10 min at 4 °C to allow cell attachment to the PDL-coated coverslip. (B) Peel off the top coverslip with tweezers and fix the membrane sheet attached to the PDL pre-coated coverslip. (C) Image the samples with TIRFM and PALM. Please click here to view a larger version of this figure.
Figure 3. PI(4,5)P2 spatial organization is similar between membrane sheets and intact live cells under conventional TIRF microscope. (A) A typical TIRF image of membrane sheets fixed at 4 °C with 4% PFA and 0.2% GA. PI(4,5)P2 was labeled with PI(4,5)P2 specific antibody. (B) A membrane sheet from the INS-1 cell expressing iRFP-PAmCherry1-PHPLCδ1. (C) TIRF image of two intact live cells expressing EGFP-PHPLCδ1 at different levels. Scale bars: (A): 3 µm; (B) and (C): 5 µm. Please click here to view a larger version of this figure.
Figure 4. PI(4,5)P2 spatial organization in membrane sheets is sensitive to common fixation conditions. (A) Membrane sheets fixed at 37 °C with 4% PFA alone and labeled with PI(4,5)P2 specific antibody (as in Figure 3A). (B) Membrane sheet fixed at RT with PFA alone and labeled with PI(4,5)P2 specific antibody. Note that the dense clusters of PI(4,5)P2 probes are clearly visible under TIRF microscope, in contrast to the much more even fluorescence images shown in Figure 3A. Scale bars: A-B: 3 µm. Please click here to view a larger version of this figure.
Figure 5. PALM imaging of PI(4,5)P2 probes reveals their nanometer scale distribution in an INS-1 cell PM. (A) iRFP TIRF image of a membrane sheet from an INS-1 cell expressing iRFP-PAmCherry1-PHPLCδ1. (B) Corresponding PALM image of PM in the same region based on PAmCherry1 signal reconstruction. Note the homogeneous PI(4,5)P2 spatial distribution in major PM regions and several PI(4,5)P2 microdomains. (C) An enlarged view of the boxed region in (B). Arrows indicate sparsely distributed PM microdomains enriched with PI(4,5)P2 probes. Scale bars: A and B: 3 µm; C: 500 nm. Please click here to view a larger version of this figure.
Figure 6. PALM imaging in live INS-1 cells. (A) TIRF image of PI(4,5)P2 in a live INS-1 cell expressing mEos3.1-PHPLCδ1. The image was rapidly acquired in the green channel before PALM acquisition (35 °C). (B) Sequential live-cell PALM images at 10 sec interval. Insets show the intensity profiles of the local PI(4,5)P2 density along the same straight line 1 position at different times in (B) and (C). Note their large local intensity changes within 10 sec. (D) Time course of the average intensity changes of PI(4,5)P2 in the large area (box2, 3×3 µm) and small circles (3, 4, and 5, 500 nm diameter) in (B) during 5 min of PALM imaging (frame/10 sec). Note the rapid intensity fluctuations of local PI(4,5)P2 probes (circle 3, 4 and 5) compared to very small changes in the broad area (box 2). (E) Enlarged PALM images of the box2 region in (B) at indicated times. Arrows indicate PI(4,5)P2 enriched membrane patches under physiological conditions. Scale bars: C: 3 µm; E: 500 nm. Please click here to view a larger version of this figure.
For troubleshooting, two processes need extra attention: membrane sheet production and sample fixation. As described in the protocol, the incubation time of the coverslips in step 1.1.6 is important for membrane sheet production. Optimal incubation time under our experimental condition is 7-10 min (Figure 2). Longer than 10 min incubation will produce intact cells instead of the membrane sheets on PDL coverslips and shorter incubation will lead to less or no membrane sheets on the PDL-coated coverslips. As described in the protocol, the fixatives and temperature during fixation are critical for maintaining the PI(4,5)P2 distribution in the PM. Fixation at RT or use of 4%-PFA alone could distort the normal lipid distribution in the PM.
By applying PALM microscopy to membrane lipid research, we are able to observe the nanometer scale distribution of PI(4,5)P2, a key phosphoinositide that mediates many fundamental cellular activities. This spatial distribution of PI(4,5)P2 with limited concentration gradients in INS-1 cells provides a framework for rethinking lipid-protein interactions and local signaling events of PI(4,5)P2 in these cells. Moreover, the methods developed in this work can also be applied to other membrane phospholipid research with proper probes, thereby, offering novel tools to study phosphoinositides in biological processes.
The use of membrane sheets in this work bypasses two major concerns in phospholipid morphological studies: detergent treatment and potential signal contamination from the cytosol. Detergents often cause clustering and significant loss of PM phospholipid signal. The contamination of cytosolic signal is particularly problematic in the case of low abundance phosphoinositides on the PM23, such as PI(3,4,5)P3 and PI(3,4)P2. Membrane sheet samples are able to circumvent these problems without significantly disrupting structures associated with the plasma membrane, such as cortical actin meshwork and clathrin-coated pits42,43. The relative homogenous distribution of PI(4,5)P2 in the PM is in good agreement with other quick freezing EM studies using GST-PHPLCδ1 probes in the fibroblast membrane44.
It is important to note that improper sample processing conditions can generate misleading results. First, it is critical to perform the fixation steps at a lower temperature (4 °C) and use the fixative GA for membrane sheet production. As shown in Figure 3, warm temperature and PFA fixation without GA is not sufficient to fix phosphoinositides in cell PM. This could distort the intact PI(4,5)P2 distribution and generate sharp clusters that are not observed in live cells under physiological conditions. Second, the use of PAmCherry1 as the SMLM probe, rather than other probes, is pivotal for quantitative PALM imaging. The benefit of PAmCherry1 application comes from its well-characterized single molecular photo-physical properties35,40,45, such as brightness, high photo-activation efficiency and most of all, very limited photo-blinking. These properties enable us to eliminate potential cluster artifacts from photo-blinking and quantitatively analyze the molecular density of membrane PI(4,5)P2.
This approach also has its limitations. First, the membrane sheet method used in this study may not fully mimic the physiological distribution of PI(4,5)P2 because the cell is disrupted before imaging. However, our live PALM imaging shows similar relatively homogeneous distribution of PI(4,5)P2, supporting the results observed with membrane sheet samples. Second, as we discussed in our previous work23, SMLM requires imaging expertise and extra attention to avoid imaging artifacts that might arise from different processes, including the probes used, sample preparation and fixation, image sampling and reconstruction. Lastly, though the PH domain based probes and antibodies have been widely used in phosphoinositide studies46,47 it remains possible that not all PI(4,5)P2 in the membrane can be detected by this approach. For example, PI(4,5)P2 bound by other endogenous proteins may not be accessible to PH probes or antibodies, and this may cause an underestimation of PI(4,5)P2 due to the space hindrance of probe themselves. An alternative way of labeling PI(4,5)P2 would be using Top-Fluor PI(4,5)P248, a pre-labeled PI(4,5)P2 analog with a modification on the tail of original PI(4,5)P2. However, it can be converted rapidly into other phosphoinositide subtypes by fast live cell metabolism since its inositol ring is the same as endogenous PI(4,5)P2. This raises the concern whether this pre-labeled PI(4,5)P2 analog in live cells can faithfully represent PI(4,5)P2 rather than its metabolic products. Therefore, despite some limitations, PH domain based probes are still among the best probes that have been widely used to monitor PI(4,5)P2 distribution and dynamics on the PM of cells.
The future application of this methodology can be extended to other phosphoinositide studies, such as PI(3,4,5)P3 and PI(3,4)P2. In summary, the novel SMLM approach used here opens new ways to study phosphoinositide in cells. Using PI(4,5)P2 as an example, we demonstrate the unique properties of PALM imaging in the morphological and quantitative study of cell membrane molecules, as well as its drawbacks. This approach can be adapted to other molecules of interests and will have wide applications in cell biology.
The authors have nothing to disclose.
This work is supported by the National Institutes of Health (NIH) grants R01DK093953 (X.L.), P30NS069271, BRFSG-2014-07 (X.L.). C.J. is partially supported by American Heart Associate pre-doctoral fellowship (14PRE20380168, to C.J.).
Microscope | Nikon | Ti-U | |
sCMOS camera | Andor | Neo | |
Spinning disk | Yokogawa | CSU X-1, 10,000 rpm | |
High Power Monolithic Laser Combiner SP with 405, 488, 561 and 642 nm lasers. | Agilent | MLC400 | laser original powers & powers at the end of optical fiber) 405nm: 15 mW & 13 mW; 488nm: 45 mW & 42 mW; 561nm: 45 mW & 40 mW; 640nm: 35 mW & 16 mW. |
100X Objective | Nikon | APO 100X Oil, NA 1.49, WD 0.12mm | |
Nikon acquisition and analysis software(NIS Element) | Nikon | ||
Matlab | MathWorks | ||
Coverslip | Warner | 64-0714 | Round 18mm coverslip, #1.5 |
Fibronectin | Millipore | FC010-5MG | |
Lipofectamin 3000(transfection reagent) | Invitrogen | L3000-008 | |
Glutaraldehyde | Electron Microscopy Science | #16120 | |
PI(4,5)P2 1st antibody | Santa Cruz | sc-53412 | |
secondary antibody F(ab’)2-goat-anti-mouse Alexa Fluor 647 | Invitrogen | A-21237 | |
Tetraspeck beads | Invitrogen | T7279 | |
magnetic quick release imaging chamber | Warner Instruments | Cat#641994 | |
temperature controller | Warner Instruments | TC-344C | |
Poly-D-Lysine | Sigma | P6407 | |
Phosphate buffer saline | Sigma | P4417-100TAB | |
EGTA | Sigma | 3780 | |
NH4Cl | Sigma | A0171 | |
Goat serum | Sigma | G9023 | |
bovine serum albumin | Sigma | A7906 | |
NaCl | Sigma | S5886 | |
KCl | Sigma | P5405 | |
CaCl2*2H2O | Sigma | 223506 | |
MgCl2*6H2O | Sigma | M9272 | |
Glucose | Sigma | G7021 | |
HEPES | Sigma | H4034 | |
Paraformaldyhyde | Sigma | P6148 |