Measuring Lipid Packing of Model and Cellular Membranes with

Jun 6, 2014 - Max Planck Institute of Molecular Cell Biology and Genetics, 01307 ... biological membranes and is involved in many membrane processes...
1 downloads 0 Views 2MB Size
Article pubs.acs.org/Langmuir

Measuring Lipid Packing of Model and Cellular Membranes with Environment Sensitive Probes Erdinc Sezgin,*,† Tomasz Sadowski,† and Kai Simons* †

Max Planck Institute of Molecular Cell Biology and Genetics, 01307 Dresden, Germany S Supporting Information *

ABSTRACT: The extent of lipid packing is one of the key physicochemical features of biological membranes and is involved in many membrane processes. Polarity sensitive fluorescent probes are commonly used tools to measure membrane lipid packing in both artificial and biological membranes. In this paper, we have systematically compared eight different probes to measure membrane lipid ordering. We investigated how these probes behave in small unilamellar liposomes, phase-separated giant unilamellar vesicles, cellderived giant plasma membrane vesicles, and live cells. We have tested the order sensitivity of a variety of measurable parameters, including generalized polarization, peak shift, or intensity shift. We also investigated internalization and photostability of the probes to assess probe potential for time-lapse live cell imaging. These results provide a catalogue of properties to facilitate the choice of probe according to need.



INTRODUCTION The membrane raft concept posits a structural and functional lateral heterogeneity in eukaryote cell membranes.1 The existence and functionality of membrane subcompartmentalization are no longer controversial; however, the exact physicochemical nature of membrane domains in specific contexts is still unclear. Membrane heterogeneity in terms of lipid packing can be quantitatively investigated by the use of probes the fluorescent properties of which are sensitive to membrane order. Laurdan is the most commonly used environment-sensitive dye whose emission depends on the polarity of the environment.2−5 It has two emission peaks with maxima at 440 and 490 nm whose relative intensity changes in ordered versus disordered membranes. This bimodal nature of Laurdan fluorescence emission is quantified by a relative index called generalized polarization (GP),6 which is a relative measure of membrane lipid packing/order. C-Laurdan (CL), a derivative of Laurdan, was developed as a more photostable alternative for membrane order measurement.7 Later, a carboxylated Laurdan derivative, CL2, was presented as a “raft-on” probe, which emits only when embedded in ordered membrane regions and increases emission intensity with increasing lipid packing.8 Finally, a sulfonated CL2 derivative (SL2) was developed as a “raft-on” probe with minimum internalization.9 In parallel, Di-4-ANEPPDHQ, a probe designed initially as a voltage-sensitive dye, has also been applied to measure lipid packing.10 Recently, three different derivatives of Di-4ANEPPDHQ (Di4), namely, Di-4-ANEQ(F)PTEA (F), Di-4AN(F)EPPTEA (FE), and Di-4-ANEP(F2)PTEA (F2) (collectively referred to as ANEP), were produced and used to measure the lipid packing of the plasma membrane.11 © 2014 American Chemical Society

As the number and diversity of polarity sensitive probes has increased in recent years, it has become necessary to make a systematic study of the applications and limitations of these tools.12 Here we investigate how the environment-sensitive dyes behave in small liposomes, phase-separated giant unilamellar vesicles (GUVs), giant plasma membrane vesicles (GPMVs), and live cells. We have tested the sensitivity to membrane composition/packing of three different measures: (i) GP, (ii) maximum peak shift, and (iii) maximum peak intensity sensitivity. In addition, we have tested the cellular internalization and the photostability of the probes to determine their potential for time-lapse live cell imaging.



RESULTS AND DISCUSSION We compared Laurdan, C-Laurdan (CL), C-Laurdan2 (CL2), S-Laurdan2 (SL2), Di-4-ANEPPDHQ (Di4), Di-4-ANEQ(F)PTEA (F), Di-4-AN(F)EPPTEA (FE), and Di-4-ANEP(F2)PTEA (F2) whose structures are shown in Figure 1. First, we checked the emission spectra of the eight tested probes in liposomes of different compositions (varying phospholipids and cholesterol content). The liposome compositions tested included homogeneous membranes composed of a single lipid specie [DOPC (1,2-dioleoyl-snglycero-3-phosphocholine) and POPC (1-palmitoyl-2-oleoylsn-glycero-3-phosphocholine) for different disordered membranes, DPPC (1,2-dipalmitoyl-sn-glycero-3-phosphocholine) for the gel-phase membrane], laterally homogeneous binary mixtures of a phospholipid with cholesterol (Chol) [such as DPPC:Chol and SM (N-palmitoyl-D-erythrosphingosylphosReceived: April 1, 2014 Revised: June 5, 2014 Published: June 6, 2014 8160

dx.doi.org/10.1021/la501226v | Langmuir 2014, 30, 8160−8166

Langmuir

Article

cholesterol was assessed by comparing probe emission spectra of pure DOPC liposomes to DOPC liposomes with 25% or 50% cholesterol. The temperature dependence of emission spectra was also explored at 20, 30, and 40 °C. We fixed the most disordered lipid environment (pure DOPC liposomes) as our reference system, to which we compared the probe response in other lipid environments. Several aspects of dye emission changed as a function of membrane environment, including the intensity of the maximum peak, the maximum peak wavelength, and the relative intensity of the order versus disordered peak. We have quantified and compared all of these changes between the reference and the tested environment: (1) GP shift = GPlipid − GPDOPC, (2) peak shift (Δλmax) = λmax,lipid − λmax,DOPC, and (3) intensity maximum ratio = Imax,lipid/ Imax,DOPC GP Sensitivity of Order-Sensitive Dyes in Various Membranes. Generalized polarization (GP) has commonly been used to measure the lipid packing of the biological membranes.5,13−17 For Laurdan, it is calculated as GP =

I440 − I490 I440 + I490

(1)

with 440 nm as the maximum wavelength in the gel phase (such as DPPC) and 490 as the maximum wavelength in the disordered membranes (such as DOPC).6 Thus, GP compares the relative packing of a given membrane as the ratio of the emission at these two wavelengths. GP can be applied to any probe if the maximum peaks are determined properly. Although each probe will yield different numerical GP values, the GP sensitivity could still be measured for each probe by comparing the GP shift with respect to the GP of a reference membrane system (DOPC in our case). For calculation of GP for the various dyes, we chose the two wavelengths at which a given probe exhibited the most prominent differences between the most disordered and the most ordered biologically relevant lipid environment [DOPC and DPPC:Chol (50:50), respec-

Figure 1. Structure of the tested environment sensitive probes.

phorylcholine):Chol for ordered membranes], and a ternary mixture (DOPC:DPPC:Chol) that would be expected to show ordered/disordered phase coexistence. The influence of

Figure 2. GP sensitivity of environment-sensitive probes to lipid composition, cholesterol, and temperature variations. (a) Exemplary normalized spectra of CL in POPC, DPPC:Chol (50:50), DPPC, SM:Chol (50:50), DOPC:DPPC:Chol (40:40:20), and DOPC liposomes. (b) GP difference between the given liposome system and DOPC liposomes. (c) Exemplary normalized spectra of CL in pure DOPC, DOPC:Chol (75:25), and DOPC:Chol (50:50) liposomes. (d) GP difference between the given liposome system and pure DOPC liposomes. (e) Exemplary normalized spectra of CL in DOPC:DPPC:Chol (40:40:20) liposomes at 20, 30, and 40 °C. (f) GP difference between 20 °C and the given temperature. Bars are standard errors from at least three measurements. (*Negative GP values in Di4 and FE in DPPC membrane is not reflective of the real order of the DPPC membrane, as these probes are likely excluded from the gel phase. It could possibly be reflective of a probe micelle/domain formation or the probe emission in aqueous solvent.) 8161

dx.doi.org/10.1021/la501226v | Langmuir 2014, 30, 8160−8166

Langmuir

Article

Almost all tested probes could distinguish between DPPC and SM in binary mixtures. We conclude that, generally, CL appears to be the most sensitive dye for measuring GP changes in membranes with varying acyl chain compositions. Laurdan and SL2 follow CL, while the others show poorer sensitivity. Cholesterol Variations and GP. We compared the GP sensitivity of the probes to differing amount of cholesterol by measuring the GP value in DOPC, DOPC:Chol (75:25), and DOPC:Chol (50:50) membranes. Raising cholesterol content in a membrane resulted in increased GP values for all tested probes, as expected from the known ordering effect of cholesterol.13 In all cases the response is proportional to the cholesterol concentration (see Figure 2c,d). GP shifts are at least twice as high for DOPC:Chol (50:50) than the ones for DOPC:Chol (75:25). The strongest response was again measured with CL. Laurdan, CL, and SL2 also showed high sensitivity to cholesterol amount. The ANEP dyes showed poor sensitivity, with F2 being the weakest. Temperature Variations and GP. We compared the sensitivity of the probes to temperature variations. To do this, we measured the GP value of the phase-separated ternary DOPC:DPPC:Chol system at 20, 30, and 40 °C (Figure 2e), in addition to DOPC and DPPC vesicles (Supplementary Figure S3, SI). All probes showed a similar trend of GP decrease with increasing temperature (Figure 2e,f) with different sensitivities. Laurdan and CL show the strongest response. The other probes showed poor sensitivity. Sensitivity of Maximum Emission to Membrane Order. Conventionally, changes in the maximal emission wavelength have not been used a measure of lipid packing/ ordering, due to the usefulness of the ratiometric quantification of the bimodal emission of Laurdan. However, many ordersensitive probes show a continuous λmax shift in addition to (or instead of) reciprocal intensity changes between two discrete peaks, suggesting that the extent of the λmax shift could reliably report membrane packing. Therefore, we investigated the change in λmax of all probes relative to the λmax of that probe in pure DOPC membranes, as a function of acyl chain compositions, cholesterol levels, and temperature. Acyl Chain Variations and λmax Shift. Di4, F, F2, and FE produced a variety of λmax values (Figure 3a), in contrast to Laurdan and CL, which showed a large, discrete shift between ordered and disordered membranes (see Figure 2a for exemplary CL spectra), as expected.18 However, in contrast to the discrete 50 nm peak shift between ordered (λmax = 440) and disordered membranes (λmax = 490) that might be expected from a perfect bimodal dye, we observed small λmax variations between different membranes. A small shift, for instance, was observed in POPC liposomes relative to DOPC liposomes (Figures 2a and 3b). Additional λmax variations between the different ordered membranes (Figures 2a and 3b) demonstrate that λmax of Laurdan and CL could report small, but potentially important, membrane packing differences. Considering that CL2 and SL2 have been described as intensity-based probes, we did not expect to see a peak shift response for these probes and did not observe one in any lipid environment, except those containing SM [both CL2 and SL2 reproducibly showed a λmax shift of ∼10 nm in SM:Chol (50:50) and DOPC:SM:Chol (40:40:20)] (Figure 3b). The λmax of Di4, F2, and FE was red-shifted in a manner roughly proportional to the number of unsaturated acyl chains in the lipid mixtures, being lowest for POPC, intermediate for

tively]. We tested probe responses to variations in acyl chain, cholesterol content, and temperature. Acyl Chain Variations and GP. The emission curves of Laurdan and CL in both DOPC (two unsaturated acyl chain) and POPC (one unsaturated acyl chain) were clearly distinguishable from each other (Figure 2a,b). POPC has a higher GP (i.e. higher order/packing), as expected from the number of unsaturated bonds within the fatty acid chains in these lipids (Figure 2a,b). Both probes are also sensitive to the differences between two different liquid-ordered membranes (DPPC:Chol and SM:Chol). Both Laurdan and CL can partition into gel-phase membranes;13 surprisingly, the gelphase DPPC membrane showed a lower or equal GP to liquidordered DPPC:Chol. Although both probes were broadly similar across membrane systems, CL appeared to be a more sensitive probe, as the GP shift (GPlipid − GPDOPC) with CL is higher than with Laurdan (Figure 2b). CL2 and SL2 were initially introduced as “raft on” probes.8 It was shown that the intensity at the peak wavelength increases with increasing order of the membrane, and these probes do not yield a notable signal in disordered membranes. Being intensity based, they are not meant to be evaluated via GP, a ratio-based and intensity-independent quantity. However, we observed that their spectra also consist of two emission maxima with the red-shifted, disordered peak, being relatively weak even in disordered membranes [Supplementary Figure S1, Supporting Information (SI)]. Despite this weakness, this peak was detectable and could thus be used to calculate to measure GP. Figure 2a,b show that CL2 and SL2 show GP sensitivity to lipid composition, comparable to Laurdan. In contrast to the Laurdan derivatives, Di4 and its derivatives yielded nearly undetectable GP shifts in POPC liposomes relative to DOPC, implying that these dyes cannot detect the order difference between these two membrane systems. In DPPC liposomes, Di4, FE, and F2 exhibit GP shifts close to 0 (even negative values for Di4 and FE), presumably due to much lower signal-to-noise ratio of the spectra of these probes (see Supplementary Figure S2a, SI). There are two possible explanations for this observation, either (1) these probes are physically excluded from gel-phase membranes or, alternatively, (2) these probes are present in the gel phase, but their emission is somehow quenched in higher order membranes. Even at higher temperature (30 °C), the intensity of these probes (especially of Di4 and FE) was negligible in DPPC membranes. At 40 °Cwhere the DPPC membrane becomes fluidthe emission intensity increases remarkably (Supplementary Figure S2b, SI). Despite the structural similarity, F showed notable signal in DPPC vesicles. In binary liposomes (DPPC:Chol and SM:Chol, each 50:50), GP shifts of Laurdan and derivatives were both in the range of the GP shift in pure DPPC liposomes and slightly higher in SM:Chol liposomes. Surprisingly, the emission spectra of Laurdan, CL, CL2, and SL2 were indistinguishable between the liquid-ordered DPPC:Chol (50:50) membranes and pure DPPC gel-phase membranes. Meanwhile, for Di4 derivatives the presence of 50% cholesterol in a DPPC membrane makes a dramatic difference, presumably due to the fact that the probes are able to be incorporated into the membrane and/or have a higher emission intensity as cholesterol fluidizes the DPPC membrane (Supplementary Figure S2, SI). Unlike the Laurdan derivatives, the GP shifts for Di4 derivatives in SM:Chol (50:50) were slightly lower than in DPPC:Chol (50:50). 8162

dx.doi.org/10.1021/la501226v | Langmuir 2014, 30, 8160−8166

Langmuir

Article

and DOPC:DPPC:Chol (40:40:20) on the other. However, while CL2 and SL2 in SM:Chol (50:50) yielded the greatest GP value, their Imax change is only intermediate between those of POPC and the other ordered membrane mixtures. This behavior mirrors the λmax response of CL2 and SL2 in SMcontaining lipid environments and suggests that these dyes are generally sensitive to membrane order but also specifically the presence of sphingomyelin. The Imax values of Laurdan and CL were not as responsive to membrane order as CL2 and SL2. In this modality, POPC could not be distinguished from DOPC (Figure 4a,b) and the

Figure 3. Peak shift sensitivity of environment sensitive probes to lipid composition, cholesterol, and temperature. (a) Exemplary normalized spectra of Di4 in POPC, DPPC:Chol (50:50), DPPC, SM:Chol (50:50), DOPC:DPPC:Chol (40:40:20), and DOPC liposomes. (b) Peak shift between the given liposome system and DOPC liposomes. (c) Exemplary normalized spectra of Di4 in pure DOPC, DOPC:Chol (75:25), and DOPC:Chol (50:50) liposomes. (d) Peak shift between the given liposome system and pure DOPC liposomes. Bars are standard errors of at least three measurements.

the binary and ternary mixes, and highest for DPPC:Chol (50:50). Cholesterol Variations and λmax Shift. The λmax shift of ANEP derivatives to cholesterol variation was slightly weaker than for variations in acyl chain composition, albeit still progressive, reproducible, and statistically significant (Figure 3c,d). For Laurdan and CL, the λmax shift in the presence of cholesterol was expectedly large but not bimodal, with 25% cholesterol inducing an intermediate shift to 50% cholesterol (Figure 3c,d). λmax shift was negligible for SL2 and CL2 (Figure 3c,d). Temperature Variations and Peak Shift. Surprisingly, no significant peak shift in response to temperature changes can be observed in any lipid composition (Supplementary Figure S4, SI). Emission Intensity as a Function of Membrane Order. Laurdan is used almost exclusively for GP measurements, which have the great advantage of being ratiometric and intensityindependent. Intensity-based measurements are inherently more error prone, as they require tight control of membrane:dye ratio, excitation light intensity, and a variety of optical settings to ensure that intensity differences are reporting the physicochemical nature of the membranes. However, the raft-on probes CL2 and SL2 allow clean intensity-based order analysis due to their high sensitivity and moderate signal-tonoise ratio.8,9 Although these probes can be used to evaluate GP (Figure 2), their utility as intensity-based probes would be an additional advantage. To this end, we tested all order probes for maximal emission intensity (Imax) as a function of membrane environment. We calculated the intensity change as percent difference relative to a pure DOPC membrane. Acyl Chain Variations and Imax. As expected, the strongest response in maximum emission peak intensity was observed for CL2 and SL2. Consistent with GP measurements (Figure 2), a disordered POPC membrane was clearly discernible from DOPC on the one hand and from DPPC, DPPC:Chol (50:50),

Figure 4. (a) Exemplary spectra of SL2 in POPC, DPPC:Chol (50:50), and DOPC liposomes. (b) Percent intensity change between the given liposome system and DOPC liposomes. (c) Exemplary spectra of SL2 in pure DOPC, DOPC:Chol (75:25), and DOPC:Chol (50:50) liposomes. (d) Percent intensity change between the given liposome system and DOPC liposomes. Bars are standard errors of at least three measurements.

intensity changes did not reflect the membrane composition in terms of acyl chain saturation. Similarly, the emission intensity response of ANEP to varying lipid environments is small and somewhat inconclusive because of large standard errors in the measurements (Figure 4a,b). Cholesterol Variations and Maximum Peak Intensity. Cholesterol variations could also be probed with CL2 and SL2 reproducibly by using Imax. Surprisingly, Di4, F2, FE, and Laurdan also showed remarkable intensity changes, albeit with large standard errors (Figure 4c,d). CL showed no response while F responded to increased cholesterol with decreased emission intensity, unlike all other probes (Figure 4c,d). Probe response to temperature was inconclusive and mostly nonreproducible (Supplementary Figure S5, SI). Emission of the Probes in Phase-Separated Vesicles. GUVs of ternary mixtures (DOPC:SM:Chol or DOPC:DPPC:Chol) are often used as models for biological membrane heterogeneity, as they show macroscopic two-phase coexistence.7,10,11 Recent studies showed that these coexisting phases in GUVs are highly ordered/disordered,16 to an extreme extent relative to the more moderate differences between phases in biological membranes.13,19 Since small differences in membrane order between coexisting phases might be highly biologically relevant, it was important to compare the ordersensitivity of the probes in phase-separated GUVs and cellderived GPMVs. To probe microscopic variation in dye spectral properties, we employed a 32-channel gallium arsenide 8163

dx.doi.org/10.1021/la501226v | Langmuir 2014, 30, 8160−8166

Langmuir

Article

Figure 5. Spectral imaging of phase-separated GUVs and GPMVs. (a) Montage of CL-doped phase-separated GUV and GPMV images taken with a 32-channel spectral GaAsP detector with 8.8 nm wavelength interval. (b) λ-Coded GUV image of and the spectrum obtained from the montage shown in part a. (c) λ-Coded CL-doped GPMV image of and the spectrum obtained from the montage shown in part a. (d) CL-doped GUV image taken by standard confocal microscopy filters and PMT detector. (e) CL-doped GPMV image taken by standard confocal microscopy filters and PMT detector.

phosphide (GaAsP) detector array to carry out the spectral imaging instead of filter-based confocal microscopy (see SI for details). Figure 5a shows the montage of images at various emission wavelengths of GUVs and GPMVs doped with CL. In GUVs, at lower emission wavelength, the ordered phase is very prominent. As the emission wavelength increases, the disordered phase becomes clear, thus allowing separate visualization and quantification of the two complementary phases. The profiles were slightly different in GPMVs, which showed clear signal in both phases through a large band of emission wavelengths. Figure 5b shows the spectral unmixing and the spectra of the CL-doped GUVs shown in Figure 5a. It is clear that two phases are present. Figure 5c shows the same for GPMVs. The spectral unmixing looks similar to that of GUVs, though with more ordered-like (i.e. blue-shifted) disordered domains. The spectra of two phases are clearly separable, but not as distant as in GUVs [see Supplementary Figure S6 (SI) for the spectra of all the probes in phaseseparated GUVs and GPMVs]. Parts d and e of Figure 5 show the confocal images using conventional confocal filters of the CL-doped GUV and GPMV, respectively. As seen from the figure, in GUVs, two channels show counter-staining, while in GPMVs, more costaining was observed. Internalization of the Probes in Live Cells. When studying the plasma membrane with fluorescent probes, it would be important to avoid internalization of the probes to confine probe signal exclusively to the plasma membrane. Thus, probes that do not permeate the plasma membrane should be used. On the other hand, while studying the intracellular membrane structures such as mitochondrial or ER membranes, it is necessary for the probes to get internalized. Therefore, we explored the cellular internalization of the probes in live cells. Figure 6 shows three exemplary probes from high to minimum internalization. CL, for instance, internalizes massively (Figure 6a), while FE remains plasma-membrane-enriched (Figure 6b). SL2 appears to be the most efficient probe to stain the plasma membrane with minimum internalization (Figure 6c; see Table 1 for the internalization of other probes). Probe Photostability. Temporal, live cell imaging is essential to study the dynamics of cellular membranes; therefore, photostability of fluorescent probes is vital for dynamic studies. We tested the photostability of the probes in

Figure 6. Internalization of the probes: (a) Laurdan, C-Laurdan, CL2, and F show high internalization; (b) FE, F2, and Di4 show intermediate internalization; and (c) SL2 shows almost no internalization.

vesicles. We scanned a line through the equatorial plane of GUVs composed of DOPC with a single photon laser and recorded the intensity vs time profile for 1 min. From this scan, by using a previously reported scanning FCS fitting algorithm,20 we obtained the bleaching profile of the probes (see Supplementary Figure S7 of the SI for the depiction of the measurement). Figure 7a shows the exemplary bleaching profiles of FE, CL, and CL2. As seen from Figure 7a,b, some of the probes such as Laurdan, CL2, and SL2 have very low photostability; thus, they cannot be used for time-lapse imaging with confocal microscopes. However, it is important to note that they may still be amenable to two-photon imaging. CL, Di4, and F showed intermediate photostability, which suggests that they can be used with confocal microscopy with moderate laser powers. FE and F2 appeared to be the most photostable probes and the best candidates for long-term time-lapse imaging (Figure 7b). 8164

dx.doi.org/10.1021/la501226v | Langmuir 2014, 30, 8160−8166

Langmuir

Article

Table 1. Summary of the Probe Properties

a

peak shift sensitivity

intensity sensitivity

gel-phase exclusion

temperature sensitivity

λexc

λordered

λdisordered

Laurdan

385

440

490

high

N/Aa

intermediate

no

high

CL

385

440

490

high

N/Aa

intermediate

no

high

CL2 SL2 Di4 F

385 385 488 385

480 480 565 430

590 590 605 470

high high intermediate intermediate

noneb noneb high intermediate

high high low low

no no yes partial

intermediate intermediate intermediate intermediate

F2 FE

488 488

610 565

635 610

low intermediate

low intermediate

low low

yes yes

low low

GP sensitivity

usability with fluorescent proteins RFP, mCherry, mKate RFP, mCherry, mKate mKate mKate CFP RFP, mCherry, mKate CFP, GFP CFP

internalization

photostability

high

low

high

intermediate

high minimum intermediate high

low low intermediate intermediate

intermediate intermediate

high high

Not applicable. bExcept with SM.

Figure 7. Photostability of the tested probes.



CONCLUSION Here, we systematically tested a number of probes used to measure membrane lipid packing in synthetic liposomes, giant unilamellar vesicles, giant plasma membrane vesicles, and live cells. We first identified several metrics by which to probe lipid packing, namely, generalized polarization, emission maximum wavelength shift, and maximal emission intensity. With each measure, we tested how the probes respond to variations in temperature and membrane composition with respect to acyl chain saturation and cholesterol amount. Next, we checked how well the dyes probe membrane heterogeneity in GUVs and cellderived GPMVs. Finally, we investigated internalization in live cells, as well as the photostability of the probes for temporal imaging (please see Table 1 for a summary). It is important to start with the fact that the emission characteristics of the tested probes varied dramatically; Laurdan and C-Laurdan are bimodal, CL2 and SL2 changed their maximum emission peak intensity considerably, and ANEP derivatives changed the maximum emission wavelength continually. Although some probes performed consistently better than others, the optimal probe choice depends to some extent on the experimental modality being used. In the liposome experiments, we found that CL is the most sensitive probe for GP measurements, although SL2 also successfully probed acyl chain and cholesterol variations in the membranes while using GP calculations. ANEP probes showed low GP sensitivity; however, it must be noted that the wavelength selection is crucial for these probes, since they are not bimodal; i.e., reference wavelengths for GP calculation are not obvious, unlike with Laurdan or CL. Thus, wavelength selection may vary the GP sensitivity of ANEP probes dramatically. For measuring the order-dependent shift in λmax, Di4, FE, and F

showed the greatest segregation between the various membrane compositions. Finally, SL2 and CL2 were optimal for intensitybased order measurements, although these measurements produced by far the greatest experimental error, as expected (see Figure 4). An important observation was that Di4 and FE showed no signal in the gel phase, which makes them potentially interesting probes for studies involving coexistence of gel/liquid phases. To probe lateral order variation in phase-separated membranes, we applied GaAsP detector spectral imaging with our panel of polarity-sensitive probes. Using this powerful technique, we have made the first microscopically resolved observations of order-sensitive dye spectra on a single vesicle and lateral membrane domains within a vesicle. Comparing the spectra of the probes between coexisting phases of cell-derived GPMVs (Figure 5 and Supplementary Figure S6, SI) suggests that CL, FE, and Di4 can successfully probe cellular membrane heterogeneity and therefore that this spectral imaging approach should have wide utility for live cell membrane studies. Internalization into cells is an important issues for probes designed for plasma membrane imaging. On the basis of our observations, SL2 is the best probe with regard to plasma membrane staining (Figure 6). However, photostability, another issue for these probes, makes SL2 less preferable for confocal microscopy, due to its fast photobleaching (Figure 7). This systematic study presents a comprehensive report on the behavior of the polarity-sensitive probes in different environments spanning from simple liposomes to live cells (Table 1). All existing probes have benefits and limitations: (1) CL is a good candidate to measure GP, but high internalization makes it difficult to use it with live cell experiments; (2) SL2 is a very efficient membrane label, but it suffers from lower photostability and spectral sensitivity; and (3) Di4 and FE are 8165

dx.doi.org/10.1021/la501226v | Langmuir 2014, 30, 8160−8166

Langmuir

Article

probe for lipid raft imaging: C-Laurdan. ChemBioChem 2007, 8 (5), 553−559. (8) Kim, H. M.; Jeong, B. H.; Hyon, J.-Y.; An, M. J.; Seo, M. S.; Hong, J. H.; Lee, K. J.; Kim, C. H.; Joo, T.; Hong, S.-C.; Cho, B. R. Two-photon fluorescent turn-on probe for lipid rafts in live cell and tissue. J. Am. Chem. Soc. 2008, 130 (13), 4246−4247. (9) Lim, C. S.; Kim, H. J.; Lee, J. H.; Tian, Y. S.; Kim, C. H.; Kim, H. M.; Joo, T.; Cho, B. R. A two-photon turn-on probe for lipid rafts with minimum internalization. ChemBioChem 2011, 12 (3), 392−395. (10) Jin, L.; Millard, A. C.; Wuskell, J. P.; Dong, X. M.; Wu, D. Q.; Clark, H. A.; Loew, L. M. Characterization and application of a new optical probe for membrane lipid domains. Biophys. J. 2006, 90 (7), 2563−2575. (11) Kwiatek, J. M.; Owen, D. M.; Abu-Siniyeh, A.; Yan, P.; Loew, L. M.; Gaus, K., Characterization of a new series of fluorescent probes for imaging membrane order. PLoS One 2013, 8 (2). (12) Klymchenko, A. S.; Kreder, R. Fluorescent probes for lipid rafts: From model membranes to living cells. Chem. Biol. 2014, 21 (1), 97− 113. (13) Kaiser, H.-J.; Lingwood, D.; Levental, I.; Sampaio, J. L.; Kalvodova, L.; Rajendran, L.; Simons, K. Order of lipid phases in model and plasma membranes. Proc. Natl. Acad. Sci. U. S. A. 2009, 106 (39), 16645−16650. (14) Owen, D. M.; Rentero, C.; Magenau, A.; Abu-Siniyeh, A.; Gaus, K. Quantitative imaging of membrane lipid order in cells and organisms. Nat. Protoc. 2012, 7 (1), 24−35. (15) Sezgin, E.; Kaiser, H.-J.; Baumgart, T.; Schwille, P.; Simons, K.; Levental, I. Elucidating membrane structure and protein behavior using giant plasma membrane vesicles. Nat. Protoc. 2012, 7 (6), 1042− 1051. (16) Sezgin, E.; Levental, I.; Grzybek, M.; Schwarzmann, G.; Mueller, V.; Honigmann, A.; Belov, V. N.; Eggeling, C.; Coskun, U.; Simons, K.; Schwille, P. Partitioning, diffusion, and ligand binding of raft lipid analogs in model and cellular plasma membranes. Biochim. Biophys. Acta-Biomembr. 2012, 1818 (7), 1777−1784. (17) Yu, W. M.; So, P. T. C.; French, T.; Gratton, E. Fluorescence generalized polarization of cell membranes: A two-photon scanning microscopy approach. Biophys. J. 1996, 70 (2), 626−636. (18) Bagatolli, L. A.; Sanchez, S. A.; Hazlett, T.; Gratton, E. Giant vesicles, Laurdan, and two-photon fluorescence microscopy: Evidence of lipid lateral separation in bilayers. Biophotonics, A 2003, 360, 481− 500. (19) Sezgin, E.; Schwille, P. Model membrane platforms to study protein−membrane interactions. Mol. Membr. Biol. 2012, 29 (5), 144− 154. (20) Mueller, P.; Schwille, P.; Weidemann, T. Scanning fluorescence correlation spectroscopy (SFCS) with a scan path perpendicular to the membrane plane. In Fluorescence Spectroscopy and Microscopy: Methods and Protocols; Engelborghs, Y., Visser, A., Eds.; Humana Press: Totowa, NJ, 2014; Vol. 1076, pp 635−651.

good probes to exploit the continuous peak shift as a measure of lipid packing and they can be used in confocal microscopy due to their fairly high photostability. Further, their spectral separation in GPMVs proves that they are able to probe the cell membrane heterogeneity. The disadvantage of Di4 and FE is their broad emission spectrum, making combinations with other fluorescent labels (especially commonly used fluorescent proteins like GFP and RFP) problematic. Further, they have only a 40−45 nm shift between the most ordered and disordered membranes, thus showing a relatively small difference between phases and thus limited order resolution. To sum up, the existing measures/probes have their advantages and shortcomings; thus, continuing efforts should be employed in finding more sensitive, photostable, and preferably noninternalized probes, as well as new techniques/ measures to investigate membrane lipid packing.



ASSOCIATED CONTENT

S Supporting Information *

Supporting figures and experimental procedures. This material is available free of charge via the Internet at http://pubs.acs. org/.



AUTHOR INFORMATION

Corresponding Authors

*E.S. e-mail: [email protected]. *K.S. e-mail: [email protected]. Author Contributions †

These authors contributed equally.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank Huw Colin-York and Ilya Levental for careful reading of the manuscript, Dr. Bong Rae Cho for providing us CL2 and SL2 and Dr. Leslie M. Loew for providing us F, FE, and F2 probes. We thank Paul Mueller for providing us the photobleaching analysis tool. We also thank Light Microscopy Facility at Max Planck Institute for Cell Biology and Genetics.



REFERENCES

(1) Lingwood, D.; Simons, K. Lipid rafts as a membrane-organizing principle. Science 2010, 327 (5961), 46−50. (2) Golfetto, O.; Hinde, E.; Gratton, E. Laurdan fluorescence lifetime discriminates cholesterol content from changes in fluidity in living cell membranes. Biophys. J. 2013, 104 (6), 1238−1247. (3) Lim, C. S.; Cho, B. R. Two-photon probes for biomedical applications. BMB Rep. 2013, 46 (4), 188−194. (4) Sachl, R.; Stepanek, M.; Prochazka, K.; Humpolickova, J.; Hof, M. Fluorescence study of the solvation of fluorescent probes prodan and Laurdan in poly(ε-caprolactone)-block-poly(ethylene oxide) vesicles in aqueous solutions with tetrahydrofurane. Langmuir 2008, 24 (1), 288−295. (5) Sanchez, S. A.; Tricerri, M. A.; Gratton, E. Laurdan generalized polarization fluctuations measures membrane packing micro-heterogeneity in vivo. Proc. Natl. Acad. Sci. U. S. A. 2012, 109 (19), 7314− 7319. (6) Parasassi, T.; Gratton, E.; Yu, W. M.; Wilson, P.; Levi, M. Twophoton fluorescence microscopy of Laurdan generalized polarization domains in model and natural membranes. Biophys. J. 1997, 72 (6), 2413−2429. (7) Kim, H. M.; Choo, H.-J.; Jung, S.-Y.; Ko, Y.-G.; Park, W.-H.; Jeon, S.-J.; Kim, C. H.; Joo, T.; Cho, B. R. A two-photon fluorescent 8166

dx.doi.org/10.1021/la501226v | Langmuir 2014, 30, 8160−8166