Laurdan: Clarifying Photophysics and Advancing the Characterization of Extracellular Vesicles

Abstract

The environment‐sensitive fluorescence emission of Laurdan (6‐lauroyl‐2‐dimethylaminonaphthalene) underlies its widespread application in studies of biomolecular assemblies. The first part of this invited contribution clarifies the relevant photophysical concepts of Laurdan. Based on an overview of the literature on the characterization of extracellular vesicles (EVs) using Laurdan, the second part advances Laurdan's use by introducing time‐resolved emission spectra (TRES) for EV characterization. As diagnostic applications require distinguishing between EV subtypes, and given that tetraspanins are major protein constituents of EVs, we characterized Laurdan fluorescence in EVs isolated from various tetraspanin knockout cell lines. To resolve differences between these EVs, we recorded TRES and analyzed them using the standard time‐dependent fluorescence shift approach. In addition, we applied spectral phasor analysis to the TRES data—a combination that, as far as we are aware, has not previously been explored. This combined approach offers clear advantages over previously employed steady‐state methods, as it can distinguish EVs with different protein profiles and provides comprehensive information on both compositional heterogeneity and lipid phase state. Inspection of the TRES phasor trajectories further suggests that Laurdan is sensitive not only to the lipid matrix but also to nonlipid constituents of EVs, such as membrane‐associated proteins.

Biophysical properties of extracellular vesicles (EVs) reflect their composition and membrane organization. We introduce a novel approach using time‐resolved emission spectra (TRES) phasor analysis of the polarity‐sensitive probe Laurdan to characterize membranes with complex compositions. TRES‐phasor reveals differences between EVs derived from wild‐type and tetraspanin knock‐out cells invisible to conventional steady‐state methods.

1. Introduction

Laurdan was first introduced by Georges Weber in 1979 as a polarity‐sensitive fluorescent probe designed to investigate the physical state of lipid membranes [1]. Due to its pronounced environment‐sensitive emission, especially the strong Stokes shift in response to membrane phase and packing, Laurdan quickly became a cornerstone in membrane biophysics. Since then, it has likely become the most widely used fluorescent dye for probing lipid phase behavior and membrane headgroup dynamics in both model systems and biological membranes. Its broad adoption is reflected in a range of reviews [2, 3, 4]. Laurdan fluorescence can be analyzed using direct spectral measurements (absorption, emission) or the time‐dependent fluorescence shift (TDFS) approach [5]. Semiempirical tools like generalized polarization (GP) [6] and phasor analysis [7] simplify the interpretation of spectral shifts or lifetime changes. Regardless of the approach, meaningful analysis depends critically on two factors: (a) the location and orientation of the chromophore within the bilayer and (b) its underlying photophysics. Laurdan's depth and orientation are membrane phase‐dependent. In fluid‐phase bilayers, it localizes near the sn‐ 1 carbonyl region [8] and shifts slightly toward the interface upon excitation [9]. Simulations show deeper and more restricted conformations in gel‐phase membranes [10], consistent with pressure‐induced depth shifts [11]. Bacalum et al. recently confirmed multiple ground‐state populations using polarization‐resolved GP and time‐resolved fluorescence [12]. Despite Laurdan's widespread use, some fundamental aspects of its photophysics reflected in its absorption and emission behavior seem unresolved. These include phase‐dependent absorption features, interpretation of spectral shifts, and the mechanistic basis of its environment sensitivity.

This motivates the first part of this invited contribution, in which we review the literature on the absorption properties of Laurdan and reconcile a new photophysical model of Laurdan's spectral relaxation based on picosecond time‐resolved infrared (TRIR) [13] spectroscopy with previous TDFS‐based conclusions on membrane structure and hydration [3]. In the second part, we demonstrate the enhanced sensitivity of TRES‐based approaches for the characterization of extracellular vesicles compared with commonly used steady‐state methods. Our data indicate that the TRES approach: (a) can distinguish between EV subtypes, (b) gives information about the EV membrane heterogeneity, and (c) reveals the impact of proteins on Laurdan's readout.

2. Part A: Clarifying Laurdan's Photophysics

2.1. Absorption Properties of Laur‐Dan

Understanding dan dye absorption behavior begins with their response in isotropic environments. In the gas phase and in nonpolar solvents, these dyes exhibit broad, featureless absorption bands centered around 350–370 nm. Detailed spectral decomposition by Viard et al. identified three underlying electronic transitions contributing to this band: two allowed π–π* transitions (labeled La and Lb) and one symmetry‐forbidden π–π* transition [14]. Their temperature‐dependent analysis in frozen ethanol revealed shifts in intensity and spectral position among these transitions, illustrating how the local environment modulates absorption. Notably, by decreasing the temperature to −77°C, a shoulder near 390 nm becomes visible, attributed to the Lb transition. This decomposition describes Laurdan's behavior well in isotropic solvents and fluid‐phase lipid membranes, where the chromophore resides in a disordered and hydrated lipid matrix. However, this framework fails for Laurdan in more ordered and less hydrated gel‐phase membranes, where the 395 nm band not only intensifies but becomes dominant (see Figure 3 in Parassasi et al. [6])—an effect not explained by solvent polarity or isotropic models. The emergence of this sharp feature suggests the presence of a specific ground‐state conformer stabilized by the ordered lipid environment. Such an interpretation is in line with Osella et al., who used molecular dynamics simulations of Laurdan in 1,2‐dipalmitoyl‐sn‐glycero‐3‐phosphocholine (DPPC) bilayers to identify two stable ground‐state conformers [15]. These differ in the orientation of the lauroyl carbonyl relative to the naphthalene core and adopt distinct alignments in the gel phase. This conformational heterogeneity was experimentally supported by Bacalum et al., who resolved multiple Laurdan populations in gel‐phase vesicles [12]. As previously suggested [16], a subset of chromophores that are both immobilized and preoriented within the tightly packed and less hydrated headgroup region of the gel membrane enable transitions that are spectrally unfavorable or suppressed in disordered phases to become prominent. Notably, this interpretation diverges from the traditional assignment of the ∼395 nm band to the Lb π–π* transition, as observed in solvents. In gel membranes, the emergence of this band reflects not only the nature of the electronic transition but also orientational selection by the hydrated lipid matrix. It is thus better understood as the signature of a stabilized, preoriented ground‐state conformer with a distinct absorption cross‐section—rather than a universal phase‐independent spectral feature.

2.2. Emission Properties of Dan Dyes

The environment‐sensitive Stokes shift of dan dyes is best characterized using TRES. The evolution of the emission maximum extracted from TRES provides the TDFS. Unlike steady‐state measurements, TDFS directly quantifies both the magnitude and kinetics of spectral relaxation following excitation, offering a dynamic readout of the local nanoenvironment. A body of our work has established that in the case of Laurdan the total Stokes shift (Δ$\nu$) reflects the polarity and hydration level of the probe's environment— most notably, the degree of headgroup hydration in fluid lipid membranes—while the relaxation time (τ) is sensitive to the local mobility of hydrated lipid acyl groups, particularly of the sn‐1 chain (please see the Method section for the parameters definition). This dual sensitivity has made TDFS a powerful tool for mapping both structural and dynamical features of biomembranes. A comprehensive summary of this approach and its implications for Laurdan and other solvatochromic dyes is provided in Scollo et al. [3]. A key experimental finding supporting this model is the pronounced isotope effect observed in the hydration layer when H₂O is replaced by D₂O. In fluid membranes, the nanosecond relaxation time (τ) increases in D₂O, consistent with longer nanosecond average lifetimes of single H/D‐bonds between a water molecule and a sn‐1 carbonyl oxygen, along with longer lasting water bridges between lipid molecules in D₂O [17]. Beranová et al. demonstrated this convincingly in phosphatidylcholine bilayers, providing strong evidence that Laurdan's TDFS is directly coupled to the timescale of local hydrogen bond dynamics [18]. The D₂O effect disappears in bilayers containing positively charged headgroups paired with fluoride (F^‐) counterions. This striking absence suggests a disruption of the stable water‐lipid hydrogen bonding network at the membrane interface, thereby eliminating the coupling between duration of hydrogen bonds and Laurdan's TDFS [19].

Initially, these findings were interpreted within the framework of classical solvent relaxation, in which excitation polarizes the surrounding medium, which then responds through dipolar reorientation. Our recent reinterpretation is grounded in picosecond TRIR spectroscopy using two dan dyes, which provides direct structural insight into the chromophore's vibrational evolution following excitation [13]. Regardless of the solvent, electron density changes upon vertical excitation are delocalized over the whole molecule and do not entail any significant charge separation. In polar/H‐bonding solvents, this optically prepared S1(ππ*) state then undergoes a solvent‐driven intramolecular electron‐density redistribution from NMe2 to the C=O group accompanied by a lowering of the S1 excited‐state energy. Ultimately, the system arrives at a relaxed planar intramolecular charge transfer (ICT) S1 state. Experimentally, this relaxation process in polar/H‐bonding solvents is manifested by an emergence, growth and small upshift of an excited state $\nu$(C=O) IR band and by a TDFS, which both occur on a picosecond to nanosecond time scale with comparable kinetics. At the same time, the ca. 80 cm^–1 downshift of the $\nu$(C=O) IR band relative to the ground state demonstrates the ICT character of the relaxed lowest excited state. Both TRIR relaxation kinetics and TDFS are strongly solvent‐dependent, reflecting changes in solvation and dielectric response as well as viscosity. The relaxation dynamics in slowly responding, structured environments—such as glycerol or the hydrated acyl chain region of fluid‐phase bilayers—reveal the presence of a slow step that is observable by TDFS but not by TRIR. It is conceivable that the last stages of slow dielectric relaxation keep decreasing the S1(ICT)→ S0 transition energy through electrostatic interactions without significantly changing the electronic structure of the S1(ICT) state. This interpretation differs from the classical view of the dynamic Stokes shift, which assumes a fluorescent excited state whose electronic and molecular structures do not change in the course of relaxation, but the emission energy decreases as the system evolves along the solvation coordinate and the fluorescent state gets stabilized by outer‐sphere interactions with the relaxing solvent. However, relaxation of dan dyes involves profound intramolecular reorganization of electron density within the lowest excited state, whose dynamics and magnitude are governed by the solvent polarity together with the dynamics of solvation and hydrogen‐bond formation. Consequently, although intramolecular in origin, the relaxation kinetics is predominantly determined by the overall environment (solvent) relaxation that includes polarization and reorientation of solvent molecules, hydrogen‐bond formation of the solvent OH groups with the C=O group, and structuring of the first solvation shell. Thus, while the primary origin of the excited‐state energy shift is intramolecular, the kinetics remain, governed by the external polar and hydrogen‐bonding environment. This reconciles the new photophysical model with previous TDFS‐based conclusions confirming that Laurdan´s emission remains a quantitative probe of phospholipid headgroup mobility and hydration—now understood with greater mechanistic clarity.

3. Part B: Applications of Laurdan in Characterization of Extracellular Vesicles

3.1. What Are Extracellular Vesicles?

Extracellular vesicles (EVs) are membrane‐enclosed particles released by all cell types into the extracellular space, having a growing inventory of fundamental biological functions, including intercellular communication, removal of cellular waste or modulation of immune responses in both physiological and pathological contexts [20]. EV secretion is a highly regulated process, resulting in heterogeneous populations of complex membrane systems whose molecular composition is dynamic and responsive to environmental stimuli and cellular signals, thereby reflecting the physiological state of the donor cell [21, 22, 23].

Based on their size, molecular composition, and mechanism of biogenesis, EVs are classified into several subtypes [24]. These include exosomes, ectosomes, and apoptotic bodies along with smaller groups of vesicles with specialized functions, such as oncosomes [25, 26]. Exosomes are the smallest EVs, typically ranging from 30 to 150 nm in diameter. Their formation is a multistep process involving the inward budding of the limiting membrane of endosomes, leading to the formation of multivesicular bodies (MVBs) containing intraluminal vesicles (ILVs). Upon fusion of MVBs with the plasma membrane, ILVs are released into the extracellular space as exosomes. Secreted exosomes can be internalized by recipient cells, thereby modulating cellular responses and phenotypes [24, 27]. Ectosomes (also known as microvesicles), typically ranging from 100 to 1000 nm in diameter, are generated through the outward budding and fission of the plasma membrane. Their cargo primarily consists of cytosolic components [28]. Apoptotic bodies represent the largest group of EVs with a size ranging from 1 to 5 μm. They are formed upon apoptosis by cell fragmentation into membrane‐bound vesicles that are consequently removed by phagocytic cells [29].

Concerning lipid composition, EV membranes are enriched in cholesterol, with 15–60% of total lipid content reported in EVs isolated from mammalian cells [30]. Other major lipid classes include phosphatidylcholine (11%–33%), sphingomyelin (8%–23%), phosphatidylserine (1%–16%), and phosphatidylethanolamine (1%–24%). Minor lipid species such as ceramides, hexosylceramides, phosphatidylinositol, and diacylglycerols have been detected in amounts up to ∼1%. In addition, ether lipids have also been reported, contributing up to 5% of the total lipid content. This lipid composition suggests a relatively rigid EV membrane, primarily due to the high content of lipids that tend to form ordered phases, namely cholesterol and sphingolipids. Disease‐associated alterations in EV lipid content, particularly in cholesterol and sphingolipids, have been reported in neurodegenerative disorders, cancer, viral infections, and metabolic diseases [31, 32, 33, 34, 35]. Such changes are expected to directly affect membrane rigidity and hydration, yet these parameters remain poorly characterized using standard biochemical approaches.

In addition to the lipids, EV membranes are characterized by a high abundance of integral membrane proteins, most notably members of the tetraspanin family, including CD9, CD63, and CD81 [24, 36, 37]. Beyond their widespread use as EV markers, tetraspanins actively contribute to membrane organization through the formation of tetraspanin‐enriched microdomains (TEMs), mediated by interactions with cholesterol, sphingolipids, and partner proteins [38, 39]. As multipass transmembrane proteins, tetraspanins are expected to influence local lipid packing, membrane stiffness, and interfacial hydration. However, their specific contribution to the biophysical properties of EV membranes remains largely unexplored.

A wide range of analytical techniques has been developed to characterize the structural, molecular, and functional attributes of EVs, as outlined in the “Minimal Information for Studies of EVs” guidelines (MISEV 2023 [40]). These methods aim to define key parameters such as particle size and concentration (e.g., flow cytometry, nanoparticle tracking analysis), morphology (e.g., transmission electron microscopy, cryo‐electron microscopy (cryo‐EM)), surface marker expression, and cargo composition (e.g., Western blotting, mass spectrometry, nucleic acid sequencing). Although essential for EV identification and classification, these approaches provide limited insight into membrane biophysical properties and do not readily allow the contributions of individual membrane components to be disentangled. However, the considerable potential of EVs for biomedical applications, particularly in diagnostics and therapeutic delivery [27, 41, 42, 43, 44], underscores the need for a comprehensive understanding of how pathological conditions influence the biochemical and biophysical properties of EVs. Here, we propose that environment‐sensitive fluorescent probes provide a direct experimental route to address these questions.

3.2. What Is Known in the Literature on Applications of Laurdan for the Characterization of EVs?

As outlined above, there is a fundamental need to characterize both the fluidity and—if possible—the hydration of EV membranes. Owing to its amphiphilic nature, Laurdan readily incorporates into EV membranes, and direct spectral measurements are straightforward to perform. To the best of our knowledge, the first application of Laurdan to probe EV membrane fluidity was reported by Parolini et al. in 2009 [45]. Specifically, they used Laurdan GP and its dependence on excitation wavelength to characterize EVs purified from the medium of metastatic melanoma cells (please see the Method section for the GP definition). By comparing EVs released under neutral and acidic conditions with parental cell membranes, Parolini et al. showed that EVs exhibit higher GP values, consistent with their enrichment in sphingomyelin, ganglioside GM3, and cholesterol. Additionally, the excitation dependence of the GP was examined. In both cells and EVs, in either buffered or acidic conditions, the GP decreased with increasing excitation wavelength, indicating that the membranes were in the liquid–crystalline, fluid phase (Ld) state. However, the authors admit that the obtained GP values larger than 0.25 for the EVs might indicate the existence of liquid‐ordered (Lo) microdomains. This work provided early evidence that EV membranes are more ordered than their parent plasma membranes.

Following this pioneering study, several studies have employed Laurdan fluorescence—along with its derivative, C‐Laurdan—to investigate the biophysical properties of EV membranes. Due to its simplicity, GP derived from steady‐state emission spectra has remained the most commonly used readout. These studies are summarized in Table 1. It should be emphasized that GP is a ratiometric, empirical parameter that does not capture the full complexity of the emission spectrum and is sensitive to instrumental settings. Consequently, GP values should be interpreted through comparison with appropriate reference systems measured under identical experimental conditions, rather than on the basis of absolute values alone.

TABLE 1.

Applications of Laurdan and C‐Laurdan GP for the characterization of extracellular vesicles (EVs). Two studies were excluded from the table: Seegobin et al., which examined bovine milk‐derived EVs, a source already represented by other included publications [46]; and Bonanno et al., which utilized liposomes composed of tetraether lipids that exhibit biophysical properties not characteristic of eukaryotic membranes [47]. RT ‐ room temperature, PM ‐ plasma membrane, PSM ‐ palmitoylsphingomyelin.

Reference	Parolini 2009 [45]	Simbari 2016 [48]	Suga 2021 [49]	Yasuda 2022 [50]	Karam 2022 [51]	Wu 2022 [52]	Peruzzi 2025 [53]
EVs´ origin	Tumor cells	Nematode parasite(1) MODE‐K(2) Raw264.7(3)	Bovine milk	Prostate cancer cells(1) A549(2) Milk(3)	Malaria parasites; EVs sub‐population(1) and(2)	Bovine milk	HEK293FT; EVs subpopulation(1) and(2)
GP calculated, nm	435/490	440/490	440/490	440/490a	440/490	435/490	439/483a
Excitation, nm	320–420	not specified	340	385	340	340	350
EVs GP values	0.43–0.26 (pH 6) 0.37–0.26 (pH 7.4)	0.52(1) 0.48(3) 0.42(2)	0.38	0.16(1) 0.14(2) 0.12(3)	0.35(1) 0.03(2)	0.42 (pH 7.4) 0.52 (pH 5.5)	0.10(1) 0.12(2)
Temperature	37°C	37°C	37°C	30°C	37°C	37°C	RT
Comparison (GP)	Parent cell PM (0.31–0.10)	Lipid‐mimetic vesicles +/− plasmalogens (0.38; 0.30)	Lipid‐mimetic (0.10–0.50) and DPPC (0.49) vesicles	POPC/Chol/ PSM vesicles with Lo/Ld phase (0.14)	Relative comparison of isolated EV fractions	Relative pH‐based comparison of EVs	Parent cell PM (−0.15), control Ld (−0.24) and Lo liposomes (0.28)
Concluded phase state	Ld (possibly with Lo domains)	Not concluded	Lo + Ld	Lo + Ld	Not concluded	Not concluded	Lo + Ld

Note: In the case of multiple comparisons, the relevant data are marked as ^{(1), (2)} or ⁽³⁾ respectively.

^a Measured by C‐Laurdan.

Beyond GP analysis, phasor‐based approaches have been applied to Laurdan emission spectra to enable a more holistic, model‐free comparison of membrane states [54]. In this framework, the full emission spectrum is mapped to a single point in a two‐dimensional phasor space, which encodes both spectral position (center of mass) and spectral width, defined by the angular and radial coordinates (Please see the Method section for the mathematical definition of the spectral phasor). Using this approach, Laurdan phasor plots of EVs derived from breast cancer cells were compared with those obtained from cell‐derived plasma membrane vesicles and from model lipid systems representing distinct phases—liquid‐disordered (Ld: DOPC (1,2‐dioleoyl‐sn‐glycero‐3‐phosphocholine), DOPC/Chol (cholesterol), and liquid‐ordered (Lo: DOPC/DPPC/Chol). Consistent with GP‐based studies by Parolini [45], Suga [49], Yasuda [50], and Peruzzi [53] (see Table 1), this comparison enables an assessment of membrane packing, placing EVs between Ld and Lo phase regimes.

Notably, across all reported studies, EV membranes consistently appear more rigid than both parent cell membranes (or vesicles artificially derived from them) and synthetic mimetic vesicles reconstructed based on EV lipid compositions identified in lipidomic studies. This increased rigidity is commonly attributed to enrichment in sphingolipids, cholesterol, and phosphatidylserine. In addition, the presence of membrane‐associated proteins likely contributes significantly to EV membrane organization. Integral membrane proteins such as tetraspanins (e.g., CD9, CD63, CD81), integrins, and flotillins are abundant in EVs and are expected to influence local lipid packing. Consistent with this interpretation, we have previously demonstrated that the incorporation of integral proteins increases membrane rigidity as reported from Laurdan's read‐out [55, 56, 57].

3.3. Can Laurdan Distinguish EVs with Different Tetraspanin Profile?

The ultimate aim of biophysical characterization of EVs is to develop analytical tools capable of distinguishing EVs that differ in their lipid, protein, and other biomolecular content, which presumably reflects metabolic differences of the parent cells. As summarized above, Laurdan steady‐state spectra appear to be a straightforward and promising tool for EV characterization. While the influence of lipid composition—particularly variations in glycosphingolipids, sphingomyelin, cholesterol, and phosphatidylserine—on Laurdan fluorescence is well established, predicting how changes in membrane protein content affect Laurdan's readout remains challenging. This motivated us to investigate whether differences in tetraspanins, which are known to be among the major protein constituents of EVs, influence Laurdan's readout.

For these measurements, EVs were isolated from DU145 prostate cancer cells, either from wild‐type (WT) cells or from cells lacking individual tetraspanins (CD9 KO, CD63 KO, or CD81 KO). EV isolation and validation are summarized in Figure 1. Cryo‐electron microscopy revealed predominantly spherical vesicles with diameters characteristic of small EVs and no obvious contamination by cellular debris or protein aggregates (Figure 1A). Size distributions and particle concentrations were determined by dynamic light scattering (DLS) using a multiangle detection approach (MADLS) (Figure 1B,C). All samples displayed narrow and highly comparable size distributions, with mean hydrodynamic diameters (Z‐average ± SD) of 141.7 ± 4.2 nm (WT), 146.0 ± 5.3 nm (CD9 KO), 142.0 ± 2.6 nm (CD63 KO), and 140.3 ± 10.2 nm (CD81 KO) (Figure 1B). Polydispersity indices were low for all samples (PI ≤ 0.18), indicating relatively homogeneous EV populations with minimal aggregation. These data suggest that deletion of individual tetraspanins does not significantly affect EV size, secretion or gross heterogeneity under the isolation conditions used. Particle concentrations ranged from 7.98 × 10⁹ to 3.92 × 10¹⁰ particles mL⁻¹ and showed larger variability between three preparations, as expected for EV isolations. Importantly, Laurdan fluorescence analyses were performed using ratiometric or normalized readouts and are therefore not affected by differences in EV concentration. Western blot analysis confirmed enrichment of canonical EV‐associated proteins HSP70, flotillin‐1, and flotillin‐2 in all samples (Figure 1D). Importantly, CD9, CD63, and CD81 were selectively absent in the corresponding knockout‐derived EVs, while remaining detectable in WT EVs, validating successful gene deletion and preservation of EV integrity.

FIGURE 1.

Characterization of extracellular vesicles (EVs) isolated from wild‐type (WT) and tetraspanin knockout (CD9 KO, CD63 KO, CD81 KO) DU145 cells. (A) Representative cryo‐electron microscopy (cryo‐EM) images of EV preparations. Vesicles are predominantly spherical, 50–150 nm in diameter, with no visible contamination by nonvesicular structures. Scale bars: 50 nm. (B) Size distribution profiles determined by multiangle dynamic light scattering (MADLS), showing narrow distributions and comparable EV populations across all conditions. (C) Summary of MADLS parameters: Z‐average (mean particle size), polydispersity index (PI), and particle concentration (particles ml⁻¹). Data represent the mean ± SD of three independent EV isolations, each measured in six technical replicates. (D) Western blot analysis of EV markers and tetraspanins. HSP70, flotillin‐1, and flotillin‐2 confirm EV enrichment, while the absence of CD9, CD63, or CD81 in the corresponding KO samples validates successful gene deletion. Molecular‐weight markers are shown on the left.

Having established that EV populations derived from WT and tetraspanin KO cells are comparable in size and overall homogeneity, we next examined whether differences in tetraspanin composition are reflected in Laurdan fluorescence readouts of membrane packing and hydration. As previous studies suggested that EVs´ membrane phase state exhibits features lying between Lo and Ld phases (references in Table 1), we used liposomes with compositions characteristic of Ld (DOPC; POPC (1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphocholine); POPC:Chol (Cholesterol), 9:1 (‘POPC10Chol’) and Lo (POPC:Chol, 1:1 (“POPC50Chol”) and DPPC:Chol, 1:1 (“DPPC50Chol”)) as benchmarks for the lipid context of EVs. All samples were measured at 23°C and 37°C under steady‐state conditions and at 37°C in time‐resolved mode.

3.4. GP and Spectral Phasor Plot Analysis of Steady‐State Spectra Hint at Differences between EVs with Different Tetraspanin Profile

Measured steady state spectra together with respective GP values and spectral phasor data are shown in Figure 2A–C respectively, and summarized in Table 2. As expected, the reference liposomes (i.e. pure lipid membranes, without proteins) showed clear differences in emission spectra in both analytical approaches. Higher lipid packing in the order DOPC (Ld [58]) < POPC (Ld [59]) < POPC10Chol (Ld [59]) < POPC50Chol (Lo [59]) < DPPC50Chol (Lo [60]) corresponds to more rigid membranes for both examined temperatures.

FIGURE 2.

Comparison of different analytical approaches for EV characterization. Laurdan was embedded either in liposomes (100 nm in diameter) with a composition of DOPC; POPC; POPC:Chol, 9:1 (POPC10Chol) and 1:1 (POPC50Chol) and DPPC:Chol, 1:1 (DPPC50Chol) or in the EVs (wild‐type ‐ WT or knock outs for different tetraspanins ‐ CD9 KO, CD63 KO and CD81 KO). (A) Laurdan steady state spectra, measured for 23°C and 37°C. (B) GP calculated from steady state spectra using values at 440 and 490 nm. (C) Spectral phasor plots calculated for 23°C (stars) and 37°C (circles). Dashed line indicates the “phasor semicircle” with diameter of 0.5. (D,E) Time‐dependent fluorescence shifts of Laurdan measured for 37°C in liposomes and EVs. (D) Position of spectral maxima and (E) full width at a half maximum (FWHM) plotted against the respective time after excitation. Both parameters were determined by fitting the time resolved emission spectra with a log‐normal function. Gray point in (D) corresponds to the position of the estimated “tau‐zero spectrum” (t = 0). (D) The inset magnifies the very early stage of the relaxation process (up to 1 ns). Experimental errors are calculated as a standard deviation from three biological replicates (EVs).

TABLE 2.

Parameters calculated from the steady state and time‐resolved experiments. GP together with angle and modulus (M) from phasor plot were calculated from steady state emission spectra. The time $\tau (FWHM)$ at which the FWHM time‐evolution reaches its maximum was extracted from the TRES. All parameters were calculated for benchmark liposomes as well as for EV samples. Measurements were done at 37°C. Errors are calculated as standard deviations from 3 replicates.

Sample	GP	Steady state phasor		$\tau (FWHM)$, ns
		Angle	M
DOPC	−0.45	172	0.43	0.25
POPC	−0.36	166	0.42	0.46
POPC10Chol	−0.25	157	0.39	0.91
POPC50Chol	0.18	120	0.37	5.06
DPPC50Chol	0.36	105	0.49	NAa
EV WT	0.23 ± 0.06	101 ± 0.4	0.38 ± 0.01	8.7 ± 1.8
EV CD9 KO	0.26 ± 0.04	100 ± 0.4	0.39 ± 0.01	12.9 ± 3.5
EV CD63 KO	0.22 ± 0.07	103 ± 0.5	0.39 ± 0.00	7.0 ± 3.0
EV CD81 KO	0.27 ± 0.04	98 ± 3.3	0.39 ± 0.01	15.1 ± 1.2

^a Not Applicable.

The steady state spectra, as well as the calculated parameters (GP and phasor's angle and modulus ‐ M), demonstrate that the EVs are pronouncedly more rigid than the POPC liposomes with the highest cholesterol content (50 mol% “POPC50Chol”; Lo). However, the GP values indicate that the EVs have similar (23°C) or somewhat higher (37°C) fluidity as the rigid, DPPC50Chol liposomes, still representing a Lo phase. The phasor plots even stronger emphasize the rigid character of the examined EVs. In summary, a comparison of the steady‐state readouts classifies EVs as a rigid membrane system, with fluidity confined to the Lo phase region. Notably, this conclusion differs from the EVs listed in Table 1, which were concluded to be significantly more fluid, mostly proposed to be a combination of Ld and Lo phase (though the reported GP values are often higher than the ones in the present study). At 23°C, the EVs are highly rigid and possible differences between the four different types of EVs appear to be impossible to resolve. Differences between individual EV samples might be glimpsed at 37°C on the spectra, though calculated GP values, as well as phasor plots, remains on the border of experimental error. The changes are small, but one might conclude that GP shows similar values for WT and CD63 KO, while slightly higher GP was found for EVs isolated from CD9 and CD81 KOs, indicating their lower fluidity. Spectral phasor plot angle at 37°C shows similar positions for WT and CD63 KO and a slight shift for the CD9 and CD81 samples to lower wavelengths (smaller angle values), again indicating a more rigid nature of the latter.

3.5. TDFS Method Resolves Significant Differences between EVs with Different Tetraspanin Profiles

While steady‐state fluorescence yields only averaged information, time‐resolved measurements reveal a more detailed description of Laurdan's spectral relaxation. Using the TDFS method, one can separate the dynamics from the magnitude of the process, i.e. retrieve information about the lipid fluidity as well as about the hydration/polarity in the probe vicinity [5]. The TDFS method was described in detail previously [5]. Briefly, TRES are reconstructed from a series of measured fluorescence decays and represent original, nonfitted data (see Figure 3, upper graph). In the standard procedure, fitting TRES with log‐normal functions yields the position of the spectral maximum ($\nu$) and the spectral width (full width at half maximum, FWHM), which can then be plotted as a function of time. The time‐dependent shift of the spectra provides two parameters—the average relaxation time ($\tau$) and the total spectral shift ($\mathrm{\Delta }\nu$), see Method section for parameter definition. We have previously shown for fluid‐state (Ld) membranes [5] that these experimental readouts are directly correlated with mobility and hydration of lipid acyl groups, respectively. As the measurement itself is considerably more demanding than steady‐state experiments (a series of 14 fluorescence decays recorded at different wavelengths), it is mostly restricted to bulk spectroscopic experiments. While these are difficult to achieve in live‐cell imaging, they are perfectly appropriate for particles, including EVs. To the best of our knowledge, the time‐resolved approach has not been used to characterize EVs before.

FIGURE 3.

Time‐resolved emission spectra (TRES) phasor plot. Illustrative example of TRES‐phasor construction, showing POPC liposomes measured at 37°C. The phasor circle covers wavelengths from 400 to 540 nm. TRES were reconstructed from a series of measured fluorescence decays. For each spectrum, the phasor coordinates G and S were calculated. Spectral evolution was determined from t = 0 (blue) to t = 20 ns (red). Arrows indicates the shift to longer wavelengths (along the circle) and direction of spectra broadening (toward the center).

We employed the standard TDFS approach and retrieved the time‐dependence of the spectral maxima ($\nu (t)$) and the FWHM from the log‐normal fits to the TRES (Figure 2D,E, respectively, shown for 37°C). For the three liposome samples representing the Ld phase, the time‐evolution of the spectral maxima levels off at a constant and low value. This indicates that the entire spectral relaxation occurred within the window of Laurdan's fluorescence. Integration of $\nu (t)$ provides the timescale of the relaxation process ($\tau$), which were found to be 0.47, 0.63, and 0.99 ns for the liposomes composed of DOPC, POPC, and POPC10Chol, respectively, all measured at 37°C. The $\mathrm{\Delta }\nu$ are 4039, 4008, and 4056 cm⁻¹ for the respective liposomes. The observed trends in $\tau$ and $\mathrm{\Delta }\nu$ reflect expected [3, 5] changes in sn−1 acylgroup mobility and hydration due to changes in headgroup fluidity and hydration for the examined lipid compositions, respectively.

For the remaining samples it is apparent from the evolution of the position of the spectral maxima $\nu (t)$ and the FWHM that the spectral relaxation does not reach its energetic minimum during the time course of Laurdan's fluorescence. In these cases, $\tau$ and $\mathrm{\Delta }\nu$ values obtained by the standard TDFS approach are parameters with limited physical meaning [3, 5]. Nevertheless, visual inspection of the time‐dependence of the spectral maxima as well of the FWHM a) provides a detailed comparison with the benchmark liposomes and, most importantly, b) clearly shows distinct differences between individual EVs. Specifically, we conclude that the TRES of EVs are considerably broader when compared to liposomes (higher position of maxima, Figure 2E), reflecting the greater compositional heterogeneity of EVs. The overall spectral evolution of the maxima and FWHM (Figure 2D,E) indicate that, in terms of fluidity, EVs most closely resemble Lo‐phase POPC50Chol liposomes. In order to semiquantitatively compare the TDFS for all samples, we use the time $\tau (FWHM)$ at which the FWHM time‐evolution reaches its maximum. We have shown before that $\tau (FWHM)$ is a good estimate of the average time taken to complete the dipolar relaxation process and therefore can be used in place of the integrated relaxation time, $\tau$, in cases the quantitative standard TDFS approach is not applicable [61]. Comparison of $\tau (FWHM)$ unambiguously show, as already suggested by the steady state readout, that EVs isolated from CD9 and CD81 KO cells are clearly more rigid than EVs from WT and CD63 KO cells (Table 2).

Additionally, two key differences emerge from TDFS experiments. First, a larger fraction of the initial spectral shift is captured within the resolution of our TDFS experiment for EVs compared to liposomes (Figure 2D, inset). This means that the initial phase of the relaxation process is considerably slower in the EVs when compared to liposomes. Second, compared to the Lo phase liposomes, a substantial larger fraction of the spectral shift in EVs occurs on timescales slower than the 20 ns fluorescence time‐window of Laurdan. Together, these findings point to higher rigidity of the structures in the vicinity of Laurdan when compared to liposomes. Moreover, the data suggest that EVs derived from CD63 KO cells are the least heterogeneous among all samples, as indicated by the lowest maxima position (Figure 2E), whereas EVs from CD9 and CD81 KO cells exhibit markedly greater heterogeneity than both WT and CD63 KO EVs, reflected by the highest maxima positions (Figure 2E).

The compositional heterogeneity of EVs is clearly reflected in the TDFS analysis (Figure 2D,E). Moreover, examination of the log‐normal fits to the TRES of EVs—and even to the benchmark Lo systems—shows that a single‐component fit does not adequately capture the data at later times after excitation when the emission spectra broaden significantly. Attempts to fit these spectra with two log‐normal functions led to unstable results due to over‐parameterization. The fact that those TRES were not well fitted by the log‐normal distribution motivated us to implement the above‐described model‐free spectral phasor analysis for TRES.

3.6. Time‐Resolved Emission Spectral Phasor Plots as a Promising Tool for EVs Characterization

Phasor plots are particularly useful when conventional fitting approaches are challenging, for instance, when steady‐state spectra cannot be satisfactorily described by commonly used models. In our case, log‐normal fits of the TRES of EVs using a single‐component model did not adequately describe the data at later times after excitation, when the emission spectra broaden. This limitation motivated us to calculate phasor plots for each TRES reconstructed from a series of fluorescence decays of the EV samples, as well as the benchmark lipid systems (Figures 3 and 4). As described above, the spectral phasor plot represents the center of mass of the spectrum in wavelength space (its position along the phasor circle) and the relative spectral width (its distance from the origin [0,0], i.e. the modulus ‐ M) [4, 7]. For the TRES phasor plot, the spectral shift travels around the phasor circle from shorter to longer wavelengths (400–540 nm, the time parameter is not plotted directly) and the circular shape is deformed by changes in spectral width, with points closer to the center representing wider spectra (Figure 3).

FIGURE 4.

Time‐resolved emission spectra (TRES) phasor trajectories of (A) liposomes (100 nm in diameter) with composition of DOPC; POPC; POPC:Chol, 9:1 (POPC10Chol), 1:1 (POPC50Chol), and DPPC:Chol, 1:1 (DPPC50Chol) and (B) EVs wild‐type—WT or KO for different tetraspanins—CD9 KO, CD63 KO, and CD81 KO measured at 37°C. The position of the estimated “tau‐zero spectrum” (t = 0) is shown as red point.

In Figure 4A TRES‐phasors are calculated for all the examined liposomes measured at 37°C. While all trajectories start at a similar angular position, distinct patterns are observed for the Ld (three blue traces) and Lo (two purple traces) phases. The extent of the relaxation process captured is represented by the endpoint of the trajectories at 20 ns. TRES phasor shows longer trajectories for Ld‐phase liposomes, where spectral relaxation was fully completed and much shorter TRES‐phasor trajectories for the more rigid Lo phase liposomes, where only partial relaxation was observed, as shown in the TDFS results (Figure 2D,E). Additionally, for the liposomes in Lo phase, a noticeable kink appears in the curve, which is sharper in the more rigid DPPC50Chol.

Comparison of TRES‐phasor trajectories between liposomes and EVs reveals one striking difference, the EVs‘ phasors start at significantly lower wavelength than the liposomes‘ ones (Figure 4). Remarkably, the spectral relaxation of Laurdan in EVs starts at an angular position similar to that of the center of mass of the tau‐zero spectrum, which represents the estimated Laurdan emission spectrum at t = 0, i.e., before any relaxation occurs (Figure 4A,B, red dot). In the case of pure lipid vesicles, independently of the phase state or lipid composition, there is a minor fast relaxation component sensed by Laurdan that cannot be captured by the time‐resolution of the time‐correlated single photon counting experiment (about 50 ps). This is obvious from the present TDFS and TRES‐phasor data when comparing the start of the relaxation trajectory with the corresponding value obtained by the estimated tau‐zero spectrum, indicated by a red point. The presence of such a minor component not resolvable with a 50 ps resolution is a general feature of Laurdan's spectral relaxation in lipid bilayers as described in previous TDFS contributions (e.g. Rieber et al. [62], or Sýkora et al. [63]). The absence of that fast component in the case of EVs suggests that Laurdan senses, besides lipids, also nonlipid components, most likely membrane proteins. Two additional features are apparent in the TRES‐phasor trajectories of EVs: (1) all curves exhibit a kink similar to that observed for the Lo phase and (2) as in the Lo liposomes, the relaxation process is not fully captured by Laurdan's fluorescence, which is demonstrated by the trajectories having the 20 ns end point at positions comparable to those of the POPC50Chol liposomes. These observations support the interpretation that EVs possess a relatively rigid lipid organization.

Though the general shapes of the TRES‐phasor trajectories appear similar for all EV samples, individual differences can be observed. These differences become even more apparent when plotting the modulus M (“relative width”) against time after excitation (Figure 5A). This plot is the phasor analysis analog to the spectral FWHM derived from the log‐normal fits (Figure 2E). As pointed out above, the phasor analysis calculates the center of the mass of the spectra (not the spectral maximums of the standard TDFS approach) and the relative width takes the shape of the spectra in a model‐free manner into account. We can see a separation into two groups, WT/CD63 KO and CD9 KO/CD81 KO, as already shown by the TDFS analysis (Figure 2D,E, Table 2). For the WT/CD63 KO, the maximum of the spectral width is slightly narrower (as indicated by the lower peak in Figure 5A) and the peak is observed at earlier times as compared to CD9 KO/CD81 KO. Similarly, to the TDFS analysis, we conclude that EVs isolated from CD9 and CD81 KO cells are more rigid and heterogeneous than EVs from WT cells. EVs from CD63 KO cells do have similarities to the WT system, being somewhat more fluid.

FIGURE 5.

Parameters obtained from TRES phasor. (A) Distance of the phasor plot from the center [0,0] is determined as modulus M (which reflects the TRES spectral width) plotted against time. Smaller distances (toward the top of y axis) correspond to larger widths. (B) The smallest distance of phasor plots from the center [0,0] is determined as modulus (M) for each sample plotted against the corresponding time. Data point closest to the center (with shortest M) of the TRES phasor trajectory corresponds to the widest TRES. For easier comparison with TDFS data, the y axis is reversed, as the M is inversely proportional to the spectral width. Please note that the TRES phasor is based on an nm scaling, while the TDFS is presented in cm⁻¹.

The latter conclusions are visualized in detail in Figure 5B, where the M of the widest spectrum of the TRES is plotted versus its corresponding time point. We suggest the “TRES phasor summary plot” 5B to serve as a simplified read‐out from the model‐free phasor analysis of experimentally determined TRES spectra. Although loosing quite some information gained form the detail analysis, it is evident from this plot that the examined EVs differ in terms of fluidity and heterogeneity from the classical lipid phases, having closest similarities with Lo phase POPC50Chol liposomes. Moreover, evidently changing the protein profile by using different tetraspanin KO cells leads to distinct changes in fluidity and heterogeneity of Laurdan's nanoenvironment. We show that such an analysis of Laurdan's fluorescence is a meaningful tool for distinguishing EVs that differ in their lipid, protein, and other biomolecular content, which presumably reflects metabolic differences of the parent cells.

4. Conclusions

For the last three decades, Laurdan fluorescence has been widely used to characterize many kinds of bio‐assemblies. Inevitably, such widespread use has sometimes led to rather superficial or inconsistent naming of the physical origins behind Laurdan's parameters. Here, we clarify the following cornerstones of Laurdan photophysics in lipid bilayers: (A) the absorption maximum at 395 nm observed in the lipid gel phase arises from a stabilized, preoriented ground‐state conformer, rather than from Laurdan's Lb (π–π*) transition; (B) Laurdan's fluorescence in lipid bilayers is predominately sensitive to the local mobility of hydrated lipid acyl groups, and attributing Laurdan's read‐out to the “mobility of water molecules” is physically incorrect, since water molecules are orders of magnitude more mobile [64]; and (C) the excited‐state energy shift originates primarily from intramolecular reorganization of electron density within the lowest excited state, rather than from a classical outer‐sphere solvent relaxation. The kinetics of this intramolecular charge separation, however, remain governed by the overall environment relaxation that includes reorientation of solvent molecules, hydrogen‐bond formation of the solvent OH groups with the C=O group, and structuring of the first solvation shell.

EVs hold great promise for biomedical applications, particularly in diagnostics and therapeutic delivery. Their biophysical characterization is essential both for understanding their functional properties and as a potential diagnostic tool. Motivated by Laurdan's widespread use in membrane biophysics, several groups (Table 1) have employed its steady‐state fluorescence parameters, GP and spectral phasor, to assess membrane packing, typically placing EVs between the liquid‐disordered (Ld) and liquid‐ordered (Lo) phase regimes. As diagnostic applications will require distinguishing between EV subtypes, and given that tetraspanins are major protein constituents of EVs, we characterized the GP and spectral phasor properties of EVs isolated from WT cells and from knockout cell lines lacking specific tetraspanins. As steady‐state measurements did not reveal substantial differences among the individual EV samples, we recorded time‐resolved emission spectra (TRES) and analyzed them using the standard TDFS approach [5]. TDFS analysis reveals that differences in tetraspanin composition among EVs are associated with distinct membrane biophysical properties. Because EV heterogeneity is reflected in broad TRES, which complicates fitting to analytical functions, we additionally applied a model‐free spectral phasor analysis to the TRES data. Both time‐resolved methods give complementary information on compositional heterogeneity and on the lipid phase state of the individual EVs, placing them to Lo region with fluidity close to the less rigid POPC50Chol liposomes. Finally, the TRES–phasor approach, supported by TDFS, revealed an unexpected pattern. In EVs, the ultrafast spectral relaxation component commonly observed in pure lipid vesicles, independent of their phase state, is absent and instead replaced by an initial relaxation process on the sub‐nanosecond timescale. We suggest that this early component of the TRES‐phasor trajectory could provide a novel means to probe nonlipid constituents of EVs, such as membrane‐associated proteins.

In summary, the TRES approach has enhanced sensitivity and gives richer information when compared to the steady‐state readouts of Laurdan fluorescence for EV characterization. Ideally, complementary analysis of TRES data using both TDFS and spectral phasor provides comprehensive information. Nevertheless, we believe that spectral phasor analysis of TRES has the greater potential for future applications, for the following reasons: (a) it is model‐free, which is advantageous for samples with high compositional heterogeneity; (b) the graphical readouts of TRES phasors display characteristic patterns for different bio‐nanoparticles; and (c) heterogeneity and fluidity can be summarized in a single, intuitive graph (Figure 5B).

5. Materials and Methods

5.1. Materials

1,2‐dioleoyl‐sn‐glycero‐3‐phosphocholine (DOPC), 1‐palmitoyl‐2‐oleoyl‐sn‐glycero‐3‐phosphocholine (POPC), 1,2‐dipalmitoylsnglycero‐3‐phosphocholine (DPPC) and cholesterol were from Avanti Polar lipids (USA, Alabaster). Laurdan (6‐dodecanoyl‐2‐dimethylaminonaphthalene) was acquired from Invitrogen (California, USA). Chloroform and methanol, both HPLC grade, were purchased from Merck (Darmstadt, Germany).

5.2. Preparation of Large Unilamelar Vesicles (LUVs)

Appropriate amounts of lipid stock solutions in chloroform were mixed in a glass tube together with Laurdan dissolved in methanol. The Laurdan to lipid ratio was 1:100. Organic solvents were evaporated under a nitrogen steam and tubes were placed for at least 2 h in the desiccator to evaporate traces of remaining organic solvents. Lipid films were dissolved in 1.5 mL of phosphate buffer by vortexing; the final lipid and Laurdan concentrations were 0.5 mM and 5 μM, respectively. LUVs were formed by extrusion through a 100 nm pore diameter membrane filters (Avestin, Ottawa, Canada).

5.3. Isolation and Characterization of Extracellular Vesicles (EVs)

The CRISPR/Cas9 technique was employed for targeted knock‐out (KO) of tetraspanin genes CD9, CD63, and CD81 in the DU145 cell line (RRID:CVCL_0105). Preparation of vesicle‐free medium for cell culture, Cell culture for Conditioned media production/EVs isolation, Ultracentrifugation with sucrose cushion, cryo‐EM, DLS, SDS‐PAGE and Western Blotting were done essentially as described in Drápela et al., 2025 [65] and shown in Figure 1.

5.4. Steady State Spectroscopy

EVs were labeled by external addition of Laurdan in DMSO with a final concentration of 1 μM (∼100 μl of EV solution, containing ∼10⁵ particles). LUVs or EVs solutions were transferred to quartz cuvettes and measured using FS5 spectrofluorometer (Edinburgh Instruments, UK). The temperature was set to 23°C or 37°C. Steady state spectra were measured using 370 nm excitation and were recorded from 400 to 600 nm. GP was calculated from emission spectra as follows:

$$\mathrm{GP}=\frac{I_{440}-I_{490}}{I_{440}+I_{490}}$$

where I ₄₄₀ and I ₄₉₀ are the fluorescence intensity recorded at 440 and 490 nm, respectively.

Transformation of steady state spectra to spectral phasor coordinates G and S was done according to

$$G=\frac{\sum _{\lambda }I\left(\lambda \right).\cos \hspace{0.17em}\left(2\pi n(\lambda -{\lambda }_i)/L\right)}{\sum _{\lambda }I\left(\lambda \right)}$$

$$S=\frac{\sum _{\lambda }I\left(\lambda \right).\sin \hspace{0.17em}\left(2\pi n(\lambda -{\lambda }_i)/L\right)}{\sum _{\lambda }I\left(\lambda \right)}$$

where $I(\lambda )$ is fluorescence intensity at the corresponding wavelength ($\lambda$), n is the number of the harmonic (here n = 1), ${\lambda }_i$ is a first point of the spectral wavelength range and L is the total wavelength range of the spectrum.

5.5. Time Resolved Spectroscopy

Time‐resolved experiments were conducted with identical samples as described in the steady‐state section. Measurements were done at 37°C using FluoTime 250 spectrophotometer (PicoQuant, Germany) equipped with 373 nm laser line. Series of fluorescence decays (400–540 nm, 10 nm step) was recorded. Decays were fitted with 3‐exponential function using the iterative reconvolution procedure (EasyTau software, PicoQuant, Germany). The decays together with a steady state emission spectrum were used to reconstruct the time resolved emission spectra (TRES) as described previously [66].

5.5.1. TRES‐Phasor

Each of the generated TRES (in wavelength domain) was transformed to the G, S coordinates using the same equations as described for steady‐state spectral phasor plots. The relative spectral width is inversely proportional to the phasor modulus (M) [67], i.e., distance of the phasor data point from the center of the phasor circle defined as:

$$M=\sqrt{G^2+S^2}$$

The spectral phasor coordinates at t = 0 were calculated as described above from the estimated tau‐zero spectrum determined using absorption and emission spectra of Laurdan measured in cyclohexane and PC liposomes [66, 68].

5.5.2. TDFS Method

Each of the generated TRES was fitted with log‐normal function in order to determine position of their maxima ($\nu$) and their width. Time dependent fluorescence shift method monitors the spectral relaxation of the Laurdan polarity‐sensitive probe. TRES possess the information about the polarity and viscosity of the microenvironment of Laurdan, which can be quantified by the total spectral shift ($\text{Δ}\nu$) and solvent relaxation time $\tau$. In lipid bilayers the first parameter reflects the extent of hydration, while the latter the mobility of hydrated lipid molecules [66]. $\text{Δ}\nu$ is calculated as $\text{Δ}\nu ={\nu }_0-{\nu }_{\mathrm{\infty }}$, where ${\nu }_0$ is the position of the spectrum right after the excitation and ${\nu }_{\mathrm{\infty }}$ is position of spectra after the spectral relaxation process is completed. ${\nu }_0$ was estimated to be 23 800 cm⁻¹ for Laurdan [66, 68]. The second parameter, relaxation time ‐ $\tau$, is determined as the mean integrated time as follows:

$$\tau ={\int }_0^{\mathrm{\infty }}\frac{{\nu }_t-{\nu }_{\mathrm{\infty }}}{\text{Δ}\nu }\mathrm{d}t$$

The parameters for liposomes in the Ld phase were analyzed with 10 ns time window. For Lo liposomes (and other rigid systems, e.g. So phase liposomes), the spectral relaxation does not finish within the Laurdan fluorescence time window, therefore ${\nu }_{\mathrm{\infty }}$ cannot be determined and the $\text{Δ}\nu$ and $\tau$ parameters cannot be calculated correctly. Therefore, an alternative parameter was used for semiquantitative comparison, namely τ(FWHM). The FWHM was extracted from log‐normal fits of the individual TRES and plotted as a function of time after excitation. τ(FWHM) was defined as the time at which the time evolution of the FWHM reaches its maximum.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary Material

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.