Squaraine Dyes Exhibit Spontaneous Fluorescence Blinking That Enables Live-Cell Nanoscopy

Abstract

Hampered by susceptibility to nucleophilic attack and chemical bleaching, electron-deficient squaraine dyes were long considered unsuitable for biological imaging. This study unveils a surprising twist: in aqueous environments, bleaching isn’t irreversible but rather a reversible spontaneous quenching process. Leveraging this new discovery, we introduce a novel deep-red squaraine probe tailored for live-cell super-resolution imaging. This probe enables Single-Molecule Localization Microscopy (SMLM) under physiological conditions without harmful additives or intense lasers, and exhibits spontaneous blinking orchestrated by biological nucleophiles such as glutathione or hydroxide anion. With a low duty cycle (~0.1%) and high emission rate (~6 × 10⁴ photons/s under 400 W/cm²), the squaraine probe surpasses the benchmark Cy5 dye by four-fold and Si-rhodamine by a factor of 1.7 times. Live-cell SMLM with the probe reveals intricate structural details of cell membranes, demonstrating the high potential of squaraine dyes for next-generation super-resolution imaging.

Graphical Abstract

Fluorescence microscopy is an essential tool for studying biological interactions and cellular functions.^1,2 However, it faces a significant challenge due to the diffraction of light, which limits the spatial resolution.³ Super-resolution nanoscopy has emerged as a revolutionary breakthrough, overcoming the diffraction limit and providing a comprehensive understanding of complex biomolecular dynamics.^4–6 Among the various super-resolution imaging techniques, single-molecule localization microscopy (SMLM) stands out for its distinct advantages, including high resolution, minimal reliance on complex hardware, and straightforward practicality. As a result, SMLM is increasingly valuable for exploring subcellular organelles, observing various organelle interactions, and investigating diverse biological processes.^7–10.

SMLM requires methods that selectively activate, detect, and localize a fraction of fluorophores, followed by deactivation or photobleaching while another subset is switched on. Repeating this process allows the construction of super-resolution images from a large number of individual molecular localizations.^11–13 Effective SMLM requires the majority of fluorophores to exist in a non-fluorescent state while retaining the ability to briefly transition to a fluorescent state.¹⁴ Currently, the commonly used dyes in SMLM include rhodamine and cyanine dyes, whose on-off states can be manipulated by photons, pH levels, thiols or other chemicals.^15–17 Betzig et al. pioneered a technique called photoactivated localization microscopy (PALM), where light triggers the transition of fluorophore from the off-state to the on-state.¹⁸ Zhuang et al. have employed cyanine dyes for super-resolution imaging, which can be driven into dark states by exposure to a high concentration of strong nucleophile, such as 140 mM of 2-mercaptoethanol.^17,19,20 These dark states subsequently undergo random transitions to a bright state, enabling super-resolution imaging. However, the imaging buffers used for cyanine dyes, including nucleophilic additives and a deoxygenation system, can lead to cell toxicity and may not always be suitable for live cell imaging.^21,22 Urano et al., Johnsson et al., and Lavis et al. engineered rhodamine derivatives to demonstrate spontaneous blinking through the reversible formation of an intramolecular spiro-ring, thereby eliminating the need for toxic imaging buffers.^23–40 Urano et al. also reported a principle of spontaneous blinking in living cells, utilizing the reversible ground-state nucleophilic attack of intracellular glutathione (GSH) upon a xanthene fluorophore.⁴¹ These innovative approaches have been successfully used for dynamic super-resolution imaging of microstructures within living cells.⁴² The advancement of high-performance, spontaneously blinking fluorescent dyes holds great promise for further development of next-generation super-resolution imaging techniques.^14,43–45.

In this study, we describe a novel series of fluorescent squaraine dyes that exhibit spontaneous blinking in physiological environments. Squaraines are resonance-stabilized dyes and typically consist of an electron-deficient central four-membered ring and two electron-donating groups arranged in a donor-acceptor-donor (D-A-D) configuration.^46,47 Squaraine dyes have attractive spectral characteristics including narrow and intense absorption/emission bands in the deep-red and near-infrared spectral range, a highly favorable wavelength region for bioimaging applications.^48,49 However, due to the electron-deficient central core, squaraine dyes, especially those with 4-aminophenyl rings as the donor groups, are susceptible to chemical bleaching caused by nucleophilic attack.⁵⁰ This vulnerability has historically restricted squaraine utility in bioimaging.^51,52 In this context, we strategically leverage the inherent electrophilicity to achieve spontaneous blinking in a physiological environment. We report that squaraine dyes with 4-aminophenyl rings as donor groups exhibit reversible on-off fluorescence when exposed to biological nucleophiles, such as OH⁻ and GSH. We exploit this effect to design a new fluorescent probe that exhibits fluorescence blinking at the single molecule level with an impressive 0.1% duty cycle under physiological conditions, without the need for any additives. Furthermore, the squaraine probe produces exceptional brightness, emitting close to 6 × 10⁴ photons per second under 400 W/cm² illumination. Leveraging the high photon emission rates, we achieved a remarkable 5.5 nm localization precision. Finally, we utilize the squaraine probe for super-resolution imaging of living cell membranes, highlighting the favorable properties for SMLM.

Squaraine dyes with 4-aminphenyl rings as donors have an electron-deficient central core that is susceptible to nucleophilic attack, resulting in a non-fluorescent state (Figure 1a). We hypothesized that structural modifications of the 4-aminphenyl rings could alter squaraine electrophilicity and potentially convert nucleophilic addition into a reversible process. This in turn would enable fluorescence switching and produce spontaneous blinking at the single molecule level.

Figure 1.

Reaction of squaraine dye with nucleophile. (a) Equilibrium for reversible nucleophilic addition to squaraine dye. (b) Absorption spectra of SQ1 in DMSO solution before and after addition of aqueous NaOH (6 × 10⁻⁴ mmol) solution, followed by NaOH + HCl (7.7 mmol) and waiting for different times. The absorbance is normalized relative to unreacted SQ1. (c) Plot and fitting curve for SQ1 absorbance at 647 nm as a function of pH in solutions composed of 1:1 THF: buffer, where the buffer is: pH < 5 citric acid/Na₂HPO₄, pH 5–9 KH₂PO₄/Na₂HPO₄, pH > 9 Na₂CO₃/NaHCO₃. The pH value is the final value after mixing a buffer solution with a dye dissolved in THF. (d) Reactive sites within squaraine SQ1, green surfaces represent electrophilic sites, blue surfaces represent nucleophilic sites. (e) ¹H NMR spectra of SQ1 in THF-d₈ in the absence (top) and presence (bottom) of OH^–. (f and g) MALDI-TOF mass spectrum of the dye SQ1 in the absence and presence of NaOH.

To test our hypothesis, we successfully synthesized a series of squaraine dyes by condensing the relevant aniline precursors with squaric acid under established literature conditions.^48,49 The initial compound, SQ1 (see Figure 1), exhibited narrow and intense absorption and emission bands in the deep-red range with high fluorescence quantum yields in organic solvents (Figure 2). We identified the reactive sites on SQ1 by conducting theoretical calculations that employed the Dual Descriptor and the Fukui function methodologies. As depicted in Figure 1d, the green isosurfaces represent electrophilic sites, while blue isosurfaces represent nucleophilic sites. We reasoned that OH⁻ should act like thiol and attack the electrophilic core of SQ1.⁵³ Indeed, when OH⁻ was introduced to SQ1, the characteristic absorption peak at 647 nm rapidly disappeared (Figure 1b, FigureS1). Covalent bonding of OH⁻ to SQ1 was confirmed by MALDI-TOF MS analysis, which identified the mass peak associated with the expected adduct (Figure 1f, Figure 1g). The reaction product was also analyzed by ¹H NMR spectroscopy in THF-d₈ (Figure 1e). The parent squaraine structure exhibited a highly symmetric spectral pattern, and upon addition of NaOH, there was a loss in spectral symmetry that was consistent with nucleophilic addition of OH⁻ to the squaraine core.⁵⁴ However, complete atom assignment was challenging due to the broadened and partially overlapping aromatic peaks. Therefore, we conducted OH⁻ addition experiments using a simplified model squaraine dye of EtSQ (see Figure S4) that lacked the N-benzyl groups and observed unambiguous ¹H NMR and ESI-MS evidence for nucleophilic addition to the squaraine core (Figure S4). Moreover, the NMR spectra of the adducts did not change over several days at room temperature indicating high molecular stability.

Figure 2.

Chemical structure of squaraine dyes and optimization of optical properties in bulk solution. (a) Normalized absorption and fluorescence spectra of SQ1 and SQ3 in CHCl₃. (b) Plots of maxima peak absorbance for SQ1, SQ3 and SQ8 as a function of pH and associated fitting curve.

To assess the reversibility of nucleophilic addition, an excess of HCl was subsequently introduced to a solution of bleached SQ1 containing OH⁻. Remarkably, the characteristic blue squaraine color gradually returned and after 12 minutes of HCl treatment, the absorption intensity of SQ1 reached 90% of its initial value, indicating reversion of the OH⁻ addition to SQ1. The stability of the SQ1-OH adduct was judged to be quite high, as more than 40% of SQ1 was successfully regenerated after 6 months storage at room temperature in the dark (Figure S2). Subsequently, we investigated the pH sensitivity of SQ1 in solutions containing 1:1 THF: buffer and spanning a pH range from 3 to 11. Under acidic conditions, SQ1 exhibited a strong absorption band centered at 647 nm, with absorption diminishing as pH values increased. By analyzing the distinct pH-dependent absorption profiles, we determined a pK_sq value of 7.1 for SQ1 (Figure 1c).

To systemically adjust the photophysical properties and pK_sq values, we synthesized a series of squaraine dyes with various 4-aminophenyl rings (Figure 2). The different N-amino substituents included N-benzyl derivatives with electron-withdrawing groups that diminished the nitrogen atom’s electron-donating capability, thus making the squaraine core more electron-deficient and lowering the pK_sq value. Figure 2 includes a summary of the photophysical properties of eight squaraine dyes in bulk solution, including absorption and emission wavelengths, molar absorption coefficients, and fluorescence quantum yield. The highest measured pK_sq value was 7.9 for SQ3 which should be primarily in a fluorescence-on state at physiological pH of 7.4, and the lowest pK_sq value was 6.7 for SQ8 which should be primarily in a fluorescence-off state at pH 7.4. In summary, we successfully adjusted the squaraine pK_sq value by introducing various substituents on the 4-aminophenyl donor rings of the dye structure.

In addition to the eight squaraine dyes with 4-aminophenyl rings as the donors, we synthesized another common type of squaraine dye with quinolinium heterocycles as donors, referred to as SQ9. We evaluated its responsiveness to OH⁻. In line with previous literature reports,^55,56 SQ9 demonstrated high chemical stability, and we did not observe significant changes in absorbance in the presence of nucleophiles (Figure S14). Theoretical calculations indicated that the squaraine core of SQ9 is much less electrophilic than SQ1 (Figure S3). The high chemical stability of squaraines with quinolinium or indolium heterocycles as donors and their resistance to nucleophilic attack means they do not exhibit spontaneous blinking, but they are good choices as dyes for single molecule tracking experiments.

GSH, a tripeptide with a thiol group, is a potent biological nucleophile.^57,58 As expected, treatment of squaraine SQ1 with GSH led to loss of the characteristic squaraine absorption peak (Figure 3a and S16). Subsequent addition of excess of HCl to the bleached solution of SQ1 plus GSH led to gradual recovery of absorbance, indicating reversion of the reaction between SQ1 and GSH. However, the recovery rate for SQ1 in the presence of GSH was relatively slow and after 40 minutes of exposure to HCl, the absorption reached 82% of its original value. This observation is attributed to the strong nucleophilic nature of thiol compounds.⁵⁹ Furthermore, we investigated the relationship between SQ1 absorbance and GSH concentration, revealing a gradual decrease in the characteristic absorption as the GSH concentration increased. The dissociation constant for GSH binding to SQ1 (K_d,GSH) was determined through a concentration-response curve (Figure 3b), yielding a value of 7.3 μM which is consistent with previously reported values for reversible addition of thiol nucleophiles to a squaraine dye.⁶⁰ This dissociation constant is considerably lower than typical physiological GSH concentrations, suggesting that most of SQ1 would be in a non-fluorescent state when in physiological conditions. The addition of GSH to the model squaraine dye EtSQ was further validated through ¹H NMR and ESI-MS analyses (Figure S4).

Figure 3.

GSH response of SQ1 in bulk solution and single molecule imaging of SQ-F. (a) The absorption spectra of SQ1 before and after addition of aqueous of GSH (6 × 10⁻⁴ mmol) solution followed by GSH +HCl (7.7 mmol) and waiting for different times. Time 0 minutes is the initial moment after adding HCl into the mixture. (b) Plot of normalized SQ1 absorbance at 647 nm as a function GSH and fitting curve that provided K_d,GSH. (c-h) Single molecule images at 5 different time points and overlay image of 4000 frames. Power density: 100 W cm⁻². (i) Typical intensity time traces of SQ-F under 400 W/cm². The blue line represents the brightness versus time, and the red line represents the bright state the molecule, which is determined by Hidden Markov Model (HMM). (j) The accumulated number of localizations versus time with or without addition of GSH (pH 7.4 PBS buffer with 5.0 mM of GSH) under different power densities. (k) Localization precision distributions of SQ-F under different power densities, the median values are 6.0 nm at 100 W cm⁻², 5.8 nm at 200 W cm⁻², 5.5 nm at 400 W cm⁻². (l) On-time distributions at 100 W cm⁻² with or without addition of GSH, the time constant was obtained by double exponential fitting. (m, n) Number of blinking times (m) and duty cycles (n) of SQ-F with or without addition of GSH under different power densities.

The relatively small size and polarity of SQ1 made it challenging to immobilize the dye on a surface for single molecule studies. Consequently, we synthesized an amphiphilic version of the squaraine dye, denoted as SQ-F (Scheme S2 and Figure 4a). The synthetic introduction of two long-chain alkyl groups enabled surface immobilization or cell membrane targeting,^61–63 while the added benzenesulfonyl groups provided partial water solubility.⁶⁴ Before embarking on cell imaging studies, we conducted single-molecule imaging experiments to assess the blinking performance of SQ-F. Using a wide-field total internal reflection fluorescence (TIRF) microscope, we determined the photophysical characteristics of immobilized SQ-F in PBS at pH= 7.4. As depicted in Figure 3c–3g, only a sparsely distributed subset of molecules exhibited fluorescence in each frame. Over time, we observed a continuous increase in the total count of molecules in the bright state, indicating distinct molecular entities were occupying this state in each frame (Figure 3h). We proceeded to generate fluorescence time traces. An exemplary example in Figure 3i. illustrates the intrinsic spontaneous blinking behavior of SQ-F in the PBS buffer, where most of the fluorophore remains in a non-fluorescent state. Furthermore, we examined the single molecule blinking behavior under different power densities (100, 200, and 400 W/cm²). As power increased, the population of molecules in the bright state decreased, a pattern consistent with other types of blinking dyes, where higher irradiance leads to more molecules in the off-state and a lower duty cycle (Figure 3j). Additionally, we introduced physiological concentrations of GSH into the buffer solution, which also resulted in more molecules in the off-state. These results agree with the bulk solution data (Figure 3a and 3b) and they are consistent with nucleophilic attack of SQ-F by GSH.

Figure 4.

Squaraine dye SF-Q for super-resolution imaging in live cell membranes. (a) Super-resolution and diffraction limited image of Hela cell membrane. (b) Intensity distribution at the cross section of sub area 2 in (a), black dots and red dots fitted by single gaussian functions correspond to diffraction limited image and super-resolution image, respectively. (c) Magnification of diffraction limited image (left) and super-resolution image (right) of sub area 1 in (a). (d) Intensity distribution at the cross section of sub area 1 in (a), the black column corresponds to the diffraction limited image, and the red dots fitted by two gaussian functions correspond to the super-resolution image. (e, f) Single molecule brightness (e) and localization precision (f) distributions throughout the imaging process with exposure time of 40 ms. (g) FRC analysis of the super-resolution image resolution, correlation = 1/7 was selected as the threshold. The irradiance for live-cell super-resolution imaging is 300 W/cm².

We conducted an in-depth analysis of the single molecule imaging data, elucidating six key switching properties: (i) photon emission rate, (ii) average on-time per switching event, (iii) localization precision, (iv) number of switching cycles, (v) on-off duty cycle, and (vi) total emitted photons before bleaching.^65,66 At 400 W/cm², we achieved an impressive emission rate up to 56,840 photons per second and an exceptional localization precision of 5.5 nm for 50 msec exposure (Figure 3k). SQ-F exhibited a brightness that is approximately four times greater than that of Cy5 and 1.7 times greater than that of Si-rhodamine (Table S1), accompanied by superior localization precision when compared to the commonly used SMLM probes. SQ-F displayed an average on-time of 0.25 s for each switching event. The introduction of GSH reduced the on-time to 0.17 s (Figure 3l) but increased the frequency of blinking (Figure 3m), resulting in extended total on-time for the molecules (Table S1). Therefore, the addition of GSH results in heightened blinking frequency and enhanced photostability of molecules, albeit with a slight reduction in their brightness. This observed decrease in brightness with GSH aligns with prior findings⁴¹, highlighting the role of thiols as general fluorescence quenchers. The increased blinking frequency is conducive to the multi-localization of individual molecules. Additionally, the duty cycle, regardless of GSH presence, remained between 0.1% and 0.4% (Figure 3n) under laser irradiation at 100–400 W/cm². These values are comparable to those of conventional fluorophores used for SMLM (such as Alexa 647) when exposed to higher laser intensity (1.5 kW/cm²) in an optimized buffer solution containing β-mercaptoethylamine (MEA) and an enzymatic oxygen scavenging system (GLOX).^66,67 In assessing the photostability of our developed SQ-F, we quantified the survival fraction following 100 seconds of irradiation. As illustrated in Figure S18 and detailed in Table S2, we found that more than 50% of SQ-F molecules remained intact after this period, highlighting their outstanding photostability. In brief, SQ-F displays outstanding brightness and a desirable spontaneous blinking behavior at lower power density within a physiologically buffer solution, suggesting it has great potential for effective SMLM-based super-resolution imaging in live cells.

Building upon the insight gained from the surface-immobilized single molecule characterization, our next task was to apply spontaneously blinking SQ-F to super-resolution imaging of live cell membranes. SQ-F exhibited robust spontaneous blinking behavior within the live cell environment. The dynamic behavior closely resembled the blinking patterns observed in previous single-molecule experiments (Movie S1). We recorded a substantial series of 20,000 consecutive images, and employed them for reconstruction of super-resolution images. The resulted reconstruction image (Figure 4a) unveiled intricate details of the cellular membrane structure, offering unparalleled clarity compared to conventional epifluorescence images. This led to a super-resolved membrane with a cross-sectional fullwidth at half maximum (FWHM) of 112 nm, a marked improvement over the traditional projection image’s FWHM of 1170 nm (Figure 4b). This advancement in spatial resolution effectively surpasses the optical diffraction limit.

Quite apparent in our super-resolution cell images are different types of pseudopodia, including filopodia (area 1 in Figure 4a) and lamellipodia (area 3 in Figure 4a). Pseudopodia are plasma membrane protrusions rich in actin filaments, which act as cellular sensory appendages and facilitate environmental surveillance.^68–70 The depicted filopodia exhibit varying lengths within the micrometer range, highlighting their remarkable diversity, consistent with literature findings (Figure 4a).^71–78 In addition, we successfully distinguished two closely positioned filopodia, separated by approximately 232 nm (Figure 4c and 4d), well beyond the capabilities of conventional fluorescence imaging.

Further analysis revealed an average single-molecule brightness of 511 photons per frame (Figure 4e), with an exposure time of 0.04 s, and an average localization precision of 19.1 nm (Figure 4f) in live cells. The single-molecule localization precision within the cellular environment is less accurate than on coverslips, due to several differences in the cell imaging conditions such as shorter exposure times, the complexity of imaging environment, the highly dynamic molecular behavior in membrane, and elevated level of background autofluorescence. Fourier ring correlation (FRC) analysis demonstrated that the reconstructed image has a resolution of approximately 69 nm (Figure 4g).⁷⁹ Analysis of rolling Fourier ring correlation⁸⁰ (rFRC) on reconstructed images reveals varying local resolutions with a median value of 43 nm (Figure S20), potentially attributed to differing focal planes of the inner membrane and pseudopodia, probe signal strength variations, and membrane fluidity affecting localization accuracy in live cells. In brief, SQ-F based super-resolution imaging of living cell membranes visualized the ultrastructure of pseudopodia. These findings validate the capability of SQ-F as an effective deep-red fluorescent probe for SMLM in the realm of super-resolution imaging of living cells.

In summary, we leveraged a new understanding of the reversible reaction of squaraine dyes with the biological nucleophiles, OH⁻ and GSH, to create a novel deep-red squaraine-based fluorescent probe that exhibits spontaneous blinking under physiological conditions. The mesmerizing single molecule blinking phenomenon is borne from the reversible disruption of squaraine dye conjugation, producing fluorescence on-off cycles. Through systematic molecular design, we achieved a favorable duty cycle ranging from 0.1% to 0.4%, and an impressive single molecule brightness of 56,840 photons per second at relatively low irradiance of 400 W/cm². Compared to other spontaneous blinking dyes,⁴⁵ our squaraine dyes stand out for their exceptional brightness, efficient duty cycles at low irradiance, and favorable deep-red spectral range. Notably, this brightness greatly surpasses that of Cy5, a benchmark deep-red fluorophore for SMLM. We developed a new squaraine-based fluorescent probe for super-resolution imaging of living cell membranes, without the need of cytotoxic additives and high-power laser irradiation. The live-cell SMLM provided nanoscopic images of cell membranes that revealed intricate membrane structures and pseudopodia details, providing new insight into cell morphology and physiological behavior. Future work is needed to confirm the generality of squaraines for super-resolution imaging of other cell substructures, however, the ease of squaraine dye synthesis will facilitate these studies. This advancement in fluorophore design has the potential to broadly impact the field of SMLM by providing new classes of high-brightness blinking fluorescent probes for super-resolution imaging.

Table 1.

Chemical structures and optical properties of squaraine dyes. Owing to limited solubility, the pK_sq value of SQ2 could not be obtained. Fluorescence quantum yields were determined using 4,4-[bis(N,N-dimethylamine)phenyl] squaraine dye as the standard (ϕ = 0.70 in CHCl₃).

	R1	R2	name	λaba (nm)	λema(nm)	pKsq	τ	ϕa	ϕb
			SQ1	624	643	7.1	2.67	0.80	0.47
	H	H	SQ2	625	645	-	2.50	0.70	0.67
			SQ3	629	650	7.9	2.85	0.60	0.69
			SQ4	626	646	7.3	2.77	0.71	0.56
			SQ5	624	643	7.1	2.61	0.85	0.56
			SQ6	623	641	6.9	2.67	0.84	0.74
			SQ7	625	643	7.0	2.66	0.82	0.53
			SQ8	621	639	6.7	2.73	0.75	0.41

^[a]CHCl₃ solution,

^[b]THF solution, Error ≤ 5%.

ABBREVIATIONS

SMLM

Single-Molecule Localization Microscopy

PALM

photoactivated localization microscopy

GSH

glutathione

D-A-D

donor-acceptor-donor

HMM

Hidden Markov Model

TIRF

total internal reflection fluorescence

MEA

β-mercaptoethylamine

FWHM

fullwidth at half maximum

FRC

fourier ring correlation