A general strategy to develop fluorogenic polymethine dyes for bioimaging

The study addresses a key challenge in live-cell fluorescence imaging: creating bright, specific, and fluorogenic small-molecule probes that minimize background while enabling labeling of intracellular targets. While rhodamine-based fluorogenic dyes have been widely optimized via an open/closed spirocyclization equilibrium to enable no-wash labeling, they are limited by lengthy, low-yielding syntheses and difficulty extending emission into the near-infrared (NIR) region; many red-shifted rhodamines remain non-fluorescent even upon binding. Polymethine dyes (for example, carbocyanines such as Cy dyes, squaraines, flavylium-based dyes, and coumarin-hemicyanine hybrids) offer simple, modular synthesis and tunable emissions from green to shortwave infrared, but lack an intrinsic cyclization equilibrium to confer robust fluorogenicity. Previous attempts appended nucleophiles (alcohols, amines, thiols) to induce intramolecular cyclizations (for example, 5- or 6-endo-trig) with limited efficiency. The authors hypothesize that enforcing a favorable 5-exo-trig intramolecular cyclization, guided by Baldwin’s rules, will yield robustly fluorogenic polymethine dyes that switch from a non-fluorescent closed form to a fluorescent open form upon target binding. The goal is to develop a general, modular strategy applicable across the polymethine family (Cy3, Cy5, Cy7) for no-wash live-cell imaging, super-resolution microscopy, and multiplexing.

• Rhodamine dyes: Exhibit an equilibrium between a fluorescent zwitterionic open form and a non-fluorescent spirocyclic closed form that can be tuned by electronic substituents and intramolecular nucleophiles. Systematic optimization has produced visible-range fluorogenic probes suitable for no-wash multicolor imaging, but red/NIR rhodamines are synthetically challenging and often remain dark upon target binding.
• Polymethine dyes: Indoleninium-based carbocyanines (Cy), squaraines, flavylium-based dyes, and coumarin-hemicyanine hybrids are widely used for cell and in vivo imaging due to modular synthesis, high extinction coefficients, biocompatibility, and broad spectral tunability. Classic examples include ICG and Cy5 derivatives (for example, Alexa Fluor 647) heavily used in SMLM.
• Prior fluorogenic strategies for polymethines: Installation of nucleophilic side chains (alcohols, amines, thiols) to form oxazines, diazines, and thiazines; coumarin-hemicyanine scaffolds with p-nitrophenol groups enabling reversible photoinduced opening, or hydroxyethyl ring-closing motifs for esterase-activatable probes. However, these cyclizations (often 5-endo-trig or 6-endo-trig) have underperformed relative to rhodamines, lacking robust fluorogenicity.
• Design heuristic: Baldwin’s rules predict 5-endo-trig cyclizations are disfavored whereas 5-exo-trig are favored for polar cyclizations, motivating a switch in cyclization topology to achieve efficient fluorogenic polymethine dyes.

Design and computation:

• Applied Baldwin’s rules to favor 5-exo-trig over 5- or 6-endo-trig intramolecular cyclizations in polymethine dyes.
• Performed DFT calculations (Gaussian 09) at B3LYP/DGTZVP and M06-2X/DGTZVP with IEFPCM (water) to estimate ring-opening and ring-closing energies for HMSiR (reference 5-exo-trig rhodamine) and Cy5 derivatives designed for 5-endo-trig (2), 6-endo-trig (3), and 5-exo-trig (4a/4b). Stationary points were validated by frequency analysis; transition states had one imaginary frequency along C–O bond elongation.

Synthesis:

• Prepared indoleninium building blocks 5, 6, 7a–d and linker 8 per literature. Constructed Cy5 5-endo-trig probes 2a/2b via microwave-assisted steps; 5-exo-trig Cy5 probes 4a/4b via ester intermediates 9a–d, then LiAlH4 reduction. Diversified substituents (for example, CF3) to tune electronic properties.
• For fluorogenic Cy5 with amide nucleophile: Converted 9c methyl ester to N-methyl amide 12 (5-exo-trig), also synthesized matched 5-endo-trig indoleninium 13 and Cy5 probe 14; conjugated alkyne-bearing dyes to benzyl guanine (10) to yield SNAP-tag probes 15 (5-exo-trig) and 16 (5-endo-trig).
• Prepared SNAP-reactive 11 by clicking azide-modified SNAP ligand 10 (from diazotization reagent FSO2N3) to alkyne-modified 4c.
• Extended scaffold to Cy3 (19) and Cy7 (20) by selecting linkers and adjusting substituents: CF3 on capping indoleninium to favor closed form in Cy3; electron-deficient amide to facilitate ring opening in Cy7.

Spectroscopy and physicochemical profiling:

• pH titrations (2–13) in buffered solutions; dielectric constant titrations (water–dioxane mixtures) to assess cyclization equilibria. Determined ring-opening pKa values and propensity to cyclize in low dielectric media.
• Absorbance and fluorescence spectra collected in PBS; extinction coefficients from Beer–Lambert plots; absolute quantum yields using integrating sphere.

Protein/biomolecule binding assays:

• SNAP-tag binding: Incubated probes (2.5 μM) with purified SNAP-tag protein (5 μM) in PBS, 1.5 h, 37 °C; measured absorbance and fluorescence turn-ons.
• Actin binding: 17-actin (2.5 μM) with G-actin under polymerization conditions; measured turn-ons.
• DNA binding: 18-DNA (1 μM) with 50 μM hairpin dsDNA in TBS; measured turn-ons.

Cell imaging:

• HeLa culture and transient transfection with SNAPf fusions (for example, H2B-SNAPf-mTurquoise2, LifeAct-SNAPf-mTurquoise2, TOM20-SNAPf-mTurquoise2). No-wash imaging after incubation with probes (typical 50–250 nM for SNAP probes; 250–500 nM for small-molecule-targeted probes). Spinning disk confocal microscopy at 37 °C, 5% CO2.
• Untargeted dye localization: Compared 5-endo-trig (2a/2b) versus 5-exo-trig (4a/4b) Cy5 derivatives.

Super-resolution and multiplexing:

• SMLM: Expressed β-tubulin-SNAP in HeLa; incubated with 11 (100 nM); imaged at 638 nm without blinking buffers; localized molecules with ThunderSTORM.
• FLIM: Co-transfected H2B-SNAPf-mTurquoise2 and TUBB5-Halo; imaged Cy3 19 and JF549-Halo; phasor analysis to separate species by lifetime.

Stereochemical studies:

• Resolved enantiomers (+)-6 and (−)-6 by chiral HPLC; assigned absolute configurations via ECD and TD-DFT (CAM-B3LYP/DGTZVP). Prepared enantiopure (R)-15 and (S)-15; compared SNAP-tag fluorogenicity and cell labeling.

Photostability/brightness:

• Measured brightness (εφ) and photobleaching kinetics of SNAP conjugates 15, 19, 20 in PBS at excitation maxima with controlled power (1.2 mW).

• Design validation by computation: DFT showed 5-exo-trig cyclizations (HMSiR; Cy5 4a/4b) have larger ring-opening and smaller ring-closing barriers than 5- or 6-endo-trig designs (Cy5 2, 3), predicting greater stability of the closed (non-fluorescent) state and efficient ring opening upon binding.
• pH-dependent equilibria: 5-endo-trig Cy5 probes had high ring-opening pKa (2a: 11.1; 2b: 8.5), implying they remain mostly open and fluorescent in cells; 5-exo-trig probes had lower pKa (4a: 6.4; 4b: 5.7), favoring the closed form under physiological conditions and minimizing background.
• Cellular background: Untargeted 5-endo-trig dyes (2a, 2b) showed bright mitochondrial fluorescence (non-specific) due to delocalized positive charge; 5-exo-trig dyes (4a, 4b) were faint, mainly lysosomal, confirming low background from stable closed forms in cells.
• SNAP-tag fluorogenicity and spontaneous blinking (Cy5): Probe 11 (5-exo-trig, from 4c + SNAP ligand) showed 10-fold absorbance and 21-fold fluorescence turn-on with purified SNAP-tag. In live-cell SMLM without additives, 11 displayed spontaneous blinking, enabling ~60–65 nm microtubule FWHM and single-molecule localization precision of 18 ± 6 nm.
• No-wash fluorogenic Cy5: Probe 15 (5-exo-trig N-methyl amide) showed 6-fold absorbance and 19-fold fluorescence turn-on with SNAP-tag; performance comparable to JF646-BG in no-wash live-cell imaging, with bright nuclear labeling and low background. The 5-endo-trig matched control 16 showed weak turn-on (1.4× absorbance, 2.5× fluorescence), poor membrane permeability, and vesicular signal.
• Generality to other targets: 17-actin (jasplakinolide-linked) exhibited 11-fold absorbance and 313-fold fluorescence turn-on upon binding actin filaments; 18-DNA (Hoechst-linked) showed 7-fold absorbance and 32-fold fluorescence turn-on upon binding dsDNA. Both specifically labeled their targets in live HeLa cells.
• Spectral extension (Cy3 and Cy7): Cy3 probe 19 and Cy7 probe 20 were fluorogenic with SNAP-tag, showing 6× and 11× absorbance increases and 28× and 124× fluorescence turn-ons, respectively. No-wash imaging confirmed specific nuclear labeling.
• Brightness and photostability (SNAP conjugates): εφ (M⁻¹ cm⁻¹) in PBS: 19 = 4,931; 15 = 36,432; 20 = 27,347. Probes 15 and 20 showed good photostability; 19 was less bright and less photostable but potentially tunable by substitution/rigidification.
• Multiplexing: Orthogonal multicolor imaging with 15 (SNAP) and JF549-Halo demonstrated compatibility with rhodamine-HaloTag dyes. FLIM enabled lifetime-based separation of spectrally overlapping Cy3 19 (2 ns) and JF549-Halo (4 ns) signals using phasor analysis.

The results validate the central hypothesis that enforcing a 5-exo-trig intramolecular cyclization creates a robust closed (dark) state in polymethine dyes that can be selectively opened and brightened upon target binding. Computation and experiments align: 5-exo-trig designs feature favorable energy landscapes (high ring-opening barriers, low ring-closing barriers) versus disfavored 5-endo-trig topologies. Consequently, 5-exo-trig Cy dyes remain non-fluorescent as free molecules under physiological conditions, drastically reducing cellular background, yet display substantial fluorescence turn-on upon binding to protein (SNAP-tag) or small-molecule targets (actin, DNA). The approach generalizes across the polymethine family (Cy3, Cy5, Cy7), spanning visible to near-infrared emission, and yields probes with strong turn-on ratios, high brightness (notably Cy7 20), and practical photostability. This strategy addresses long-standing limitations of fluorogenic rhodamines in the NIR by leveraging the modular synthesis of polymethines while retaining high performance in live-cell no-wash imaging and enabling advanced modalities (SMLM via spontaneous blinking with Cy5 11; multicolor and lifetime multiplexing with rhodamine-HaloTag dyes). Preliminary modeling suggests that binding-induced ring opening may be mediated by specific interactions (for example, hydrogen bonding) between lactam and protein amides or DNA phosphates, offering a path to tune fluorogenic responses through target-interaction design.

The study establishes a simple, general 5-exo-trig ring-closure design to confer binding-induced fluorogenicity to polymethine dyes. Using modular syntheses, the authors created: (i) a spontaneously blinking Cy5 dye (11) for additive-free live-cell SMLM; (ii) a no-wash fluorogenic Cy5 dye (15) with performance comparable to JF646-BG; (iii) targeted fluorogenic probes for actin (17-actin) and DNA (18-DNA) with large fluorescence turn-ons; and (iv) fluorogenic Cy3 (19) and bright NIR Cy7 (20) probes, enabling multicolor imaging and lifetime-based multiplexing. Cy7 20 is particularly notable for its high brightness (εφ ≈ 27,300 M⁻¹ cm⁻¹) and long emission wavelength (λem ≈ 784 nm), outperforming fluorescent proteins in a similar spectral range. Future directions include elucidating the molecular mechanism of binding-induced ring opening (for example, molecular dynamics and site-directed mutagenesis), tuning photophysical properties and fluorogenicity via substituents and linker chemistry, and extending the strategy across the polymethine spectrum toward green (Cy1) and shortwave infrared (Cy9) dyes for deeper tissue imaging and broader multiplexing.

• Mechanism of fluorescence turn-on remains to be fully elucidated; current evidence suggests roles for hydrogen-bonding interactions with proteins or nucleic acids.
• Cy3 derivative 19 is less bright and photostable than Cy5 15 and Cy7 20, indicating a need for further optimization (for example, substituents, chain rigidification).
• Some 5-endo-trig controls (for example, 16) showed poor membrane permeability and vesicular accumulation, highlighting that topology strongly affects cellular behavior; while 5-exo-trig reduces background, uptake and permeability may vary by design.
• Cellular uptake of probe 11 was not higher than its 5-endo-trig analogue, though labeling specificity improved; optimization of delivery could further enhance performance.
• Fluorogenic response and equilibrium depend on environmental pH and polarity; performance in diverse biological milieus may require tuning of substituents.