A General Strategy for Development of Near-Infrared Fluorescent Probes for Bioimaging

Abstract

Near-infrared (NIR) fluorescent dyes with favorable photophysical properties are highly useful for bioimaging, but such dyes are still rare. The development of a unique class of NIR dyes via modifying the rhodol scaffold with fused tetrahydroquinoxaline rings is described. These new dyes showed large Stokes shifts (> 110 nm). Among them, WR3, WR4, WR5, and WR6 displayed high fluorescence quantum yields and excellent photostability in aqueous solutions. Moreover, their fluorescence properties were tunable by easy modifications on the phenolic hydroxy group. Based on WR6, two NIR fluorescent turn-on probes, WSP-NIR and SeSP-NIR, were devised for the detection of H₂S. The probe SeSP-NIR was applied in visualizing intracellular H₂S. These dyes are expected to be useful fluorophore scaffolds in the development of new NIR probes for bioimaging.

Fluorescent probes are essential chemical tools for bioimaging owing to their high sensitivity, fast response, and ease of use. To date, a variety of fluorescent dyes have been employed as signaling units to construct fluorescent probes.^[1] Among these, xanthene-based dyes are highly popular owing to their excellent photophysical properties, including high extinction coefficients and quantum yields, excellent photostability, and water solubility. However, these traditional xanthene dyes usually emit green to red fluorescence (approximately 500–600 nm) with small Stokes shifts (typically less than 30 nm).^[2] Fluorescence emission in this region suffers from limited tissue penetration and low signal-to-noise ratios. On the other hand, near-infrared (NIR) emission (650–900 nm) has distinct advantages including low photodamage to biological samples, deep tissue penetration, and minimum interference from background autofluorescence in living biosystems.^[3] Furthermore, small Stokes shifts can result in serious self-quenching of fluorescence and measurement error from excitation backscattering.^[4] Accordingly, the development of xanthene-based NIR fluorescent dyes with large Stokes shifts is highly desirable and valuable for bioimaging.

So far, a number of modified xanthene dyes have been reported. Common strategies include the extension of π-conjugation system and the replacement of the endocyclic oxygen with other elements (such as C and Si).^[5] These pioneering works have resulted in the discovery of dyes with relatively long wavelengths or NIR emission. However, xanthene-based NIR dyes possessing large Stokes shifts, high fluorescence quantum yields in aqueous solution, and optically tunable groups that can be modified to develop reaction-based probes are still lacking. To this end, we have initiated a program to develop novel NIR xanthene dyes, attempting to address aforementioned problems. Herein, we report the development of a new class of rhodol-based NIR dyes, named the Washington Red (WR) series. Their application in developing reaction-based fluorescent probes is also reported.

Two factors were considered in our design of new NIR xanthene dyes: 1) The free hydroxy or amino group in the xanthene core are known to be excellent optically tunable groups and their substitution can lead to fluorescence quenching.^[2] We decided to keep a hydroxy group in the xanthene core so it could serve as a useful “on/off” switch for the development of reaction-based fluorescent probes; 2) it is known that the spectroscopic properties of rhodamine and rhodol dyes can be significantly influenced by the substituents on the nitrogen atom.^[6] Increasing the electron-donating power of the substituents or introducing rigid rings generally induced bathochromic shifts in dyes. As rhodols contain a free hydroxy group and a substitutable amino group (Scheme 1), their core structure was selected as our research framework. We expected the incorporation of one or more heteroatom containing fused rings would improve their spectroscopic properties. As the first proof-of-concept, tetrahydroquinoxaline was selected owing to synthetic feasibility. Two rhodol derivatives (WR1 and WR2) with the tetrahydroquinoxaline unit fused to the amino side were prepared (Scheme 1).

Scheme 1.

The structures of rhodols, WR1, and WR2.

The photophysical properties of WR1 and WR2 were tested in 50 mm PBS buffer solution (pH 7.4) containing 15% ethanol. The results are compiled in Table 1. The solutions of WR1 and WR2 displayed pink color because of their large molar extinction coefficients (ε_max). It should be noted that the maximum absorption wavelengths of rhodols with two N-alkylations are usually located in the range of 518–525 nm, and their maximum emission wavelengths are in the range of 544–555 nm.^[6b] As expected, the absorption and emission maxima of WR1 and WR2 displayed distinct bathochromic shifts (Figure 1). This unprecedented NIR fluorescence emission with large Stokes shifts (> 120 nm) should be attributed to the excited-state intramolecular charge transfer between the electron donor and acceptor. However, WR1 and WR2 showed weak fluorescence and low fluorescence quantum yields in buffer solutions, which is presumably due to the formation of a twisted internal charge transfer (TICT) state.^[7]

Table 1.

Photophysical data of WR1–WR10 in 50 mm PBS buffer solution (pH 7.4) containing 15% ethanol.

Dye	λabs[nm]a	εmaxb	λem[nm]c	Stokes shiftd	Φfe
WR1	539	31000	668	129	0.06
WR2	545	30000	668	123	0.07
WR3	540	35500	656	116	0.20
WR4	543	44500	656	113	0.21
WR5	545	32500	662	117	0.19
WR6	551	42000	662	111	0.20
WR7	539	29500	676	137	0.06
WR8	545	30000	676	131	0.05
WR9	557	39000	706	149	0.04
WR10	542	30500	666	124	0.06

^aThe maximal absorption of the dye.

^bUnits: Lmol⁻¹cm⁻¹.

^cThe maximal emission of the dye.

^dUnits: nm.

^eΦ_f is the relative fluorescence quantum yield estimated by using seminaphthofluorescein-10 (Φ_f = 0.35 in pH 11 Glycine-NaOH buffer) as a reference standard.^[8]

Figure 1.

A) Normalized absorption and B) fluorescence emission spectra of WR1 (gray) and WR2 (black).

Recent studies found that replacing the N,N-dimethyl group in tetramethylrhodamine (TMR) with a rigid four-membered azetidine or a five-membered pyrrolidine ring could mitigate TICT and enhance quantum yields. The latter could also induce a slight bathochromic shift.^[9] We reasoned that this strategy could resolve the deficiency of WR1 and WR2. Therefore, we synthesized analogues of WR1 and WR2 that incorporate a pyrrolidine ring. A small library of compounds (WR3–WR10, Scheme 2) were prepared to investigate the location of pyrrolidine and other nitrogen-containing rings or influence of substituents on the spectroscopic properties of the dyes. The photophysical properties of WR3–WR10 were studied (Table 1) under identical conditions used for WR1 and WR2. These dyes all exhibited a pink color and intense maximum absorption bands in the range 539–557 nm (Supporting Information, Figure S1). Their maximal fluorescence emissions were still located in NIR region (656–706 nm; Supporting Information, Figure S2). The Stokes shifts of WR3–WR10 were between 111 nm and 149 nm, which is superior to those of the classic xanthene dyes. This feature could avoid the limitations of small Stokes shift and improve the sensitivity of fluorescence detection. These results led us to believe that the tetrahydroquinoxaline unit was crucial to large Stokes shifts and NIR emission. Furthermore, TD-DFT (time-dependent density functional theory) calculation results are consistent with the measured absorption spectrum of these dyes (Supporting Information, Table S1 and Figure S3).

Scheme 2.

The structures of WR3–WR10.

To understand how the fused ring of WR3–WR10 regulates fluorescence intensity, we also compared their fluorescence quantum yields with WR1 and WR2. When the pyrrolidine ring was located at the meta-position of endocyclic oxygen atom, the condensed ring apparently enhanced their fluorescence quantum yields (for WR3–WR6). In these dyes, different alkylations on the para-nitrogen atom only had minor effects on quantum yields. Briefly, WR3–WR6 had similar fluorescence quantum yields at about 0.2, which is quite high for NIR dyes. However, when the pyrrolidine ring was located at the para-position of the endocyclic oxygen atom (WR7 and WR8), or was replaced by a six-membered piperidine ring (WR9), the resultant dyes showed obvious bathochromic shifts in emission wavelengths but their quantum yields were not improved. Adding another condensed ring substituent (WR10) or introducing a chloride to the xanthene core (WR6, WR9) also did not display the regulatory function for enhancing fluorescence quantum yields.

Among these new dyes, WR3–WR6 appeared to be most promising due to their high quantum yields. They were therefore selected for further evaluation. In the design of reaction-based fluorescent probes, the –OH group in these dyes served as the optically tunable platform. The deprotonation/protonation of the phenolic hydroxy group would dramatically affect their fluorescence properties. We then investigated the effects of pH on their fluorescence profiles (Figure 2). WR3–WR5 displayed weak fluorescence in pH range from 2.5 to 5.0 while the increment of pH from 5.0 to 7.4 resulted in significant fluorescence enhancement. The pK_a values of WR3–WR5 were determined to be 6.2. These results demonstrate that these dyes are suitable for effective fluorescence detection under physiological pH. Interestingly, the introduction of a Cl to the dye, for example, WR6, was found to decrease the pK_a to 4.7. This dye could display strong fluorescence even under acidic pH (4.0), suggesting it has a broader pH application range. Furthermore, the photo-stability and cytotoxicity of WR3–WR6 also were evaluated (Supporting Information, Figures S4, S5). The results showed that these dyes were stable to continuous irradiation for 3 hours and they had no obvious toxicity. In all, the data indicates that WR3–WR6 can be useful fluorophores for the design of NIR fluorescent probes by modification on the phenolic hydroxy group.

Figure 2.

pH-dependence of the fluorescence intensity of WR3–WR6.

As a proof-of-concept we next applied WR dyes in the development of novel NIR probes for the detection of hydrogen sulfide (H₂S). H₂S is a gasotransmitter that modulates diverse cellular functions. Although many fluorescence probes for H₂S have been reported in the past several years, few can be applied to real biological detection.^[10] In 2011, our laboratory developed the first H₂S nucleophilic reaction-based probes WSP series (Scheme 3).^[11] A drawback of the WSP series is that the probes can be consumed by biothiols and thus high probe loading is needed when biothiols are presented. To solve this problem we developed diselenide-based SeP probes in 2015.^[12] The unique reaction mechanism between SeP and H₂S (Scheme 3) eliminated the problem associated with the presence of biothiols. However, in SeP probes two identical fluorophores are needed in one molecule, which does not allow much flexibility to tune their reactivity toward H₂S. That also results in water solubility problem. Additionally, the maximum emission wavelengths of those probes are below 600 nm with small Stokes shifts. These problems could be addressed by the use of our new WR dyes.

Scheme 3.

Sensing mechanisms of WSP/SeP probes.

We decided to use WR6 to construct two NIR probes for comparison (Scheme 4): WSP-NIR adopts the WSP series template; SeSP-NIR adopts a modified SeP template, which mimics the intermediate II of SeP reacting with thiols. We expected SeSP-NIR should react more effectively with H₂S and retain the high selectivity. With a reactive Se–S bond in the structure, SeSP-NIR could also react with thiols. However, the product should still possess the –Se–S–linkage and can be easily turned on by H₂S. As such the probe should not be consumed by thiols. This new design made the probes more synthetically feasible and structurally tunable.

Scheme 4.

The structures of new NIR fluorescent probes for H₂S.

The syntheses of both probes are detailed in the Supporting Information. With them in hand, we first studied their fluorescence properties and responses to H₂S. As expected, WSP-NIR and SeSP-NIR showed non-fluorescent (quantum yields < 0.01), indicating the protection of the hydroxy of WR dyes can effectively quench their fluorescence. Upon the treatment of H₂S, the two probes gave dramatic fluorescence enhancements, concomitant with a distinct color change from nearly colorless to pink (Supporting Information, Figure S6). The maximum emission intensities of WSP-NIR and SeSP-NIR were reached within 2 min and 8 min, respectively, indicating their fluorescence turn-on was a fast process.

To examine the selectivity of the two probes for H₂S, WSP-NIR and SeSP-NIR were separately treated with a series of reactive sulfur species including glutathione (GSH), cysteine (Cys), homocysteine (Hcy), glutathione disulfide (GSSG), Na₂S, SO₃²⁻. and S₂O₃²⁻. As shown in Figure 3, no significant fluorescence increase was observed except for Na₂S. The responses of the probes to other common amino acids were also tested, and no response was found. We also tested the responses of the probes to H₂S when high concentrations of biothiols were present. As seen in Figure 3, WSP-NIR showed dramatically decreased fluorescence. In contrast, SeSP-NIR still elicited strong fluorescence (Figure 3; Supporting Information, Figure S7). These results clearly demonstrate the advantage of using the new Se–S bond based template in the development of H₂S probes, for example, avoiding probe consumption by biothiols.

Figure 3.

Fluorescence response of WSP-NIR (2 mm) or SeSP-NIR (2 mm) in the presence of various reactive sulfur species and amino acids: 1) probe alone; 2) 100 μM GSSG; 3) 100 μM Na₂S₂O₃; 4) 100 μM Na₂SO₃; 5) 100 μM alanine; 6) 100 μM arginine; 7) 100 μM glutamine; 8) 100 μM glycine; 9) 100 μM proline; 10) 100 mm serine; 11) 2 mm GSH; 12) 10 mm GSH; 13) 1 mm Cys; 14) 1 mm Hcy; 15) 2 mm GSH+50 μM Na₂S; 16) 1 mm Cys+ 50 μM Na₂S; 17) 1 mm Hcy +50 μM Na₂S; 18) 50 μM Na₂S.

Based on the excellent sensing properties of SeSP-NIR for H₂S, the fluorescence responses of SeSP-NIR to a series of varied concentrations of Na₂S (0–20 mm) also were tested. The maximum emission intensity increased linearly with Na₂S in the concentration range of 0–3 μM (Supporting Information, Figure S8). The detection limit was calculated to be 6 nM, indicating a high sensitivity. To clarify the fluorescence turn-on mechanism of SeSP-NIR by H₂S, we studied the model reaction between phenyl 2-((pyridin-2-ylthio)selenyl)benzoate and Na₂S in a mixed solution of THF/PBS (pH 7.4, 1:1 v/v). As expected, 2,1-benzothiaselenol-3-one and phenol were isolated with high yields. This result confirmed the H₂S-mediated nucleophilic substitution-cyclization mechanism in Scheme 3.

Finally we wondered whether SeSP-NIR could be used for imaging endogenously produced H₂S in cells. As shown in Figure 4 and the Supporting Information, Figure S9, HeLa cells were first treated with a H₂S biosynthesis inhibitor propargylglycine (PAG) or a H₂S donor N-(benzoylthio)benzamide.^[13] Then SeSP-NIR (4 mm) was loaded in the cells for 20 min at 378C. A significant reduced fluorescence was observed in cells treated with PAG (Figure 4b; Supporting Information, Figure S9b) as compared to control cells (Figure 4a; Supporting Information, Figure S9a). In contrast, cells treated with the H₂S donor showed substantial fluorescence enhancement (Figure 4c; Supporting Information, Figure S9c). These results indicate SeSP-NIR has good cell permeability and is sensitive to detect endogenous H₂S. Furthermore, the cell viability assay demonstrated that SeSP-NIR has almost no cytotoxicity to cells (Supporting Information, Figure S10).

Figure 4.

Fluorescence imaging of H₂S in HeLa cells. Cells were imaged after incubation with 4 μM SeSP-NIR for 20 min after pretreatment with a) no pretreatment, b) 60 μM PAG for 40 min, or c) 100 μM H₂S donor for 60 min.

In summary, we report in this study the design, synthesis, and evaluation of a unique class of rhodol-based NIR dyes with large Stokes shifts. Among them, WR3–WR6 displayed high fluorescence quantum yields and excellent photostability in aqueous solutions. Moreover, the fluorescence properties of WR3–WR6 were found to be tunable by easy modifications on the phenolic hydroxy group. To further show the value of these new dyes, two probes, WSP-NIR and SeSP-NIR, were devised for the detection of H₂S by using WR6 as the fluorophore. SeSP-NIR showed much improved sensitivity to H₂S, and overcame a major limitation of previously reported WSP and SeP series probes. SeSP-NIR was also applied in visualizing intracellular H₂S. The dyes presented herein are expected to be useful fluorophore scaffolds in the development of new NIR probes for bioimaging.