Synthetic green fluorescent protein chromophore analogues with a positive charge at the phenyl-like group

Abstract

Synthetic green fluorescent protein (GFP) chromophore analogues with a positive charge at the phenyl-like group have the highly electrophilic amidine carbon, smaller LUMO–HOMO energy gap, red-shifted electronic absorptions and fluorescent emissions, and accelerated E-Z thermoisomerization rates. They are water-labile and their hydrolysis results in ring-opening of the imidazolinone moiety with a half life around 25–37 h in D₂O at 25 °C.

Introduction

Green fluorescent protein (GFP) and its mutants have attracted interest as fluorescent biological labels in recent two decades (Meech 2009; Zimmer 2002). The wild-type GFP (wtGFP) has a globular protein of 238 amino acids that are folded into an 11-stranded β-barrel (Ormo et al. 1996; Yang et al. 1996; Suhling et al. 2002). Its chromophore, p-hydroxybenzylideneimidazolinone (p-HBDI), sits at the center of the β-barrel and is biosynthesized from Ser(65)-Tyr(66)-Gly(67) (Ormo et al. 1996; Yang et al. 1996; Suhling et al. 2002). The wtGFP has a major electronic absorption at 395 nm, which is called A band and attributed to the electronic absorption of the neutral form of the p-HBDI chromophore and a minor electronic absorption at 475 nm, which is B band and assigned to the electronic absorption of the anionic form of the p-HBDI chromophore (Heim et al. 1994; Chattoraj et al. 1996). On excitation of the A band, the A* state undergoes a rapid excited-state proton transfer (ESPT) (Tolbert et al. 2002; Han et al. 2011; Wan et al. 1993; Formosinho et al. 1993) to form the anionic I* state, which emits dominant GFP fluorescence at 508 nm with 80% quantum yield. After proton recombination, the ground-state I form quickly reforms the ground-state A form. On excitation of the B band, the B* state emits fluorescence at 503 nm. Conversion between the I and B states might involve reorganization of the surrounding protein matrix, such as isomerization of the Thr203 residue, but it occurs on a much slower timescale (Heim et al. 1994; Chattoraj et al. 1996; Brejc et al. 1997).

Due to the nature of the space and shape limitations of the protein environment, the GFP chromophore (p -HBDI) inside the β-barrel can hardly decay through Z-E photoisomerization by 180° rotation around the I-bond (Chen et al. 2001), resulting in a high fluorescence quantum yield of 0.8 in the wtGFP. However, p -HBDI significantly loses fluorescence after it is stripped of the protein environment (Meech 2009; Zimmer 2002). Hence, many synthetic GFP chromophore analogues have been prepared to improve fluorescence quantum yield and/or emission properties (Yang et al. 2008; Chen et al. 2013; Lo et al. 2013; Wu et al. 2008; Huang et al. 2012; Chuang et al. 2011; Follenius-Wund et al. 2003). For our laboratory, we took advantage of low fluorescence quantum yield of p -HBDI and its analogues to develop a sensor for DNA/RNA. Because the backbone of DNA/RNA carries negative charges, we prepared synthetic GFP chromophore analogues with a positive charge at the phenyl-like group as a sensor for DNA/RNA. Meanwhile, to tune the properties of the GFP chromophore, we also liked to know what will happen if synthetic GFP chromophore analogues carry a positive charge at the phenyl-like group. Even though many analogues of the GFP chromophore have been prepared and studied, to our knowledge, synthetic GFP chromophore analogues with a positive charge at the phenyl-like group have not been reported yet.

After we prepared synthetic GFP chromophore analogues with a positive charge at the phenyl-like group (2a and 2b) and tested their DNA/RNA sensitivity in water, we found that they are labile in water. The information about how a positive charge on the phenyl-like group makes the GFP chromophore analogues labile in water and affects their various properties may be important to chemists who develop synthetic GFP chromophore analogues.

Materials and methods

Computational details

All the calculations were performed with the Gaussian09 program (Frisch et al. 2009). Geometry optimization of p-HBDI, 1a and 2a was carried out at the B3LYP/6-31+G* without any symmetry restriction. After the geometric optimization, analytical vibration frequencies were calculated at the same level to determine the nature of the located stationary point. The electrostatic potential maps for the electronic ground states of p-HBDI, 1a, and 2a were calculated at the B3LYP/6-31+G* level.

Method for measurement of fluorescence quantum yields of 1a,b and 2a,b

The fluorescence quantum yields of 1a,b and 2a,b were determined by the comparative method (Williams et al. 1983), and measured by comparing the wavelength-integrated intensity of the test samples to a standard. The standard used in this article is quinine sulfate in 0.1 M H₂SO₄ with the known fluorescence quantum yield of 0.577 under the excitation wavelength of 350 nm (Eastman 1967). The fluorescence quantum yields of 1a,b and 2a,b were calculated using

$$ Q \, = \, Q_{R} \left( {I/I_{R} } \right)\left( {{\text{OD}}_{R} /{\text{OD}}} \right)\left( {n^{2} /n_{R}^{2} } \right), $$

where Q is the fluorescence quantum yield, I is the integrated fluorescence intensity, n is the refractive index of the solvent, and OD is the optical density. The subscript R refers to the reference fluorophore with known fluorescence quantum yield. In this article, the reference fluorophore is quinine sulfate in 0.1 M H₂SO₄. In this method, solutions of quinine sulfate, 1a,b and 2a,b were illuminated with the light of the same excitation wavelength (350 nm), where quinine sulfates, 1a,b and 2a,b, have significant absorption. Solutions of 1a,b and 2a,b were prepared with optical densities in the range of 0.1–0.01 for the sake of accuracy.

General method to prepare 1a and 1b

8 mL of acetic anhydride was added isonicotinaldehyde or nicotinaldehyde (10 mmol), N-acetylglycine (1.17 g, 10 mmol), and sodium acetate (0.82 g, 10 mmol). The mixture was stirred in a nitrogen atmosphere at 100 °C for 75 min. After reaction, the solution was cooled by ice water. The solution was stirred until most of brown solid precipitated. After filtration, the brown solid was purified by recrystalization in aqueous EtOH.

The brown solid in 12 mL of chloroform was slowly added aniline or benzylamine (10 mmol). The mixture was stirred in a nitrogen atmosphere at room temperature for around 4.5 h until the reaction was complete. Then, the solution was concentrated by a rotary evaporator, and pyridine (7.5 mL) was added to the residue. The mixture was stirred in a nitrogen atmosphere at a reflux condition for 5 h. After reaction, the solution was cooled down and some ice was added to the solution. The solution was stirred until most of solid precipitated. The crude product was purified by column chromatography with hexane/ethyl acetate (1:1) as a mobile phase to get the product of 1a or 1b.

(Z)-2-Methyl-1-phenyl-4-(pyridin-4-ylmethylene)-1H-imidazol-5(4H)-one (1a). color: orange yellow; yield: 50%; ¹HNMR (400 MHz, CDCl₃) δ 2.30 (s, 3H), 7.06 (s, 1H), 7.23–7.26 (m, 2H), 7.45–7.55 (m, 3H), 8.01 (d, J = 6.1 Hz, 2H), 8.69(d, J = 6.1 Hz, 2H); ¹³C NMR (100 MHz, CDCl₃) δ 16.6, 124.1, 125.2, 127.2, 129.1, 129.8, 133.0, 140.9, 141.6, 150.4, 164.4, 169.4; HRMS [FAB, (M + H)⁺] m/z calcd for C₁₆H₁₄N₃O 264.1137, found 264.1142.

(Z)-1-Benzyl-2-methyl-4-(pyridin-4-ylmethylene)-1H-imidazol-5(4H)-one (1b) color: orange yellow; yield: 70%; ¹HNMR (400 MHz, CDCl₃) δ 2.29 (s, 3H), 4.83 (s, 2H), 7.03 (s, 1H), 7.22–7.34 (m, 2H), 7.37 (m, 3H), 7.95(d, J = 5.9 Hz, 2H), 8.67(d, J = 5.9 Hz, 2H); ¹³C NMR (100 MHz, CDCl₃) δ 16.2, 44.0, 123.8, 125.2, 127.0, 128.1, 129.0, 135.6, 140.9, 141.8, 150.2, 165.1, 170.1; HRMS [FAB, (M + H)⁺] m/z calcd for C₁₇H₁₆N₃O 278.1290, found 278.1291.

General method to prepare 2a and 2b

1a or 1b (2.5 mmol) in 7 mL of CHCl₃ was added 0.74 mL (12 mmol) of methyl iodide. The mixture was stirred in a nitrogen atmosphere at room temperature overnight until the reaction was complete. After filtration, the solid product was washed with CHCl₃. The product was purified by recrystallization in CHCl₃ to get the product of 2a or 2b.

(Z)-1-methyl-4-((2-methyl-5-oxo-1-phenyl-1H-imidazol-4(5H)-ylidene)methyl)pyridinium iodide (2a). color: dark brown; yield 85%;¹HNMR (400 MHz, D₂O) δ 2.24 (s, 3H), 4.28 (s, 3H), 7.15 (s, 1H), 7.30 (d, J = 6.5 Hz, 2H), 7.51–7.55 (m, 3H), 8.44 (d, J = 6.5 Hz, 2H), 8.68 (d, J = 6.5 Hz, 2H); ¹³C NMR (100 MHz, DMSO) δ 17.2, 48.1, 117.6, 128.2, 128.4, 129.7, 130.1, 133.2, 145.9, 146.2, 149.4, 169.1, 170.2; HRMS [FAB, M⁺] m/z calcd for C₁₇H₁₆N₃O 278.1293, found 278.1293.

(Z)-1-methyl-4-((2-methyl-5-oxo-1-benzyl-1H-imidazol-4(5H)-ylidene)methyl)pyridinium iodide (2b). color: dark brown; yield 85%; ¹HNMR (400 MHz, D₂O) δ 2.31 (s, 3H), 4.27 (s, 3H), 4.83 (s, 2H), 7.12 (s, 1H), 7.23 (m, 2H), 7.32 (m, 1H), 7.35 (m, 2H), 8.39 (d, J = 6.6 Hz, 2H), 8.65 (d, J = 6.6 Hz, 2H); ¹³C NMR (100 MHz, CDCl₃) δ 16.7, 44.3, 49.1, 117.0, 127.1, 128.4, 128.8, 139.2, 135.0, 145.0, 146.6, 150.0, 169.4, 170.4; HRMS [FAB, M⁺] m/z calcd for C₁₈H₁₈N₃O 292.1450, found 292.1454.

General method for hydrolysis of 2a and 2b

2a or 2b (2 mmol) was added 5 mL of water. The mixture was stirred at room temperature until the reaction was complete. After reaction, the solution was concentrated by a rotary evaporator. The solid product was purified by recrystallization in water and acetonitrile to get the product of 3a or 3b.

(Z)-4-(2-acetamido-3-oxo-3-(phenylamino)propenyl)-1-methyl-pyridinium iodide (3a). yield: quantitative; ¹HNMR (400 MHz, D₂O) δ 2.11 (s, 3H), 4.26(s, 3H), 7.03(s, 1H), 7.24(t, J = 6.76 Hz, 1H), 7.36–7.40(m, 4H), 7.98(d, J = 6.24 Hz, 2H), 8.63(d, J = 6.2 Hz, 2H); ¹³C NMR (100 MHz, DMSO) δ 21.7, 47.2, 120.4, 122.2, 126.0, 126.5, 128.8, 135.6, 136.1, 144.5, 150.1, 164.7, 173.2; HRMS [FAB, M⁺] m/z calcd for C₁₇H₁₈N₃O₂ 296.1399, found 296.1395.

(Z)-4-(2-acetamido-3-oxo-3-(benzylamino)propenyl)-1-methyl-pyridinium iodide (3b). yield: quantitative; ¹HNMR (400 MHz, D₂O) δ 2.05 (s, 3H), 4.24 (s, 3H), 4.42 (s, 2H), 6.95 (s, 1H), 7.27 (m, 2H), 7.29 (m, 1H), 7.33 (m, 2H), 7.92 (d, J = 6.6 Hz, 2H), 8.61 (d, J = 6.6 Hz, 2H); ¹³C NMR (100 MHz, D₂O) δ 22.0, 43.4, 47.6, 121.3, 126.9, 127.2, 127.5, 128.6, 128.8, 129.0, 136.1, 137.4, 144.9, 150.5, 166.4, 173.7; HRMS [FAB, M⁺] m/z calcd for C₁₈H₂₀N₃O₂ 310.1558, found 310.1560.

Results and discussion

Synthesis and structure

Synthesis of the compounds 2a and 2b, which are synthetic GFP chromophore analogues with a positive charge at the phenyl-like group, is shown in Scheme 1. Reaction of isonicotinaldehyde with N-acetylglycine in the presence of NaOAc/Ac₂O was carried out under reflux, followed by reaction with aniline or benzylamine and cyclization in pyridine, to produce 1a or 1b. Methylation of 1a and 1b with methyl iodide generated 2a and 2b, respectively. According to ¹H and ¹³C NMR spectra of 1a, 1b, 2a, and 2b, they all stay in single configuration. Further studies on their configuration with single-crystal X-ray diffraction confirm that they all stay in Z configuration (Fig. 1).

Scheme 1

Synthesis of the model compounds 2a and 2b

Fig. 1

Single-crystal X-ray diffraction structures of 1a, 2a, 1b, and 2b with thermal ellipsoid at the 50% probability level

Photophysics

As shown in Table 1 and Figs. 2 and 3, the lowest energy electronic absorptions of 1a and 1b in acetonitrile are located at 345 (ε = 7.5 × 10³ M⁻¹ cm⁻¹) and 353 (ε = 1.3 × 10⁴ M⁻¹ cm⁻¹), respectively. They are supposed to involve π → π* electronic absorption. Their fluorescent emissions are located at 461 and 450 nm, respectively. Like p-HBDI (Meech 2009; Zimmer 2002), fluorescent quantum yields of 1a and 1b are as low as 4.5 × 10⁻² and 3.0 × 10⁻². In comparison with 1a and 1b, the lowest energy electronic absorptions and fluorescent emissions of 2a and 2b in acetonitrile are all shifted to longer wavelengths, indicating that the positively charged pyridinium group in 2a and 2b decreases the energy gap of the chromophore between the lowest unoccupied molecular orbital (LUMO) and the highest occupied molecular orbital (HOMO).

Table 1 Electronic absorption λ _A (molar absorptivity ε), fluorescence emission λ _F (fluorescence quantum yield ϕ _f), and band gap energy ΔE of 1a, 1b, 2a, and 2b

Fig. 2

¹H NMR spectra of 2a in D₂O before hydrolysis (top), 21 h after hydrolysis (middle), and at the end of hydrolysis (bottom)

Fig. 3

Single-crystal X-ray diffraction structure of 3a with thermal ellipsoid at the 50% probability level

Reactivity

The compounds 2a and 2b are stable in anhydrous acetonitrile and DMSO, but they are labile in alcohols and water, which are very weak nucleophiles. The whole hydrolysis processes of 2a and 2b were monitored by their ¹H NMR spectra, and only one product was formed for each of them (Fig. 2 and Scheme 2). The structure of the hydrolyzed product (3a) from the compound 2a was confirmed by single-crystal X-ray diffraction (Fig. 3), and it proves that the amidine carbon is the only site that is attacked by very weak nucleophile of water and the rest of the structure of 2a remains intact, resulting in ring-opening of the imidazolinone moiety. Even though the benzylidene double bond of p-HBDI analogues might be susceptible to a nucleophilic attack, according to a series of ¹H NMR spectra in Fig. 2, water nucleophile does not attack the benzylidene double bond of 2a and 2b.

Scheme 2

Hydrolysis of the compounds 2a and 2b

Hydrolysis rates of the compounds 2a and 2b in D₂O were measured by monitoring growth of 3a and 3b or decay of 2a and 2b with ¹H NMR spectrometer at 25 °C, and their rate constants are shown in Table 2. The half lives of the hydrolysis reactions of 2a and 2b in D₂O were measured to be around 25–37 h.

Table 2 Hydrolysis rate constants [k _D2O (s⁻¹)] of 2a,b in D₂O at 25 °C, ¹³C NMR chemical shifts [δ(ppm)] of the amidine carbons of 1a,b and 2a,b, and APT charges (Q) of the amidine carbons of 1a and 2a

According to the computation calculated at the B3LYP/6-31+G* level, the atomic polar tensor (APT) charge (Q) of the amidine carbon of 1a is 0.939, and it turns to 1.412 after 2a is produced by methylation of 1a (Table 2). It indicates that the compound 2a has been turned highly electrophilic at the amidine carbon in comparison with 1a. For the ¹³C NMR spectra of 1a,b and 2a,b, the resonance signals of the amidine carbons of 2a,b are around 5 ppm downfield in comparison with those of 1a,b (Table 2). This evidence confirms that the amidine carbons of 2a,b become highly electrophilic after they are produced by methylation of 1a,b.

It is interesting to know that synthetic GFP chromophore analogues with electron-withdrawing phenyl-like group, such as p-CF₃-phenyl and p-NC-phenyl groups, are stable in water (Follenius-Wund et al. 2003). Even though p-NO₂ group (σ_p for p-NO₂ = 0.78) is highly electron-withdrawing, p-NO₂-phenyl group (σ_p for p-NO₂-phenyl = 0.26) become weakly electron-withdrawing (Hansch et al. 1991). The same thing also happens to p-CF₃-phenyl and p-NC-phenyl groups. However, p-pyridinium group (σ_p for p-pyridinium = 0.58) is still highly electron-withdrawing (Hansch et al. 1991). Hence, it makes sense that 2a and 2b are labile in water, while synthetic GFP chromophore analogues with electron-withdrawing p-CF₃-phenyl and p-NC-phenyl groups are still stable in water.

Electronic structures

To confirm the reactivity of the positively charged p-HBDI chromophore, we calculated the electrostatic potential maps of p-HBDI, 1a, and 2a at B3LYP/6-31+G* level. In the electrostatic potential maps, red color represents an electron-rich area, blue color indicates an electron-deficient area, and other colors represent areas with intermediate electron density. As shown in Fig. 4, the phenol moiety of p-HBDI is electron rich and its amidine carbon is not electron deficient at all. Like p-HBDI, the electrostatic potential map of 1a shows that its pyridine moiety is also electron-rich and its amidine carbon is not electron-deficient, either. After methylation of 1a, not only the pyridinium moiety but also the amidine carbon become electron deficient according to the electrostatic potential map of 2a, indicating that the amidine carbon of the positively charged p-HBDI chromophore with a positive charge at the phenyl-like group becomes highly electrophilic in comparison with 1a and p-HBDI.

Fig. 4

Electrostatic potential maps of p-HBDI, 1a, and 2a at the B3LYP/6-31+G* level

Z-E photoisomerization and ground-state E-Z thermoisomerization

After 20 min of irradiation with 350 nm UV light in a photo-reactor at room temperature, some of 1a, 2a, 1b, and 2b were converted to their respective E-isomers (Figs. 5, 6). These E-isomers are characterized by the downfield-shifted vinyl protons, and this is consistent with the literature (Yang et al. 2008). These E-isomers are less stable than their respective Z-isomers and will be thermally isomerized back to the respective Z-isomers. The Z-E photoisomerization and ground-state E-Z thermoisomerization are reproducible.

Fig. 5

¹H NMR spectra of Z-1b in CD₃CN a before UV irradiation, b right after 20 min of irradiation with 350 nm UV light, and c 8 days after UV irradiation, and the ground-state E-Z thermoisomerization decay rate profile of E-1b in CD₃CN at 25 °C (the bottom figure)

Fig. 6

¹H NMR spectra of Z-2b in CD₃CN a before UV irradiation, b right after 20 min of irradiation with 350 nm UV light, and c 3 days after UV irradiation, and the ground-state E-Z thermoisomerization decay rate profile of E-2b in CD₃CN at 25 °C (the bottom figure)

Ground-state E-Z thermoisomerization of E-1a,b and E-2a,b in CD₃CN or CDCl₃ was followed by monitoring decay of their E-isomers or growth of their Z-isomers with ¹H NMR spectroscopy (Figs. 5, 6). The E-Z thermoisomerization rate constants are shown in Table 3. The E-Z thermoisomerization rate constants of E-1a,b in CDCl₃ are larger than those in CD₃CN, indicating that the transition states of the E-Z thermoisomerization of E-1a,b are less polar than E-1a,b. The E-Z thermoisomerization rate constants of E-2a,b, which carry a pyridinium group, are larger than those of E-1a,b in CD₃CN, indicating that an electron-withdrawing group such as a pyridinium group makes the E-Z thermoisomerization rate of the p-HBDI analogues faster. This result is consistent with what Tolbert et al. found (Dong et al. 2008).

Table 3 E-Z thermoisomerization rate constants (k) of E-1a,b and E-2a,b at 25 °C

Conclusion

Like p-HBDI, its 4-pyridinyl analogues (1a,b) are stable and not easy to be hydrolyzed. After 2a,b are produced by methylation of 1a,b, the amidine carbons of 2a,b become highly electrophilic in comparison with 1a,b and labile to water and alcohols. Both the ¹³C NMR chemical shifts and APT charges of the amidine carbons of 2a,b confirm that their amidine carbons are highly electrophilic. Hydrolysis of 2a,b results in ring-opening of the imidazolinone moiety. The amidine carbon of 2a,b is the only site that is attacked by very weak nucleophile of water and the rest of the structure of 2a,b remains intact. The half lives of hydrolysis of 2a,b are around 25–37 h.

In comparison with 1a,b, the pyridinium group in 2a,b accelerates the E-Z thermoisomerization rate, and also decreases the LUMO–HOMO energy gap of the chromophore, making their electronic absorptions and fluorescent emissions shifted to longer wavelengths.