Oligoethylene Glycol-substituted Aza-BODIPY Dyes As Red Emitting ER-Probes

Abstract

This study features aza-BODIPY (BF₂-chelated azadipyrromethene) dyes with two aromatic substituents linked by oligoethylene glycol fragments to increase hydrophilicity of aza-BODIPY for applications in intracellular imaging. To prepare these, two chalcones were attached α,ω onto oligoethylene glycol fragments, then reacted with nitromethane anion. Conjugate addition products from this reaction were then subjected to typical conditions for synthesis of aza-BODIPY dyes (NH₄OAc, ⁿBuOH, 120 °C); formation of boracycles in this reaction was concomitant with creation of macrocycles containing the oligoethylene glycol fragments. Similar dyes with acyclic oligoelythene glycol substituents in the same position were used to compare the efficiencies of the intra- and inter-molecular aza-BODIPY forming reactions, and the characteristics of the products. All the fluors with oligoethylene glycol fragments, ie cyclic or acyclic, localized in the endoplasmic reticulum of a fibroblast cell line (WEHI-13VAR), the human pancreatic cancer cell line (PANC-1, rough ER predominates) and human liver cancer cell line (HepG2, smooth ER prevalent). These fluors are potentially useful for near IR (λ_{max emis} at 730 nm) ER staining probes.

INTRODUCTION

The most prevalent applications of BODIPY dyes are for aspects of intracellular imaging of various organelles,^1–3 but aza-BODIPY dyes have attracted far less attention as organelle-stains.^4–7 There may be several reasons for this discrepancy, but one is that 1,3,5,7-tetraaryl aza-BODIPYs are much more accessible than non-aryl substituted analogues,⁸ and those aromatic substituents tend to make the fluors very hydrophobic. Hydrophobic dyes, such as aza-BODIPY dyes, are difficult to solubilize for cell-based experiments, and tend to accumulate non-selectively in hydrophobic membranes. One way to counter this trend is to attach oligoethylene glycol fragments to the aryl substituents of aza-BODIPY dyes. Thus, last year, Perez-Inestrosa and co-workers reported aza-BODIPY systems with two, or with four, acyclic triethylene glycol substituents; their dyes were water-soluble, and permeated into the cytoplasm of HeLa cells.⁹ Overall, this approach should give probes that behave differently in intracellular imaging relative to aza-BODIPY dyes that are made more hydrophilic via introduction of charged groups.

Research described in this paper features the aza-BODIPY structures 1 and 2 having cyclic or acyclic oligoethylene glycol substituents, respectively. Several obvious questions might be asked regarding fluors of this kind, but in the event the most interesting ones concerned their roles as near-infrared (near-IR) probes for intracellular imaging. Specifically, we report that all four compounds 1 and 2 that were investigated in this study accumulate quite selectively in rough and smooth endoplasmic reticula (ER) of several cell types.

RESULTS AND DISCUSSION

Syntheses Of The Strapped Fluors 1 and Controls 2

Scheme 1 describes syntheses of the strapped systems 1. Reaction of the chalcone 3 with the ethylene glycol containing fragments shown afforded the α,ω-functionalized systems 4. These Michael electrophiles were reacted with the enolate of nitromethane, and the conjugate addition products 5 formed were condensed with ammonium acetate to form the azadipyrromethenes, then complexation of those with boron trifluoride gave the target molecules 1.

Scheme 1.

Syntheses of the strapped fluors 1.

The non-macrocyclic control molecules 2 were prepared via a route similar to Scheme 1, but featuring ω-methyl oligoethylene glycol derivatives (Scheme S1 and S2). Isolated yields for compounds 2a and 2b were 83 and 74 %, respectively; these are significantly higher than those shown in Scheme 1 for 1a and 1b. A HPLC study was performed to confirm that this observation was not an artifact of the isolation procedure. Thus purified samples of the intermediate conjugate addition products 5b (strapped) and 6b (acyclic control) were mixed in a 1:2 ratio with excess NH₄OAc then heated to 120 °C for 24 h. HPLC was used to compare the relative intensities of the corresponding azadipyrromethenes with purified authentic samples of those (Figure S1 and S2); the acyclic azadipyrromethene was the dominant product indicating that the two strap lengths examined were detrimental to the yield.

Physiochemical Characteristics Of The Strapped Fluors 1 and The Acyclic Controls 2

Acetonitrile dissolved all the compounds, as did dichloromethane except 1a which had limited solubility in that solvent. None of the products 1 or 2 were significantly soluble in pure water, water with approximately 0.1 % Triton X, or aqueous sodium chloride at various concentrations.

The fact that 1 and 2 were not soluble well in aqueous media made it difficult to determine how they might complex metal salts. When 1 and 2 (at micromolar concentrations) were dissolved in acetonitrile or dichloromethane and treated with any of a variety of metal salts (including NaCl, KCl, LiCl, ZnCl₂, CaCl₂, CuCl₂, MgCl₂, HgCl₂, CoCl₂; at ~33 μM, see supporting) then there was no significant effect on the fluorescence wavelengths of the fluors. Micromolar concentrations of Cu(ClO₄)₂ diminished the fluorescence of compounds 1 but this is unlikely to be due to strong complexation because the fluorescence of the non-macrocyclic controls 2 were similarly decreased.

Even aside from solubility issues, the experiments described above seem to indicate that the strapped aza-BODIPY dyes 1a and 1b are not good chelating agents. Consequently, some calculations were performed to try to understand this behavior. Thus the molecules were first minimized via molecular mechanics in a continuous dielectric of 80, then further optimized using density functional theory {B3LYP density functional, and 6-31G(d’) basis set}. This routine was first applied to 15-crown-5 then Na⁺•15-crown-5 as a control. Without sodium the crown converged to irregular conformations, but the expected complex was formed with the metal (see supporting). Application of this strategy to 1a and 1b led to the calculated energy minimized conformations shown in Figure 1. Sodium cation complexation appears to be possible for both compounds. Predictably, there is less space in the cavity of 1a and, perhaps less obvious, the constraints caused by the aryl groups in 1a do not allow so much movement in the ethylene glycol-derived strap. Together these phenomena mean that if the sodium were sequestered by 1a then the metal would be drawn towards the aza-BODIPY nitrogen. For 1b however, four of the oxygen atoms in the strap can orientate to complex the Na⁺ atom, but the fifth oxygen is unsuitably aligned.

Figure 1.

Optimized minimum energy conformations of: a, 1a and Na⁺•1a; and, b, 1b and Na⁺•1b. DFT B3LYP/6-31G(d′) throughout.

All compounds in the series 1 and 2 have very similar absorption and fluorescence maxima. Moreover, the quantum yields of all the compounds were similar (in the 0.26 – 0.29 range), hence restriction of the 1,7-aryl substituents with an ethylene glycol linker did not increase the quantum yields of 1 over 2. We were somewhat surprised by this observation, so checked it by making solutions of all the compounds with almost identical absorbance at the λ_max wavelength (~688 nm) and measuring their fluorescence intensities (Figure 2a). In the event, these solutions had almost the same fluorescence intensities, consistent with their having almost identical quantum yields as asserted above.

Figure 2.

a Quantum yields of all four compounds 1 and 2 are very similar; however, b the strapped system 1b is more brilliant because it has a higher molar extinction coefficient. CH₂Cl₂ solutions throughout.

Figure 2b shows UV absorbance and fluorescence spectra for equimolar solutions of the four compounds 1 and 2, where fluor 1b is significantly more brilliant than the others. The quantum yields of all the compounds are almost the same (see above) so the relative brilliance of 1b can be attributed to its higher extinction coefficient (Table 1).

Table 1.

Electronic spectra of compounds 1 and 2 in dichloromethane (9.2 × 10⁻⁶ M): salient characteristics.

Compound	λmax abs (nm)	εmax (104 M−1cm−1)	λmax emis (nm)1	Φ2fluor
1a	688	2.55	725	0.26 ± 0.003
1b	686	8.22	725	0.29 ± 0.001
2a	690	2.51	725	0.29 ± 0.004
2b	688	5.75	725	0.28 ± 0.004

¹Excited at 650 nm.

²Zn-phthalocyanine is a standard (Φ_{fluor =} 0.30).

Behaviour Of Fluors 1 and 2 As Near-IR ER-stains

Localization of fluors 1 and 2 in WEHI-13VAR fibroblast cells was investigated first. Incubation with 10 μM of each sample for 3 h, for all the fluors, showed the fluorescence localized in the ER, and co-localized with a blue ER-tracker dye (Figure 3 for 1b and S10 for all the other probes). That blue standard probe is sold by Life Technologies who also market a BODIPY based ER-tracker that emits around 615 nm (https://www.lifetechnologies.com/order/catalog/product/E34250).¹⁰ All four aza-BODIPY dyes in this study localize in the same organelle as that ER-tracker, and it is potentially useful that they emit at a complementary λ_{max emis} (ca 730 nm) that could be observed simultaneously.

Figure 3.

Images of WEHI-13VAR cells. a Under a bright-field (DIC image); b with ER tracker; c with fluor 1b; and, d overlay of images a – c.

Incubation of probes 1 and 2 with two other cell lines was investigated to explore the generality of their affinity for the ER. PANC-1 human pancreatic cancer cells are known to have a prevalent rough ER,¹¹ whereas smooth ER predominates for human liver cancer HepG2 cells.^12,13 Figure 4 shows both these cell lines were stained just as the fibroblast cells by illustrative probe 1b (the other three probes gave similar images, Figure S11), except there was one difference. The smooth ER of the HepG2 liver cells appeared as just punctates, just as others have observed for the smooth ER.¹²

Figure 4.

Parallel images for staining PANC-1 (left) and HepG2 cells (right) with: a, b ER tracker; c, d with fluor 1b. e,f Shows the overlay of images for each cell type.

All the probes 1 and 2 were tested in all three cell types (the fibroblasts, pancreatic cancer, and liver cancer cells) for colocalization with LysoTracker Green, and with MitoTracker Green, but these did not colocalize. Moreover, there was no significant accumulation of the probes in the cell membranes.

CONCLUSION

Oligoethylene glycol substituents on the aza-BODIPY probes 1 and 2 probably increase their hydrophobicity enough to overcome non-selective accumulation in all hydrophobic membranes, and promote accumulation in the rough and smooth ER. This observation is different to those made by X. Chen et al⁴ for (positively charged) ammonium salts of aza-BODIPY dyes that stained cell membranes, the cytoplasm, and the ER. It is also different to the observations of Perez-Inestrosa’s triethylene glycol substituted aza-BODIPY dyes that are neutral like the ones reported here, but that localize in the cytoplasm of HeLa cells.⁹ Overall, these comparisons between different aza-BODPY dyes reflect their potential to be directed to different organelles via relatively small changes to the periphery of the dye core. Probes 1 and 2 are complementary to ER stains that fluoresce in the blue and green region,^11–14 and they should facilitate clear observations of blue- and green-labeled molecules in the ER without significant cross-talk.

In this work the oligoethylene glycol substituents used contained four or less repeats. This was not enough to render the fluors significantly water-soluble, and the size and shape of the crown-like cavities was not conducive to complexation of metals, especially relative to conventional crown ethers (cf stability constant of Na⁺•15-crown-5 is ca. 6.2).¹⁵ However, since the organelle staining properties of fluorophores tend to be governed by lipophilicity and charge characteristics, we predict variations on organelle-staining properties could be induced by varying the length of the ethylene glycol substituents, and conceivably by their acyclic/cyclic characteristics.