The quantitative determination of pH in the intracellular space can provide invaluable information on the many processes responsible for regulating cell functions (Weisz, 2003; Boron 2004; Casey et al., 2010). However, the microscaled dimensions of cells in conjunction with the need to ensure minimal structural perturbations restrict the number of analytical methods compatible with these measurements significantly (Loiselle and Casey, 2010; Demuth et al., 2016). In this context, the noninvasive character, micrometer resolution, fast response, and inherent sensitivity of fluorescence measurements (Lakowicz, 2006) and optical microscopy (Murphy, 2001) are particularly convenient. Indeed, membrane-permeable dyes with pH-sensitive emission can be exploited to probe pH in the cytosol and in specific organelles of live cells (Han and Burgess, 2010; Wang et al., 2010; Li et al., 2014; Ni and Wu, 2014; Hou et al., 2017; Specht et al., 2017). Generally, the fluorescence quantum yield of the sensitive molecular probe is designed to change with pH; therefore, the emission intensity measured in a given intracellular region of interest can be correlated quantitatively with pH. Nonetheless, concentration and optical effects can also influence the magnitude of the detected emission intensity, complicating the quantitative determination of pH. These limitations can be overcome with ratiometric measurements, if the sensitive dye is engineered to produce dual emission in resolved spectral windows (Huang et al., 2018; Bigdeli et al., 2019). Under these conditions, the intensity of one emission can be measured relative to the other in order to eliminate any concentration or optical artifact. In addition to the inherent challenge in designing molecular constructs with dual emission, a further complication in the development of such ratiometric probes arises from the need to adjust their sensitivity to a relatively wide range of pH values. In fact, the pH in the intracellular environment can vary from ca. 4 in the acidic lysosomal compartments to as much as ca. 8 in the basic mitochondria (Weisz, 2003; Boron 2004; Casey et al., 2010). Thus, modular structural designs that can easily be modified with the aid of chemical synthesis to ensure ratiometric fluorescence response and tunable pH sensitivity would be especially valuable.

Our laboratories identified a mechanism to activate the fluorescence of a coumarin fluorophore in response to pH (Deniz et al., 2012; Swaminathan et al., 2012). It is based on the opening of a 2H,4H-benzo [1,3]oxazine (oxazine) ring upon acidification with the concomitant bathochromic shift in the coumarin absorption. Selective excitation of the protonated species then results in intense fluorescence at ca. 660 nm. Furthermore, the covalent connection of such a pH-activatable probe to folate ligands allows the activation of fluorescence in acidic endosomal compartments of cancer cells and their discrimination from normal cells (Tang et al., 2017). However, a significant increase in emission intensity is observed only at pH values lower than ca. 5. The need to tune the pH sensitivity of this molecular probe and impose ratiometric response on it suggested the possibility of exploring structural modifications of the original oxazine heterocycle and extending these operating principles to 2H,3H,4H,5H-[1,3]oxazole (oxazolidine) rings, which are also known to open in response to acidification (Szalóki et al., 2015; Fujioka et al., 2020). This article reports a detailed spectroscopic analysis of four members of this family of molecular switches together with representative imaging experiments in model preparations.

Compounds 1Cl and 2Cl incorporate an oxazine heterocycle and differ in the position of a nitro substituent (Figure 1). Compounds 3Cl and 4Cl have an oxazolidine heterocycle in place of the oxazine ring and differ in the substituent on 2H,3H-indole (indole) heterocycle. Compounds 1Cl and 3Cl were prepared according to literature protocols (Deniz et al., 2010; Mazza et al., 2019). Compounds 2Cl and 4Cl were synthesized in one and three steps, respectively (Supplementary Figures S1,S2), and their structural identities were confirmed by high-resolution ESIMS and ¹H NMR spectroscopy.

The absorption spectra (Figures 2, 3) of acetonitrile solutions of 1Cl–4Cl show an intense band at 399–412 nm (^Clλ_Ab in Table 1) associated with the coumarin chromophore in all instances. Upon addition of increasing amounts of trifluoroacetic acid (TFA), this absorption decreases in intensity with the concomitant appearance and growth of a new band at 576–594 nm (^OpHλ_Ab in Table 1). This additional absorption resembles that of model compound 5 (Supplementary Figure S3) and is indicative of the opening and protonation of the oxazine or oxazolidine heterocycle to form 1OpH–4OpH. Indeed, this structural transformation converts the carbon atom at the junction of the indole and either the oxazine or the oxazolidine heterocycle from sp³ to sp² and brings the coumarin component in conjugation with the resulting 3H-indolium cation. As a result of the extended electronic delocalization, the main absorption of the coumarin chromophore shifts bathochromically, in agreement with literature on related compounds (Deniz et al., 2012; Swaminathan et al., 2012; Tang et al., 2017).

The emission spectra (Figures 2, 3) of acetonitrile solutions of 1Cl–4Cl reveal an intense band at 454–480 nm (^Clλ_Em in Table 1) for the coumarin chromophore in all instances. However, the fluorescence quantum yields (^Clϕ in Table 1) of the oxazine derivatives are significantly lower than those of the oxazolidine counterparts. Upon acidification, this emission band decreases in intensity with the concomitant appearance and growth of a new band at 647–666 nm (^OpHλ_Em in Table 1) for 1OpH–4OpH. Once again, the fluorescence quantum yields (^OpHϕ in Table 1) of the oxazine derivatives are significantly lower than those of the oxazolidine counterparts. Presumably, electron transfer from the excited coumarin chromophore to the nitro group is responsible for suppressing the ability of the ring-closed and protonated ring-open forms of 1 and 2 to be emitted.

Compounds 1Cl–4Cl are relatively hydrophobic and essentially insoluble in aqueous environments. However, they readily dissolve in the presence of appropriate amounts (>1 mg mL⁻¹) of Pluronic 123 and retain their photophysical properties as long as their concentration is maintained in the micromolar regime (<50 μM) to avoid aggregation. Indeed, this particular amphiphilic polymer is known to form micelles capable of capturing hydrophobic molecules in their interior and solubilizing them in aqueous environments (Pitto-Barry and Barry, 2015; Jarak et al., 2020). Consistently, DLS measurements on the resulting solutions indicate the formation of nanoparticles with an average hydrodynamic diameter of ca. 80 nm in all instances. Absorption spectra (Supplementary Figures S4,S5), recorded at different pH values, are consistent with the spectral changes observed in acetonitrile upon acidification (Figures 2,3). Specifically, the absorption band of the ring-closed form at 398–412 nm (^Clλ_Ab in Table 1) decreases and that of the protonated ring-open form at 565–608 nm (^OpHλ_Ab in Table 1) increases upon acidification in all instances. Analyses of the pH dependence of the absorbance at ^OpHλ_Ab suggest that pK _a is 5.15 for 1; 3.51 for 2; 5.97 for 3; 5.28 for 4. Comparison of the values estimated for 1 and 2 demonstrates that the relocation of the nitro group from the para- to the meta-position, relative to the oxygen atom, on the phenylene ring fused to the oxazine heterocycle hinders protonation and ring opening. Comparison of the values determined for 1 and 3 shows that the transition from an ozaxine to an oxazolidine ring has instead the opposite effect.

The pH dependence of the equilibrium between the ring-closed and protonated ring-open forms of 1 to 4, evident from the absorption spectra, translates into dual emission (Supplementary Figures S4,S5). Specifically, the emission band of one species at 465–478 nm (^Clλ_Em in Table 1) decreases and that of the other at 647–671 nm (^OpHλ_Em in Table 1) increases with acidification in all instances, consistently with the behavior observed in acetonitrile (Figures 2, 3). As a result, the ratio between the emission intensities of the ring-closed form (I _Cl) and that of the protonated ring-open species (IOpH) increase monotonically with a rise in pH. The corresponding plots (Figure 4) show that 1, 3, and 4 respond to pH predominantly between values of ca. 4 and ca. 8. The largest change in the magnitude of the ratio between the two emission intensities is observed for 4. Instead, 2 responds to pH predominantly between values of ca. 2 and ca. 5, demonstrating that the position of the nitro group on the benzoxazine heterocycle has a pronounced influence on the equilibrium between two interconverting species. These observations suggest that the nature and position of the substituents on the pH-sensitive benzoxazine ring can be exploited to tune the response of these molecular switches.

The resolved emission bands of the ring-closed and protonated ring-open forms of 1‐4 provide, in principle, the opportunity to probe pH ratiometrically with the acquisition of fluorescence images in separate detection channels of the same microscope. Alginate hydrogels are optimal biocompatible model systems (Seliktar 2012; Rosales and Anseth, 2016; Zhang and Khademhosseini, 2017) to explore these possibilities and allow the identification of appropriate imaging conditions to capture and differentiate the fluorescence of the two interconverting species. Specifically, alginate beads of millimeter dimensions can be doped with nanoparticles of Pluronic 123 containing 4, dispersed in water with a given pH, and imaged. The fluorescence of 4Cl can be collected between 450 and 600 nm (yellow channel in Figure 5) and that of 4OpH can be recorded between 600 and 770 nm (green channel in Figure 5). The resulting images show a fluorescence increase in the yellow channel (a–d in Figure 5) with a concomitant emission decrease in the green channel ( e – h in Figure 5) as the pH rises from 4.0 to 7.4. Plots ( i in Figure 5) of the emission intensities measured in the two channels further confirm that they are pH-dependent. In fact, the ratio ( j in Figure 5) between them increases monotonically with pH, by analogy with the spectroscopic measurements (Figure 4), providing the opportunity to quantify this parameter ratiometrically. Additionally, storage of the sample at a fixed pH does not cause any change in the detected intensities (Supplementary Figure S6), demonstrating that the sensitive probes neither undergo hydrolytic degradation nor photobleaching under these experimental conditions.

The covalent integration of a coumarin chromophore and either an oxazine or an oxazolidine heterocycle within the same molecular skeleton provides the opportunity of assembling molecular switches with ratiometric fluorescence response to pH. The oxazine or oxazolidine heterocycle opens upon acidification to establish an equilibrium between a ring-closed and a protonated ring-open form. In the latter species only, the coumarin chromophore can extend electronic delocalization over an indolium auxochrome. As a result, the protonated ring-open form is absorbed and emitted at wavelengths that are significantly longer than those where the ring-closed species is absorbed and emitted. This behavior enables the selective excitation of the two interconverting forms at distinct wavelengths and the collection of their emissions in resolved spectral windows. Furthermore, the relative amounts of the two equilibrating species and, hence, their relative emission intensities change with pH. It follows that separate detection channels of the same microscope can be exploited to collect the fluorescence of the two components independently and reconstruct a ratiometric profile of the pH distribution of any sample of interest labeled with these compounds.

A significant advantage of this family of molecular switches over existing compounds with ratiometric fluorescence response to pH is their synthetic accessibility and versatility. Three synthetic steps from readily available precursors are sufficient to assemble the parent molecular scaffold of both oxazine and oxazolidine derivatives, offering the opportunity to prepare relatively large amounts of these compounds. Additional synthetic modifications permit the manipulation of the substituents on the indole heterocycle and on the phenylene ring fused to the oxazine ring. These structural manipulations can be exploited to adjust the pH response and, in principle, to introduce hydrophilic chains for aqueous solubility (Cusido et al., 2012) or reactive functional groups for the covalent connection of biomolecules (Cusido et al., 2016; Zhang et al., 2018) or synthetic polymers (Tang et al., 2017). In fact, the fascinating prospect of incorporating targeting agents to direct these molecular switches into specific intracellular components and allow local pH probing can definitely be envisaged. Thus, the combination of photophysical and structural properties that can be engineered into these molecular switches can lead to the realization of versatile fluorescent probes for the intracellular visualization of pH distributions at the microscale with ratiometric imaging schemes.