Decoupling Photophysical Brightness and Cellular Imaging Efficiency in 4‐ANI‐Based Fluorophores: The Critical Role of Sterically Hindered N‐Substituents

Abstract

In this article, we report the screening of a library of twenty‐one 4‐ANI‐based fluorophores, leading to the identification of a novel cycloheptylamine‐containing derivative with outstanding performance for cellular imaging. This compound exhibits fluorescence intensities more than tenfold higher than those of most analogs, enabling its use at submicromolar concentrations with minimal perturbation of cellular function. Beyond this lead discovery, our results highlight a critical disconnect between photophysical properties in aqueous media and imaging efficiency in biological systems. By combining multiple analytical techniques, including photophysics, fluorescence microscopy, and flow cytometry as well as computational DFT calculations, we demonstrate that sterically hindered N‐substituents play a critical role by combining efficient cellular uptake and good solubility, two essential features that govern probe performance for cell imaging. These findings challenge the conventional reliance on intrinsic brightness as a design criterion and emphasize the necessity of integrating biological context into the evaluation and development of fluorescent imaging agents.

A screening of twenty‐one 4‐ANI fluorophores led to the identification of an N‐cycloheptylamine derivative with >10‐fold enhanced cellular fluorescence, enabling submicromolar imaging with minimal cell perturbation. Integrated photophysics, microscopy, flow cytometry, and DFT reveal that sterically hindered N‐substituents optimize uptake and solubility. These findings challenge brightness‐driven design and highlight biological context as essential to probe development.

1. Introduction

Fluorescence imaging has become an indispensable tool for the study of biological systems, enabling highly sensitive visualization of the localization, dynamics, and interactions of biomolecules within living cells. Advances in this field have been accompanied by an increasing demand for organic fluorophores exhibiting properties tailored to modern experimental constraints [1]. Among the desired characteristics are high brightness [2], excellent photostability [3], a low cytotoxicity, efficient cellular uptake [4, 5], and labeling performance [6, 7]. Despite substantial progress in this area, the identification of fluorophores that simultaneously combine favorable photophysical properties with optimal performance in biological environments remains challenging. The development of new fluorescent probes often relies on structural modifications of well‐established fluorophores [8]. In many cases, the optically active molecular core is preserved, while modifications are introduced at peripheral positions of the scaffold. This strategy enables systematic investigation of the influence of specific substituents on parameters such as solubility [9, 10], lipophilicity [11], overall charge, and subcellular targeting preferences [12, 13]. Intuitively, modifications at peripheral positions are not necessarily expected to significantly affect the intrinsic photophysical properties of the fluorophore. However, since solubility is a key parameter governing fluorescence efficiency in organic fluorophores, such modifications can nevertheless lead to measurable variations. This is particularly evident when fluorescence properties are evaluated in aqueous media, which closely mimic more closely biological environments. In this situation, precipitation and/or aggregation of polyaromatic compounds may occur and often results in a substantial decrease in fluorescence quantum yield through aggregation‐caused quenching (ACQ). Recent studies have further demonstrated that subtle structural modifications of fluorophore scaffolds, particularly cyanines, are critical for suppressing aggregation and thereby maximizing their fluorescence performance [14, 15]. More generally, the development of fluorophores that remain highly emissive in water, such as Alexa Fluor dyes [16], remains a major objective in the design of probes for biological applications. However, strong fluorescence in aqueous media alone is not sufficient to guarantee optimal performance in cell imaging. Although certain physicochemical parameters are known to influence cellular uptake, intracellular distribution, and retention of fluorophores, these factors do not always allow a reliable prediction of their performance for cell imaging. This limitation continues to hinder the rational design of fluorophores optimized for specific applications. Our group has been working for several years on 4‐aminonaphthalimide (4‐ANI) derivatives for the development of fluorogenic probes [17, 18, 19, 20]. When equipped with a free aniline in the 4‐position, this fluorophore family is widely used in the literature for the construction of sensors across multiple fields [21, 22, 23, 24, 25]. However, numerous reports in the literature, together with our own laboratory experience, consistently indicate that these fluorophores generally need to be used at relatively high concentrations (5–10 µM) to achieve high‐quality fluorescence microscopy imaging. This phenomenon is primarily attributable to the push–pull character of this family of intramolecular charge–transfer (ICT) fluorophores. Their low molar extinction coefficient (ε) leads to markedly reduced brightness relative to other fluorophore families, such as cyanines, rhodamines, and BODIPYs. In this context, we have undertaken the study of a series of fluorophores sharing the same 4‐aminonaphthalimide core but differing on the peripheral structural motif attached to the imide nitrogen atom. In this article, we present a systematic investigation of a library of 21 fluorophores, including comprehensive photophysical characterization, confocal fluorescence microscopy studies, and flow cytometry analysis. This study enabled us to identify a specific analog whose fluorescence imaging performance is significantly superior to that of the other members of the series.

2. Results and Discussion

2.1. Rational Choice of 4‐ANI‐Based Dyes

For this study, we selected several fluorophores that have already been described and used in the literature [17, 18, 20, 26, 27]. In addition, we included a series of original fluorophores that we synthesized using the classical method based on the ring‐opening and ring‐closing reactions of 4‐amino‐1,8‐naphthalic anhydride (detailed synthesis protocols are accessible in the Supporting Information).

We thus obtained a library of 4‐ANI derivatives by systematically varying the substituents on the phthalimide moiety (Figure 1). The primary objective of this study was to investigate the cellular uptake of fluorophores lacking intrinsic targeting motifs for specific subcellular compartments. For this reason, substituent selection was guided by the need to minimize preferential localization. In particular, cationic groups and protonatable amine functionalities, well‐documented to promote accumulation in mitochondria [28] and lysosomes [29], respectively, were deliberately excluded. Instead, we focused on incorporating neutral or anionic substituents. In total, 21 fluorophores were considered for this study. Alongside commonly employed derivatives, such as the n‐butyl analog 1 and the unsubstituted compound 2, we introduced relatively short substituents (compounds 5 and 6), including heteroatom‐containing chains, as exemplified by the methoxyethyl (3), hydroxyethyl (4), and aminoethyl (19) derivatives. Cycloalkylamine substituents were also explored, with ring sizes ranging from four‐ to seven‐membered systems (compounds 7–10), as well as the N‐phenyl derivative 11. Sulfonic acid and sulfonamide derivatives (12 and 13) were obtained in good yields of 60% and 68%, respectively (see the Supporting Information). The synthesis of the carboxylic acid derivative 14 further enabled access to the corresponding formamide 15 and methyl ester 16 analogs. To extend the structural diversity, longer and more flexible water‐solubilizing substituents were introduced, yielding methyl‐PEG₄ (17), methyl‐PEG₈ (18), an amino‐PEG₂ derivative 20, as well as a polysarcosine‐based analog 21 considered here as a water‐solubilizing alternative to pegylated compounds [30]. Synthetic procedures and full characterization data for all compounds are provided in the Supporting Information. Each fluorophore was characterized using standard analytical techniques, including NMR spectroscopy and mass spectrometry, unambiguously confirming its chemical structure.

FIGURE 1.

Chemical structures of 4‐ANI fluorophores 1–21.

2.2. Photophysical Characterization of 4‐ANI Based Dyes

The photophysical properties of the synthesized fluorophores were systematically investigated to enable a comparative evaluation of their individual performance. To this end, measurements were conducted in both DMSO and phosphate‐buffered saline (PBS) aqueous solutions (pH = 7.4). We carefully performed the measurements of each compound in parallel over a short time period in order to ensure methodological consistency, thereby enabling the most accurate possible comparison of the results from one fluorophore to another. For each compound, absorption, emission, and excitation spectra were recorded in both media. All recorded spectra are provided in the Supporting Information (Figure S1). In addition, molar extinction coefficients and fluorescence quantum yields were determined. The complete set of photophysical data is summarized in Table 1. Regarding the absorption and emission maxima, the values vary only marginally from one fluorophore to another, both in DMSO and in PBS buffer. This result was expected given the structural similarity of the polyaromatic scaffold constituting the fluorescent core of the different compounds. In DMSO, the absorption maxima show only limited variation, ranging from 432 to 438 nm, while the emission maxima similarly fluctuate only slightly between 527 and 537 nm. The measurements were then carried out in PBS containing 1% DMSO without any noticeable solubility issues. A slight hypsochromic shift of the absorption maxima (~5 nm) is observed, with values remaining within a narrow range of 427–433 nm. In contrast, a more pronounced bathochromic shift (~15 nm) is observed for the emission maxima, although these still vary only modestly between 546 and 563 nm. Compound 21 stands out slightly, exhibiting a lower emission wavelength recorded at 532 nm. The limited variation in both absorption and emission maxima consequently results in only minor differences in the calculated Stokes shifts, which are centered around 4000 cm⁻¹ in DMSO and closer to 5000 cm⁻¹ in PBS. From the perspective of excitation and emission wavelengths, all fluorophores investigated here are, therefore, highly similar. In contrast, the fluorescence quantum yield (QY_FL) measurements reveal greater disparities.

TABLE 1.

Photophysical properties of 4‐ANI based dyes 1–21.

Dye	In DMSO						In PBS buffera
	λ max Absb, nm	ε max c, M−1.cm−1	λ max Em, nm	Stokes shift, cm−1	QYFL d	B (ε × QYFL)	λ max Absb, nm	ε max c, M−1.cm−1	λ max Em, nm	Stokes shift, cm−1	QYFL d	B (ε × QYFL)
1	438	13,100	530	3963	0.73	9563	433	12,200	552	4979	0.17	2074
2	432	12,500	527	4173	0.70	8750	432	12,500	555	5130	0.17	2125
3	437	14,200	532	4086	0.70	9940	432	13,100	557	5195	0.19	2489
4	436	13,800	533	4174	0.64	8832	430	13,200	555	5238	0.21	2772
5	435	12,900	532	4192	0.69	8901	433	12,300	552	4979	0.19	2337
6	436	11,900	528	3996	0.70	8330	432	11,100	553	5065	0.15	1665
7	436	14,300	535	4244	0.68	9724	427	12,700	560	5562	0.19	2413
8	437	13,400	535	4192	0.70	9380	430	12,300	551	5107	0.20	2460
9	436	13,300	533	4174	0.69	9177	430	12,300	558	5335	0.17	2091
10	436	12,600	529	4032	0.68	8568	433	11,300	557	5141	0.14	1582
11	436	14,300	536	4279	0.67	9581	432	13,900	550	4966	0.17	2363
12	436	13,900	531	4103	0.71	9869	429	12,900	558	5389	0.16	2064
13	438	12,600	537	4209	0.64	8064	432	11,800	552	5032	0.12	1416
14	438	11,900	532	4034	0.71	8449	432	12,600	556	5163	0.19	2394
15	437	13,300	531	4051	0.72	9576	432	12,800	556	5163	0.24	3072
16	436	11,000	532	4139	0.68	7480	432	9500	555	5130	0.13	1235
17	437	13,300	531	4051	0.65	8645	432	12,400	546	4833	0.20	2480
18	437	13,800	533	4122	0.69	9522	433	12,800	552	4979	0.23	2944
19	439	16,600	537	4157	0.55	9130	431	12,800	563	5440	0.09	1152
20	439	11,300	536	4122	0.65	7345	433	10,200	553	5012	0.14	1428
21	437	16,200	530	4015	0.67	10,854	432	15,500	532	4351	0.26	4030

^a Measurements performed in a PBS buffer (1X, pH = 7.4) containing 1% DMSO (v/v).

^b Absorbance maxima were measured at the S0 → S1 transition.

^c Molar extinction coefficient related to S0 → S1 transition, using the absorbance value of a 10⁻⁵ M solution.

^dRelative fluorescence QY was determined at 25 °C using coumarin 153 as a reference (QY_FL = 0.53 in EtOH) [31].

In DMSO, the values are overall very similar, reaching approximately 70% for the majority of the compounds, while the molar absorption coefficients are likewise comparable, preventing any particular fluorophore from clearly standing out in terms of brightness in DMSO. However, the variability of the fluorescence QY_FL is more pronounced in PBS buffer. Notably, compounds 15, 18, and 21 display relatively high QY_FL values of 24%, 23%, and 26%, respectively, resulting in a substantial improvement in brightness and therefore displaying noticeable improvement compared to reference compound 1. This enhancement may likely be attributed to the presence of hydrophilic functionalities on the nitrogen substituent of the phthalimide moiety, which improve compound solubility and thereby limit quenching arising from possible partial aggregation (ACQ). Taking into account the greater variability in molar absorption coefficients also observed in PBS, these same fluorophores appear to distinguish themselves by higher brightness in an aqueous medium, suggesting more promising performance for cellular imaging applications. To verify this hypothesis, we carried out a series of complementary cellular experiments involving all the fluorophores from the tested library.

2.3. Confocal Microscopy in Live Cells

We first carried out a series of imaging experiments using confocal microscopy in living cells. All fluorophores were incubated at a concentration of 5 µM in live A549 cells at 37 °C for 3 h. The cells were subsequently imaged by confocal microscopy while maintaining rigorously identical microscope settings in order to enable a direct comparison of the efficiency of each fluorophore for fluorescence imaging in living cells. The obtained results are presented in Figure 2. Surprisingly, the results obtained are in complete disagreement with the conclusions drawn from the photophysical study presented above. Indeed, compounds identified as promising candidates (15, 18, and 21) exhibited extremely weak fluorescence signals, remaining undetectable under the experimental conditions employed. Overall, the intracellular fluorescence intensity remains relatively low for the majority of the tested compounds. Only the introduction of bulky lipophilic substituents led to a substantial enhancement of the intracellular fluorescence signal. For example, compounds 9 and 10, bearing N‐cyclohexyl and N‐cycloheptyl substituents, respectively, displayed a particularly strong fluorescence signal in comparison with the one of reference dye 1 (N‐nbutyl).

FIGURE 2.

Confocal imaging of fluorophores 1 – 21 in live A549 cells. 4‐ANI‐based dyes were incubated at 5 µM for 3 h at 37 °C, acquisition was performed using λ _Exc: 440 nm and λ _Em: 500–650 nm. The choice of the pseudocolor image look‐up table (LUT) was made deliberately in order to better visualize and appreciate the differences in the observed fluorescence intensities. Scale bars: 30 µm.

The presence of a cyclic substituent, therefore, appears to be favorable, although the N‐cyclobutyl (7), N‐cyclopentyl (8), and N‐phenyl (11) derivatives exhibited detectable intracellular fluorescence that remains weaker than dyes 9 and 10. It is also noteworthy that the only noncyclic derivative providing a sufficiently intense signal was the isopentylamine derivative 5. In addition, ester derivative 16, initially designed to improve fluorophore permeability compared to the corresponding carboxylic acid compound 14, exhibited only a weak fluorescence signal, suggesting that this strategy is not effective in the present case for enhancing the cellular uptake of the dye. Altogether, these observations suggest that substitution of the phthalic nitrogen atom of 4‐ANI‐based dyes with a sufficiently bulky lipophilic group is beneficial for the development of efficient fluorophores for cellular imaging experiments. Furthermore, the images provide unequivocal evidence that the introduction of water‐solubilizing groups, whether through anionic (12 and 14) or neutral moieties (17–21), is not an effective strategy for generating efficient 4‐ANI‐derived dyes for cell imaging, and this is even when compounds 18 and 21 displayed promising brightness characteristics in aqueous environments.

2.4. Flow Cytometry Quantitative Analysis of Intra‐Cellular Signal

To obtain an accurate quantification of the intracellular fluorescence signal, we set up and conducted a flow cytometry (FACS) study.

A549 cells were incubated with the 4‐ANI derivatives (1–21) at a concentration of 5 µM at 37 °C for 3 h. For each fluorophore, the experiment was performed in triplicate. Following a PBS washing step, the cells were treated with trypsin, collected, and analyzed by FACS. A total of ~20,000 cells were counted for each experiment, ensuring statistically relevant measurements, and the mean fluorescence intensity (MFI) was measured and is reported in Figure 3. The results presented in Figure 3A confirm the observations made during the confocal microscopy imaging experiments (all cytometry plots are accessible in Figure S2). Compounds 5 and 7 displayed intracellular fluorescence signals comparable to those of reference compound 1. In contrast, functionalization of the nitrogen atom of the phthalimide moiety with more hindered cyclic lipophilic moieties (dyes 8, 9, 10, and 11) substantially enhanced the intracellular fluorescence signal (Figure 3B). Notably, only the N‐cyclobutyl derivative 7 failed to show a significant improvement. To better evaluate the impact of this functionalization, an “enhancement factor” relative to the intracellular signal of the reference compound 1 was calculated using the MFI values (Figure 3C). These results revealed signal enhancements by factors of ×6.4 for compound 9 and ×12.2 for compound 10, thereby confirming the strong performance observed for these two compounds during confocal microscopy experiments. It is noteworthy that compounds 15, 18, and 21, which exhibited the most promising fluorescence properties in PBS buffer, ultimately proved to be among the least effective for cellular imaging, displaying fluorescence intensities approximately twofold lower than that of the reference compound 1. Furthermore, additional measurements in PBS supplemented with 5% BSA (Figure S3) did not reveal any compound‐dependent photophysical differences that could account for the distinct fluorescence signals observed in cells. This result may likely be attributed to the poor cellular uptake of these compounds. Overall, the compound incorporating the cycloheptylamine moiety 10 clearly stands out from the other fluorophores owing to its superior performance as an imaging agent for live‐cell fluorescence microscopy. The results obtained from the cellular imaging experiments do not appear to be attributable to the intrinsic fluorescence properties of the investigated compounds. One of the key parameters governing the performance of an imaging agent is its ability to efficiently penetrate cells. To assess this aspect, we estimated the logP values of all compounds using the SwissADME platform [32].

FIGURE 3.

Flow cytometry (FACS) analysis of fluorophore 1 – 21 in live A549 cells. (A) Average intracellular green fluorescence signal intensity measured after incubation at 5 µM for the various 4‐ANI analogs (1–21). Dye 0 means absence of dye. Mean fluorescence intensity (MFI). (B) Overlay of MFI from dyes 1, 5, 7, 8, 9, 10 and 11. (C) Table presenting improvement factor of intracellular green fluorescent signal in comparison with N‐nbutyl‐4‐ANI 1.

The estimated values are summarized in the table presented in Figure 4. In addition, the intracellular fluorescence intensities measured by fluorescence‐activated cell sorting (FACS) were plotted as a function of the corresponding logP values (Figure 4, bottom). A clear trend can be observed. Compounds exhibiting very low intracellular fluorescence also display relatively low logP values (<2.5). Compounds with logP values ranging from 2.5 to 3.35 exhibit intracellular fluorescence intensities that are comparable to that of reference compound 1 used in this study. Finally, 4‐ANI‐based dyes 9 and 10, which produced the highest intracellular fluorescence intensities, also possess the highest logP values (3.40 and 3.74, respectively). Collectively, these findings suggest that logP is a critical parameter to consider in the design of fluorescent probes for cellular imaging applications. Nevertheless, compound 10, which exhibited an outstanding overall performance, contains an uncommon cycloheptylamine moiety. We therefore sought to determine whether the cyclic nature of this substituent contributes to enhanced intracellular fluorescence independently of its log‐P value. To address this question, we synthesized analog 10n (estimated logP = 4.12), in which the cycloheptylamine group was replaced with a linear seven‐carbon alkyl chain (the synthesis and structural characterization of this compound are provided in the Supporting Information). The photophysical properties of compound 10n were evaluated in both DMSO and PBS buffer. Subsequently, cellular imaging experiments were performed (Figure 5). The fluorescence properties of compound 10n in DMSO were very similar to those of the other derivatives investigated in this study, with absorption and emission maxima at 437 and 539 nm, respectively. Its fluorescence quantum yield was among the highest of the series, reaching 73%. In contrast, measurements performed in PBS revealed extremely weak fluorescence, with a QY_FL estimate to be below 1%. This strong decrease is likely attributable to the poor aqueous solubility of compound 10n. This hypothesis is supported by the profiles of the absorption, excitation, and emission spectra recorded in PBS (Cf Figure S1, compound 10n). The first observation is the very low fluorescence intensity, accompanied by a distorted emission spectrum. Furthermore, although the absorption spectrum could be recorded by performing the measurement immediately after solution preparation, the presence of a long‐wavelength tail nevertheless suggests the onset of precipitation. Finally, the poor overlap between the absorption and excitation spectra further supports an aggregation behavior consistent with incipient precipitation. To investigate this hypothesis, we monitored the absorption spectra of 10 µM solutions prepared in PBS containing 1% DMSO over a period of 30 min. The same experiment was performed with the cycloheptylamine derivative 10, and the results are presented in Figure 5B. A rapid distortion of the absorption spectrum, typically observed during the precipitation process, was observed for compound 10n, whereas no spectral changes occurred for compound 10 over the same time interval. Analysis of the absorbance at λ _Max as a function of time revealed that compound 10n, bearing the linear heptyl substituent, underwent complete precipitation within only 5 min at this concentration. By contrast, the absorbance of the cyclic derivative 10 remained unchanged throughout the 30 min incubation period. These findings highlight the advantage of incorporating a bulky cyclic substituent, whose steric hindrance is expected to reduce intermolecular π–π stacking interactions, thereby preventing fluorophore aggregation and subsequent precipitation in aqueous media. Despite the limited solubility of compound 10n at 10 µM, confocal fluorescence imaging experiments were performed at a lower concentration (1 µM), at which precipitation appeared to be less pronounced.

FIGURE 4.

Estimated logP values of the tested compounds and variation of intracellular fluorescence intensity as a function of logP.

FIGURE 5.

Comparison between performances of 10 and 10n. (A) Photophysical properties of ANI‐based dye 10n at 25 °C. (B) Evolution of absorbance of compounds 10 and 10n in PBS buffer at 25 °C over time. (C) Confocal imaging of 10 and 10n (1 µM in RPMI) in live A549 cells. 4‐ANI‐based dyes were incubated at 1 µM for 3 h at 37 °C, acquisition was performed using λ _Exc: 440 nm, λ _Em: 500–650 nm. Scale bars 30 µm. (D) Zoomed pictures of regions of interest (ROIs) for compound 10n. Scale bars: 10 µm. Arrows indicate lipid droplets.

As shown in Figure 5C, compound 10n produced a substantially weaker fluorescence signal than the cyclic derivative 10. More importantly, the magnified images presented in Figure 5D revealed a tendency of compound 10n to accumulate within spherical vesicular structures resembling lipid droplets, indicating a heterogeneous intracellular distribution. To further investigate this observation, a series of colocalization experiments were performed to assess the subcellular localization of compounds 10 and 10n using confocal fluorescence microscopy.

2.5. Subcellular Localization of 10 and 10n

To further investigate the intracellular distribution of compound 10n, cells were first coincubated with 10n (1 µM) and the commercial red fluorescent lipid droplet marker LipidSpot 610. Representative confocal microscopy images are shown in Figure 6A.

FIGURE 6.

Intracellular distribution of 10 and 10n using confocal microscopy. (A) Imaging of 4‐ANI‐based dye 10n using a 63× oil immersion objective at 1 μM after 3 h of incubation at 37 °C (λ _Exc: 440 nm, λ _Em: 450–550 nm). Scale bar 30 μm. Colocalization experiments were performed using LipidSpot 610 (λ _Exc: 600 nm, λ _Em: 620–700 nm) and associated plot profile. (B) Colocalization experiments with 4‐ANI‐based dye 10 using MitoView 633 (λ _Exc: 620 nm, λ _Em: 640–700 nm), LipidSpot 610 (λ _Exc: 600 nm, λ _Em: 620–700 nm), NucSpotlive 650 (λ _Exc: 640 nm, λ _Em: 660–700 nm) and LysoView 633 (λ _Exc: 630 nm, λ _Em: 650–700 nm). PCC calculations were performed using the JAcOP plugin of ImageJ. Scale bar: 30 µm.

Partial colocalization of the two fluorescence signals was observed, with a Pearson's correlation coefficient (PCC) of 0.46. Plot‐profile analyses further confirmed that the vesicular structures identified in the preliminary imaging experiments corresponded to lipid droplets. These results demonstrate that compound 10n displayed a stronger fluorescence in lipid droplets, although a detectable fluorescence signal was also observed throughout the remainder of the cell. To determine whether this behavior was specific to the linear analog, similar costaining experiments were performed with the cyclic derivative 10 using a set of organelle‐specific fluorescent markers, including MitoView 633 for mitochondria, LipidSpot 610 for lipid droplets, LysoView 633 for lysosomes, and NucSpot Live 650 for the nucleus. As shown in Figure 6B, the consistently low PCC values obtained with all four markers indicate the absence of significant colocalization, demonstrating that compound 10 displays a diffuse, nonspecific intracellular distribution rather than preferential accumulation within a particular organelle.

2.6. Rationalization of Fluxionality and Biological Relevance Through Computational Calculation

The experimental results described above revealed a marked propensity of compound 10n to precipitate in aqueous media, whereas no such behavior was observed for compound 10. Because this difference is likely to have a major impact on the biological performance of these fluorophores, the structural determinants underlying their distinct behaviors were investigated by density functional theory (DFT) calculations.

The equilibrium geometries of both compounds were first optimized using water as an implicit solvent. To evaluate their conformational flexibility, potential energy scans were then performed by systematically varying the C—N—C—C dihedral angle (θ). As shown in Figure 7, both derivatives displayed nearly identical rotational energy profiles around the C—N bond. Interestingly, DFT calculations yield similar rotational barriers for both 10 and 10n compounds (ΔE ≈ 32–33 kJ.mol⁻¹). As defined by Eliel et al. [33], these values categorize both substituents as fluxional at room temperature, precluding any “conformational locking”. Although DFT calculations are performed at 0 K, the rotational barrier of ~33 kJ.mol⁻¹ is significantly lower than the available thermal energy at physiological temperature (310 K). The interconversion rate between conformers follows Arrhenius kinetics, where the rotational barrier acts as the activation energy (E _a). Given our calculated barrier and a physiological temperature of 310 K, the ratio of E _a/RT is approximately 12.8. According to the transition‐state theory [34], this barrier is sufficiently low to ensure rapid conformational exchange on the nanosecond timescale. This confirms that the observed steric protection against π–π stacking for the cycloheptyl derivative is not a static result of a “frozen” conformation but an emergent property of the molecule's fluxional interconversion under cellular conditions. On the other hand, a striking difference emerged from the analysis of the optimized conformations. At θ = 0°, compound 10n readily adopted a fully planar geometry, whereas the bulky cycloheptyl substituent of compound 10 consistently prevented complete planarization because of steric constraints (Figure 7, bottom). This conformational restriction is expected to substantially reduce the effective overlap between neighboring aromatic cores, thereby disfavoring intermolecular π–π stacking interactions. By contrast, the planar structure accessible to 10n is likely to facilitate molecular stacking, providing a plausible explanation for its rapid precipitation in an aqueous solution. Taken together, these theoretical results provide a compelling molecular rationale for the experimental observations and highlight the critical role of steric engineering in the phthalimide nitrogen substituent. Beyond simply increasing lipophilicity to promote cellular uptake, the introduction of a bulky cyclic substituent effectively suppresses precipitation or aggregation by preventing molecular planarization. This preserves aqueous solubility while maintaining favorable physicochemical properties for biological imaging applications.

FIGURE 7.

In silico conformational study. Top: The graphs show the relative energy (ΔE in kJ/mol) as a function of the dihedral angle (θ). The lowest energy state on the potential energy surface was used as the reference (ΔE = 0 kJ/mol). Rotational barrier estimates were derived from the energy difference between this global minimum and the highest energy point obtained along the relaxed rotation path. Bottom: Conformations obtained from DFT calculations: front view of stable conformation and side views of lowest and highest energy conformations.

2.7. Balancing Probe Concentration and Cell Viability

Given the strong performance of compound 10 in microscopy experiments, this fluorophore was incubated at several concentrations ranging from 0.1 to 5.0 µM in A549 cells in order to evaluate its minimum effective working concentration.

Similarly to the previous confocal microscopy experiments (Figure 2), the microscope settings and image analysis parameters were rigorously maintained identically throughout the study to enable direct comparison of the results. The corresponding images are presented in Figure 8. These results demonstrate that fluorophore 10 can be effectively used at lower concentrations as a detectable fluorescence signal was observed even at 0.1 µM. However, based on these observations, we estimate that a minimum working concentration of 0.5 µM is required to obtain a sufficiently intense and practically exploitable fluorescence signal. The choice of the working concentration is a critical parameter for preserving cellular integrity during fluorescence imaging experiments. We evaluated the cytotoxic effects of the different compounds on A549 cells.

FIGURE 8.

Confocal imaging of fluorophore 10 at different concentrations ANR‐11‐IDEX‐003‐02 in live A549 cells. 4‐ANI‐based dye 10 was incubated at concentrations ranging from 0.1 to 5.0 µM for 3 h at 37 °C. Acquisition was performed using λ _Exc: 440 nm and λ _Em: 500–650 nm. The choice of the LUT was made deliberately in order to better visualize and appreciate the differences in the observed fluorescence intensities. Scale bars: 30 µm.

To this end, all compounds were incubated at concentrations ranging from 0.5 to 10 µM, and cell viability was determined after 72 h of incubation. The results for the most promising dyes 1, 5, 7, 8, 9, 10, and 11 are presented in Figure 9 (effects on cell viability of other dyes are available in Figure S4). The results revealed that most of the dyes don’t display significant toxicity. However, a slight decrease in cell viability is observed at high concentrations of 5 or 10 µM. These findings confirm that 4‐ANI derivatives may exert deleterious effects on cell viability at high concentrations [35, 36]. This observation provides an additional valuable argument for the use of these fluorophores at lower concentrations. Since compound 10 can be effectively employed at submicromolar concentrations, it is not expected to exert any effect on cell viability during imaging experiments, constituting another advantage for its use. Furthermore, the compounds displaying the most favorable emissive properties in PBS buffer, as reported in Table 1, exhibited no measurable cytotoxicity (Figure S3), even at the highest concentrations investigated. This finding is consistent with the hypothesis that these compounds have a limited capacity for cellular uptake. Such poor intracellular accumulation may likewise explain the absence of a detectable fluorescence signal in the confocal microscopy studies.

FIGURE 9.

Cell viability of A549 cells treated with dyes 1, 5, 7, 8, 9, 10, and 11 at concentrations ranging from 0.5 to 10 µM, as determined using the CellTiter‐Glo assay.

3. Conclusion

In conclusion, screening a library of twenty‐one 4‐ANI‐based fluorophores led to the discovery of a structurally original derivative 10 bearing a cycloheptylamine motif, which displays outstanding potential for cellular imaging applications. Despite average fluorescence properties measured in vitro, this compound stands out by exhibiting fluorescence intensities in cells more than ten times higher than those of most analogs evaluated in this study. We have demonstrated that these improved performances result from the introduction of sterically hindered lipophilic substituents, which enhance cellular uptake by increasing LogP while maintaining good aqueous solubility through the prevention of intermolecular π–π stacking interactions. Its high performance enables its use at submicromolar concentrations, thereby minimizing any perturbation of cellular functions. Altogether, these features establish it as a compelling scaffold for the future development of fluorogenic probes based on naphthalimide derivatives. More broadly, this work challenges the common assumption that photophysical performance in an aqueous solution is a reliable predictor of cellular imaging efficiency. By integrating photophysics, fluorescence microscopy, flow cytometry, and computational study, we demonstrate that the brightest emitters in vitro are not necessarily the most effective imaging agents in a biological context. Instead, cellular performance emerges from a more complex interplay of photophysical and physicochemical properties, among which membrane permeability and intracellular accessibility appear to be decisive yet difficult‐to‐anticipate parameters. This study, therefore, emphasizes a conceptual shift: the evaluation of imaging probes cannot be reduced to intrinsic photophysical descriptors alone but must necessarily incorporate the full biological context in which these molecules operate.

4. Experimental Section

4.1. General Procedure for the Synthesis of Undescribed ANI‐Based Dyes

To a suspension of 6‐amino‐1H,3H‐benzo[de]isochromene‐1,3‐dione (50.0 mg, 0.235 mmol, 1.00 eq.) in absolute EtOH (2.35 mL, 0.1 M) was added the corresponding amine. The number of equivalents was adjusted depending on the amine employed (see the Supporting Information). For amine hydrochloride salts, Na₂CO₃ (1.0 equiv relative to the amine) was added. The resulting reaction mixture was heated at 90 °C until the completion of the reaction. After the reaction was completed (monitored by TLC), the crude mixture was evaporated to dryness and purified by flash chromatography on silica gel using an appropriate eluting solvent system to furnish pure dye. In some cases, recourse to reverse purification was necessary (see the Supporting Information). Complete experimental procedures for the synthesis of each ANI‐based dye, together with their full characterization and copies of the corresponding NMR and MS spectra, are provided in the Supporting Information.

4.2. Photophysical Measurements

UV–vis absorption measurements (scan mode) were conducted on a Cary UV 60 spectrophotometer (Varian) using rectangular 10 mm path length quartz cuvettes at 25 °C. Fluorescence spectroscopic studies (scan mode) were performed with a Spectrofluorimeter FS 5 (Edinburgh) at 25 °C, with rectangular 10 mm path length quartz cuvettes from Hellma. The emission and excitation spectra were recorded with a diluted solution lower than 0.1 absorbance values. For the excitation and emission spectra, the slit widths were adjusted to 1.0 nm. All fluorescence spectra were corrected for the lamp fluctuations and the apparatus response fluctuations. Fluorescence quantum yields (QY_FL) were measured at 25 °C using three diluted solutions (A < 0.1). For each QY_FL measurement, slit width, excitation wavelength, scan rate, integration time, and emission range were kept the same for the reference and the sample. The fluorescence quantum yield was determined thanks to the following formula:

$${\mathrm{QY}}_{\mathrm{FL}}^i=\frac{S_i}{S_s}\frac{n_i^2}{n_s^2}{\mathrm{QY}}_{\mathrm{FL}}^s$$

where S is the slope of the linear plot of the integrated fluorescence intensity in function of the absorbance value, n is the refractive index of the solvents (at 25 °C) used in measurements, and the subscripts s and i represent standard and unknown, respectively. Coumarin 153 (QY_FL = 0.53 in EtOH) was used as the standard [31].

4.3. Confocal Microscopy

After adequate confluence was obtained in an Ibidi μSlide 8 Well Glass Bottom plate, the medium was removed and replaced by a solution of ANI‐based dye (1–21) in RPMI medium, prepared at the desired concentration from a stock solution of fluorophore in DMSO (concentration of DMSO was kept under 1% v/v). After 3 h of incubation, the dye‐containing medium was discarded and replaced by fresh dye‐free RPMI medium before imaging. For colabeling experiments, cells were subjected to an additional coincubation step with the desired markers (MitoView633, NucSpot live 650, LysoView 633, or LipidSpot 610). Detailed experimental procedures for each marker are provided in the Supporting Information. Fluorescence images were acquired using a Leica SP8‐X inverted confocal microscope with a 40× or 60× oil or immersion objective (HC PL APO CS2 Leica). Excitation was performed using a diode of 440 nm LDH‐P‐C‐440B pulsed at 40 MHz or using a white light laser pulsed at 80 MHz. Detection was carried out by using a PMT detector (Hamamatsu 6357) collecting photons over the appropriate emission wavelength window.

4.4. Flow Cytometry (FACS)

Experiments were conducted on A549 cells. Cells were seeded in 12‐well plates at a density of 50,000 cells per well and maintained at 37 °C in a humidified atmosphere containing 5% CO₂ for 24–48 h. Once satisfactory confluence was reached, the culture medium was removed and replaced with 1.5 mL of a 5 µM solution of ANI‐based dye in RPMI medium, prepared from a fluorophore stock solution in DMSO (the final DMSO concentration was kept below 1% v/v). After 3 h of incubation, the dye‐containing medium was removed. The wells were washed twice with PBS and then treated with 400 µL of trypsin. Cells were collected in centrifuge tubes and centrifuged. The resulting cell pellets were resuspended in 500 µL of PBS and analyzed by flow cytometry. For each dye, incubations were performed in triplicate. At least 15,000 cells were acquired using a CytoFLEX S (Beckman‐Coulter) cytometer, driven by CytExpert 2.5 software. Fluorophores were excited by a 488 nm laser, and green fluorescence was collected through a 525/40 nm band pass filter. Data analysis was made using an R‐script. A first gate was made on an FSC‐height/SSC‐height dot plot to select cells, and doublets were discarded using a gate on the SSC‐area/SSC‐height dot plot. MFIs were then measured on a distribution histogram. Overlay presentation was made using Kaluza 2.1 software (Beckman‐Coulter). Different conditions were measured with a minimum of three independent experiments. Differences between means were evaluated by the Wilcoxon–Mann–Whitney T‐test with p < 0.05 being taken as the level of significance.

4.5. DFT Calculation

All quantum chemistry calculations were performed using the Gaussian 16 software package [37]. Conformational analysis was carried out by DFT using the B3LYP hybrid functional [38] in combination with the 6‐31G(d) basis set [39] a standard level of theory for substituted aromatic systems. To best mimic the biological environment, solvent effects (water) were implicitly incorporated using the integral equation formalism polarizable continuum model (IEFPCM) [40]. The rotational potential energy surface around the N—C bond was explored via a relaxed scan without symmetry constraints. Specifically, the C–N–C–C dihedral angle (θ, defined in Figure 6) was systematically varied in specific increments, allowing all other internal coordinates to fully relax at each step to account for steric hindrance.

Funding

This work was supported by Agence Nationale de la Recherche (ANR‐21‐CE18‐0005‐01, ANR‐18‐CE09‐0023‐01, ANR‐23‐CE07‐0041‐01, ANR‐11‐IDEX‐003‐02, and ANR‐24‐CE44‐5747‐02).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary Material

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.