Incorporating a Piperidinyl Group in the Fluorophore Extends the Fluorescence Lifetime of Click-Derived Cyclam-Naphthalimide Conjugates

Ligands incorporating a tetraazamacrocycle receptor, a ‘click’- derived triazole and a 1,8-naphthalimide fluorophore have proven utility as probes for metal ions. Three new cyclam-based molecular probes are reported, in which a piperidinyl group has been introduced at the 4-position of the naphthalimide fluorophore. These compounds have been synthesized using the copper(I)-catalyzed azide-alkyne Huisgen cycloaddition and their photophysical properties studied in detail. The alkylamino group induces the expected red-shift in absorption and emission spectra relative to the simple naphthalimide derivatives and gives rise to extended fluorescence lifetimes in aqueous buffer. The photophysical properties of these systems are shown to be highly solvent-dependent. Screening the fluorescence responses of the new conjugates to a wide variety of metal ions reveals significant and selective fluorescence quenching in the presence of copper(II), yet no fluorescence enhancement with zinc(II) as observed previously for the simple naphthalimide derivatives. Reasons for this different behaviour are proposed. Cytotoxicity testing shows that these new cyclam-triazole-dye conjugates display little or no toxicity against either DLD-1 colon carcinoma cells or MDA-MB-231 breast carcinoma cells, suggesting a potential role for these and related systems in biological sensing applications.

We have recently developed a novel class of fluorescent probes for Zn 2+ (Figure 1) by attaching the 1,8-naphthalimide fluorophore to a tetraazamacrocycle scaffold via copper(I)-catalyzed azidealkyne Huisgen cycloaddition (colloquially known as the click reaction). [19][20][21][22] The click-generated triazole is a linker but also acts as a coordination site, thus playing a role in the metal ion binding and detection. Compounds 1-4 signal the binding of Zn 2+ to the tetraazamacrocycle-triazole moiety with a multifold increase in fluorescence emission of the pendant 1,8-naphthalimide. Reversing the triazole topology in the cyclam-triazole-naphthalimide system (3 vs. 1) gives a 10-fold brighter fluorescence response to Zn 2+ in HEPES buffer (10 mM, pH 7.4). [22] Furthermore, the cyclam-based probe 1 has been used to detect the cellular Zn 2+ flux during apoptosis in vitro, [21] and the cyclen-based probe 2 has been applied in vivo to image Zn 2+ in zebrafish. [19] In a related approach, tethering a second pendant group (biotin) to the zinc(II) complex of compound 1 afforded a fluorescent 'allosteric scorpionand' probe 5 that visualizes the binding of the pendant biotin to the cognate biomolecule avidin. [23] Replacement of the 1,8-naphthalimide dye in compounds 1 and 3 with the coumarin fluorophore provided probes 6 and 7 ( Figure 1) that respond selectively to Cu 2+ and Hg 2+ [22,24].
To minimize cell damage and interference from background autofluorescence in cell-based assays, the absorption and emission spectra of the fluorescent probe should be as close as possible to the red end of the visible spectrum. [25] In this regard the spectral characteristics of probes 1-7 are sub-optimal (l abs ,320-360 nm, l em ,380-460 nm in aqueous buffer). [19][20][21][22][23][24] Previous studies have shown that introducing alkylamino groups at the naphthalene moiety of 1,8-naphthalimide induces such a bathochromic shift. [11,[26][27][28][29] To this end, we designed three new cyclampiperidinylnaphthalimide conjugates 8-10 ( Figure 2). A phenyl linker was used in compounds 8 and 10 to connect the cyclamtriazole moiety to the piperidinylnaphthalimide fluorophore, while compound 9, containing a flexible ethylene chain, was designed as a control to verify the importance of conjugation. The metal-ion  responsiveness, fluorescence quantum yields and decay times, and cytotoxicity of these new conjugates were investigated to explore their potential for application as metal ion probes in vitro and in vivo.

(a) Synthesis
Synthesis of the cyclam-piperidinylnaphthalimide conjugates 8-10 required the preparation of precursors 13, 17 and 20 ( Figure 3). Azide 17 [30,31] and alkyne 20 [29,30,32] were successfully synthesized according to literature procedures, whereas the preparation of azide 13 proved challenging. Conversion of bromide 12 to the corresponding azide 13 was initially attempted with sodium azide in the presence of sodium ascorbate, copper(I) iodide and N, N9-dimethylethylenediamine (DMEDA) at reflux in either an ordinary round-bottomed flask or a pressure tube. [33][34][35] A solvent screen including methanol/water, ethanol/water or dimethyl sulfoxide (DMSO)/water (7:3 in all cases) failed to afford the desired azide 13, giving instead full recovery of starting material 12; this outcome may be attributed to the extraordinarily low solubility of bromide 12 in these solvent combinations. Switching to tetrahydrofuran (THF)/water (7:3), all reactants and reagents were dissolved at reflux in the pressure tube and reaction proceeded to give azide 13 in 50% yield. The corresponding amine was also detected by LCMS analysis of the reaction mixture, consistent with previous observations that both azide and amine may be generated through a copper-assisted aromatic substitution reaction with sodium azide. [33] Reacting each of the three precursors 13, 17 and 20 individually with the complementary propargyl-tri-Boc cyclam [23,36] or azidoethyl-tri-Boc cyclam [24] under standard click conditions [24,36] yielded the Bocprotected cyclam-piperidinylnaphthalimide conjugates 14, 18 and 21 respectively in good to excellent yields. Removal of Boc groups from these conjugates was effected in a mixture of TFA/DCM/ H 2 O (90:5:5), [23,36,37] followed by basification to recover the corresponding free amines 8-10. However, the outcome of the basification step was contingent on the base used. Addition of 2 M sodium hydroxide solution [36,38] or saturated sodium carbonate solution [36] resulted in decomposition of the desired free amines or incomplete removal of trifluoroacetate counter ions respectively (indicated by analysis with 1 H and 13 C NMR spectroscopy). Successful isolation of the pure amines 8-10 was achieved using excess Ambersep 900 (hydroxide form) in methanol.

(b) Photophysical Properties
i) Steady-state photophysical properties. The steady state photophysical properties of cyclam-piperidinylnaphthalimide conjugates 8-10 were investigated using both UV-Vis and fluorescence spectroscopy. The UV-Vis absorption spectra of 8-10 in HEPES buffer (10 mM, pH 7.4) are almost identical, with the lowest-energy absorption (l abs ) centered at 41562 nm and stretching out to 500 nm ( Figure 4). The fluorescence emission spectra of 8-10 are only slightly shifted giving a broad emission band ranging from 500 to 700 nm, centered around 545-558 nm (l em ) ( Figure 4). Introduction of the piperidine to the naphthalimide fluorophore not only leads to a red-shifted emission maximum but also to a broadening of both the absorption and emission bands. The similarity of these spectra in aqueous buffer is remarkable, and implies i) the role of the linker (phenyl 8 vs ethyl 9) exerts minimal influence and ii) the triazole connectivity (8 vs 10) does not have a significant impact on the UV-Vis absorption and fluorescence emission of these conjugates. The fact that the psystem is not extended in 8 or 10 by conjugation of the phenyl group with the 1,8-naphthalimide core can be rationalized by considering a twisting of the two aromatic planes to minimize adverse steric interactions. This effect may be enhanced after excitation of the probe, giving rise to charge separated states and significant solvent-dependent variation in spectral properties.
Screening the spectral properties of 8-10 in various solvents spanning a wide range of polarities revealed a solvent-dependent shift in absorption, and -to a much larger extent -emission maxima (Table 1). Comparing measurements made in aqueous buffer versus non-polar toluene shows that the impact of solvent polarity is less in the case of ethyl-linked 9, where the emission in HEPES buffer (545 nm) shifts less than 40 nm in toluene (507 nm). In the emissions of 8 and 10, a blue-shift of nearly 60 nm is seen in toluene relative to HEPES buffer. The Stokes shifts (D n n) (calculated from the difference of the absorption and emission maxima) allow easier comparison: the Stokes shifts of ligands 8 and 10 respond similarly throughout the solvent screen; the slight differences that are observed between the two ligands can be attributed to the effect of the different triazole connectivity (further evidence for the minor impact this structural change exerts on the spectral properties). More importantly, there is a distinct decrease in the Stokes shift of both compounds when moving from HEPES buffer into less polar solvents, suggesting that charge separation in the excited state is most likely linked to conformational changes. In the ethyl-linked analogue 9, the effect of the solvent is weaker, indicating that the excited state of 9 incorporates a much smaller charge separation. Lippert-Mataga plots [39][40][41] (Figures S1-S3 and Text S1 in File S1) were constructed to build a picture of solvent-fluorophore interactions. The Stokes shifts of all three conjugates in the hydrogen bonding solvents (e.g. alcohols) are typically greater than those in solvents that less readily form hydrogen bonds (e.g. toluene); such behavior can be attributed to protic solvent-fluorophore hydrogen bonding and has been observed for other fluorophores [42,43].
ii) Response to metal ions. The UV-Vis and fluorescence responses of conjugates 8-10 to a wide variety of metal ions (Ag + , Ba 2+ , Ca 2+ , Cd 2+ , Co 2+ , Cu 2+ , Fe 2+ , Fe 3+ , Hg 2+ , K + , Li + , Mg 2+ , Mn 2+ , Na + , Ni 2+ , Pb 2+ , Rb + and Zn 2+ ) were assessed in HEPES buffer (10 mM, pH 7.4 -see File S1). Of the metals tested, only Cu 2+ triggered a significant response, quenching the fluorescence of all three conjugates (Figures S4-S6 in File S1). This response is consistent with previous observations that Cu 2+ quenches the fluorescence of derivatives 1, 3, 6 and 7, [20][21][22] and may be due to paramagnetic or heavy atom effects; [44][45][46][47] work is underway to determine the mechanism of Cu 2+ -mediated fluorescence quenching in these systems. However none of the cyclampiperidinylnaphthalimide conjugates 8-10 show any meaningful response to either Zn 2+ or Hg 2+ , in contrast to the previouslyreported cyclam-naphthalimide conjugates 1, 3 and 4 which exhibited fluorescence increases in the presence of Zn 2+ and quenching in response to Hg 2+ respectively. [20][21][22] Addition of Co 2+ , Fe 2+ and Fe 3+ each triggered a small to moderate reduction in the fluorescence of 10, but had no effect on the fluorescence of 8 or 9. Taken together, these fluorescence results imply that both the nature of the pendant fluorophore and the connectivity between the fluorophore and metal-cyclam complex play a role in the metal-ion responsiveness of these conjugates. The addition of these metal ions had no significant effect on the UV-Vis absorption spectra of all three conjugates ( Figures S7-S9 in File S1).
To investigate the effectiveness of 8-10 as probes for Cu 2+ in the presence of competing metal ions, competitive binding experiments were conducted using Zn 2+ . Thus a 10 mM solution of 8-10 in HEPES buffer (10 mM, pH 7.4) was combined with 50 equivalents of Zn 2+ , followed after approximately 3 minutes by 1 equivalent of Cu 2+ . In all cases, much weaker fluorescence quenching was observed than in the experiments in which the two metal ions were added in the reverse order, or a premixed Cu 2+ / Zn 2+ (1:50) solution was added ( Figure 5). These results show the effectiveness of 8-10 as Cu 2+ -probes but indicate a limitation in the presence of high Zn 2+ concentrations.
The fluorescence responses of probes 8-10 were evaluated over a wide pH range, both in the absence and presence of Cu 2+ (Figures S10-S12 in File S1). These experiments indicate optimum responsiveness to Cu 2+ at neutral pH. At low pH, the fluorescence responses of the free ligands 8-10 change little in the presence of Cu 2+ , presumably due to inhibition of metal coordination when the cyclam amine groups are protonated. The fluorescence of the free ligands 8-10 is diminished at high pH, as previously observed with probe 1. [21] However, it is the absence of any protonationinduced fluorescence enhancement with probes 8-10 that is more significant. This indicates that the photoinduced electron transfer (PET) from the cyclam-triazole moiety to naphthalimide observed with probes 1-3 [22] does not occur with the piperidinylnaphthalimide fluorophore incumbent. This in turn means that the full fluorescence response of 8-10 is turned 'on' in the free ligands, eliminating the possibility of a fluorescence 'turn-on' pathway Extending the Fluorescence Lifetime of Cyclam-Naphthalimide Conjugates PLOS ONE | www.plosone.org upon protonation or metal binding. The fact that PET is not favoured with probes 8-10 can be rationalized by considering the push-pull-character of the 4-aminonaphthalimide, where the electron acceptor is located at the amine and the electron donor at the imide. In the excited state, the resultant negative charge density on the imide inhibits acceptance of an additional electron via PET when the electron donor is connected at this position on the fluorophore [48].
iii) Time resolved photophysical properties and fluorescence quantum yields. Fluorescence quantum yields were acquired in three representative solvents (HEPES buffer, ethyl acetate and acetonitrile) to investigate the intrinsic photophysical properties in more detail ( Table 2). In HEPES-buffer and acetonitrile, the quantum yields and the fluorescence decay times of the free ligands 8-10 are generally low, although ligand 9 gives a significantly longer decay time (,t. = 4.86 ns) in buffer compared to ligands 8 and 10 (,t. = 2.47 and 2.42 ns respectively). The quantum yields in ethyl acetate (0.25-0.50) are at least one order of magnitude higher than in acetonitrile (0.036-0.052) and about twice as high as in buffer (0.005-0.009). Clearly solvent has a strong influence on the photophysical properties of these probes. Strong solvent-dependence was also observed in the decay time profile of all three ligands. In ethyl acetate, ligands 8 and 10 decay with a single exponential profile, ligand 9 with a bis-exponential profile. In acetonitrile, decay times for all three ligands are fitted with two exponentials. In buffer, three exponentials give the best fit in all three cases. These multiexponential fits indicate the presence of multiple excited species in these solvents. The additional components observed in aqueous buffer over the organic solvents can be rationalized by considering changes in ligand protonation, which give rise to new species that are absent in the aprotic organic solvents. Interestingly, the photophysical properties of 8, 9 and 10 change relative to each other with changes in solvent: while similar values for quantum yields and decay times are observed for all three ligands in acetonitrile, only 8 and 10 afford similar data in ethyl acetate and buffer, and the values recorded for 9 are appreciably different. In ethyl acetate, longer decay times for 8 and 10 (5.73 and 6.36 ns respectively) and higher quantum yields (0.44 and 0.50) are found compared to ligand 9, but in buffer ligand 9 gives higher values than ligands 8 and 10. Notably, the averaged decay of 9 is particularly long (,5 ns) in the aqueous solvent. The long decay times may render these new probes suitable for time-correlated assays, e.g. fluorescence lifetime imaging (FLIM) techniques in biological samples.

(c) Biological Evaluation
The cytotoxicity of cyclam-piperidinylnaphthalimide conjugates 8-10 and Boc-protected precursors 14, 18, and 21 was assayed against DLD-1 colon carcinoma cells and MDA-MB-231 breast carcinoma cells (Table 3). Cisplatin was used as a positive control in this cell viability study; its cytotoxicity against DLD-1 cells was found to be 11.260.3 mM, consistent with literature value of 11.861.2 mM. [49] Five of the six cyclam-piperidinylnaphthalimide compounds did not display any significant cytotoxicity against either cell line, with a safe dosage level of 20 mM. The single exception was conjugate 10, which showed moderate activity against both carcinoma cell lines. In general, IC 50 values for all cyclam-piperidinylnaphthalimide conjugates 8-10 were lower than those for the corresponding Boc-protected counterparts 14, 18 and 21.

Conclusions
We have reported the synthesis of three cyclam-piperidinylnaphthalimide conjugates 8-10 which respond to the presence of copper(II) with a significant decrease in fluorescence. Despite the different triazole connectivities and the variation of the pendant alkyl arm length, these probes exhibit remarkably similar photophysical properties. However, these photophysical properties are highly dependent on solvent, as seen in the UV-Vis and fluorescence spectra, quantum yields and decay times of all three ligands. The influence of the flexible ethyl linker is reflected in the long averaged fluorescence decay time of compound 9 in HEPES buffer, which is twice as long as those of ligands 8 and 10. None of the probes display significant cytotoxicity to mammalian cells, supporting the potential suitability of this new probe class for sensing, labeling or imaging studies in biological systems.

(a) General Materials
All reactions except azidation of 12 were carried out with continuous magnetic stirring in ordinary glassware; azidation of 12 was performed in a 15 mL Ace pressure tube, purchased from Sigma-Aldrich. Heating of reactions was conducted with a paraffin oil bath or a water bath. All reagents and solvents were purchased from Sigma-Aldrich, Alfa Acer, Merck, or Ajax Finechem. Reagents were used as received unless otherwise specified. Hexane and ethyl acetate were distilled before use. Dichloromethane and ethanol were distilled over calcium hydride and stored over activated 4 Å molecular sieves. Tetrahydrofuran was distilled over sodium wire/benzophenone. Methanol and acetonitrile were collected freshly from a PureSolv MD 7 solvent purification system having been passed through anhydrous alumina columns.   Table 1. Photophysical properties of 8-10 in various solvents with decreasing polarity from aqueous (HEPES buffer) to toluene.

(c) Synthesis
See the Supporting Information for synthetic experimental procedures of known compounds (Text S2 and Figures S13-S15 in File S1) and 1 H and 13 C NMR spectra of novel compounds (Figures S16-S33 in File S1).

(e) Cell Viability Assay
The cytotoxicity of compounds 8-10, 14, 18 and 21 was evaluated using cell viability assay as described previously. [50] Compounds 8-10, 14, 18 and 21 were prepared as a 10 mM stock solution in DMSO and diluted with growth medium (2% FCS and 1% glutamine) to give rise to a range of concentrations (0-200 mM). DLD-1 colon carcinoma cells and MDA-MB-231 breast carcinoma cells were cultured as monolayers in Advanced DMEM, supplemented with 2% FCS, 1% glutamine and 1% antibiotic/antimycotic (A/A). Cells were incubated at 37uC with 5% CO 2 in a humidified incubator, seeded at 1610 4 cells per well of a 96-well plate in 100 mL of growth medium, and allowed to adhere for 15 h. Growth medium was removed, and 100 mL of compounds at different concentrations were added in triplicate. The plates were incubated for 72 h. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT, 20 mL), a water-aqueous soluble yellow tetrazole compound, was added to a final concentration of 1 mM per well and the plates were incubated for 4 h. Growth medium was removed, and DMSO (150 mL) was added to dissolve the water-insoluble purple formazan crystals. The plates were shaken until all the crystals were dissolved. The absorbances at 600 nm were read by a microplate reader (Victor, PerkinElmer) and averaged for each concentration. IC 50 value was determined by the concentration of the compound at which the absorbance was half of that of the cells grown in only growth medium. The average of three independent IC 50 values for each concentration was used to calculate the standard error of the mean.