In vivo NIR-to-NIR imaging with bispidine–lanthanide(iii) complexes

Abstract

Two-photon luminescent bioprobes have emerged as powerful tools for in cellulo and in vivo imaging. Here, we report a family of bispidine-based chelators L^1a, L^1b and L^1c bearing push–pull two-photon antennas for efficient sensitization of lanthanide(iii) ions. The ligands and their complexes are readily accessible in good yields of 16–22% and 39–87%, respectively. The resulting complexes [EuL^1a]⁺, [SmL^1a]⁺, [DyL^1b]⁺, [TbL^1b]⁺ and [YbL^1c]⁺ display two-photon excitability combined with high quantum yields, long luminescence lifetimes, and high brightness, making them highly suitable for bioimaging applications. Importantly, the probes [EuL^1a]⁺, [SmL^1a]⁺, [DyL^1b]⁺, and [TbL^1b]⁺ show no apparent nonspecific binding in cellulo and can be readily washed out, indicating minimal interaction with cellular membranes or the intracellular matrix. Furthermore, [EuL^1a]⁺ and [YbL^1c]⁺ were successfully applied for in vivo imaging in zebrafish larvae, confirming their non-specificity and demonstrating their potential for angiographic imaging. The vascular system was clearly visualized, with sharp visible emission observed for [EuL^1a]⁺ and a completely dark background obtained for [YbL^1c]⁺. These results establish bispidine-based lanthanide(iii) complexes as robust and versatile lanthanide(iii) luminescent probes for biological imaging.

Two-photon luminescent bioprobes have emerged as powerful tools for in cellulo and in vivo imaging.

Introduction

Lanthanide complexes are of interest for various types of imaging, in particular for biological materials and medicinal applications, including radionuclide-based methods, magnetic resonance imaging with lanthanide-based contrast agents and optical microscopy imaging.^1–3 For all these applications, efficient complexation, high complex stability and in particular inertness are of importance. Open-chained ligands with rigid scaffolds and cavity shapes and sizes well adapted to the specific metal ion often outperform ligands based on macrocyclic platforms in terms of complexation efficiency with still high complex stability and inertness.⁴ Among the large variety of polydentate and macrocyclic ligands reported in the literature,^5–7 nonadentate bispidine derivatives (bispidine = 3,7-diazabicyclo[3.3.1]nonane, see Scheme 1 for the structure and numbering of L¹ and for complex structures) with planar bi- and tridentate donor groups at N3 and N7 have been shown to (i) have an ideal cavity size for lanthanide(iii) ions (predicted by molecular mechanics-based hole size calculations^8–10); (ii) fully encapsulate the lanthanide(iii) ions (confirmed by a range of single crystal X-ray structures, see Scheme 1 for an overlay plot of the available molecular structures⁸); (iii) lead to fast and efficient complexation of lanthanide(iii) ions (supported by studies of radiolabeling with ¹³³La^III and ¹⁷⁷Lu^III 8); and (iv) have very high thermodynamic stabilities and be kinetically inert (confirmed by potentiometric titrations and challenge experiments of the radioprobes with human serum as well as competing metal ions and small molecule ligands^8,11). Preliminary experiments and computational studies suggest that complex formation and decay are not microscopically reversible,¹² and the two main features are that the metal ion is captured by the tridentate donor attached to N3 (note that, for ligands with two different substituents, it is of importance which pendant group is attached to N3 and N7, but for the current study this is not relevant), and that the last step involves removal of the H⁺ in the bispidine cavity connected to N3 and N7 atoms that coordinate the metal ion (swinging-in of the metal ion into the cavity). Thus, the lone pairs at N3 and N7 are blocked from attack by protons and metal ion loss,⁸ leading to extremely stable and inert metal complexes.

Scheme 1. Structure and atom numbering of L¹, and calculated hole size of L¹ together with an overlay plot of the experimental structures of the complexes with Eu^III, Gd^III, Tb^III, Yb^III, and Lu^III,^8,11,13 and arrows indicating the size of the respective cations.

In short, lanthanide(iii) complexes of L¹ are predicted to have a well-defined rigid structure without any interaction of the central metal ion with H₂O solvent molecules, to be very stable and inert and therefore be ideal as luminescent probes. Indeed, aqueous solutions of [Tb(L¹)]³⁺, [Eu(L¹)]³⁺ and [Yb(L¹)]³⁺ have well-resolved optical spectra in the visible or NIR spectral range, respectively, with excited state lifetimes in the millisecond range and quantum yields up to 70% for the terbium complex.¹³ The bispidine platform L¹ is therefore perfectly suited for the design of lanthanide luminescent bioprobes (LLB) for two-photon (2P) microscopy imaging.^14–16 To that end, the functionalization of the picolinate moieties by π-conjugated antenna to improve the 2P-cross-section was envisaged.^17–20 This strategy, initially developed for triaza-cyclononane-based macrocyclic complexes of Yb^III,²¹ Eu^III,²² Sm^III 23 and Tb^III or Dy^III 24 has been successfully exemplified using various macrocyclic (cyclen,^25–27 cyclam,²⁸ pyclen²⁹) and polydentate complexes.^30,31 It is worth noting that combination of 2P excitation with a NIR emitting lanthanide (Sm, Yb or Er) enables 2P microscopy imaging in a NIR-to-NIR configuration, that is, with both excitation and detection in the favourable biological transparency window ideal for in-depth thick-tissue imaging.^{15,21,32–35}

Here, we present the synthesis and characterization of three derivatives of the nonadentate bispidine L¹ noted as L^1a–c (Scheme 2) with intraligand charge transfer (ILCT)-optimized antennas combining donor-π-conjugated acceptor chromophores suitable to ensure efficient 2P-excitation as well as the isolation and photophysical characterization of their Eu^III, Sm^III (L^1a), Tb^III, Dy^III (L^1b) and Yb^III (L^1c) complexes. Also reported are 2P bioimaging studies performed either in HEK cells or zebrafish embryos as a small animal model.

Scheme 2. Synthesis of the three antenna-derived nonadentate bispidine ligands L^1a, L^1b and L^1c.

Results and discussion

Synthesis

For the ligand synthesis, we followed routes developed for other multidentate bispidines with orthogonal protection of the amines of the diazaadamantane scaffold,^13,36–38 where, in contrast to the previous preparative scheme for L¹, the substituent at N3 was first introduced in order to obtain a common intermediate for the three derivatives L^1a, L^1b and L^1c with 2P antennas at the picolinate donor appended to N7 (see Scheme 2 and the SI for experimental details). The three methyl ester-protected picolinate derivatives were obtained as described in the literature^21,22,24 and used in the following deprotection reactions without further purification. The antenna-derived chelators were synthesized in a one-pot synthesis approach with L^1a–c in 16–22% yield (SI Fig. S1 and S7–S14).

The lanthanide(iii) complexes (Eu^III, Sm^III: L^1a; Tb^III, Dy^III: L^1b; Yb^III: L^1c) were obtained from equimolar amounts of the corresponding lanthanide nitrates and the ligands L^1a–c dissolved in methanol/water (1 : 1). The complexes were purified by means of reverse-phase column chromatography, which upon removal of the solvents under reduced pressure produced solid samples of the complexes in 39–87% yield (SI Fig. S2–S6). The NMR spectra of the complexes show distinct paramagnetically shifted resonances for all described complexes (except Dy^III). However, complete assignment was not possible, most likely due to the presence of multiple enantiomeric species in solution, as previously observed for the parent system (Fig. S15–S18).¹³ No X-ray single crystal structures of the complexes with the 2P-antenna ligands L^1a, L^1b and L^1c were obtained, possibly due to the presence of the small hydrosolubilizing polyethylene glycol (PEG) substituents, but the rigidity of the ligand does not allow for any conformational flexibility, cleavage of a single metal–ligand bond or addition of a solvent molecule (see also Scheme 1),^8,11–13 which suggests that these structures are identical to those of the parent ligand. This is corroborated by a DFT geometry optimization of the yttrium(iii) complexes with the three ligands L^1a, L^1b and L^1c in comparison with that of L¹ and the X-ray structure of the corresponding Gd^III complex (see the SI for details and Fig. S19 and Table S1).

Optical properties

Solution absorption and emission spectra of the five complexes with the charge transfer antennas are shown in Fig. 1, and all photophysical data are assembled in Table 1. The absorption spectra of the complexes are dominated by intense, broad and structureless transitions centered at around 323 nm for L^1a, 310 nm for L^1b and 390 nm for L^1c-based complexes, assigned to intraligand charge transfer (ILCT) transitions from the oxygen or nitrogen donor substituent at the antenna to the electron-withdrawing picolinate function. The first excited states of Sm^III (⁴G_5/2: 17 900 cm⁻¹) and Eu^III (⁵D₁: 19 000 cm⁻¹, ⁵D₀: 17 400 cm⁻¹) are comparable in energy and, therefore, can be sensitized by the same antenna, featuring the alkoxyphenylethynyl group (L^1a).^23,29 Similarly, the almost isoenergetic relevant excited states of Tb^III (⁵D₄: 20 500 cm⁻¹) and Dy^III (⁴F_9/2: 21 100 cm⁻¹ respectively) are well sensitized by the twisted charge transfer antenna (L^1b),^24,29 and the relatively low energy excited state of Yb^III is well suited for L^1c with the strongest dialkylamino donor group²⁹ (see SI Fig. S20 for the corresponding absorption spectra).

Fig. 1. Normalized absorption (dashed lines), excitation and emission spectra (solid lines) of [EuL^1a](NO₃), [SmL^1a](NO₃), [TbL^1b](NO₃), [DyL^1b](NO₃), and [YbL^1c](NO₃) at 298 K in H₂O showing the ligand-based emission at high energies, the metal ion-centered emission of the complexes in the visible region and the NIR region. Excitation wavelengths were 335 nm for [SmL^1a](NO₃) and [EuL^1a](NO₃), 306 nm for [TbL^1b](NO₃) and [DyL^1b](NO₃), and 390 nm for [YbL^1c](NO₃). Emission wavelengths were taken as the maxima of the metal-based luminescence signals (* indicates the Raman scattering signal of the solvent, here water). The experimental conditions for luminescence measurements are not identical for all complexes and have been optimized for each detector (visible and NIR).

Table 1. Selected solution photophysical properties of the lanthanide(iii) bispidine complexes of the fully deprotected ligands [EuL^1a](NO₃), [SmL^1a](NO₃), [TbL^1b](NO₃), [DyL^1b](NO₃), and [YbL^1c](NO₃) at 298 K alongside data of the parent complexes [EuL¹](CF₃CO₂), [TbL¹](NO₃) and [YbL¹](NO₃)^a,b.

Compound	Media	λ abs max (nm)	ε (L mol−1 cm−1)	τ (µs)	Φ c , d	B (L mol−1 cm−1)
[EuL1a][NO3]	H2O (D2O)	323	22 000	1230 (1640)	12%	2600
	MeOH			1580	27%
	Bufferb			1210	17%
[Sm1a][NO3]	H2O	323	23 000	21	0.29%	67
	MeOH			28	0.53%
[TbL1b][NO3]	H2O (D2O)	310	14 500	1840 (2010)	51%	7400
	MeOH			1300	34%
	Bufferb			1740	52%
[DyL1b][NO3]	H2O	310	19 000	<10	0.37%	7
	MeOH			<10	0.39%
[YbL1c][NO3]	H2O	390	22 000	3.4	0.15%e	33
	MeOH			5.8	0.36%e (0.67%f)
[EuL1][CF3CO2]a	H2O (D2O)	314	8000	1510 (2150)	35%
[TbL1][NO3]a	H2O (D2O)	313	9000	1950 (2150)	68%	6120
[YbL1][NO3]a	H2O	312	9500	3.6	n.d.	—

^aData for the L¹-based complexes from ref. 13.

^bBuffer used: HEPES pH 7, 0.1 M.

^cPrecision ±5%.

^dUsing quinine bisulfate in 0.5 M H₂SO₄ as the standard (Φ = 0.546). Excitation wavelengths for quantum yields were 335 nm for [EuL^1a], 335 nm for [SmL^1a], 306 nm for [TbL^1b], and 306 nm for [DyL^1b].

^eUsing Yb(TTA)₃Phen as a reference for NIR emission quantum yield in toluene (Φ = 1.1%), λ_ex. = 350 nm.^39,42

^fLigand-centered luminescence.

Upon excitation of the ILCT transition, the respective complexes are highly emissive, producing well-resolved emission spectra typical of Ln^III ions, with characteristic bands corresponding to the lanthanide-centered energy levels (red emission for [EuL^1a]⁺ (⁵D₀ → ⁷F_J, J = 0–4), orange emission for [SmL^1a]⁺ (⁴G_5/2 → ⁶H_J, J = 5/2–11/2 in the visible and ⁴G_5/2 → ⁶F_J, J = 3/2–9/2 in the NIR region), green emission for [TbL^1b]⁺ (⁵D₄ → ⁷F_J, J = 1–6), yellow-white emission for [DyL^1b]³⁺ (⁴F_9/2 → ⁶H_J, J = 15/2–5/2), and NIR emission for [YbL^1c]⁺ (²F_5/2 → ²F_7/2)). Quantum yields were determined under optically dilute conditions using the external standard method (standard = quinine bisulfate in 0.5 M H₂SO₄, Φ = 54,6% for visible emission and Yb(TTA)₃Phen in toluene, Φ = 1.1% for the Yb^III NIR emission,³⁹ Fig. S21–S23), and luminescent lifetimes were estimated using mono-exponential fits of the luminescence decays (Table 1). Compared to the non-functionalized complexes with ligand L¹,¹³ the introduction of a π-conjugated substituent onto the picolinate significantly decreases the quantum yield of the Eu^III and Tb^III complexes and redshifts the excitation wavelength. However, due to increased extinction coefficients, their brightness remains comparable, meaning that the complexes of L¹ and the modified compounds exhibit similar performance. The major advantage of the new complexes lies in their potential for 2P-excitation due to the presence of the well-known 2P-absorbing antennas.²⁹ The hydration number of the complexes was determined for both [EuL^1a]⁺ and [TbL^1b]⁺ by measuring the luminescence lifetimes in heavy water.^40,41 As expected, no water molecules were found to coordinate to the metal centers. This further supports the structural similarity of these complexes to their parent compounds, featuring ligand L¹ without the appended antenna.

The nonlinear optical properties of the complexes have been determined using the two-photon excited fluorescence method (TPEF) using a Ti-sapphire source in water solution. For all complexes (except Yb^III), the quadratic dependence of the lanthanide emission on the incident laser power was checked under 700 nm 2P-excitation (Fig. S24). For [YbL^1c](NO₃), only the residual ligand centered emission was used because our setup is not able to detect NIR emission.

The 2P-absorption spectra and 2P-cross-sections were then measured for the complexes featuring a quantum yield higher than 2% to limit the experimental error (Fig. S25–S27), here [EuL^1a]⁺ and [TbL^1b]⁺ (Table 1). The 2P absorption spectrum of the Eu^III complex matches well the red tail of the wavelength-doubled trace, reaching a maximum at 680 nm, the limit of the Ti:sapphire laser range, with a cross-section of 40 GM, as expected for this type of antenna.^17,43,44 In the case of Tb^III, a modest 2P-cross-section of 6 GM is reached at 680 nm that does not correspond to the maximum located at higher energy, out of the laser range.^24,25

Cellular imaging

The [SmL^1a]⁺, [EuL^1a]⁺, [TbL^1b]⁺, and [DyL^1b]⁺ probes were used for live-cell and fixed-cell imaging of HEK293 cells (human embryonic kidney cells). For this purpose, 100 µM solutions of the complexes were prepared in water. The cells were first incubated with the 2P-LLBs in Leibovitz medium (with phenol red) for 2 hours at 37 °C. A representative cell imaging experiment performed in the case of [EuL^1a]⁺ is represented in Fig. 2. After 1 h incubation in the presence of living cells, no internalization of the complex was observed, the cells remained perfectly dark, and the complexes stayed in the intercellular medium (Fig. 2A). The cells were then fixed with a 4% paraformaldehyde solution in PBS (pH 7.4) for 30 min in order to make their membranes permeable. Addition of the complexes resulted in homogeneous staining of the cells and the surrounding medium (Fig. 2B), but after rinsing, the complexes are completely removed from the extracellular medium and also from the cells. This result is in marked contrast to what was previously observed with other 2P-LLBs. Generally, cationic complexes are spontaneously internalized in living cells, staining mostly endosomes,^28,43,44 whereas anionic or neutral complexes are unable to stain living cells but accumulate in lipophilic sub-cellular compartments in fixed cells.^29,30 The bispidine compounds [LnL^1a–c]⁺ are formulated as cationic but this does not consider the two carboxylates at C1 and C5, which are not coordinated to the lanthanide ions. Under the staining conditions (pH = 7.4), it is reasonable to consider that one or both of these carboxylic functions are deprotonated, generating neutral or monoanionic structures (Scheme 3). The neutral form will have an asymmetric charge distribution, akin to zwitterions. It is well established that zwitterionic functions (like sulfo-betaine for instance) strongly limit the non-specific interactions with the lipophilic biological surrounding⁴⁵ compared to other neutral (poly-ethylene glycol) or charged (sulfonate, carboxylate, or ammonium) hydrosolubilizing functions.^30,46 In conjunction with nanoparticles^47,48 or organic fluorophores,^49,50 zwitterionic functions strongly improve hydrophilicity, and consequently hydrosolubility, ultimately limiting non-specific hydrophobic interactions with biological material (membrane, protein, etc.). This strategy was tested ten years ago by decoration of a lanthanide complex with peripheral sulfo-betaine fragments, and exactly the same behavior as described for the bispidine complexes was observed (i.e. reversible staining of fixed cells).⁴⁵

Fig. 2. 2P-microscopy imaging of living (A) and PFA-fixed (B) HEK cells incubated with [EuL^1a]⁺ (100 µM, λ_exc = 700 nm). Left panel: luminescence image recorded with 420–680 nm bp APD detection (arbitrarily colored in red); middle panel: differential interference contrast (DIC) image; right panel: merge. Scale bars correspond to 10 µm. The upper line is the image of the cells after 1 h incubation and the lower line after rinsing the cells with PBS.

Scheme 3. Acid–base equilibria in aqueous solution of the Ln^III complexes studied.

Partition coefficients of [EuL^1a]⁺ and [TbL^1b]⁺ were determined using 1-octanol/buffer (HEPES, pH 7, 0.1 M) distribution experiments and determining the quantity of the complex with UV/vis spectroscopy. These data were compared with different antenna-derived structures (Fig. S28 and Table S2). The partition coefficients (log P < −2) for all bispidine complexes indicate high hydrophilicity, contrasting with previously studied TACN and pyclen-based lanthanide complexes, which typically have less negative log P values ranging from −1.6 to −0.3.⁵¹ While all compounds are hydrophilic, this finding helps to explain the results of initial cell-staining experiments, where no nonspecific binding was observed for the probes [SmL^1a]⁺, [EuL^1a]⁺, [TbL^1b]⁺, and [YbL^1c]⁺, strengthening the hypothesis of a zwitterionic or anionic form at pH 7.4 (Scheme 3).

Zebrafish imaging

The fact that the functionalized bispidine complexes do neither stain nor interact with living cells is of course discouraging from the cellular bioimaging point of view but is a real advantage for angiography, an imaging technique consisting of the imaging of blood capillary vessels in thick tissues and, even better, in small animals. The requirement for angiographic probes is to circulate in the blood vessel without sticking at their cellular periphery. Zebrafish (Danio rerio) embryos are well-established small animal models and are recognized as a valuable alternative to small mammals.⁵² Such imaging experiments have already been performed with organic dyes^53–56 and, in a few cases, with europium^29,57 but also NIR-emitting ytterbium and erbium molecular complexes.^21,32,33,58

To evaluate the in vivo imaging capabilities of the [EuL^1a]⁺ and [YbL^1a]⁺ probes, zebrafish larvae were intravenously injected with 10 nL of 10 mM concentrations of the respective compound and subsequently exposed to 2P-imaging at 720 nm and 800 nm excitation wavelengths, respectively. Imaging was first conducted on both control and [EuL^1a]⁺-injected embryos (Fig. 3). The results demonstrate efficient distribution of the [EuL^1a]⁺ probe within the circulatory system, as evidenced by red fluorescence localized in the heart, aortic arches, dorsal longitudinal anastomotic vessel, caudal vein, caudal artery, and intersegmental vessels (Fig. 3B). In contrast, control embryos exhibited weak red fluorescence restricted to the yolk (Fig. 3A). To determine whether the observed red fluorescence originated from the europium complex or endogenous autofluorescence, spectral imaging was performed. Injected embryos displayed two distinct emission peaks at 580 nm and 620 nm, confirming the presence of [EuL^1a]⁺ (Fig. 3B, far right). Conversely, control embryos exhibited a broad emission peak at 450 nm (Fig. 3A, far right), indicative of autofluorescence. Notably, all injected embryos remained viable four days post-injection.

Fig. 3. 2P-microscopy imaging of 72 hours post-fertilization (hpf) Casper zebrafish embryos with or without intravenous injection of [EuL^1a]⁺. Observation was performed 2 h after injection at 720 nm (n = 5 embryos). Scale bar 500 µm. From left to right: bright-field image, 2P-microscopy image (red color is arbitrary), merge, 2P-spectral microscopy (detected with a PMT array, the colors are real); the white arrows indicate regions where emission spectra were recorded; 2P-excited emission spectra at the white arrow. (A) Control images without injection. (B) Zebrafish injected with [EuL^1a]⁺.

Subsequently, imaging was performed on both control and [YbL^1c]⁺-injected zebrafish larvae following fixation and excitation in the near-infrared (NIR) at 800 nm (Fig. 4). Images were acquired using a custom-built imaging setup with detection in either the visible range (470–610 nm; NIR-to-visible configuration) or the NIR range (850–1050 nm; NIR-to-NIR configuration).^21,28,59,60 In the conventional NIR-to-visible configuration, images of control and [YbL^1c]⁺-injected larvae appeared similar, primarily due to autofluorescence and residual ligand-centered fluorescence (see Fig. 1 for the emission spectrum of [YbL^1c]⁺). In contrast, in the NIR-to-NIR configuration, control images were completely dark, confirming the absence of autofluorescence within this spectral range. Notably, images of [YbL^1c]⁺-injected larvae exhibited a high signal-to-noise ratio, facilitated by the dark background, allowing clear detection of the Yb^III emission signal. As observed previously with [EuL^1a]⁺, the entire vascular system of the embryo was distinctly visualized, underscoring the effectiveness of NIR-to-NIR imaging as a powerful approach that remains unaffected by autofluorescence in biological media.

Fig. 4. 2P-microscopic images of 72 hours post-fertilization (hpf) Casper zebrafish embryos with or without intravenous injection of [YbL^1c]⁺. Observation was performed after embryo fixation. (A) Representative transmission image (bright-field differential interference contrast (DIC) image). (B) Control images without staining. (C) Zebrafish stained with [YbL^1c]⁺. 2P-microscopy images under 800 nm excitation with detection either in the visible (left: 470–610 nm) or in the NIR (right: 850–1050 nm).

Subsequently, imaging was performed on both control and [YbL^1c]⁺-injected zebrafish larvae following fixation and excitation in the near-infrared (NIR) at 800 nm (Fig. 4). Images were acquired using a custom-built imaging setup with detection in either the visible range (470–610 nm; NIR-to-visible configuration) or the NIR range (850–1050 nm; NIR-to-NIR configuration).^21,23,28,59 In the conventional NIR-to-visible configuration, images of control and [YbL^1c]⁺-injected larvae appeared similar, primarily due to autofluorescence and residual ligand-centered fluorescence (see Fig. 1 for the emission spectrum of [YbL^1c]⁺). In contrast, in the NIR-to-NIR configuration, control images were completely dark, confirming the absence of autofluorescence within this spectral range. Notably, images of [YbL^1c]⁺-injected larvae exhibited a high signal-to-noise ratio, facilitated by the dark background, allowing clear detection of the Yb^III emission signal. As observed previously with [EuL^1a]⁺, the entire vascular system of the embryo was distinctly visualized, underscoring the effectiveness of NIR-to-NIR imaging as a powerful approach that remains unaffected by autofluorescence in biological media.

Conclusion

We report the first two-photon-excitable bispidine lanthanide(iii) complexes and demonstrate their suitability as two-photon luminescence imaging probes. These complexes combine high quantum yields, long luminescence lifetimes, and high brightness with pronounced kinetic inertness and structural rigidity. Luminescence data indicate complete encapsulation of the Ln^III ions, with no inner-sphere water coordination, in agreement with structures of the parent L¹ complexes. Their excellent imaging performance is attributed to the rigid bispidine scaffold, the close match between the ligand cavity and the size of the lanthanide(iii) ions, and the resulting suppression of conformational dynamics. Together with the absence of solvent coordination and the comparable structural features across the lanthanide series, these properties provide a robust platform for luminescence imaging in biological environments. In cellulo imaging shows no detectable interaction with cellular membranes or intracellular organelles. Similarly, in vivo experiments in zebrafish larvae reveal no apparent nonspecific tissue interaction. This inert biological behavior, combined with the generally hydrophilic character of the complexes, makes them attractive candidates for targeted bioimaging. Moreover, the established C9-functionalization chemistry of the bispidine scaffold, including previously demonstrated coupling strategies,^60,61 enables conjugation to biological vectors and opens a route toward targeted two-photon lanthanide probes for various imaging applications.

Author contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. PC and OM designed and supervised the project. PAC and MR performed the synthesis; PDE carried out the theoretical calculations; PAC, MR, SRK, LP and OM carried out the luminescence measurements and PAC, MR, LMAA, SB, AG and MGB performed the different bioimaging experiments.

Conflicts of interest

The authors declare no conflict of interest.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information: complete experimental procedures, details on the photophysical measurements and in cellulo as well as in vivo imaging. See DOI: https://doi.org/10.1039/d6sc05072g.