All five heteroleptic complexes 1a–e have an intense blue colour in solution (Figure 2). Their spectroscopic properties were investigated in a nonpolar solvent, such as toluene (PhMe), and in a polar aprotic solvent, such as dimethyl sulfoxide (DMSO), which will be used for the preparation of the biological assays. In order to understand their photophysical properties, their relative homoleptic complexes were also characterised (see Figure 3 and Supplementary Table S1 in ESI). The UV-vis absorbance spectra of complexes 1a–e have shared features, as shown in Figures 2, 3.

From the absorption spectra, we identify three main electronic transition bands in the heteroleptic complexes 1a–e (Table 1). The broad band at high energy centred at ca. 360 nm is attributed to the electronic transitions localised on the naphthyl vinyl moieties of the π-extended dipyrrins, as they are absent in the plain dipyrrins. It is worth to notice that, in complex 1b, the characteristic structured band of the anthracenyl moiety is not visible as it is hidden by the aforementioned naphthyl vinyl absorption. This is not the case for the homoleptic complex 2b, in which spectrum the vibronic structure of the anthracene absorption is clearly visible. The other two main bands are in the visible region, and their profile is reminiscent of the absorption of the dipyrrin ligands. Between these two bands, the one at highest energy presents a shoulder at 465 nm and a relative maximum at ca. 490 nm (e.g.: ε (1e) = 2.2 10⁴ cm⁻¹ M⁻¹).

This absorption is attributed to the singlet ligand centred (¹LC) π→π* transition localised on the plain dipyrrin (¹L_pC in Figure 4). At longer wavelengths, a very intense absorption at ca. 620 nm (e.g.: ε (1e) = 3.3 10⁴ cm⁻¹ M⁻¹) is present with a shoulder at ca. 575 nm, which is assigned to the π→π* transition and its vibronic coupling localised on the π-expanded dipyrrin (¹L_πC in Figure 4). The Zn^II centre is a d¹⁰ metal, and it is not involved in the transitions. Furthermore, it is reasonable to expect that the dipyrrinato ligands are almost orthogonal to each other with a weak if not absent exciton coupling (Telfer et al., 2011; Trinh et al., 2014). Each complex shows fluorescence in the far-red region (emission centred at 635 nm) and a lower intensity shoulder in the near-infrared region up to 800 nm. The Φ values were measured by the relative method in two solvents: dimethyl sulfoxide (E_T ^N: 0.44) and toluene (E_T ^N: 0.099) (Reichardt, 1994). In a nonpolar solvent such as toluene, the zinc bis(dipyrrinato) complexes have the highest Φ, with values ranging from 18% for complex 1c to 70% for complex 1a. The difference in Φ among complexes 1a–e has to be ascribed to the distinct aryl group in meso-position of the plain dipyrrins. It has been previously proven that the aryl group rotates in respect to the plain of the dipyrrin, allowing nonradiative deactivation unless bulky substituents impede this rotation (I.V. Sazanovich, 2004). In addition to that, the planarity of the chelating dipyrrin might change upon functionalisation, which also influences the rigidity and, therefore, the radiative transitions of the systems (Tungulin et al., 2019). The π-extended dipyrrin is the same for the heteroleptic compounds, besides complex 1e, which possess a hydroxyl group in position 4 to the 2,6-dimethylphenyl substituent, which is in para to the dipyrrin. The presence of this extra hydroxyl group in 1e with respect to 1d might induce additional nonradiative processes since Φ in 1e is slightly lower than the one in 1d (Φ = 45% and Φ = 63%, respectively). The consistently lower emission efficiency in 1c is related mainly to the increased rotational freedom of the aryl group in the meso-position of the plain dipyrrin. Furthermore, the electron-donating groups such as methoxy and hydroxyl groups induce an additional quenching effect. The homoleptic derivatives are emissive only in toluene (see Supplementary Table S1), although their Φ is much lower than their heteroleptic counterparts (e.g. Φ (2a): 18%). By comparing the emissions in a polar aprotic solvent such as DMSO, it is possible to assert that the emission energies are not affected by changing the polarity of the medium. In fact, LC transitions are not influenced by different polarities. Heteroleptic zinc bis(dipyrrinato) complexes are not symmetric in their ground and excited states. Therefore, the non-emissive symmetry breaking charge transfer state (SBCT) is not present, which is favoured in the case of homoleptic complexes instead (cf. Figure 4 and Supplementary Figure S5). The intensity of the emission of 1a–e in DMSO, although reduced in comparison to the values obtained in PhMe, is still very strong with Φ of ca. 40%, except for complex 1c (Φ = 5%). These values are incredibly appealing for far-red/near-IR emitters, especially because by lowering the emission energies, the nonradiative deactivation paths are much more probable to occur.

The fluorescence decays are monoexponential, and the lifetimes (τ) are close to 3 ns (in DMSO) and 4 ns (in PhMe), with minor differences among the complexes. Radiative rate constants (k _r) are comparable among the heteroleptic Zn^II complexes and are higher than the nonradiative ones (k _nr) in PhMe and lower in DMSO (except for complex 1c). Excitation spectra of the investigated complexes show a precise comparison with their relative absorption spectra, meaning that the far-red emission centred on the π-extended dipyrrinato moiety also occurs upon excitation of the plain dipyrrinato moiety (Figures 2B, 3B, and Supplementary Figure S1). Thus, the excited state ¹L_pC undergoes a rapid interligand nonradiative transition to populate the lower-lying ¹L_πC (Figure 4). Therefore, upon excitation at shorter wavelength (470 nm), the detected emission is at lower energies (emission maximum at ca. 635 nm). This effect prompts a pseudo-Stokes shift of more than 4,600 cm⁻¹ (Table 1). As the quenching of the plain dipyrrin is total, Nishihara and coworkers suggested a 100% efficient energy transfer from the donor L_p to the acceptor L_π. (Kusaka et al., 2012; Sakamoto et al., 2016). Advanced studies are necessary in order to elucidate the photophysical pathways of these heteroleptic complexes and, currently, we are investigating the involved nonradiative processes by means of transient absorption spectroscopy, which are beyond the scope of the present work.

Further experiments were done by measuring the fluorescence lifetimes of the heteroleptic complexes 1a–e, using three different excitation wavelengths (455, 570, and 625 nm). For each complex, the obtained decays show identical fittings independently from the excitation energy used (Supplementary Figure S1).

The emission colors of these new heteroleptic Zn^II bis(dipyrrinato) complexes, as well as their high quantum yields also in polar solvents and their large pseudo-Stokes Shifts are appealing properties for their exploitation in bioimaging. Before evaluating their biocompatibility, we analysed their stability in different aqueous environments. The UV/vis absorption spectra of the complexes 1a–e were recorded in Dulbecco Modified Eagle Medium (DMEM) and in deionized water (Supplementary Figure S3), using the same concentrations adopted for the confocal laser microscopy, in order to evaluate these the new dyes in media close to the cellular environment. The absorption profiles are stable. The same is true also in aqueous solutions at pH 3.3 and 5.0 (Supplementary Figure S4). These conditions were chosen based on the typical pH gradient in endocytic compartments of cells and DMEM is tipically used as cellular medium for cell culture applications. Emission profiles of the compounds in DMEM overlay very well with those measured in organic solvents, while there is a bathochromic shift in water, where the emission maxima are at about 670 nm (Supplementary Table S1). Quantum yields of the compounds in aqueous media reflect the extremely large polarity (E_T ^Nof H₂O: 1.00), since the values are up to 3.1% in DMEM and up to 1.7% in water. Furthermore, it should be noted that water causes an additional quenching effect due to hydrogen-bond-assisted nonradiative deactivation (Maillard et al., 2021).

The stability of these complexes was tested at increasing temperatures (Supplementary Figure S6. The emission of the complexes is only slightly reduced when going from 20 to 50°C, and this can be ascribed to the increasing collisions with solvent molecules followed by an increase of nonradiative deactivation processes. Thus, our far-red emissive bis(dipyrrinato) zinc complexes are stable in an aqueous solution at different pH values and temperatures. In order to test their biocompatibility in living cells, cell viability and cellular uptake were determined in four different cell types, including primary somatic cells such as human dermal fibroblasts (NHDF), a mouse cell line from embryonic fibroblasts (NIH3T3), and two human cancer cell lines (HeLa, and MCF7). To test the viability, MTT assays were performed by treating 10⁴ cells of the respective cell type with different concentrations of the complexes 1a–e for 72 h at 37°C. For all complexes, the LD₅₀ values were >20 μM, showing high biocompatibility (Supplementary Figure S7). On the contrary, MTT assays of the single dipyrrin ligand showed a decreased viability already at concentrations lower than 7 µM (Supplementary Figure S8). For all these results, we assume high stability of these heteroleptic complexes in the cellular environments.

Since all the complexes showed only negligible toxicity when used to treat the different cell lines, a concentration of 20 µM was chosen for the cellular uptake experiments and the live-cell fluorescent imaging (Figure 5, 6 and Supplementary Figures S11–17). With the best performing complexes 1d and 1e we tested their cellular uptake also at different concentrations, such as 1, 10, and 20 µM (Supplementary Figures S9, S10) It was assumed that due to their hydroxyl groups, a higher water solubility was achieved with complexes 1c, 1d, and 1e. As expected, due to the low Φ of 1c, this complex is hardly detectable. In contrast, complexes 1d and 1e display improved cellular uptake with respect to DMSO controls by virtue of their higher solubility in aqueous solution. Complexes 1a and 1b showed decreased cellular uptake. All complexes were taken up by endocytosis at the respective concentrations, leading to an accumulation in the endosomal/lysosomal compartment, which was proven by the counterstaining with Lysotracker Green^TM (Figure 5) (Canton and Battaglia, 2012; Kolmel et al., 2012).

As a further analysis, confocal live-cell fluorescence microscopy was performed with Mitotracker™ Green. However, no counterstaining with mitochondrial markers was observed (Supplementary Figures S12–14). Figures 5A–C shows the counterstaining experiments in HeLa cells with Lysotracker Green with Pearson coefficients of 0.85 for 1e, 0.60 for 1d, and 0.24 for 1a (Bolte and Cordelières, 2006). To further test the suitability of these heteroleptic Zn^II bis(dipyrrinato) complexes for bioimaging applications, the correlation of incubation time with signal intensity in live-cell imaging and the stability after fixation of cells were also investigated. While differences in the obtained signal intensity could be detected when incubating cells for a period of 0.5, 1.5, and 6 h (Supplementary Figure S17, compound 1e), the complexes showed no decrease in fluorescence intensity after fixation (Figure 6). This proves the versatility of these complexes, as they can be used in live-cell imaging and in fixed-cell experiments, e.g. for immunocytochemistry. Moreover, given the photochemical properties of the compounds and the significant pseudo-Stokes shift, the excitation of complexes was possible at multiple wavelengths, allowing for the simultaneous excitation of two fluorophores at 488 nm and the detection of their emission at different wavelengths.