Elizabeth J.
New
,
Aileen
Congreve
and
David
Parker
*
Department of Chemistry, Durham University, South Road, Durham, UK DH1 3LE. E-mail: david.parker@dur.ac.uk
First published on 12th May 2010
A series of experiments has been undertaken in order to gain a greater understanding of the cellular uptake and localisation behaviour of emissive lanthanide complexes as cellular stains or probes. Out of a large number of structurally related complexes characterised recently, a set of seven representative examples has been examined in detail, containing either tetraazatriphenylene or azaxanthone-based sensitising chromophores. Intracellular localisation profiles and cellular uptake and egress behaviour have been studied by microscopy and flow cytometry. Typically, the maximum intracellular concentration was of the order of 0.4 mM, or about 109 complexes per cell. The complexes studied were generally not toxic and did not perturb the mitochondrial membrane potential. A common uptake mechanism of macropinocytosis has been identified. A generalisation of trends in behaviour, and structure–activity relationships is presented, and the implications for future probe design discussed.
A common theme of the study of luminescent lanthanide complexes has been the design of systems which can report on their environment. To this end, systems have been reported which are sensitive to pH,9 endogenous anions such as bicarbonate,10 citrate11 and urate,12 and metal ions such as zinc,13 copper(I)14 and potassium.15 In addition, lanthanide complexes have been developed as potential probes for sub-cellular events, including redox state,16 peptide phosphorylation17 and singlet oxygen.18 Most valuable amongst such complexes are systems which can act as ratiometric probes, and can therefore provide information about the target analyte independent of probe concentration and the local environment. In general, studies of these complexes have been performed in vitro, and have not demonstrated in cellulo applicability.
In order for a sensitised lanthanide complex to have use as a cellular probe, it must exhibit certain biological properties. Notably, the complex must readily cross the cell membrane, and localise in a region of interest within the cell. For example, a pH sensor might be best localised in the lysosomes, in which acidity can signify endosome age or health.19 Similarly, a probe for phosphorylation might be best directed to the mitochondria, where numerous phosphorylation events take place. In addition, a complex that is intended to act as a probe of cellular function should only minimally perturb the cellular homeostasis, and in particular should not induce cell proliferation or lead to death at working concentrations.
The importance of these biological properties has led to recent efforts to characterise the sub-cellular behaviour of a large number of sensitised lanthanide complexes.20 This led to the identification of a number of important structural effects; the variation of sub-cellular localisation with the nature of the sensitising chromophore;21 the conservation of localisation following changes to either the ligand pendant arms, complex charge or helicity;22a the alteration of protein binding and quenching sensitivity that occur following minor structural changes.22 In addition, most complexes were reported to be non-toxic at dosing concentrations, with only isolated cytotoxic complexes.21a,23 Every complex investigated exhibited noticeable cell uptake, and preliminary studies of a set of sensitised lanthanide complexes were consistent with an uptake mechanism of macropinocytosis.24
Here, we describe experiments performed in order to gain a greater understanding of the cellular behaviour of luminescent lanthanide complexes. Seven representative ligand sets are examined, containing tetraazatriphenylene and azaxanthone-based sensitising chromophores. A generalisation of trends in behaviour, and structure–activity relationships is presented, and the implications for future probe design discussed.
Fig. 1 The four distinct sub-cellular localisation profiles of luminescent lanthanide complexes—lysosomal, mitochondrial, nucleolar and mitochondrial-lysosomal—with representative examples of complexes exhibiting each behaviour. |
Fig. 2 Fluorescence microscopy images of CHO cells treated with various europium complexes (50 μM, 4 h) and commercially-available fluorescent stains (30 min) showing the lysosomal localisation of [Eu.L1]Cl3, the mitochondrial localisation of [Eu.L3]Cl3, the mitochondrial-lysosomal co-localisation of [Eu.L5]Cl3 and the nucleolar distribution of [Eu.L7]Cl3. (A) Europium luminescence, (B) stain fluorescence and (C) image overlay. Scale bars represent 20 μm; image brightness and contrast have been adjusted. |
In contrast to these two, well-defined classes of complexes, a third distinct group of complexes has been identified, in which luminescence is observed simultaneously in both the lysosomes and the mitochondria. In such cases, bright spots consistent with lysosomal staining can be observed, as well as the more diffuse pattern characteristic of mitochondrial localisation. The sub-cellular distribution was confirmed by co-staining with both LysoTracker and MitoTracker probes (Fig. 2). Such behaviour is demonstrated by complexes [Tb.L4]3+ and [Eu.L5]3+.
The final class of complexes is those which localised to the nucleoli, the sub-nuclear, protein-rich structure which is primarily involved in ribosome assembly.25 This distribution was observed at normal dosing concentrations for a number of complexes, including [Ln.L7]3+. In addition, nucleolar distribution at high incubation concentrations was observed for a number of complexes which, at lower concentrations, localised to the lysosomes.26 Previous work has demonstrated that this nucleolar localisation can be attributed to an enhanced cellular membrane permeability, induced by the presence of the probe itself.24
There are a number of ways in which the chromophore can be altered in the synthesis of luminescent lanthanide complexes. The nature of the chromophore appears to have an effect on the localisation of the complex. This is evident in the varying distributions of [Tb.L1]3+ and [Tb.L4]3+ which differ only in the sensitising chromophore. Small modifications to the chromophore structure, on the other hand, do not appear to influence the complex localisation.21a,22b A marked change in localisation has been observed on alteration of the linkage between the chromophore and the cyclen ring. This has previously been reported for the mitochondrially-localising [Ln.L3]3+ and its lysosomally-localising analogue in which the amide linkage is replaced by a methylene group.21a This is also evident in the case of the analogue [Eu.L6]3+, containing a pyridine linkage, which localises to the lysosomes.
In considering the effect of the pendant arms on the localisation of complexes, it is important to consider both the nature and number of pendant arms. Complexes [Ln.L1]3+ and [Ln.L2a]3+, which form part of a larger group of complexes previously discussed,22a differ only in the identity of the amide-based pendant arms, and exhibit identical lysosomal localisation. In contrast, the number of pendant arms, or the ligand denticity, does appear to influence the cellular distribution of the complex. While [Eu.L5]3+ is observed in both lysosomes and mitochondria, removal of one pendant arm gives rise to [Eu.L6]3+, which exhibits a purely lysosomal distribution. This behaviour has also been observed for the analogue of [Tb.L4]3+ bearing only two pendant arms.
In every case, Tb and Eu complexes of a common ligand exhibited identical sub-cellular localisation patterns, indicating that the nature of the lanthanide ion does not affect distribution. There is no evidence to suggest that complex charge has an effect on sub-cellular localisation: [Ln.L2a]3+ and its neutral analogue [Ln.L2b], for example, exhibit lysosomal distributions. Previous studies of complexes with differing geometries and helicities have not shown any effect on localisation profiles.22a The counterion which was most commonly employed for cationic complexes is chloride. Cellular studies were also performed in which acetate, nitrate and triflate forms of the complexes were administered to cells. These did not reveal any significant change in the sub-cellular localisation.
In summary, these investigations have revealed that the localisation of a complex is affected by the nature of the chromophore and, more significantly, its point of attachment to the macrocycle, as well as by the ligand denticity. It is therefore important to consider whether it is possible to predict the sub-cellular localisation of a novel complex. The complexes which have been investigated comprise a library of chromophores and pendant arms, which is not exhaustive. The information which has been gained from this structural study allows the prediction of the localisation of a complex composed of elements from this library. For entirely new complexes, however, which contain different chromophores or pendant arms, such prediction is less straightforward. In addition, if alternative sub-cellular localisations were desired, it is likely that more drastic structural changes would need to be made, or a targeting moiety incorporated into the complex. While the nature of the pendant arms does not appear to have any effect on the sub-cellular distribution, complexes bearing three identical pendant arms, such as [Tb.L4]3+ and [Eu.L5]3+, appear to exhibit less well-defined localisation profiles than their analogues in which one pendant arm has been removed, leaving a free cyclen NH.
Fig. 3 Uptake and egress of [Tb.L1]Cl3 (50 μM) in NIH 3T3 cells over time, showing standard deviations. (a) Shows uptake of complex over a 24 h period; (b) shows the egress of complex following 4 h incubation over a 4 h period. |
In control experiments examining the protein concentration of sets of counted cells, it was estimated that one hundred micrograms of protein corresponded to 250000 cells. Therefore if we consider an intracellular lanthanide ion concentration of 1 nmol mg−1 protein, this corresponds to 4 × 10−16 moles/cell, or 2.4 × 108 lanthanide complexes per cell. Assuming an average cell volume of 5000 cubic microns (5 × 10−9 cc), then a concentration of complex within the cell of 1 nmol mg−1 protein is of the order of 80 μM. Similar values have been estimated previously.26
The uptake profile (Fig. 3a) indicates that the intracellular concentration increases very rapidly over the first four hours, with only a slight increase after this time-point. An incubation time of 4 h was therefore selected for all subsequent uptake experiments. The egress (Fig. 3b) profile was established by incubating the cells with 50 μM complex for 4 h, after which the medium was removed and replaced by buffered saline solution. Egress was relatively rapid over the first 20 min, but by one hour appeared to reach an equilibrium value and had decreased very little at 4 h. It may be noted that the maximum intracellular concentration defined here is of the order of 0.4 mM. This lies comfortably within the target range for labelling cells with Gd complexes, for cell tracking MRI applications.
Fig. 4 Schematic diagram illustrating the three endocytotic pathways in addition to the intracellular maturation of endosomes to lysosomes. Inhibitors (−) and activators (+) of each pathway are indicated. |
The measured intracellular concentrations, relative to the control cells treated with complex alone, demonstrate behaviour that is consistent with uptake by macropinocytosis. Both wortmannin and amiloride resulted in statistically significant decreased uptake of all complexes (p < 0.01 and < 0.05 are indicated, Fig. 5), while the presence of the phorbol ester and fatty acid glycerol demonstrably enhanced uptake. The other drug treatments did not result in significant variations from the control. This finding is consistent with the observation that at short incubation times (5–20 min) these luminescent lanthanide complexes co-localise with FITC–dextran, a fluorescent marker of the macropinosomes.24
Fig. 5 Relative changes in uptake of [Eu.L1]Cl3 (blue), [Tb.L2b] (red), [Eu.L3]Cl3 (green), [Tb.L4]Cl3 (yellow), [Eu.L5]Cl3 (purple) and [Eu.L7]Cl3 (pink) in the presence of various inhibitors and activators. Statistically significant differences from controls (* p < 0.05, ** p < 0.01) calculated using a two-tailed, paired Student's t-test. |
Fig. 6 Intracellular lanthanide concentrations as a function of dosing concentrations for CHO cells incubated for 4 h with (a) [Tb.L4]Cl3 and (b) [Eu.L5]Cl3. |
Log Poct | SD | IC50/μM | SD | |
---|---|---|---|---|
a n.d. = not determined; the IC50 values obtained using the parallel WST-1 method (see ESI) for [Eu.L3]Cl3, [Tb.L4]Cl3 and [Eu.L7](OTf)3 were >200, 40 [± 6] and 59 [± 10] μM respectively, comparable to the values obtained by the MTT assay. | ||||
[Eu.L1]Cl3 | −1.34 | 0.10 | 109 | 20 |
[Tb.L1]Cl3 | n.d. | — | 131 | 27 |
[Eu.L2a]Cl3 | −2.10 | 0.40 | 144 | 12 |
[TbL2a]Cl3 | −2.06 | 0.09 | n.d. | — |
[Eu.L2b] | −1.74 | 0.37 | >240 | — |
[TbL2b] | −1.56 | 0.25 | n.d. | — |
[Eu.L3]Cl3 | −1.56 | 0.09 | 164 | 31 |
[Tb.L4]Cl3 | 0.05 | 0.04 | 40.9 | 8.9 |
[Eu.L5]Cl3 | −0.06 | 0.03 | 72.9 | 4.1 |
[Eu.L7](OTf)3 | 1.06 | 0.13 | 79.0 | 9.4 |
[Eu.L7]Cl3 | 0.49 | 0.11 | 120 | 10 |
The possibility of a passive uptake mechanism may also be addressed by examining the lipophilicities of a range of complexes. Values of log P were obtained by allowing the complex to partition between 1-octanol and water, and calculating the concentration in each layer by measuring the lanthanide emission relative to a calibration curve. The measured partition coefficients range from hydrophilic (−2.10) to lipophilic (+1.06). The nature of the lanthanide ion does not appear to affect the log P value, but variation of the counterion does have a significant effect, with the triflate form of [Eu.L7]3+ being more lipophilic than the chloride form. This is consistent with the observed lower aqueous solubility of triflate salts.
While the complexes vary greatly in partitioning behaviour, microscopy studies revealed a similar rate of uptake for each complex, with luminescence readily observed 5 min after dosing. This confirms that cell uptake is most unlikely to occur by passive diffusion as in such a case diffusion rates would be likely to mirror trends in log P.
Analysis of the resulting IC50 values (Table 1) highlights a number of trends. The replacement of Eu3+ for Tb3+ does not greatly alter toxicity, exemplified by the study of two complexes of L1. In general, it appears that neutral complexes are less cytotoxic than their cationic analogues; [Eu.L2a]Cl3, for example, exhibits an IC50 value of 144 μM, while its neutral analogue, [Eu.L2b] was non-toxic over the range studied. The nature of the sensitising chromophore also affects the cytotoxicity; while the tricationic complexes, [Eu.L1]Cl3 and [Eu.L2a]Cl3, containing a tetraazatriphenylene chromophore exhibit IC50 values of greater than 100 μM, [Eu.L5]Cl3, which contains a pyridine-linked azaxanthone sensitiser, is more toxic. Notably, it was observed that complexes containing an appended tBu group, such as [Tb.L4]Cl3 and [Eu.L5]Cl3, exhibit higher cytotoxicity. The difference in cytotoxicity between the triflate and chloride salts of [Eu.L7]3+ also indicates that the nature of the counterion may play an important role. This result, coupled to the lower solubility of the triflate salts, indicates that triflates are not the preferred salts for administration of complexes.
The cytotoxicity results indicate that complexes are generally non-toxic under the conditions which are most applicable to their use as cellular probes; incubation times of 1 to 4 h, and dosing concentrations of up to 50 μM. Previous studies have reported isolated examples of complexes which exhibited higher toxicities.21a, 23 In each case, this high cytotoxicity could be explained by demonstrable mechanisms, such as dissociation of the complex to toxic metabolites and disruption of cell integrity by insertion into the cell membrane. Such cases, however, appear to be the exception rather than the rule and can be easily avoided.
Fig. 7 (a) FL4/FL1 mean intensity from flow cytometric analysis of CHO cells with various treatments. Carbonyl cyanide p-trifluoromethoxylphenylhydrazone (FCCP) treatments are 20 μM (black) and 100 μM (grey) for 3 h; [Eu.L3]Cl3 treatment is 50 μM for 1 h (black) and 4 h (grey); [Tb.L4]Cl3 and [Eu.L5]Cl3 treatments are 50 μM for 4 h (black) and 24 h (grey). (b) and (c) flow cytometry ‘dot’ plots following staining with propidium iodide and annexin V–FITC of CHO cells treated for 24 h with (b) 200 μM [Tb.L4]Cl3 and (c) 200 μM [Eu.L5]Cl3. |
In order to further understand these effects, an apoptosis–necrosis assay was performed, in which propidium iodide (PI) was employed as a marker for necrosis, and an annexin V–FITC conjugate used to signal apoptosis.39 CHO cells were treated with 200 μM complex for 24 h, above the IC50 value of each complex. The cells were then harvested, treated with PI and annexin V–FITC, and analysed by flow cytometry, measuring red (FL4; PI) against green (FL1; FITC) fluorescence (Fig. 7b and c). The flow cytometry plots indicate that a population of cells treated with [Tb.L4]Cl3 are stained with PI but do not exhibit binding to annexin V, indicative of necrosis. In contrast, cells treated with [Eu.L5]Cl3 show both red and green fluorescence, suggesting cell death by apoptosis. These modes of cell death are consistent with the observed mitochondrial membrane changes. Mitochondrial membrane damage is an effect rather than a cause of necrosis, and as a result, the MMP should decrease over time, as observed for [Tb.L4]Cl3. Apoptosis, on the other hand, is driven by disruptions to the mitochondria, and a more rapid change in MMP for [Eu.L5]Cl3 is therefore observed. Many of these cells with dramatically perturbed MMP at 4 h would have completed apoptosis by 24 h and would be too fragmented to be measured by flow cytometry, resulting in a higher average MMP. This investigation provide further insight into the low IC50 values of [Tb.L4]Cl3 and [Eu.L5]Cl3, suggesting that their toxicity is related to a perturbation of the mitochondrial membrane potential. The results add further weight to the argument that complexes which localise in both mitochondria and lysosomes are not ideal candidates for use as cellular probes.
Sensitised lanthanide complexes appear to exhibit one of a limited number of sub-cellular distribution patterns, principally localising in the lysosomes or mitochondria. In addition, nucleolar localisation can be promoted in cases of enhanced membrane permeability. While these membrane effects may perturb the homeostasis of the cell, this technique may prove to be a valuable method by which to probe this protein-rich region of the cell.
The results presented here indicate that if lysosomal or mitochondrial localisation is required, it is not necessary to tag complexes to direct their delivery. Cells can be labelled with lanthanide complexes in concentrations of the order of 0.05 to 0.5 mM. Such levels are not only sufficient for optical signalling using one- or two-photon excitation, but also are in the target range for the cell loading levels required for tracking cell populations by MRI, assisted by Gd contrast agents.
Clear structural requirements for mitochondrial localisation have also been defined, while a lysosomal localisation appears to be the default distribution. In addition, structural features have been identified which evade the localisation of the complex in both mitochondria and lysosomes. If other localisations are desired, however, it may be necessary to investigate more drastic structural modifications, or to conjugate the complex to a moiety with a controlling influence on uptake and sub-cellular trafficking.
Footnote |
† Electronic supplementary information (ESI) available: Experimental section. See DOI: 10.1039/c0sc00105h |
This journal is © The Royal Society of Chemistry 2010 |