Non-covalent complex of quantum dots and chlorin e6: efficient energy transfer and remarkable stability in living cells revealed by FLIM

A Forster resonance energy transfer (FRET) system of semiconductor quantum dots and porphyrins represents a new promising photosensitizing tool for the photodynamic therapy of cancer. In this work, we demonstrate the ability of a non-covalent complex formed between commercial lipid-coated CdSe/ZnS quantum dots (QD) bearing different terminal groups (carboxyl, amine or non-functionalized) and a second-generation photosensitizer, chlorin e6 (Ce6) to enter living HeLa cells with maintained integrity and perform FRET from two-photon excited QD to bound Ce6 molecules. Spectroscopic changes, the highly efficient FRET, observed upon Ce6 binding to QD, and remarkable stability of the QD–Ce6 complex in different media suggest that Ce6 penetrates inside the lipid coating close to the inorganic core of QD. Two-photon fluorescence lifetime imaging microscopy (FLIM) on living HeLa cells revealed that QD–Ce6 complexes localize within the plasma membrane and intracellular compartments and preserve high FRET efficiency (∼50%). The latter was confirmed by recovery of QD emission lifetime after photobleaching of Ce6. The intracellular distribution pattern and FRET efficiency of QD–Ce6 complexes did not depend on the charge of QD terminal groups. Given the non-covalent nature of the complex, its exceptional stability in cellulo can be explained by a combination of hydrophobic interactions and coordination of carboxyl groups of Ce6 with the ZnS shell of QD. These findings suggest a simple route to the preparation of QD-photosensitizer complexes featuring efficient FRET and high stability in cellulo without using time-consuming conjugation protocols.


Introduction
2][3][4][5][6][7][8] In 2003 Samia et al suggested to exploit QD as resonance energy donors for classical photosensitizers (PS) used in photodynamic therapy (PDT) of cancer. 1 PDT is a treatment that uses a photosensitizing drug, usually porphyrin-type molecules, and the light to cure the cancer. 9Once the light is applied, the excited molecules of PS generate reactive oxygen species (ROS) that subsequently damage cancer cells.QD are particularly well suited as energy donors for PS due to their size-tunable emission spectrum, high emission quantum yield and long lifetime.Additionally, high extinction coefficient (10 5 -10 6 M -1 cm -1 ), broad absorption spectrum and minimal photobleaching enable efficient and prolonged excitation of QD.7][18][19] While significant number of studies on different non-covalent QD-PS systems has been reported to date, the majority of them have focused on assemblies in solutions, based either on electrostatic, 1,13,[20][21][22] or coordinational, 23,24 interactions.Despite the efficient FRET these complexes tend to aggregate, [20][21][22] or may lose their noncovalently bound PS.Furthermore, stability of such QD-PS complexes in cellular context is questionable and needs to be examined.Covalently coupled QD-PS systems, 14,17,25 meet the stability requirements in this respect, however the efficient FRET is hard to achieve, because PS molecules are grafted at the interface between water and the QD coating, which is relatively far from the QD core.][28] In this work, we prepared complexes of commercial CdSe/ZnS QD bearing a lipid-based coating with different terminal groups (carboxyl, amine and non-functionalized) with chlorin e 6 (Ce 6 ), a well-known second-generation photosensitizer having a high quantum yield of singlet oxygen production (Scheme 1). 29We obtained exceptionally high FRET efficiency of these complexes, suggesting that Ce 6 is firmly imbedded inside QD lipid coating close to the inorganic core.Most importantly, according to the fluorescence lifetime imaging (FLIM) with two-photon excitation, these QD-Ce 6 complexes readily entered living HeLa cells with maintained efficient FRET, which shows their remarkable stability in the intracellular media.

Aqueous solutions
All solutions were prepared in phosphate buffer of pH 7. A stock solution of 1 mM Ce 6 was freshly prepared and further diluted just before the experiments.Working solutions of 0.02 µM QD were prepared by diluting the stock solution of QD 24 hours before the experiments.
QD-Ce 6 solutions were prepared by titrating 2 µl of Ce 6 solution of appropriate concentration into 2 ml of QD solution.In these mixed QD-Ce 6 solutions, the concentration of QD was 0.02 µM, while Ce 6 concentration varied from 0.002 µM to 0.2 µM (QD:Ce 6 molar ratios from 1:0.1 to 1:10 were obtained).To allow the binding process to reach its equilibrium, the spectra of QD-Ce 6 solutions were measured 20 minutes after QD and Ce 6 were mixed together.

Characteristics of FRET
Changes in spectral properties of QD and Ce 6 upon QD-Ce 6 complex formation in aqueous solution consisted well with the features of non-radiative dipole-dipole energy transfer mechanism and were evaluated using FRET formalism. 30he efficiency of energy transfer (E) was calculated from changes in fluorescence of QD (donor) as follows, where F D and Fˈ D are the intensities of QD fluorescence in the absence and presence of Ce 6 (acceptor), respectively.<τ D > and <τˈ D > are the amplitude-weighted average lifetimes of QD fluorescence in the absence and presence of Ce 6 , respectively.The quantum yields (QY) of Ce 6 and QD fluorescence were calculated by comparison with Rhodamine B in water (QY R =31 % at λ ex = 514 nm, 31 ) (Table 1 and 2

, respectively).
Due to a very low absorbance of Ce 6 at used excitation wavelength (λ ex =465 nm) for FRET measurements, an increase in efficiency of Ce 6 fluorescence in the presence of QD was evaluated using not the QY, but the ratio F´A/F A , where F A and F´A are the integrated fluorescence intensities of Ce 6 in the absence and presence of QD (donor), respectively.In this case, the change in refractive index of Ce 6 surrounding was not reckoned in.

Spectroscopic measurements of solutions
Absorption measurements were carried out with Cary 50 spectrophotometer (Varian Inc, USA).The absorption spectra of samples were smoothed using Savitsky-Goaly filter smoothing method.
Fluorescence measurements were performed on Cary Eclipse spectrophotometer (Varian Inc., USA).For the FRET measurements within QD-Ce 6 complex, the excitation at 465 nm was used because only QD could be excited at this wavelength, while the absorption of Ce 6 is minimal (Fig. 1A, dotted arrow).Fluorescence decay was measured with F920 spectrometer (Edinburgh Instruments, UK), equipped with a single photon photomultiplier detector (S900-R).The excitation source was a picosecond pulsed diode laser (EPL-405) with a radiation wavelength at 405 nm and pulse width of 66.9 ps.
Quartz cuvettes with the optical path length of 1 cm were used for absorption and fluorescence measurements.

Fluorescence lifetime imaging microscopy in living cells
FLIM experiments on HeLa cells were performed by using a home-built two-photon laser scanning setup based on an Olympus IX70 inverted microscope with an Olympus 60x1.2NAwater immersion objective.Two-photon excitation was provided by a titanium-sapphire laser (Tsunami, Spectra Physics) that operated at 830 nm with 5 mW excitation power.Detection system consisted of Avalanche Photodiodes (APD SPCM-AQR-14-FC, Perkin-Elmer) connected to a counter/timer PCI board (PCI6602, National Instrument).The filter of 605 nm with bandwidth of 30 nm was used to exclude the fluorescence of Ce 6 .Irradiation of the samples was performed by a blue light (bandpass filter 420/50 nm) for 30 s.

Steady-state spectral characteristics in aqueous solution
Addition of Ce 6 produced significant changes in absorption and fluorescence spectra of buffered aqueous solutions of carboxyl, amine and non-functionalized QD (Fig. S1 and Fig. S2, respectively).These changes were quite similar for all three different QD.The representative absorption, fluorescence excitation and fluorescence spectra of carboxyl QD solution mixed with different Ce 6 amounts are shown in Fig. 1 A, B and  C, respectively.The absorption spectra of QD-Ce 6 solutions did not show simple superposition of corresponding free QD and Ce 6 spectra (Fig. 1A).The most pronounced difference was seen for Ce 6 Q (I) absorption band, which in the presence of QD shifted from 655 nm to 662 nm.The absorbance of this redshifted band was higher than that of free Ce 6 .Furthermore, in QD-Ce 6 fluorescence spectrum, besides QD emission at 625 nm, the fluorescence band at 670 nm appeared (Fig. 1C) which could be assigned to Ce 6 molecules bound to QD.The successive titration with Ce 6 resulted in the fluorescence intensity decrease of all three types of QD emission at 625 nm and simultaneous increase in fluorescence intensity at 670 nm (Fig. 1C and Fig. S2 B, D, G), which indicates the energy transfer from excited QD to bound Ce 6 molecules.Quite similar absorption and fluorescence characteristics of Ce 6 obtained upon binding to differently charged QD exclude the electrostatic interaction with QD lipid surface as a driving force for the QD-Ce 6 complex formation.Moreover, from the redshift of Ce 6 Q (I) absorption and fluorescence bands we can state that Ce 6 molecules within QD coating are situated in the hydrophobic microenvironment, most likely, hydrophobic part of QD lipids.This was confirmed by the absorption and fluorescence measurements of Ce 6 in the presence of 5% Triton-X 100, that is a well-known nonionic surfactant forming micelles above 0.02% (critical micelle concentration).Addition of Triton-X 100 to aqueous solution of Ce 6 produced precisely the same red-shift of its absorption and fluorescence bands as in the case of QD (Fig. S3 A and B, respectively and Table 1).The same bathochromic shift of Ce 6 fluorescence maximum to 670 nm was also reported for Ce 6 in the presence of lipid bilayers. 32,33 ing excitation at 400 nm, where both bound to QD and free Ce 6 can be efficiently excited, the red-shifted emission at 670 nm was observed for range of Ce 6 /QD ratios 0.5-5, without signs of unbound Ce 6 at 660 nm (Fig. S4).Therefore, in these conditions the binding of Ce 6 to QD is probably complete, which is in agreement with our earlier studies. 34At Ce 6 /QD ratios ≥10 a contribution of the unbound Ce 6 species at 660 nm could be detected probably due to the saturation of the QD binding sites.
Remarkably, the fluorescence intensity of Ce 6 in complex with QD was ~100-fold larger than that of free Ce 6 in buffer directly excited at the same wavelength (465 nm) (Table 1).Changes in the environment of Ce 6 from buffer to QD lipid coating cannot explain this increase, as could be seen from minor variation in QY of Ce 6 from buffer to 5% Triton-X 100 (Table 1) or lipid membranes. 32Therefore, the observed drastic fluorescence enhancement clearly points to FRET from QD, which function as an efficient energy antenna.Indeed, at 465 nm excitation wavelength, the extinction coefficient of QD is >100-fold higher than that of Ce 6 , and thus the efficient fluorescence of Ce 6 originates from the energy transferred from QD.
Moreover, the fluorescence excitation spectra of mixed QD-Ce 6 aqueous solutions registered at 670 nm displayed the contribution of both QD and Ce 6 spectra, but the intensity of QD-Ce 6 solutions was much higher than the sum of the fluorescence intensities of separate components at corresponding concentrations (Fig. 1B and Fig. S2 A, C, E), which confirmed that QD significantly contribute to the fluorescence of bound Ce 6 molecules via energy transfer.
The quenching of QD emission intensity by increasing concentration of Ce 6 was slightly faster for amine and nonfunctionalized QD than for carboxyl QD (Fig. 2A).Furthermore, none of QD intensity decrease reached a plateau even at highest used Ce 6 concentrations.In contrast, the intensity of bound Ce 6 fluorescence band at 670 nm reached its maximum around Ce 6 /QD =5-10 for three studied QD (Fig. 2B).Further increase in Ce 6 concentration resulted in a decrease in this band intensity (data not shown).The latter effect could be explained by the self-quenching of bound Ce 6 fluorescence due to its high density on the surface of QD.Thus, we consider that 5 Ce 6 molecules per QD is an optimal number to obtain QD-Ce 6 complexes with the highest energy transfer efficiency but without the negative self-quenching effect.The FRET efficiency calculated from the decrease in intensity of QD emission at Ce 6 /QD = 1 and 5 are given in Table 2.The stability of QD-Ce 6 complex over time was studied in aqueous medium of different pH, in phosphate buffer saline (PBS) and in the presence of bovine serum albumin (BSA) (QD:Ce 6 :BSA 1:5:200) (Fig. 3).In acidic medium (pH 4-6), which should mimic the endosomal/lysosomal compartments of the cells, QD(carb)-Ce 6 complex was extremely stable as the FRET efficiency of the QD(carb)-Ce 6 complex did no change for 24 hours.In phosphate buffer pH 7.0 and PBS, only a slight decrease in the FRET efficiency was observed (Fig. 3), so that after 24 hours, it retained 85% and 90% of its initial value in PB and PBS, respectively.Addition of BSA resulted in a partial release of Ce 6 from the QD(carb)-Ce 6 complex, which reduced its initial efficiency of energy transfer by 13 %.The release process continued slowly and after 24h, the FRET efficiency of ~30% was still preserved.Thus, addition of BSA produces a burst release of weakly bound Ce 6 molecules, while a significant fraction of the photosensitizer remains strongly bound to QD and thus exhibits slow release kinetics.

Fluorescence decay and FRET in QD-Ce 6 complexes
Fig. 4 shows the fluorescence decay profiles of carboxyl QD solution with increasing concentrations of Ce 6 .They were satisfactorily fitted to a three-exponential decay time model (0.98 ≤ χ² ≤ 1.16) and the obtained average lifetimes of QD decay are summarized in Table 2.The fluorescence decay profile and consequently the average lifetime of QD with different terminal groups varied only slightly: from <τ>=16.4 ns for carboxyl QD to <τ>=14.2ns for amine QD (Table 2 and Fig. S5).The increase in concentration of Ce 6 substantially shortened the fluorescence decay time of QD (Fig. 4, Fig. S5, Table 2), indicating that efficient FRET occurs.The efficiencies of FRET within QD-Ce 6 complexes calculated from the fluorescence decay lifetimes were slightly lower than those obtained from the intensity measurements (Table 2), suggesting that besides FRET some static quenching by bound Ce 6 may exist contributing to the emission intensity decrease of QD without affecting their lifetime.For this reason, we have used FRET efficiency calculated from time-resolved data to estimate the distance between QD and bound Ce 6 molecules (Table 2).
Interestingly, while FRET efficiency values at 1:1 QD:Ce 6 ratio for studied QD slightly varied, this difference disappeared at higher Ce 6 concentration (QD:Ce 6 1:5) and reached about 60 % for all three types of QD (Table 2).For comparison, in the case of covalent QD-Ce 6 conjugate where 26 Ce 6 molecules were covalently attached to peptide-coated QD only 50% FRET efficiency was achieved. 14he Forster radius (R0) and center-to-center (r) distance between QD and bound Ce 6 molecules estimated from the FRET efficiency at 1:1 QD:Ce 6 molar ratio are given in Table 2.In these calculations the value of 2⁄3 for dipole orientation factor (κ2) was used assuming that Ce 6 molecules were orientated randomly upon binding to QD.The QD-Ce6 distance ranged from 3.8 nm for non-functionalized QD-Ce 6 to 4.7 nm for carboxyl QD-Ce 6 pair (Table 2).According to literature, the diameter of QD (eFluor 625 NC) without organic coating is ∼ 7.1 nm, 35 while the thickness of PEG-lipid layer from QD hydrodynamic diameter measurements could range from 5 to 9 nm. 35,36 he center-to-center distance of 3.8-4.7 nm, estimated from FRET data, is close to the shortest possible QD-Ce 6 distance that includes ~3.6 nm of QD radius and ~0.5 nm of Ce 6 radius.Thus, we can validate that the amphiphilic Ce 6 molecules were able to penetrate inside PEG-lipid coating and localize in close proximity to the inorganic core of QD for efficient FRET to occur.We have also examined time-resolved decay of Ce 6 molecules bound to QD (Fig. S6).In the absence of donors, the decay of directly excited Ce 6 was single-exponential with the lifetime of 4.3 ns and 5.1 ns in buffer and 5% Triton-X 100, respectively.The decay of bound Ce 6 excited via energy transfer from QD became significantly longer and could not anymore be fitted by a monoexponential function (Fig. S6).Within a FRET couple, the apparent decay time of the acceptor Ce 6 should contain the decay time of the donor QD, thus explaining the observed phenomenon.Similar elongation of acceptor lifetime was described by Maliwal et al.where long-lifetime lanthanidebased luminophore (donor) resulted in a long-lived component in the covalently linked acceptor decay, which alone displayed a short lifetime.observed at the plasma membranes and inside the cells for all three types of QD, whereas the control cells without QD showed no sign of fluorescence at the same experimental conditions (data not shown).Therefore, we can conclude that these QD readily bind and enter living HeLa cells.Despite the difference in surface charge, the pattern of QD distribution inside cells was quite similar: the highest amount of QD concentrated within the plasma membrane while significant fraction of QD was located in the intracellular compartments (Fig. 5 A, E, I).No signal was obtained from the nucleus of the cells.Similarly, other studies have demonstrated that although the charge of terminal groups may determine the pathway and quantity of QD internalization, it does not affect the final intracellular distribution and localization of QD. [38][39][40] Lately, we have shown that carboxyl QD enter cells via lipid raft/caveolinmediated endocytosis, accumulate in endosomes and end-up in the multivesicular bodies. 41,42  seen from Fig 5 .A, E, I, the emission lifetime of QD in cells was a little shorter than in solutions (Table 2).Moreover, the distribution of QD emission lifetime in cells was not homogeneous.In the plasma membrane the average emission lifetime was rather long (∼12.5 ns) while in the intracellular vesicles it was by 2-6 ns shorter (Fig. 5 A, E, I).Irradiation of the cells by a blue light for 30 s enhanced the emission intensity of QD and lengthened its lifetime by ∼2 ns in average.
Remarkably, the obtained lifetime values were close to those for QD in solutions (Fig. 5 B, F, J and Table 2), thus confirming that the observed cellular fluorescence belongs to QD.Moreover, after irradiation, the difference between emission lifetime of QD in plasma membranes and those in the intracellular compartments reduced.Shortening of fluorescence lifetime of thiol-capped CdTe QD inside living cells was reported by Zhang et al, who demonstrated in solutions that both, reduction of pH and interaction with different amino acids and proteins may be responsible. 43,44 n our case, almost complete recovery of QD fluorescence lifetime after irradiation suggests that QD might be quenched by a blue light absorbing intracellular chromophores, such as NADH, flavins, quinones, and other cofactors or even endogenous porphyrins.The direct excitation by a blue light causes their photobleaching and thus recovers almost completely the QD properties.
FLIM images of QD-Ce 6 complexes (QD:Ce 6 1:5) incubated for 2h with HeLa cells before and after irradiation are shown in the third and fourth columns of Fig. 5, respectively.Similarly to QD alone, most of QD-Ce 6 complexes accumulated in the plasma membrane and fewer in the intracellular compartments.The fluorescence lifetime of QD-Ce 6 complexes was the half (5.8-6.8 ns) of the QD, which matches perfectly with the time-resolved data of these complexes in solutions (Table 2).Hence, QD-Ce 6 complexes in cells preserved the high FRET efficiency (Table 2).Photobleaching of Ce 6 (FRET acceptor) by irradiation of the cells with the blue light for 30 s resulted in the recovery of the large values of the emission lifetime (Fig. 5 D, H, L) close to that of the irradiated QD without Ce 6 .These results show that QD complexes with Ce 6 after internalization into HeLa cells remain stable in the cellular context and do not change their composition.This is a striking result, taking into account that Ce 6 molecules are known to readily bind different proteins and cellular membranes, including the plasmatic, nuclear and mitochondrial membranes. 33,45 he absence of leakage of Ce 6 molecules from QD into cellular membrane components indicates exceptionally strong binding between QD and Ce 6 .Hydrophobic with lipid coating could be one possible explanation for this phenomenon, as amphiphilic Ce 6 seems to localize in the apolar lipid environment of QD.From the simple geometric consideration, taking into account the QD diameter of 20 nm and a typical surface area per lipid of 0.7 nm 2 , the estimated number of lipids per QD is <1800, so that the lipid concentration for the 0.02 µM solution of QD was <36 µM.Taking into account the relatively low affinity constant of Ce 6 to vesicles of unsaturated lipids (dioleoylphosphatidylcholine) at pH 7.4 (6 × 10 3 M −1 ), 33 only <18 % of Ce 6 should be bound to lipids of QD at the Ce 6 /QD ratios used.However, the observation of highly efficient FRET at 1:1 complex and our titration data suggest nearly quantitative binding of Ce 6 to QD.Moreover, taking into account that the affinity of Ce 6 to BSA (1.8 × 10 8 M −1 ), 33 is about ~30000-fold higher than that to lipid membranes, the QD-Ce 6 complex should be readily destroyed in the presence of BSA excess.However, the opposite was observed, so that the release of Ce 6 from QDs to BSA was very slow and incomplete (Fig. 3), in line with the earlier work. 46Therefore, the hydrophobic interactions of Ce 6 with lipids of QD cannot be the only reason for this exceptional stability of the complexes in biological media.We speculate that due to its three carboxyl groups Ce 6 could interact directly with ZnS layer of the QD core (Scheme 1).For instance, Patel et al. showed that in oleic acid-capped ZnS semiconducting nanocrystals, the two oxygen atoms of the carboxylate were coordinated symmetrically to the surface of the nanorystals, thus providing high stability to the formed fatty acid monolayers and to the obtained nanocrystals colloids. 47uch bonding of Ce 6 carboxyl groups to ZnS layer of QD could also explain the unexpectedly high values of FRET efficiency, indicating that deeply imbedded in the QD lipid coating Ce 6 molecules situate very close to QD inorganic core.

Conclusions
The use of QD as FRET donors can drastically improve the excitation efficiency of the photosensitizer.Here, we studied formation of complexes between QD bearing neutral, carboxyl and amine functional groups with second-generation photosensitizer, chlorin e 6 .Spectroscopic changes and the highly efficient FRET, observed upon Ce 6 binding to QD, suggest that Ce 6 localizes inside lipid coating close to the inorganic core of QD.Two-photon fluorescence lifetime imaging microscopy on living HeLa cells revealed that, independently of QD surface functional groups, QD-Ce 6 complexes localize within plasma membrane and intracellular compartments and preserve ~50% FRET efficiency.This exceptional stability in cellulo of non-covalent QD-Ce 6 complexes can be explained by coordination of carboxyl groups of Ce 6 with ZnS shell of QD, in addition to hydrophobic interactions.Our data suggest that a simple protocol without chemical conjugation can lead of QDphotosensitizer complexes characterized by efficient FRET and excellent stability in cellulo.

Figure 1 .
Figure 1.A-absorption and B-fluorescence excitation spectra of 0.02 µM carboxyl QD, 0.1 µM Ce6 and corresponding mixed QD-Ce6 (0.02 µM QD : 0.1 µM Ce6) aqueous solutions.C-fluorescence spectra of 0.02 µM carboxyl QD, 0.2 µM Ce6 and mixed QD-Ce6 aqueous solutions at increasing QD:Ce6 molar ratio from 1:0.5 to 1:10.The dotted arrow in absorption spectra (A) shows the excitation at 465 nm, used for the fluorescence (C) measurements.The inset of Figure C shows the fluorescence of pure Ce6 solution at corresponding concentrations at λex =465 nm.The fluorescence excitation spectra (B) were recorded at the fluorescence maximum of QD-Ce6 complex at λem=670 nm.

Figure 2 .
Figure 2. A-Normalized emission intensity of pure QD and mixed QD-Ce6 solutions at increasing QD:Ce6 molar ratios measured at 625 nm.The normalization was performed to the maximum of carboxyl QD emission intensity.B-Absolute intensities of bound Ce6 at 670 nm in QD-Ce6 solutions.For the comparison, the rise in Ce6 intensity at 660 nm due to increasing concentration is also shown.

3. 3
Microscopy studies of QD-Ce 6 complexes in living HeLa cells Two-photon FLIM images of HeLa cells treated either with QD alone or QD-Ce 6 complexes are shown in Fig. 5.After 2h of incubation with QD alone, strong fluorescence signal was

Figure 5 .
Figure 5. Two-photon FLIM images of HeLa cells incubated for 2 hours with carboxyl (A, B), amine (E, F) and non-funcionalized (I, J) QD and their complexes with Ce6 (C, G, K and D, H, L) before and after irradiation with blue light for 30 s, respectively.The size of all images is 70x70 µm.The color lifetime scale of each image is from 4 (red) to 21 (blue) nm.<τ> indicates the lifetime of QD emission averaged through the entire image.

Table 2 .
FRET properties of QD-Ce6 complexes calculated from the steady-state and time-resolved spectral results in solutions and in HeLa cells.calculatedtaking<τ> and <τ´> values before cell irradiation, the first and third columns of Fig.4, respectively. *