A pillar[5]arene-based [2]rotaxane lights up mitochondria

Here we integrate diagnostic and therapeutic agents into a mitochondria-targeting [2]rotaxane, which can be utilized as a drug delivery platform to conjugate anticancer drugs to prepare prodrugs for efficient targeted drug delivery.

as control. To study the time-dependent cellular uptake, R1 NPs in fresh DMEM medium was incubated with cells for 0.5, 1, and 2 h. Then the cells were washed with PBS, trypsinized, and re-suspended in PBS. Flow cytometry measurements were conducted using BD FACSEALIBUR with excitation at 405 nm. The mean fluorescence was determined by counting 10,000 events.
For confocal imaging, the tested cells were cultured in the chambers at a density of 5 × 10 5 per mL for 24 h. The cells were incubated with R1 NPs at 37 °C for 2 h, followed by staining with MitoTracker Red for 30 min. Then the cells were washed with PBS and imaged immediately by confocal laser scanning microscope (CLSM, ZEISS LSM780).

Cell Imaging with Carbonyl Cyanide m-Chlorophenylhydrazone (CCCP)
Treatment. Cells were grown overnight on a 35 mm petri dish with a cover slip. The cells were incubation with 10 μM CCCP for 30 min. The CCCP treated cells were then stained by R1 for 2 h and MitoTracker Red for 30 min.
Photostability studies of R1 and MT. Continuous scanning by confocal microscope was used to quantitatively investigate the photostability of R1 and MitoTracker Red. Two dishes of HeLa cells subcultured from the same source were stained with 2 μM R1 and 100 nM MitoTracker Red, respectively. With the help of a power meter, excitation power from 405 and 560 nm channels of the microscope were unified (65 μW) and used to irradiate the R1 and MitoTracker Red stained cells. The initial intensity referred to the first scan of R1 and MitoTracker Red stained cells was normalized, and the percentage of fluorescence signal loss was calculated. S4
To determine the stoichiometries and association constants for the complexations between P5 and the guests (4 and 5), 1 H NMR titrations were done with solutions which had a constant concentration of the guest (4 or 5) (2.00 mM) and varying concentrations of host P5. By a non-linear curve-fitting method, the association constant (K a ) of P54 (or 5) was estimated. By a mole ratio plot, 1:1 stoichiometry was obtained.

S31
Further evidence for the formation of host-guest complex between P5 and 4 was obtained from electrospray ionization mass spectrometry (ESI-MS). The relevant peak of the inclusion complex P54 was found at m/z 1340.6, corresponding to [P54  PF6] + , which confirmed the 1:1 complexation between P5 and 4 (Fig. S38). However, other than the peaks at m/z 314.6, 768.5 and 773.4 corresponding to [5  PF 6 ] + , [P5 + NH 4 ] + and [P5 + Na] + , respectively, no peaks were observed related to the host-guest complex P55 (Fig. S39), in good agreement with the results obtained from 1 H NMR investigations. The ESI-MS data of R1 contained a peak at m/z 1019.5 (Fig. S20), corresponding to [R1  2PF 6 ] 2+ , which provided direct evidence for the formation of R1. From previous work, we knew that the pillar-shaped cavity moved along the long alkylene chain, S32 causing upfield shifts of the methylene protons. The pillar[5]arene ring was statistically located on the methylenes whose protons showed relatively bigger upfield shifts in solution. Therefore, we used 1 H NMR spectroscopy to determine the position of 6 in this MIM. On account of the relative high binding affinity between P5 and 4, we speculated that the cationic pyridinium ring and the adjacent methylene groups were located in the cavity of the pillar[5]arene ring. Indeed, the 1 H NMR and 2D NOESY spectra of R1 in chloroform-d confirmed our deduction (Fig. S40, a and c). As shown in Fig. S40a between protons H a , H b , and H c of the wheel and methylene protons H 9-16 of the axle, demonstrating the formation of the host-guest inclusion complex (Fig. S40c). In order to further verify this mechanically interlocked structure, the highly polar solvent DMSO-d 6 was used for proton NMR investigations. As shown in Fig. S40b, the peaks related to protons H 1 , H 2 , H 8 , H 9 and H 12-18 displayed upfield chemical shifts due to the shielding effect, indicating that the axle was in the cavity of 4. NOE correlation signals were observed between protons H a-c of the pillar[5]arene ring and protons H 9-17 and H 19 on the axle in the NOESY spectrum of R1 in DMSO-d 6 (Fig. S40d), convincingly confirming the formation of a MIM.  The free axle 7 exhibited the characteristic AIE feature. It gave very weak emission in THF where it was well dissolved. The fluorescence intensity of 7 at 550 nm increased slowly with increasing volume fraction of water (f w ) in the THF/H 2 O mixture from 0 to 80 vol%, and increased dramatically upon further enhancement of the f w value from 80 to 98 vol% (Fig. S43), consistent with other classical AIE dyes. [13] Upon formation of a MIM, R1 exhibited an enhancement of the AIE effect. It had faint fluorescence intensity when molecularly dissolved in THF, but was fluoresced intensively when the f w value increased (Fig. S43a). It should be noted that the fluorescence intensity of R1 was higher than that of 7 at the same concentration (Fig. S43b).  NPs can enter cells by several different endocytic pathways, such as phagocytosis and pinocytosis, which not only affect the uptake efficiency of NPs but also their intracellular fate, affecting the pharmacological activities of the loaded cargoes (such as dyes, drugs or genes). Phagocytosis is conducted primarily by specialized cells, such as monocytes, macrophages and neutrophils, which can clear out large particles (several micrometers) in blood. Clathrin-mediated, macropinocytosis, caveolin-mediated, and clathrin-and caveolin-independent endocytosis are the four major processes of S36 pinocytosis, which operate in all mammalian cells. The internalization pathways of the NPs self-assembled from R1 in HeLa and HEK293 cells were studied using flow cytometry by applying various endocytosis inhibitors (Fig. S45). The internalization pathways of the NPs self-assembled from R1 were studied using flow cytometry by applying various endocytosis inhibitors. Uptake of R1 NPs by the HeLa and HEK293 cells was almost completely inhibited at 4 °C (the low temperature is beneficial to minimize the metabolism of cell plasma membrane), demonstrating the energy-dependent nature of particle uptake (Fig. S45). HeLa cells treatment with sucrose resulted in a 56% decrease in the cellular uptake of R1 NPs (Fig. S45a), suggesting that R1 NPs might be mainly internalized via clathrin-mediated endocytic pathway, which generally plays an important role in the internalization of nanocontainers into cells. Additionally, a 36% decrease in the cellular uptake of R1 NPs was monitored by treating HeLa cells with amiloride-HCl, indicating that the macropinocytosis-mediated pathway also partly contributed to the internalization of R1 NPs. However, genistein, an inhibitor of caveolae-mediated endocytosis, did not show an obvious effect on the cellular uptake of R1 NPs, demonstrating that the caveolae-mediated endocytosis pathway exerted a negligible effect on the uptake of R1 NPs by the HeLa cells. Similarly, the cellular uptake of R1 NPs by HEK293 cells was effectively blocked by sucrose and amiloride-HCl, suggesting that the internalization of these particles was mainly mediated by macropinocytosis-and clathrin-mediated endocytosis rather than the caveolae-mediated pathway (Fig. S45b). These pathways allowed R1 NPs to undergo the endo/lysosomal transport for intracellular delivery of R1.

Förster resonance energy transfer in R2
According to the Förster theory, the transfer efficiency (Φ T ) can be expressed as: Where τ DA and τ D are the fluorescence lifetime of the donor in the presence and absence of the acceptor, respectively. k T is the transfer rate between an excited donor and an acceptor fluorophore.
Where r is the distance between the fluorophores.
The Förster radius (R 0 , in Ångstrom) can be calculated using: R 0 6 = 8:79*10 -5 k 2 Φ D 0 n -4 J(λ) Where Φ D 0 is the fluorescence quantum yield of the donor in the absence of the S38 acceptor, n is the refractive index of the medium, J(λ) is the overlap integral describing the degree of overlap between the donor fluorescence emission spectrum and the acceptor absorption spectrum, and k 2 is the orientation factor, a measure of the relative orientation of the transition dipole moments of the donor (emission) and the acceptor (absorption) and the vector connecting the molecules.  found an overlap between the emission spectrum of R1 and the absorption spectrum of DOX·HCl, confirming that R1 could act as a fluorescent donor for the acceptor (DOX·HCl) that absorbs maximally at 500 nm. By the reaction between DOX·HCl and R1, DOX was grafted on R1 by imine bonds formed between the amine group on the DOX and aldehyde groups on R1. As shown in Fig. S48, the NMR signals at 9.84 ppm of the aldehyde groups disappeared upon formation of R2, indicating that these groups were converted completely. Similary, NPs self-assembled from R2 in water with an average diameter of about 60 nm were obtained through a reprecipitation technique (Fig. S49).
Next, we examined the ETR-caused dual-fluorescence quench behavior in R2. As shown in Fig. S47b, the characteristic emission corresponding to the TPE-based fluorogen was not observed for R2, indicating that the AIE behavior disappeared by introducing DOX into this MIM. The disappearance of the AIE behavior for R2 was ascribed to the emissive energy transfer from the TPE-based fluorogen to DOX, because the distance between the donor and the acceptor was so short that FRET easily took place in R2. However, the ACQ effect of DOX in the aggregated state reduced the fluorescence intensity by "π-π stacking" of their rigid planar aromatic rings. To verify that the ACQ behavior occurred in R2 NPs, a solvent dependent aggregation method was employed.
The DOX chromophore in R2 exhibited strong fluorescence at 591 nm when R2 was well dissolved in THF. However, the fluorescence intensity dramatically decreased upon addition of water (Fig. S47, b and c). When the water fraction reached 98 vol %, the fluorescence intensity intensity of the DOX chromophore was nearly 19.3-fold weaker than that in pure THF (Fig. S47c), confirming that the ACQ effect of the DOX chromophore gradually increased along with the aggregation of R2. Thus, a dual-S40 fluorescence-quenched supramolecular system was prepared through ETR, in which the emission from R1 was transferred to DOX, whereas the emission of DOX was selfquenched due to the ACQ effect.
Lifetime is a key kinetic parameter for the fluorescence intensity decay. Timeresolved spectroscopy was employed to investigate the photophysical behavior of R1 and R2 (Fig. 5, d and e). The decay dynamics of R1 were better fitted by a doubleexponential function, suggestting that two relaxation pathways were involved in the decay process. For example, 92.3% (A 1 ) and 7.7% (A 2 ) of the excitons of R1 decayed via the fast and slow channels with lifetimes of 6.44 ns (τ 1 ) and 11.9 ns (τ 2 ), respectively. A possible explanation is that electron transfer to the pyridinium unit occurs in a reversible fashion. S4 For R2, the excited state decayed in a three-exponential fashion with short components τ 1 = 1.01 ns (90.3%) and τ 2 = 2.54 ns (7.1%) that predominated at short wavelengths, and a longer decay time τ 3 = 9.9 ns (2.6%). A tentative explanation for this observation is that the noncovalent interactions between the TPE-based axle and the wheel led to an exciplex or heteroexcimer. S4 The weighted mean lifetimes (τ) of R1 and R2 were calculated to be 6.86 ns and 1.29 ns, respectively. From these analyses, the efficiency of energy transfer (Φ T ) of this system was calculated to be 81%, and the Förster radius (R 0 ) and the energy transfer rate (k T ) were estimated as 1.99 nm and 0.77, respectively. In the present case, the value of Φ T was quite high, indicating that a very efficient energy transfer took place from the TPE-based chromophore to the wheel unit.
Moreover, the distance between the donor and the acceptor was calculated to be 1.56 nm, which is in good agreement with the molecular structure.   and under a UV lamp with laser excitation at 365 nm (right).

S42
The fluorescence recoveries of R2 NPs at different pH values were visualized by using a UV lamp with an excitation wavelength of 365 nm. As indicated by fluorecent spectra shown in Fig. S50 and Fig. S51, the fluorescence intensity of DOX recovered effectively when the solution pH was adjusted to 5.0, demonstrating that DOX was released from R2 NPs. Unexpectedly, no obvious fluorescence intensity recovery was observed when we monitored the fluorescence recovery of R1 by culturing the R2 NPs at pH 5.0. A possible reason is the persistence of FRET between the R1 and DOX molecules. After dialysis, the detached DOX was separated from R1, and the fluorescence intensity corresponding to R1 recovered significantly, indicating that the AIE behavior of R1 was indeed retained. This phenomenon verified that R1 remained in the aggregated state and retained its AIE behavior when DOX was released from the R2 NPs. The fluorescence intensity of R1 and DOX was recovered, affording a dual-color fluorogenic process once the imine bonds were cleaved. The intracellular microenvironment of tumor cells is typically characterized by slightly acid pH in the endosomal (5.0−6.0) and lysosomal (4.0−5.0) compartments.
When the pH was adjusted to 5.0 upon addition of DCl, the imine bonds were broken and R2 decomposed into R1 and free DOX. Actually, when the pH value was changed to 6.5, the signal at 9.84 ppm corresponding to the aldehyde group appeared again (Fig. S48), confiming the hydrolysis of R2. The release behavior of DOX from R2 NPs was carried out at pH 7.4, 6.0, and 5.0, respectively, mimicking the pH gradient from blood circulation to the endo/lysosomal compartments. As shown in Fig. S52, about 6.0% of DOX was released within 24 h at pH 7.4. However, 51.3% of DOX was released from R2 NPs after 24 h at pH 6.0 and nearly 100% at pH 5.0, respectively. The DOX release profile was clearly pH-dependent, owing to the accelerated hydrolysis of the imine bonds.
The DOX release from R2 NPs was switched off during systematic circulation (pH 7.4).
However, prompt DOX release occurred upon being entrapped in the endo/lysosomal compartments after endocytosis. As shown in Fig. S53, the NMR signal at 9.84 ppm of the aldehyde groups disappeared upon formation of the prodrugs, indicating that these groups were converted completely.

S48
The results discussed in the main text indicated that DOX was released in endo/lysosomal compartments due to the hydrolysis of Schiff base linkages in mildly acidic environments. To assess the antitumor activity of the released DOX in vitro, we evaluated its cytotoxicity towards HeLa and HEK293 cell lines by using MTT assays.
The cytotoxicity of each treatment was expressed as the percentage of cell viability relative to the untreated control cells. MTT assays were performed by exposing the cells to free DOX·HCl and R2 NPs with concentrations ranging from 5 to 25 µM. As shown in Fig. S55a, the relative cell viability of the HeLa cells incubated with R2 NPs decreased gradually from 88.7% to 14.3% upon increasing the concentration of R2 NPs from 5 to 25 µM, confirming that the released DOX retained anticancer activity. It should be noted that the relative cell viability of the HeLa cells incubating with R2 NPs was higher than the HeLa cells cultured with free DOX·HCl under the same conditions, because the cationic DOX·HCl with good water solubility diffused into the cells easily, so the concentration of the anticancer drug was higher than that released by the hydrolysis of R2 NPs. On the other hand, the relative cell viability of the HeLa cells was lower than that of the HEK293 cells at the same concentration of R2 NPs (Fig. S55b), because carcinoma cells have a higher membrane potential than normal cells, thus resulting in the improvement of the cellular uptake of R2 NPs. Another reason for the difference in cytotoxicity of R2 NPs towards HeLa and HEK293 cells was the difference in intracellular pH values. Typically, the intracellular pH in cancer cells is lower than that of normal cells. Therefore, the anticancer drug conjugated on the rotaxane was released faster in HeLa cells than in HEK293 cells. S49