Photoresponsive Nanocarriers Based on Lithium Niobate Nanoparticles for Harmonic Imaging and On-Demand Release of Anticancer Chemotherapeutics

Nanoparticle-based drug delivery systems have the potential for increasing the efficiency of chemotherapeutics by enhancing the drug accumulation at specific target sites, thereby reducing adverse side effects and mitigating patient acquired resistance. In particular, photo-responsive nanomaterials have attracted much interest due to their ability to release molecular cargos on demand upon light irradiation. In some settings, they can also provide complementary information by optical imaging on the (sub)cellular scale. We herein present a system based on lithium niobate harmonic nanoparticles (LNO HNPs) for the decoupled multi-harmonic cell imaging and near-infrared light-triggered delivery of an erlotinib derivative (ELA) for the treatment of epidermal growth factor receptor (EGFR)-overexpressing carcinomas. The ELA cargo was covalently conjugated to the surface of silica-coated LNO HNPs through a coumarinyl photo-cleavable linker, achieving a surface loading of the active molecule of 27 nmol/mg NPs. The resulting nanoconjugates (LNO-CM-ELA NPs) were successfully imaged upon pulsed laser excitation at 1250 nm in EGFR-overexpressing human prostate cancer cells DU145 by detecting the second harmonic emission at 625 nm, in the tissue transparency window. Tuning the laser at 790 nm resulted in the uncaging of the ELA cargo as a result of the second harmonic emission of the inorganic HNP core at 395 nm. This protocol induced a significant growth inhibition in DU145 cells, which was only observed upon specific irradiation at 790 nm, highlighting the promising capabilities of LNO-CM-ELA NPs for theranostic applications.

M utations in the epidermal growth factor receptor (EGFR) gene leading to the overexpression of the transmembrane protein have been associated with a variety of human carcinomas, 1 including non-small-cell lung cancer 2 (NSCLC, more than 60% of the cases), glioblastomas 3 (50%), and epithelial cancers 4 (80−100%). This led to the identification of the EGFR gene as a proto-oncogene and relevant therapeutic target for these cancer types. 5 Amplification of the EGFR protein causes its uncontrolled activation and is correlated with tumor cell proliferation, metastasis, and angiogenesis. 6 Tyrosine kinase (TK) inhibitors emerged as potential chemotherapeutics for the treatment of EGFRmutated malignancies. Their mode of action consists in blocking the adenosine triphosphate (ATP) binding site of the EGFR, resulting in the inhibition of its auto-phosphorylation and downstream signaling mechanism, thereby limiting cell growth. 7 Notably, erlotinib (Tarceva) was developed as a small-molecule TK inhibitor and approved for the treatment of NSCLC and pancreatic cancer. 8 Despite its potency and selectivity as a TK inhibitor, erlotinib exhibits a low bioavailability, which requires the daily administration of a significant dose (150 mg for NSCLC treatment), over a long period of time, to achieve therapeutic effects. 9,10 Such treatment modality may lead to many potential adverse effects, commonly including rashes, diarrhea, and fatigue, and in rarer cases, gastrointestinal tract perforations, Stevens−Johnson syndrome, or pulmonary damages, which can even be fatal. 11,12 Another important limitation to erlotinib treatments, which is common to most small-molecule TK inhibitors, is patient acquired resistance. 13 The latter most frequently results from mutation of the EGFR ATP binding pocket and typically affects patients after 9 to 15 months of treatment. 14 The development of nanoparticle-based formulations for this class of chemotherapeutics emerged as an appealing strategy to increase their efficiency at lower dosages due to enhanced drug accumulation at the target site. Various nanotechnologyoriented delivery systems were recently disclosed to increase the therapeutic efficiency of erlotinib by limiting dose-related side effects and potentially delaying or even eliminating acquired resistance. 15,16 Most of the nanodevice-based strategies reported involve encapsulation in polymer-based nanocarriers, 17,18 liposomes, 19 and lipid−polymer hybrid nanoparticles. 20 A different strategy which, to the best of our knowledge, has not yet been extensively investigated in the case of similar TK inhibitors is drug delivery through systems, which combine optically active inorganic nanomaterials [quantum dots, plasmonic nanoparticles, and upconversion nanoparticles (UCNPs)] 21−23 with photo-cleavable scaffolds (ortho-nitro-benzyl, para-hydroxyphenacyl, and coumarinyl derivatives). 24−26 This approach offers precise control over the release, both in time and space, 27 while adding complementary bioimaging capabilities for theranostic applications. 28 Most photo-responsive caging moieties rely on high-energy UV light excitation, which severely limits their application in biomedical setups due to phototoxicity and poor tissue penetration depth. 29 Nanocarriers based on lanthanide-doped UCNPs can overcome these shortcomings due to their ability to convert two or more low-energy near-infrared (NIR) excitation photons into UV or visible emission. The favorable properties of NIR light, including comparatively large penetration and decreased photodamage in living cells and tissues, has prompted the development of UCNP-based delivery systems for the in vitro and in vivo targeted release of biomolecules, 30 small interfering RNA, 31 and anticancer chemotherapeutics. 32 Harmonic nanoparticles (HNPs) 33 share with UCNPs the ability to convert excitation photons to shorter wavelengths; however, the underlying photo-physical process is substantially different. In UCNPs, the emission follows a series of sequential absorption events proceeding through the real energy levels of lanthanide ions, while harmonic generation in HNPs is an Scheme 1. Synthesis of the Caged Erlotinib Derivative ACS Nanoscience Au pubs.acs.org/nanoau Article inherently nonlinear process occurring through virtual states, which exist exclusively during the interaction of the particle with an intense electromagnetic field. The approach therefore relies on ultrafast (femtosecond) laser excitation to reach large peak intensities at comparatively low pulse energies. This stringent technical requirement definitively makes HNPs a more niche approach than UCNPs. On the other hand, HNPs provide some complementary advantages that are worth investigating. While UCNP emission is typically associated with long lifetimes, 34 the process of harmonic generation is practically instantaneous and therefore compatible with scanning (three-dimensional) approaches. The distinctive features of HNPs include their coherent emission, 35 absence of bleaching/blinking when excited within the material transparency range, and wide spectral flexibility. 36−38 The nonlinear generation efficiency is material-and wavelengthdependent; the orientation-averaged second-order susceptibility of lithium niobate (LNO) nanocrystals under 700−1300 nm excitation was reported by Riporto et al. 39 The possibility to colocalize the simultaneously emitted second and third harmonic signals upon excitation in the shortwave infrared 40 windows make HNPs a promising approach for selected imaging applications, such as cell monitoring in tissues. 41−44 The tunable response of HNPs offers an opportunity for decoupled imaging modality and photo-triggered uncaging of surface-conjugated molecular cargos. We previously demonstrated that the harmonic emission of bismuth ferrite (BiFeO 3 , BFO) or LNO HNPs can be used as a local trigger for the controlled release of caged L-tryptophan 45 and anticancer drug chlorambucil. 46 Herein, we disclosed the controlled release of an analogue of the EGFR TK inhibitor erlotinib (ELA) from the surface of LNO-based nanoconjugates. Making use of a coumarinyl (CM) tether, the ELA cargo was covalently immobilized on silica-coated LNO HNPs through copper-free azide for strained alkyne [3 + 2] cycloaddition. The ELA-functionalized nanocarriers were first assessed for their nonlinear optical imaging properties upon incubation with the EGFR-overexpressing DU145 human androgen-dependent prostate cancer cells. Then, we evaluated their cytotoxic effect in vitro by tuning the NIR irradiation to 790 nm, which induced the release of ELA and led to a marked decrease in cancer cell viability. These decoupled imaging and NIR light-triggered delivery protocols demonstrated the capability of harmonic nanoconjugates to act as efficient exogenous optical probes and photoresponsive drug delivery systems. The chemical structure of erlotinib features a 4-aminoquinazoline core and an alkyne-substituted aromatic ring, which interact with the tyrosine residue of the EGFR and the ATP receptor, respectively (Scheme 1). The design of an erlotinib analogue, amenable to covalent conjugation to a photosensitive caging group while maintaining its ability to interact with the EGFR ATP binding site, was based on the modification of the C6 and C7 aromatic substituents which are directed outside of the binding pocket. 47 Starting from the aniline derivative 1 (synthesis protocols detailed in Supporting Information, S3), condensation with dimethylformamide dimethylacetal (DMF-DMA) followed by Dimroth rearrangement in the presence of 3-ethynylaniline afforded the substituted 4-anilinoquinazoline scaffold (3) 48 in 54% yield (two steps). Methanolysis of the acetate followed by conversion of the resulting primary alcohol into a leaving group allowed for the introduction of a terminal primary amine by nucleophilic substitution, affording the erlotinib derivative ELA. Introduction of the photosensitive spacer proceeded through conjugation to the activated 7amino-coumarine derivative 5 (synthesis protocols detailed in Supporting Information, S9), in quantitative yield. Subsequent saponification of the ethyl ester and coupling with the dibenzocyclooctyne-containing unit 7 (synthesis protocols detailed in Supporting Information, S15) afforded the targeted caged erlotinib derivative CM-ELA, ready to undergo strainpromoted [3 + 2] cycloaddition reaction. For control experiments, ELA was also coupled to a precursor of compound 7 to provide DIBO-ELA (synthesis protocols detailed in Supporting Information, S28), the equivalent of the drug delivery system DIBO-CM-ELA without the photosensitive coumarinyl linker.

Preparation of LNO-Based Nanoconjugates
Surface silanization of LNO HNPs [prepared at a low hydrolysis rate of alkoxide precursors with a solvothermal route already described; 49 for transmission electron microscopy (TEM) and X-ray diffraction (XRD) characterization of LNO HNPs, see Supporting Information S31] in the presence of 4-azido-N-(3-triethyloxysilylpropyl)butanamide provided LNO-N 3 NPs (for the coating protocol and TEM characterization, see Supporting Information S31). The resulting suspension was subjected to the strain-promoted click reaction in in the presence of CM-ELA, under ultrasonication, to provide the functionalized LNO-CM-ELA NPs, which were stored in EtOH (2 mg/mL) (Scheme 2).
The surface loading of the erlotinib derivative ELA was estimated to be 27 nmol/mg NPs (for the quantification protocol, see Supporting Information S33, Figure S3). The coated and conjugated NPs were characterized by means of their mean hydrodynamic diameter (d), which increased upon surface silanization (from 77.8 ± 10.3 nm for bare LNO NPs to 146.3 ± 42.9 nm for LNO-N 3 NPs) and post-   Table 1). The polydispersity index (PDI) of the final nanoconjugates (0.09) indicated their monodisperse size distribution and the absence of aggregate formation. These observations are consistent with the TEM analysis performed on the LNO-N 3 ( Figure 1, for images of a larger ensemble of NPs, see Supporting Information Figure  S2A). The energy-dispersive X-ray (EDX) elemental analysis of LNO-N 3 NPs revealed an atomic ratio of silicon to niobium of 7.8% (see Supporting Information, S32, Figure S2B).
The stability of the nanoconjugate suspension was monitored by DLS upon storage of LNO-CM-ELA NPs in PBS for 24 h and 7 days at 4°C (Table 1). No significant variation of the mean hydrodynamic diameter was observed. However, the increase in the PDI values indicated the formation of some aggregates in PBS over time. The size of LNO-CM-ELA NPs was also measured in cell culture medium used during the in vitro experiments.

Nonlinear Optical Imaging of LNO-CM-ELA NPs
In the upper row of Figure 2, we show the tunable spectral response of LNO-CM-ELA NPs deposited on a substrate obtained upon NIR excitation between 800 and 1000 nm on a Nikon setup. The spectra at each wavelength are associated with the emission of the object indicated by the dashed outline in the exemplary image obtained upon 1000 nm excitation (further spectral characterizations are provided in Supporting Information S-43, Figure S13). In the lower row, we present a series of nonlinear microscopy images of DU-145 cells treated with LNO-CM-ELA NPs (50 μg/mL) acquired on a Leica setup. Upon excitation at 800 nm, the second harmonic generation (SHG) is visible in the 400 nm channel and the 4′,6-diamidino-2-phenylindole (DAPI) fluorescence from the nuclei is visible in the channel centered at 495 nm. Using the 1250 nm excitation, the system is capable to acquire both the third harmonic generation (THG) at 416 nm and the SHG at 625 nm (1/2 and 1/3 of the excitation wavelength, respectively). The comparison between the three harmonic responses highlights HNP excitation tunability beyond the NIR range. The labeling of cells is quite sparse as HNPs appear as isolated aggregates but overall sufficient for recognizing cell morphologies.

NIR-Triggered Release of ELA and in vitro Evaluation of Cytotoxic Activity
The erlotinib derivative ELA was first evaluated for its cytotoxic effect on the EGFR-overexpressing DU-145 human androgen-dependent prostate cancer cells (for EGFR expression in DU-145 cells, see Supporting Information S34, Figure S4), in comparison with the parent commercial drug. DU-145 cells were incubated with commercial erlotinib for 72 h using concentrations ranging from 0.5 to 10 μM following the protocol described in the experimental section. The impact on cell viability was assessed using 3-(4,5-dimethyl-2-thiazoyl)-2,5-diphenyltetrazolium bromide (MTT) colorimetric assay. Our findings indicate that the two compounds displayed a similar cytotoxicity profile as they were able to induce 40−57% growth inhibition at 1 μM and 88−90% at 10 μM ( Figure 3). These results suggest that the erlotinib derivative ELA conserves the ability to inhibit the growth of EGFRoverexpressing cancer cells.
The photocleavage and subsequent release of ELA from the surface of functionalized LNO NPs was first assessed under direct UV irradiation (see Supporting Information S35 for an estimation of UV flux at the sample) at 366 nm in PBS, in the absence of cancer cells (Figure 3, black curve, see Supporting Information S35 for detailed calculation and progress curve). Analysis of the medium by ultrahigh-performance liquid chromatography−mass spectrometry (UHPLC−MS) at successive time points indicated that the unique released component was ELA (assessed by its coelution with the standard after 2.0 min and accurate mass, see Supporting Information S38 for quantification protocols and calibration curve). Then, the ability of the second harmonic emission of LNO-CM-ELA NPs to trigger the uncaging of the cytotoxic cargo was investigated using the unfocused beam from an amplified Ti:sapphire pulsed laser system. In the first setting, we used an amplified laser system delivering a femtosecond pulse train at a 1 kHz repetition rate (setup 1, see Experimental Section) according to the protocols previously developed on BFO NPs. 45 The choice of an amplified laser system at a low repetition rate allows to maintain a comparatively large peak intensity (necessary for exerting nonlinear interaction) on a large sample area, which facilitates the subsequent reliable quantification of ELA release by UHPLC−MS. The low repetition rate is also advantageous for limiting heat accumulation effects and sample photodamage. 50 With setup 1, after 15 min of irradiation at 790 nm in PBS, the release of ELA reached 41% (Figure 4). This irradiation induced an SHG emission at 395 nm from the HNPs, matching with the absorption of amino-coumarin dyes reported to be centered around 380 nm. 51 The progress curves were fitted using a monoexponential function, and the initial rate (k 0 ) at 5 min was calculated to be 271 nM·min −1 (see Supporting Information S36 for the detailed calculation and progress curve).
In order to verify the dependence of SHG wavelength on the release, given that setup 1 works at fixed wavelength, we repeated the irradiation on a tunable MHz laser (setup 2, see Experimental Section). Because of the inherent characteristics of this source, we were bound to work using different peak intensities and irradiation surfaces. To partially compensate for these discrepancies, we applied longer irradiation times and we continuously moved the sample during the irradiation as detailed in the Experimental Section. We applied the irradiation treatment at three excitation wavelengths (800, 1000, and 1100 nm) for 1 h. Although the results from UHPLC−MS are less quantitatively reliable under these irradiation conditions, we could only observe ELA release upon 800 nm irradiation (corresponding to SHG emission within the absorption spectrum of CM) and basically no ELA detection at the two other wavelengths (Supporting Information S41, Figure S11).
Under NIR irradiation, the observed release rate was slower as compared to direct UV treatment as the efficiency of SHG to generate the cleaving wavelength has to be taken into account. However, the concentration of ELA reached under these conditions was in the expected range to exert cytotoxic effects on cancer cells.
Different control experiments were performed to verify that the ELA release resulted from the photocleavage of the carbamate bond of CM-ELA, primarily induced by the SHG from the HNP core (Table 2). In addition, the synthesis of LNO nanocrystals was adjusted to produce bare LNO NPs of around 100 nm mean diameter, in order to increase the SHG emission intensity and resulting uncaging capacity. Following silanization and post-conjugation to DIBO-CM-ELA and DIBO-ELA, the resulting functionalized HNPs presented ELA loading values of 30 and 38 nmol/mg for LNO-CM-ELA and LNO-ELA NPs, respectively. First, these LNO derivatives were irradiated for 15 min at 790 nm, leading to 70.1 and 7.4% release of the cargo, respectively. Then, the release of ELA was quantified from LNO-CM-ELA NPs, kept in the dark for 15 min at 22°C (6.6% release), or heated at 50°C up to 30 min (8% release). Under NIR irradiation, the temperature increase induced by exposure to the laser excitation was estimated well below this value (see Supporting Information S40, Figure S10). The release from DIBO-CM-  ELA, non-conjugated to the surface of LNO NPs, was also evaluated. The cargo release resulting from direct two-photon absorption of the photocleavable coumarinyl linker contributed to 37% of the release with respect to conjugated LNO-CM-ELA NPs. This value is not surprisingly higher than what was previously observed for BFO NPs 45 as LNO has a smaller nonlinear optical efficiency, 34 which is compensated by better size and morphology dispersion properties, leading to more robust functionalization strategies.
As one can observe, only LNO-CM-ELA NPs, exposed to NIR irradiation, induced the release of ELA at a concentration prone to exert cytotoxic effects on cancer cells. These control experiments confirmed that the cargo release took place upon cleavage of the photosensitive linker when being exposed to the proper irradiation and that any potential temperature increase induced by the irradiation did not contribute to the uncaging process.
The release pattern of ELA upon NIR irradiation of functionalized NPs and the cargo concentration reached after 15 min of exposure to the laser source prompted us to further explore the ability of LNO-CM-ELA NPs to exert cytotoxic effect on DU145 cells, under similar conditions. The cells were incubated with LNO-CM-ELA NPs for 24 h and subsequently irradiated at 790 nm for 15 min. Cell viability was evaluated 72 h after the light exposure treatment ( Figure 5). Irradiation of DU-145 cells exposed to LNO-CM-ELA resulted in 66% inhibition of cell growth, an effect comparable to the growth inhibitory action induced by 0.98 μM ELA (Figure 3). The cell viability observed was slightly lower in the case of treatment with LNO-CM-ELA NPs as compared to incubation with ELA. This observation is consistent with the calculated concentration of ELA released from LNO-CM-ELA NPs at a concentration of 150 μg/mL, which was estimated to reach 1.6 μM. Importantly, no detrimental effect was observed for nonirradiated cells or for irradiated cells incubated with LNO-N 3 NPs. This result suggests that LNO-N 3 and LNO-CM-ELA NPs do not have intrinsic cytotoxic properties and that the release of ELA upon NIR irradiation from loaded HNPs can efficiently inhibit the growth of EGFR-overexpressing prostate cancer cells.
The uncaging of chemotherapeutic compounds from the surface of functionalized inorganic NPs through photocleavable linkers was already reported for other types of materials. 52 Most of the previous studies made use of UCNPs while HNPs represent more recent systems investigating such an approach. The most frequently reported drug model is doxorubicin, which induces much more aggressive effects than erlotinib as it causes cell apoptosis. 53 In particular, the NIRtriggered release of doxorubicin from UCNPs was presented by Wong et al. 54 and resulted in cancer cell growth inhibition comparable to the effect achieved with LNO-CM-ELA NPs. However, their protocol required longer irradiation time (60 min) and significantly higher concentration of functionalized UCNPs (750 μg/mL) to trigger efficient reduction in cell viability. To the best of our knowledge, the study herein presented is the first example of the successful NIR-triggered photo-release of a TK inhibitor such as erlotinib from surfacefunctionalized inorganic NPs.

■ CONCLUSIONS
Nonlinear optical parametric signals from nanoparticles such as harmonic generation and frequency mixing, present appealing properties for imaging applications at longer wavelengths complementary to those of other fluorescence or luminescence nanoparticle probes. 55 By means of advanced surface functionalization protocols, additional modalities, such as the controlled uncaging of molecular payloads, can be activated by simple tuning of the wavelength of the pulsed laser excitation. This procedure is automatic on a number of commercial laser systems nowadays. The combined results herein presented demonstrate the multi-harmonic imaging and NIR-triggered delivery of a chemotherapeutic compound targeting EGFRoverexpressing cancer cells. A cytotoxic compound derived from the EGFR TK inhibitor erlotinib was synthesized to allow further conjugation to the surface of LNO HNPs displaying surface-reactive azide functionalities, while preserving the growth inhibitory effect of the parent drug molecule on human prostate DU145 cancer cells. Caging with a photosensitive coumarinyl-based spacer, followed by covalent immobilization on coated LNO-N 3 NPs through copper-free azide for alkyne click [3 + 2] cycloaddition, provided functionalized HNPs at a loading rate of 27 nmol/mg LNO. The colloidal stability of LNP-CM-ELA NPs upon storage in  Cell viability was evaluated using MTT assay 72 h after exposure to the laser source (gray bars). The same protocol was performed, in parallel, without laser exposure (black bars). Results are expressed as the mean ± SD of three independent experiments. Differences between irradiated and non-irradiated samples and between cells exposed to LNO-N 3 and LNO-CM-ELA NPs were statistically analyzed by a two-way ANOVA test: **p < 0.01 and ***p < 0.001.  Figure S7).

Release of ELA upon NIR Femtosecond Pulsed Laser Irradiation
Irradiation Setup 1. In this case, the irradiation source is an amplified Ti:Sapphire laser system (Astrella, Coherent) with 4.9 W average output at a 1 KHz repetition rate. The system delivers pulses centered at 790 nm with 35 nm bandwidth corresponding to 27 fs transform-limited duration. The actual pulse duration measured using a commercial second harmonic generation frequency resolved optical gating (SHG-FROG) device (PulseCheck, APE Berlin) yields approximately 60 fs. The beam diameter, determined using a beam profiler (Newport) set at the amplifier output, corresponds to approximately 11 mm at 1/e 2 . The beam is directed on the bottom surface of the multi-well plate using a 45°dielectric mirror, without focusing. The peak intensity at the sample corresponds to 87 GW/ cm 2 at 1/e 2 (Supporting Information S39, Figure S8). LNO-CM-ELA NPs (150 μg) were suspended in PBS (1 mL; pH = 7.4, 144 mg/L KH 2 PO 4 , 9000 mg/L NaCl, and 795 mg/L Na 2 HPO 4 ·7H 2 O) and placed in eight-well borosilicate cover glass polystyrene chambers (Nunc Lab-Tek II, Merck). The samples were irradiated at 790 nm.
Irradiation Setup 2. The setup for comparing the release of ELA at different wavelengths is based on a 680−1300 nm tunable ultrafast system (InSight X3, Spectra-Physics) with 1.4 W average output power at an 80 MHz repetition rate. The nominal pulse duration is approximately 120 fs. The experiment was performed at 800, 1000, and 1100 nm. The pulse temporal compression was carefully adjusted to obtain Fourier-limited duration for each condition. The beam diameter, determined using a beam profiler (Newport) set at the laser output, corresponds to approximately 1.36 mm at 1/e 2 . The beam is directed on the bottom surface of the multi-well plate with a 45°d ielectric mirror, without focusing. The peak intensity at the sample corresponds to 10 MW/cm 2 at 1/e 2 . To account for the smaller beam size compared to the size of the well (12 mm), we used a rotational motor (PRM1-Z8�Thorlabs) to move continuously the sample with respect to the beam position in a circular pattern (2π/10 min). To partially compensate for the lower peak intensity with respect to setup 1, we applied the irradiation for a longer time span (1 h instead of 15 min).
Aliquots (40 μL) of the suspension were withdrawn at the indicated time points, diluted with PBS to 200 μL to achieve a 1:5 dilution ratio, and centrifuged (10 min, 13,000 rpm). Quantification of ELA in the supernatant (triplicates) was performed by UHPLC− MS. Details of the parameters for UHPLC−ESI−HRMS analysis and calibration curves are provided in Supporting Information (S35, Figure S7).

In Vitro Evaluation of the Cytotoxic Effect of Erlotinib and ELA
DU145 cells cultured in DMEM containing 10% of FCS were exposed to increasing concentration of either ELA or commercial erlotinib for 72 h at 37°C. Cells were then washed two times with PBS and incubated for 2 h with a solution of MTT in DMEM containing 10% FCS without phenol red. Cells were then lysed by incubating them for 15 min with isopropanol/HCl 0.04 M. After homogenization, absorbance was measured at 540 nm using an Epoch 2 microplate spectrophotometer (BioTek). Experiments were conducted in triplicates and repeated two times. Means ± standard deviation (SD) were calculated.

In Vitro Evaluation of the Cytotoxic Effect of the NIR-Triggered Release of ELA
LNO-CM-ELA NPs or LNO-N 3 NPs (10 mg/mL in EtOH) were diluted to 0−200 μg/mL with DMEM containing 10% of FCS and added to eight-well borosilicate cover glass polystyrene chambers (Nunc Lab-Tek II, Merck) containing DU145 cells in DMEM (400 μL). Cells were incubated with LNO-N 3 NPs (150 μg/mL), LNO-CM-ELA NPs (150 μg/mL), ELA (0.98 μM), or DMEM containing 1.5% EtOH as the control. After 24 h incubation at 37°C, the samples were irradiated at 790 nm for 15 min using the setup described above. The samples were incubated at 37°C for 72 h. Cells were washed with PBS and incubated with a solution of MTT in DMEM containing 10% FCS without phenol red (1 mM), for 2 h. The supernatant was discarded, and cells were lysed with isopropanol/HCl 40 mM. Absorbance at 540 nm was measured using an Epoch 2 microplate spectrophotometer (BioTek) and compared to the values of control cells incubated in only DMEM containing 1.5% EtOH. Experiments were conducted in triplicate wells and repeated three times. Means ± SD were calculated.
Cell images presented in Figure 2 were acquired on a Leica SP8 DIVE FALCON upright multiphoton microscope coupled with a tunable femtosecond laser (InSight X3, Newport Spectra-Physics) ACS Nanoscience Au pubs.acs.org/nanoau Article using a NA 0.95 water-immersion objective (Leica HC FLUOTAR L 25×). The acquisition spectral range for each channel was selected using a Leica 4Tune system. Images at each excitation wavelength (SHG and DAPI fluorescence at 800 nm and SHG and THG at 1250 nm) were acquired simultaneously. The gains in detectors were adjusted for each signal. Nanoparticles. The spectral response upon NIR excitation of LNO-CN-ELA NPs was performed on a different microscopy setup (Nikon MP) characterized by a narrower excitation tunability range (750−1000 nm) but capable of hyperspectral imaging over 32 independent detection channels. From a colloidal dispersion of the compound in EtOH (2 mg/mL), an aliquot of 2.5 μL was drop-cast on a microscope coverslip. After solvent evaporation, the sample was imaged on a Nikon multiphoton inverted microscope (A1R-MP) coupled with a 80 MHz Ti:Sapphire oscillator (Mai-Tai, Newport-Spectra-Physics) through a water immersion objective (Nikon, Plan Apo IR 60×, NA 1.27). The epi-collected signal was acquired using the Nikon spectral unit (max range 400−650 nm) with grating selection being set to 6 nm resolution.
Synthesis Protocols. Designation of the compounds refers to the chemical structures presented in Schemes 1 and 2. Details on the experimental protocols and instrumentation used for characterization are given in Supporting Information, S3.
Preparation of Compound 2. To a solution of 1 (10.02 g, 20.0 mmol, 1.0 equiv) in toluene (120 mL) were added DMF-DMA (10.6 mL, 80.0 mmol, 2.0 equiv) and AcOH (0.7 mL, 12.0 mmol, 0.3 equiv). The reaction mixture was refluxed for 2 h. The mixture was cooled to rt and concentrated under reduced pressure. The resulting oil was triturated with heptane and sonicated. The precipitate was filtered and washed with heptane to afford 2 as a pale-yellow solid (9.45 g, 30.9 mmol, 77% yield). 1 13 C NMR spectra, IR data (Supporting Information S17).