Activatable cell–biomaterial interfacing with photo-caged peptides

We report an effective strategy to design activatable cell–material interfacing systems enabling photo-modulated cellular entry of cargoes and cell adhesion towards surfaces.


Introduction
The recent explosive growth of research in the eld of nanotechnology has provided a wide range of novel materials and strategies for biomedicine, including important advances in bioimaging, drug delivery, photothermal/photodynamic therapy, and gene transfection. [1][2][3][4][5][6][7] However, further translation beyond basic research is heavily hampered by the inefficient performance of nanomaterials in biological environments. Amongst the main hurdles, the hydrophobic nature of the lipid bilayer of the plasma membrane renders it impermeable to most polar, hydrophilic molecules including peptides, proteins, oligonucleotides, drugs and nanomaterials that lack specic membrane receptors or transport mechanisms. Cellpenetrating peptides (CPPs) such as the HIV transactivator of transcription (TAT) peptide and arginine oligomer proteintransduction domain have been widely used as transport vector tools for the cellular import of a variety of cargos (e.g., nanomaterials and biomolecules) through the cell membrane. 8,9 In addition, new cell-penetrating transporters have been developed such as synthetic peptides, 10,11 helical poly(arginine) mimics, 12 antibiotics, 13 poly-(disulde)s 14,15 and guanidinium-containing synthetic polymers. 16,17 In spite of their potential, the inherent non-specicity of CPPs has restricted their application in targeted delivery systems. An alternative strategy to selectively enhance cell-material interfacing lies in the design of smart systems that are triggered by external stimuli, or specic features of the target cell or local tissue physiology. For example, the development of acidactivated CPPs takes advantage of the acidic tumour extracellular environment 18,19 for tumour-targeted drug delivery. [20][21][22] Among these, pH-induced charge reversal and pH-sensitive transmembrane insertion of a low-pH insertion peptide have also been explored as novel strategies to increase the efficacy of biomaterial administration. 19 Similarly, the design of hydrogen peroxide (H 2 O 2 ) and matrix metalloproteinase-2 (MMP2) responsive cellular delivery systems have been reported. 23,24 Key attributes of such smart delivery systems are their ability to control cellular entry of biomolecules/nanomaterials and their release on demand, minimize side effects and improve the therapeutic efficacy of pharmaceuticals.
In this work, we developed a stimuli-responsive cell-material interfacing system enabling the spatial and temporal control of cellular delivery and cell attachment via photo-activation. Photo-stimulation of biomaterials is advantageous since it can be manipulated precisely, enabling ne control over irradiated volumes. [25][26][27][28][29][30][31][32][33] In the presence of photo-stimulation, it is possible to rapidly increase the concentration of the active form of molecules with strict control over the illuminated area, time, and dosage. [34][35][36][37][38] In our system, we synthesized a series of photocaged peptide ligands consisting of a cell penetrating sequence, a blocking sequence, and a photo-cleavable linker (Scheme 1), and conjugated them to various nanomaterials or surfaces. The cell-material interaction is effectively suppressed when there is no light activation, resulting in minimal cargo uptake. Upon photo-irradiation, the linker is cleaved to remove the blocking peptide, resulting in activation of the cell penetrating peptide and increased cellular uptake. This phenomenon was demonstrated with a wide range of cargos including small and large molecules, organic and inorganic nanoparticles, and selfassembled so materials.

Results and discussion
As shown in Fig. 1a, we synthesised a photo-labile ligand consisting of hepta-arginine (R7), hepta-glutamic acid (E) and a photo-linker. The positive R7 serves as the cell penetrating sequence capable of delivering cargos to cells either by direct membrane translocation or promoting endocytosis. 39 The proximal glutamic acid-rich sequence (E7) blocks the oligoarginine by electrostatic attraction, forming a U-shaped antifouling zwitterionic ligand. 40 The cell penetrating peptide (R7) and the blocking sequence (E7) are connected with a photocleavable linker, which can be cleaved at the position of the onitrophenyl group to yield two peptide chains terminated with ketone and amide groups (Fig. 1a). We monitored the photocleavage reaction with UV-vis spectroscopy over a period of 15 min (Fig. 1b). Without UV light irradiation, there are two absorption peaks at approximately 305 nm and 350 nm. Upon UV irradiation, the absorbance at 305 nm decreased and the peak intensity at 350 nm increased, while a shoulder peak emerged at 370-400 nm. Since arginine and glutamic acid do not have an absorption peak between 300 nm and 400 nm, these spectral changes originate from the photo-cleavage of the o-nitrophenyl group. The isosbestic point at 317 nm indicates no side reactions or decomposition in the process of photo-irradiation. The reaction reached a photostationary state aer 10 min, as the absorbance at 390 nm approached a plateau (Fig. 1c). Because the electrostatic interactions between heptaglutamic acid (E7) and hepta-arginine (R7) are strong due to multiple charge pairs, it is important to determine whether the negative blocking peptide will electrostatically bind to the cell penetrating peptide even when the covalent spacer is cleaved. To test this phenomenon, we developed a FRET system using a uorescein isothiocyanate-labelled, thiolated peptide (HS-R7E7-FITC) and citrate-coated gold nanoparticles ($40 nm) ( Fig. S1 † and 1d). The FITC-appended peptide was covalently bound to AuNPs via the Au-thiol bond, and the close proximity between FITC and AuNPs enabled efficient energy transfer (ET) from FITC to AuNPs, resulting in quenched FITC uorescence. Upon exposure to UV light, FITC uorescence increased with irradiation time (Fig. 1e), suggesting that the FITC-conjugated Scheme 1 (a) Schematic of light-activated cellular uptake of biomolecules and nanomaterials. (b) Light-responsive delivery systems are functionalised with a synthetic ligand consisting of a negative blocking sequence, a light-cleavable sequence, and a cell penetrating peptide (CPP). blocking peptide (E7) was released from the hepta-arginine sequence following cleavage of the nitrophenyl group.
Since the blocking peptide could be removed with 365 nm UV light, we anticipated that the bioactivity of CPP (R7) could be enhanced through controlled light irradiation. To test this idea, we synthesised a uorescein isothiocyanate-labelled photo-activatable peptide (FITC-R7E7) with FITC attached to the N-terminus (Fig. S1 † and 2a). Peptide solutions at different concentrations (100, 50, 20 and 10 mM) were either exposed to light or le unexposed and incubated with MDA-MB-231 (MDA) cells. Aer 24 h, we evaluated the level of internalized uorophores by measuring the uorescence intensity of cell samples. We found that photo-irradiation enhanced the cellular uptake by 1.1 to 3-fold for FITC-R7E7 concentrations between 10 and 100 mM, respectively (Fig. 2b). Fluorescence activated cell sorting (FACS) conrmed that cells incubated with the activated peptide showed a higher uorescence intensity than those incubated with the native construct ( Fig. 2c). In addition, confocal microscopy showed the subcellular localization of the uorophore, by staining the cell membrane with WGA (wheat germ agglutinin) and the nucleus with DAPI (4 0 ,6-diamidino-2-phenylindole, dihydrochloride). MDA cells incubated with the non-activated peptides exhibited a weak and scattered green uorescence signal ( Fig. 2d-f, arrowheads), whereas cells incubated with light-activated FITC-R7E7 ( Fig. 2g-i) exhibited high penetration of the uorophore, with a preferential perinuclear localization (co-localization with DAPI; see orthogonal views in Fig. 2h and i). In a control experiment, we demonstrated that light irradiation did not cause any obvious uorescence changes in the FITC-labelled peptide (Fig. S2 †). Similarly, we demonstrated that light irradiation enhanced cellular internalization of the uorophore with HeLa cells (Fig. S3 †). These results show that although neutral peptide-bound FITC can be taken up by the cells to some degree, light-triggered cleavage of the blocking sequence signicantly increases the quantity that translocates into MDA cells.
We further applied this strategy to control the intracellular delivery of large biomolecules. We used avidin, a di/tetrameric biotin-binding protein, 66-69 kDa in size, as a model macromolecule to test the possibility of light-activated protein delivery. To this end, we conjugated FITC-labelled avidin with 4 equivalents of the biotinylated photo-labile ligand (biotin-R7E7) via biotin-avidin affinity and incubated with MDA cells (Fig. S1 †). Aer 24 h, cell uorescence increased by up to 200% for FITC-avidin concentrations of 2.5 and 1.25 mM (Fig. S4 †), highlighting the feasibility of this strategy to deliver full-length proteins.
We subsequently applied the light-activated peptide to control intracellular entry of inorganic nanoparticles. As an example, we coupled the cleavable peptide ligand to semiconducting CdSe/ZnS QDs. Compared to organic uorophores, inorganic QDs possess a number of superior properties such as bright emission, high stability, reduced photobleaching and ease of surface functionalization. 41 For example, the biotinylated photo-cleavable ligand was attached to streptavidincoated QDs (5 nm in diameter, l em ¼ 605 nm) via streptavidin-biotin affinity interactions ( Fig. 3a and b). As shown in Fig. 3c, light irradiation increased the cellular uptake of QDs to 1015% and 573% when using 40 and 20 nM QDs, respectively. This increase in uorescence is not attributed to enhanced QD emission aer light irradiation (Fig. S5 †). It is noteworthy that the light-triggered increase in QD uptake was strikingly higher than that of FITC-avidin as shown in Fig. S4, † which is ascribed to the high number of biotin-binding sites on QDs. According to the vendor, there are 5-10 covalently bound streptavidins per QD, permitting the attachment of 20-40 biotin-R7E7 molecules on a single QD. QD uptake by MDA cells was examined with confocal microscopy ( Fig. 3d-i), where non-activated QDs were poorly taken up ( Fig. 3d-f, arrowheads), and strong uptake was observed for cells incubated with the light-activated QD-peptide ( Fig. 3g-i). To conrm the functionality of our system in a biologically relevant environment, we incubated the QD-R7E7 complexes with MDA cells and irradiated them in situ. The results were consistent with our previous observations, where FACS indicated a strong uorescence signal per cell (Fig. S6 †), indicating increased uptake of the activated QDs. There was however a notable difference in the uptake pattern between activated QD/biotin-R7E7 and activated FITC-R7E7, where cells showed a punctate and mainly cytosolic distribution of QDs in contrast with the preferentially nuclear distribution of FITC. The reasons behind this are unclear, but it is reasonable to hypothesize that differences in the cargo size and CPP concentration could affect whether the QDs are primarily taken up via endocytosis or direct membrane translocation. 39 Our light-cleavable ligands have broad applicability for improving intracellular uptake of a variety of nanomaterials, which can be bound to the peptide using well-established conjugation chemistry (Table 1). For example, carboxylated polystyrene particles modied with an aminated ligand (H 2 N-R7E7, Fig. S1 †) displayed enhanced cellular uptake aer light activation (Fig. S7 †). Branched Au nanostars with a localised surface plasmon resonance peak at 820 nm in the infrared range (so-called "biological window") were internalized with a greater efficiency following light irradiation (Fig. S8 †). The Au nanostars were conjugated via thiolated ligands (HS-R7E7). Similarly, we demonstrated light-enhanced delivery of lipid vesicles ($200 nm size) by conjugating HS-R7E7 via the thiol-maleimide reaction (Fig. S9 †). In all cases, light-triggered cleavage of the nitrophenyl group and subsequent removal of the blocking peptide greatly promoted the cellular uptake of nanomaterials.
Light-activated cellular uptake strategies enable targeted delivery of nanomaterials to diseased tissue, which could minimize off-target side effects and also potentially decrease the dosage administered to a patient. In this work, we prepared phospholipid-coated polylactic-co-glycolic acid (PLGA) nanoparticles as a drug carrier for camptothecin, a cytotoxic alkaloid employed in cancer therapy. However, this drug has poor water solubility which limits its efficacy. 42 PLGA particles are attractive polymer-based delivery systems because of their biocompatibility, efficient encapsulation of hydrophobic materials, and tunable drug release properties. We synthesized PLGA particles via emulsion/solvent evaporation methods, using lecithin and DSPE-PEG-NHS as surfactants. We further conjugated a lightactivatable peptide ligand to PLGA particles by reacting an aminated peptide (H 2 N-R7E7) with DSPE-PEG-NHS. As shown in Fig. 4a, c and d, exposing drug-loaded polymer particles to 365 nm UV light resulted in enhanced MDA cell death, as demonstrated by the statistically signicant decrease in alamarBlue® reduction activity. A similar trend was observed for HeLa cells (Fig. 4b, e and f), and it is likely that the difference in the magnitude of effect was determined from differential entry propensities and drug sensitivities between the two cell lines.
We further demonstrated the potential for activating cellnanomaterial interactions with two-photon excitation, where a high power pulsed laser with a very short pulse was applied in confocal microscopy, providing the potential for deep tissue manipulation. The high photon density in the focused region enhances the probability that a chemical group/molecule absorbs two photons quasi-simultaneously. To this end, we incorporated an o-nitrobenzyl ether moiety into the peptide chain between the blocking sequence and cell-penetrating sequence ( Fig. 5a and S1 †) and attached this peptide to the surface of streptavidin-coated CdSe/ZnS QDs via biotin-streptavidin affinity. Compared to the amide bond (Fig. 1a), the ester bond is more sensitive to light irradiation and gives a higher quantum yield of photocleavage. It was previously demonstrated that the irreversible photocleavage of an o-nitrobenzyl ether moiety into nitroso-and acid-terminated by-products via two-photon irradiation could be used to dynamically control the properties of 3D hydrogels with photocleavable crosslinkers. 43 Here, we incubated the peptide-modied QDs (20 nM) with  HeLa cells encapsulated in the 3D matrix of 8-arm polyethylene glycol (PEG) hydrogels (Fig. 5b). We prepared the hydrogels with 8% w/v 8-arm PEG acrylate, 5 mM RGD peptide (CGGRGDSP), and 6 mM PEG dithiol crosslinker, where the RGD peptides were used to promote cell viability and adhesion to the hydrogel network. To precisely locate viable cells within the 3D hydrogel, HeLa cells expressing green uorescent protein (GFP) were used. As shown in Fig. 5c-h, two-photon excitation (740 nm) was applied within precisely dened 3-dimensional regions of the hydrogel, resulting in increased intracellular red uorescence indicating the signicant enhancement of QD uptake into HeLa cells.
The application of CPPs is usually hampered by their poor selectivity towards target cells. We demonstrated enhanced uptake of photo-activated biomaterial-conjugated CPPs, revealing their great potential for controlling delivery of small molecule drugs, biomolecules and nanomaterials. Uptake can be controlled both spatially and temporally, potentially reducing side effects of toxic drugs and increasing the effectiveness of therapies. Moreover, stabilising and neutralising positively charged CPP sequences with anionic peptides until controlled photo-activation occurs could enhance the circulation time in the blood stream. By using precisely controlled and deep tissue penetrating two-photon excitation, we believe that this system will be applicable to in vivo studies providing 3D resolution for biomaterial delivery.
Switchable biological surfaces have important applications in cell-based diagnostics and tissue engineering. 44,45 The phototriggered conversion from a neutral zwitterionic ligand to  a highly positive peptide can be exploited to construct a photoresponsive surface with tuneable cell binding capacity. Hydrophilic zwitterionic peptides have been established as effective antifouling agents, 46 whereas a positive surface (e.g., polylysine) oen favours cell attachment. Here, we conjugate HS-R7E7 to an Au surface (Fig. 6a), and signicant increases in cell attachment to the light-activated surfaces were observed compared to nonactivated surfaces (Fig. 6b-k). Moreover, cells plated on the photo-activated Au surface could spread and remain attached, as depicted by their spread morphology and the relative increase in cell density aer 24 h, yet those on non-activated surfaces were unable to do so, as indicated by their rounded shape and low cell density. Therefore, our strategy can be utilized to controllably switch the surface properties of a material towards supporting cell attachment and growth, which could be used to prepare patterned biological surfaces with photomasks that regulate cell patterning and migration. 44,47,48

Conclusions
In conclusion, we report a general strategy to develop photoswitchable cell-biomaterial interfaces with a new class of photo-labile ligands conjugated onto different biomolecules (e.g., FITC and avidin) and nanomaterials (e.g., quantum dots, polystyrene particles, Au nanostars, and liposomes). We demonstrated enhanced cellular uptake of cargoes by light activation up to 9-fold, as shown for QDs. We also applied our system to demonstrate controllable cancer drug delivery and showed the possibility to engineer smart biointerfaces with tuneable cell attachment/growth properties controlled by light irradiation. The photo-reaction is highly cytocompatible (Fig. S10 †), facilitating a wide range of new technologies to regulate cell-material interactions, and control cell attachment, migration and culture in a 3D scaffold, and pave the way towards novel and exciting translatable applications. In particular, incorporating NIR-responsive crosslinkers (Fig. 5) 33,43 and employing spatiotemporally controlled two-photon excitation will address the issues of low penetration depth and the absorption of UV-light in biological tissues.  obtained from IRIS Biotech GmbH. A Qdot 605 ITK Streptavidin Conjugate Kit was obtained from Life Technologies (U.K.). Carboxylate-modied polystyrene nanoparticles (Latex beads, 20 nm, l ex $470 nm, l em $505 nm) were obtained from Sigma-Aldrich (U.K.). Lipids were purchased from Avanti Polar Lipids (Alabaster, AL). DSPE-PEG-NHS was purchased from Nanocs. 8arm PEG (40 000 MW) was purchased from Jenkem (U.S.A.). All the other chemicals were purchased from Sigma-Aldrich (U.K.) and used without further purication.

Solid phase peptide synthesis (SPPS)
Peptides were synthesized manually using standard uorenylmethoxycarbonyl (Fmoc) chemistry protocols. The Fmoc protecting group was removed with 20% piperidine/DMF. Fmoc-protected amino acids were activated with 4 molar equivalents of the Fmoc-protected amino acids, 3.95 molar equivalents of HBTU, and 6 molar equivalents of DIEA in DMF. The coupling solution was added to the resin, and the coupling reaction was allowed to proceed for two to three hours. Kaiser tests were performed aer each Fmoc deprotection and coupling step to monitor the presence of free amines. Peptides were cleaved in the mixture of triuoroacetic acid/triisopropylsilane/H 2 O (95 : 2.5 : 2.5) for four hours. The peptides were puried using reversed phase preparative high performance liquid chromatography (HPLC; Shimadzu) in an acetonitrile/ water gradient under acidic conditions on a Phenomenex C18 Gemini NX column (5 micron pore size, 110Å particle size, 150 Â 21.2 mm). The puried peptide mass was veried by matrix-assisted laser desorption spectroscopy (MALDI; Waters).

Light irradiation
The light-irradiation experiments were conducted using a UV lamp (UVLMS-38 EL Series 3 UV™, 365 nm, 1.3 mW cm À2 ) with a distance of 10 cm between the sample and the lamp. To test the light-responsiveness of peptides, 0.1 mM peptide was dissolved in phosphate buffer (10 mM, pH 7.0). The solution was exposed to UV light and the UV-vis spectra were recorded every 30 seconds.

Synthesis of Au nanostars
Gold nanostars were prepared using a seeded, HEPES/ hydroxylamine reduction approach similar to that reported by Maiorano et al. with minor adjustments. Briey, a solution, consisting of 38.5 mL Milli-Q water, 18.75 mL of 100 mM HEPES buffer at pH 9.6, 750 mL of fresh 40 mM hydroxylamine and 750 mL of the as-prepared citrate gold nanoparticle seeds, was prepared in a 100 mL conical ask. A sufficiently large magnetic stirrer bar was used to achieve effective mixing at a stirring rate of 1450 rpm. To this rapidly mixing solution, 22.5 mL of 1 mM HAuCl 4 $3H 2 O was added at a rate of 2 drops per second. Upon complete addition of the HAuCl 4 $3H 2 O, the stirring speed was reduced to 400 rpm, and the stirring was continued for a further 15 minutes, and subsequently, 80 mL of 1 wt% aqueous solution of Tween-20 surfactant was added. The particles were centrifuged at 4000 rpm for 30 minutes for 3 cycles including brief sonication to assist resuspension such that the nal concentration of Tween-20 was approximately 0.0001 wt%. The particles were then passed through a 0.2 mm PES lter membrane. The peak maximum of the longitudinal plasmon absorption was at 820 nm, and the nal concentration of gold [Au] ¼ 2.6 mM.
Synthesis of 13 nm citrate-capped gold nanoparticle seeds 180 mL of Milli-Q water in a 250 mL round-bottom ask (2-neck) was brought to reux in an oil bath and stirred gently. A solution of 79 mg of HAuCl 4 $3H 2 O in 5 mL of Milli-Q water was then added, and the stirring speed was increased as high as possible. Within 5 minutes, a 10 mL solution of 2 wt% trisodium citrate dihydrate in Milli-Q water (pre-heated to 70 C) was injected rapidly. Reux was continued for a further 5 minutes with fast stirring before being removed from the oil bath and allowed to cool to room temperature. The particles were then allowed to cool and stored at 4 C prior to use.

Surface functionalization of Au nanostars
To functionalize Au nanostars, 10 mL of 1.0 mM HS-PEG2K and 50 mL of 0.5 mM HS-R7E7 were added to 2 mL Au nanostar solution (5.5 mM). The mixture was incubated at room temperature overnight. Aer centrifugation at 10 000 rpm for 10 minutes, the supernatant was discarded, and the pellet was redispersed in phosphate buffer.

Surface functionalization of quantum dots (QDs)
Peptide conjugation on CdTe quantum dots was achieved via biotin-streptavidin chemistry. Briey, 10 mL of commercial streptavidin-coated CdTe QDs (4 mM) was incubated with 1.0 mL 10 mM biotinylated photo-cleavable peptide (biotin-R7E7) for 1.0 hour. The solution was centrifuged at 32 490 rpm for 2 hours with an ultracentrifuge to remove the excess unbound peptides. The supernatant was discarded, and the QDs were re-dispersed in phosphate buffer.

Surface modication of uorescent polystyrene (PS) nanoparticles
Typically, 80 mL of carboxylic acid-modied polystyrene particles (20 nm, 25 mg mL À1 ) were dispersed in 2 mL MES buffer (pH 5.7, 20 mM). Aer that, 20 mg of EDC and 20 mg of NHS were added to activate the carboxyl group. The mixture was incubated for 2 hours at room temperature before the addition of NH 2 -R7E7 (2 mg in 50 mM phosphate buffer, pH 8.0). Aer incubation overnight, PS particles were centrifuged, and the suspension was discarded to remove excess chemical reagents. The PS particles were washed with PBS again before the cell experiments.

Surface modication of Au plate
The Au surface was rinsed with ethanol three times and dried in air. Aer that, 1 mg mL À1 of HS-R7E7 solution was applied to the Au surface and incubated for 5 hours. The Au surface was then washed with water and ethanol several times. The Au surface was further incubated with a HS-PEG1000 solution (1 mM) to block the surface.
Characterization UV-Vis spectra were recorded with a Lambda 25 spectrometer (Perkin Elmer). Fluorescence spectra were recorded with a Fluorolog uorometer (Horiba). The uorescence measurements of cell experiments were conducted with a SpectraMax M5 plate reader.

Cell culture
All cell culture reagents were from Thermo Fisher Scientic (Loughborough, UK) unless otherwise stated. MDA-MB-231 cells were purchased from ATCC (Teddington, UK), HeLa cells from DSMZ (Braunschweig, Germany), and GFP-HeLa cells from Cell Biolabs, Inc. (San Diego, CA, USA). Cell lines were cultured under standard conditions (37 C and 5% CO 2 ) in DMEM supplemented with 10 v/v% fetal bovine serum and 1 v/v% penicillin streptomycin and split using trypsin-EDTA upon conuence. For uptake experiments (FITC, avidin, QDs, PS, Au nanostars, liposomes and camptothecin-loaded PLGA nanoparticles), cells were plated in 96 well plates at 10 000 cells per well and le to attach overnight. The next day, the medium was replaced by the corresponding constructs diluted in PBS : medium (1 : 1) and again incubated overnight. Aer 24 hours, the medium was discarded and cells were gently washed with PBS 3 times before uorescence intensity was measured or the alamarBlue® test was performed (PLGA nanoparticles).
For live activation of QD-R7E7 complexes, MDA-MB-231 cells were incubated with 20 nM QDs and irradiated for 0, 5 and 15 minutes. 24 hours later, epiuorescence images were taken, and cells were trypsinized and analyzed by FACS.
For Au attachment experiments, cells were detached using 0.05 w/v% trypsin-EDTA, counted and plated at 100 000 cells per mL in DMEM without supplements to avoid interference from the charged molecules in the serum. 30 minutes aer plating, the medium was removed and the samples were washed with PBS. Cell attachment was evaluated aer staining with calcein AM, and 4-6 pictures per sample were taken randomly. The substrates were fed with complete medium and incubated overnight. Aer 24 hours, the samples were again stained with calcein AM, and the same number of images was acquired.
Viability of the cultures aer incubation with the different nanomaterials was evaluated using either LIVE/DEAD staining or alamarBlue®, both following the manufacturer's instructions.

Fluorescence-activated cell sorting (FACS)
For FACS experiments, cells were plated in a 12 well plate at an equivalent density to the uptake experiments and le to attach overnight. The next day, they were incubated with the different compounds or nanomaterials for 24 hours and detached with 0.05 w/v% trypsin-EDTA. Non-treated cells were used as a control. Fluorescence was measured using a Fortessa cytometer (BD, Oxford, UK).

Cell encapsulation in PEG hydrogels
15 mm diameter glass coverslips were thiol-functionalized with 3-mercaptopropyl trimethoxysilane in acetone for 10 minutes, rinsed in acetone, heated at 80 C for 10 minutes, cooled and stored at À20 C. Cell-laden hydrogels were prepared with 8% w/v 8-arm PEG acrylate, 5 mM RGD peptide (CGGRGDSP), 6 mM PEG dithiol crosslinker (MW 1000), and 1 Â 10 6 cells per mL in DMEM with 20 mM HEPES, pipetted into silicone moulds with 6 mm diameter and 500 mm thickness, on top of thiolfunctionalized glass coverslips, and covered with Rain-x treated glass coverslips. Aer 12 minutes, gelation was complete, and individual gels attached to 15 mm coverslips were transferred to 24-well plates with the cell culture medium and cultured for 2 days before 2-photon photoactivation experiments.

Two-photon photoactivation
Cell-laden hydrogels were incubated with 20 nM QD-peptide in the culture medium for 3 hours, and several 3-dimensional regions of interest at the center of the cell-laden gels (300 Â 300 mm x-y, 200 mm in z with 5 mm z-spacing) were selectively exposed to multiphoton pulsed laser light (740 nm, 40 Â 0.8 NA water immersion objective, 16 mW average power at the objective) to photo cleave the QD-bound peptide, using an upright multiphoton confocal microscope (Scientica, Uckeld, UK). Hydrogels were incubated overnight to allow for cellular uptake of the photoactivated QD-peptide.

Microscopy
For conventional uorescence microscopy, cells were imaged live using an IX51 epiuorescence inverted microscope (Olympus, Southend-on-Sea, UK). For confocal microscopy, cells were plated on glass bottom microslide chambers (Ibidi, Glasgow, UK) and treated as above. All washes were performed with PBS. Aer incubation, cells or cell-laden hydrogels were washed and xed for 15 minutes in 4 w/v% paraformaldehyde, washed again in PBS and stained with WGA (where indicated, conjugated to 488-AlexaFluor for FITC-R7E7 and PS-R7E7 or 594-AlexaFluor for QD-biotin-R7E7) for 15 minutes at room temperature. Samples were then washed and incubated with DAPI for 5 minutes, washed and mounted with a Vectashield (Vector Laboratories, Peterborough, UK). Cell-laden hydrogels were mounted with a FluorSave (Calbiochem USA). Imaging was performed using a SP5 MP/FLIM inverted confocal microscope (Leica Microsystems, Milton Keynes, UK). For hydrogel experiments, GFP-positive cellular uptake of QDs was compared at the gel center (photoactivation region) versus gel edge (receiving no activation).

Statistics
Results are presented as mean AE standard deviation unless specied otherwise. Statistical analysis was performed using SPSS 22 and Prism soware. Distributions were assumed normal, and differences were analyzed using the Student's ttest. Differences were considered statistically signicant when p < 0.05 (*) and very signicant when p < 0.01 (**).

Conflicts of interest
There are no conicts to declare.