Selective photothermal killing of cancer cells using LED-activated nucleus targeting fluorescent carbon dots

The development of effective theranostic probes in cancer therapy is hampered due to issues with selectivity and off-target toxicity. We report the selective LED-photothermal ablation of cervical (HeLa) cancer cells over human dermal fibroblasts (HDF) using a new class of green-emissive fluorescent carbon dots (FCDs). The FCDs can be easily prepared in one pot using cheap and commercial starting materials. Physico-chemical characterization revealed that a surface coating of 2,5-deoxyfructosazine on a robust amorphous core gives rise to the nanomaterial's unique properties. We show that intracellular uptake mostly involves passive mechanisms in combination with intracellular DNA interactions to target the nucleus and that cancer cell selective killing is likely due to an increase in intracellular temperature in combination with ATP depletion, which is not observed upon exposure to either the “naked” core FCDs or the surface components individually. The selectivity of these nanoprobes and the lack of apparent production of toxic metabolic byproducts make these new nanomaterials promising agents in cancer therapy.

Photothermal therapy (PTT) is a promising non-invasive therapeutic strategy, in which nanoparticles embedded within tumors generate heat, typically in response to exogenously applied light, for thermal ablation of cancer cells. 1-3 PTT offers unique advantages in cancer therapy including high specicity, minimal invasiveness and precise spatial-temporal selectivity. 4 A variety of photothermal nanotherapeutics (PTN) including noble metal nanostructures, nanocarbons, carbon dots, transition metal sulde/oxide nanomaterials, and organic nano-agents have been extensively explored with encouraging results. 2, [5][6][7][8][9] However, one of the major challenges in PTT is the ability to selectively and efficiently nd a biocompatible material that can target cancerous cells while avoiding non-specic heating of healthy cells upon irradiation and that does not generate toxic byproducts. 5,10 Successful cancer therapy relies on early and accurate diagnosis, and uorescence imaging has increasingly been recognized as a viable strategy to identify a range of cancers. 11,12 Fluorescent probes that exhibit high stability, sensitivity and specicity for their target without the limitations of organic uorophores and uorescent proteins are of great interest in many areas of research, particularly in cellular biology, bioimaging and medical diagnostics. 13,14 Nanomaterials with novel optical, electronic and surface properties, have become useful platforms for a myriad of applications including imaging, drug delivery and diagnostics. [15][16][17] Fluorescent carbon dots (FCDs) [18][19][20][21][22] have emerged as a new class of non-isotopic detection labels suitable for live cell imaging, that provide a non-toxic alternative to heavy metal-containing uorescent nanomaterials (e.g. quantum dots) while offering many advantages with regard to their ease and low cost of synthesis, unique photoluminescence (PL) properties, chemical inertness, high water solubility and generally low cytotoxicity. [23][24][25][26][27][28][29] Previous work by our group [28][29][30][31][32][33][34] and others 35-38 on the use of water soluble uorescent probes for live cell imaging applications has shown that the nanoparticle type, size, shape and surface functionalization have a signicant effect on their intracellular uptake and localization. Encouraged by this, we embarked on the synthesis of a new class of uorescent carbonbased nanomaterials (FCD-3) with unique surface functionality that could be used to target cancer cells and have potential applications in PTT.
Initial efforts were devoted to preparing a green-emitting nanomaterial that could be more easily imaged directly in live cells using confocal microscopy than the more common blueemitting FCDs. 23,26,[39][40][41][42] The synthesis of FCD-3 was thus achieved in one pot aer three minutes of microwave heating (domestic 700 W MW) of glucosamine$HCl 1 and m-phenylenediamine 2 (Scheme 1A, see the ESI † for experimental details). 43 Centrifugal ltration (10 kDa molecular weight cutoff) afforded monodisperse quasi-spherical FCD-3. The nanoparticles exhibited an average diameter of 2.42 AE 0.55 nm (Fig. 1A) and an amorphous, carbonaceous core as determined by high-resolution transmission electron microscopy (HR-TEM, Fig. S2 †). Dynamic light scattering (DLS) indicated a hydrodynamic diameter of 6-9 nm (Fig. S1A †), while the zeta (electrokinetic) potential indicated a cationic surface of around 12 mV (Fig. S1B †). Thermogravimetric analysis (TGA) showed good core stability at high temperatures, while residual solvent and surface-adhered species were more loosely associated (95% of the mass of FCD-3 was retained when heating to 150 C, while a 65% mass loss was recorded when reaching 650 C, see Fig. S3 †). Fluorescence spectroscopy showed an excitationindependent emission for FCD-3 centered around 525 nm (l ex ¼ 350-500 nm, Fig. 1B), indicative of emissive center uniformity. The quantum yield for FCD-3 was calculated to be 33% relative to uorescein (Fig. S4 in the ESI †) and shown to be photostable upon continuous 60 min irradiation with no signicant decrease in emission intensity (l ex ¼ 460 nm, l em ¼ 525 nm, Fig. S5 and S6 †). 44 The photoluminescence (PL) properties remained unchanged in the presence of a different range of potential metal and organic quenchers (500 mM), conrming the stability of FCD-3 (see Fig. S7 †).
To ascertain the chemical composition of FCD-3, functional group analysis was conducted using Fourier-transformed infrared (FTIR) spectroscopy, X-ray photoelectron spectroscopy (XPS) and UV-vis spectroscopy. The FTIR signal at 3338 cm À1 was attributed to either residual carbohydrate architecture (OH groups) or N-H bonds (Fig. S8 †). Surface passivation was evidenced by the IR stretching at 1629 cm À1 (C]O amide) and minor peaks at 1016 cm À1 (C-O or C-N functionality), while the broad signal at 612 cm À1 indicated C-Cl bonds. The presence of mainly C, H, N, Cl and O in FCD-3 was also conrmed by elemental and XPS analysis (Table S1 and 45,46 NMR analysis was used to shed further light on the molecular architectures found on FCD-3 (Fig. S12 †). Pyrazine motifs were identied (three singlets at d 8.51, 8.30 and 5.01 ppm), 47 as probed by 1 H, 1 H-13 C HSQC and HMBC experiments, in addition to other complex N-hetero/aromatics akin to fructosazine and likely generated from 1,2-aminoaldose self-dimerisation 48 ( Fig. S13 and S14 †). Polyhydroxylated species (d 2.50-4.00 ppm), arising from incomplete glycan dehydration, 49 were also detected. Time-point 1 H and 13 C NMR analysis of reaction aliquots carried out at 30 s intervals up to 180 s ( Fig. S15 and S16 in the ESI †) showed loss of sugar anomeric signals from the starting carbohydrate in the spectra, indicative of pyranose ring-opening and iminium formation, followed by dehydration. 29 Other aromatic structures were also detected in the spectra (d 6.00-7.50 ppm) suggesting the presence of complex polyaromatics. Indeed, 1 H-13 C HMBC analysis demonstrated the presence of sp 2 and sp 3 -enriched surface domains.
Since nanoparticle surface composition is important for biological recognition, additional efforts were made to fully identify the surface molecular structures found on FCD-3. The material was thus subjected to Sephadex G15 size exclusion  This journal is © The Royal Society of Chemistry 2019 chromatography using methanol as the polar solvent to help separate non-covalently bound surface species from the core-FCD. In this manner, green uorescent 2,5-deoxyfructosazine 4, which is a known product of glucosamine self-condensation, 50,51 was isolated as the major component in addition to core FCD-5 (FCD without 4) and FCD-3, as determined by 1 H-NMR of the fractions (Fig. S17-S20 † and Scheme 2). 52 Zeta potential measurements of core 5 (À1.37) vs. 3 (12.05) showed a change from a slightly negative net charge to a cationic charge when 4 is conjugated to the naked core, as expected for a passivated surface. Moreover, diffusion ordered spectroscopy (DOSY) of FCD-3 showed that it had a single diffusion coefficient of 3.83 Â 10 À6 cm 2 s À1 conrming that 4 is associated with the FCD-core in FCD-3 (Fig. S20 †).
The utility of FCD-3 as a bioimaging tool was then investigated in HeLa cells (cervical cancer) and human dermal broblasts (HDF). It was found that FCD-3 (50 mg mL À1 ) was internalized in both cell lines, although the rate of uptake and intracellular localization differed between cell lines. Both active and passive cell transport mechanisms (37 C vs. 4 C) were observed, the latter accounting for 67% of uorescence in HeLa, (Fig. S21 †) but only 20% in HDF ( Fig. S24 and S25 †). Interestingly, very rapid nuclear internalization of FCD-3 was observed for HeLa cells within 1 min (Fig. S22 †), with maximum internalization at 5 min aer which nuclear uorescence decreased slightly due to increased cytosolic accumulation which plateaued aer 24 h with an average Global Pearson Coefficient (GPC) ¼ 0.62 at 2 h exposure ( Fig. 2 and 3). No signicant colocalization of FCD-3 was observed for any other organelle (for representative images see Fig. S29-S36 †). This result is signicant as previous glucosamine-based blue-emitting FCDs 26,29,33 did not target the cell nucleus. In contrast, much slower overall cellular uptake of FCD-3 was detected for healthy HDF cells (Fig. S26 and S27 †) with colocalization in the nucleus only seen aer 2 h of continuous exposure (GPC ¼ 0.50, Fig. S28 †). Similarly, only nuclear and cytosolic accumulation was detected. These results might be linked to the different cell metabolic ratios associated with cancer vs. non-cancer cell lines. 53 The nucleus contains most of the cell genetic material e.g. DNA, RNA, rRNA and proteins. 54 Thus, it was proposed that favorable electrostatic interactions between FCD-3 and the phosphate backbone of polynucleotide chains might be responsible for the nuclear uptake observed. To evaluate this, cells were incubated with FCD-3 for 2 h at 37 C as before (see the ESI † for experimental details), before Förster resonance energy transfer (FRET) between FCD-3 (FRET donor) and cell permeable DNA (NuclearID-Red 55 and DRAQ-5 (ref. 56)) or RNA (F22 RNASelect 57 ) intercalating dyes as FRET acceptors was measured using ow cytometry (Fig. S37-S43 †). For both cell lines, a FRET-associated decrease in emission was observed between FCD-3 and intercalating DNA-dyes. For instance, in HeLa cells FCD-3 donor emission decreased by 82% and 60% with Enzo NuclearID-Red and DRAQ-5 DNA dyes, respectively ( Fig. S40 and S41 †), and by 50% and 44% in HDF ( Fig. S41 and S42 †). 58 On the other hand, no RNA FRET-response was observed with F22 in HeLa and a minor reduction of 7% was seen in HDF ( Fig. S40 and S43 †), suggesting that DNA-mediated interaction with FCD-3 might be responsible for the observed internalization.
To evaluate the cytotoxicity of FCD-3, HeLa cells (cervical cancer) and human dermal broblasts (HDF) were continuously exposed to FCD-3 at concentrations from 10 À3 to 2000 mg mL À1 for 1 h, 1 day, and 3 days (Fig. 4). Metabolic competence was assessed using Alamar Blue (AB), and the number of live cells was obtained with calcein. A concentration-dependent toxic effect was observed in both HeLa and HDF treated cells. For HeLa, the lethal concentration (LC 50 ) was found to be 100 mg mL À1 aer three days of exposure, while higher concentrations of FCD-3 were tolerated at shorter exposure times (LC 50 at 1 h was 1500 mg mL À1 ). Reductive metabolism (RM) was halved aer 1 h of exposure to 500 mg mL À1 FCD-3 and to 50 mg mL À1 aer three days of exposure (Fig. S44 †). In HDF cells (Fig. S45 †), toxicity was detected at higher concentrations and/or longer exposure times to FCD-3 when compare to HeLa cells. LC 50 was 10 times higher and RM 50 was observed at 500 mg mL À1 FCD-3 but only aer three days of exposure. The observed differences in cytotoxicity for each cell line could be attributed to the different rates of cellular uptake and nuclear accumulation of FCD-3.
Next, we showed that photothermal activation of FCD-3 (500 mg mL À1 ) in deionized H 2 O and cell media DMEM (2 mL) was Scheme 2 2,5-Deoxyfructosazine 4 was isolated from FCD-3 after size exclusion chromatography. possible using blue-light-emitting diodes (LEDs), which were chosen due to their low cost, ease of use, and widespread availability (l em ¼ 460 nm, Fig. S48 †). Encouragingly, a 14 C increase in temperature was measured aer 90 min of illumination when compared to illumination of the solvent alone. Photothermal effects in FCD-3-treated cells post-illumination were then evaluated. An increase in temperature was also recorded in the wells of HeLa exposed to FCD-3 for 2 h at room temperature during the 90 min blue-LED irradiation, when compared to controls (FCD-3-treated cells kept in the dark). The temperature increase was proportional to FCD-3 concentration (Fig. S49 †). LED-irradiation of HeLa cells treated with 50 mg mL À1 FCD-3 while keeping the cells at 37 C resulted in temperatures reaching 56 C aer 90 min, while the same experiment in HDF cells only recorded 44 C. 59 Thermal increases in eukaryotic cells are known to have detrimental effects and lead to toxicity. 60 To evaluate this, HeLa and HDF cells, cultured in 96-well plates, were exposed to FCD-3 at 1, 10, 50, 100 and 500 mg mL À1 for 2 h, before irradiation with blue LEDs for 30, 60 or 90 min. The FCD-3 containing medium was then removed and fresh medium was added. Cellular health was then assessed aer 1 h, 1 day, and 3 days postillumination using AB and calcein as before and compared to controls (cells that have not been exposed to FCD-3 but were LED-irradiated for the same amount of time). Excitingly, signicant cell death and decrease in metabolism were observed for HeLa cells aer incubation with at least 10 mg mL À1 FCD-3 followed by LED-irradiation (Fig. S50-S52 †). In general, cell populations decreased by 75% and 80% aer 1 day and 3 days post 30 min illumination when using concentrations as low as 50 mg mL À1 FCD-3. Increasing the blue-LED irradiation time to 60 and 90 min increased the toxic effect in HeLa cells. Conversely, in 50 mg mL À1 FCD-3-treated HDF cells aer 30 min of illumination, RM was only diminished by 40% aer 3 days (Fig. S53 †) and only aer exposures of up to 500 mg mL À1 FCD-3 a population reduction of 60% was observed aer 1 day post-30 min LED irradiation. Interestingly, at 3 days HDF cells were  This journal is © The Royal Society of Chemistry 2019 recovered with only a 20% reduction relative to controls. This apparent recovery was observed for HDF at all FCD-3 exposure concentrations and aer all LED-irradiation times (Fig. S54 and S55 †). These differences in cellular toxicity aer LED illumination in both cell lines could be attributed to differences in photothermal activation which in turn are likely associated to the differential cellular uptake of the nanoprobes.
Since FCD-3 contains a robust FCD-core (5) coated with in situ generated 2,5-deoxyfructosazine 4, and with the knowledge that 4 is a versatile molecule with anti-diabetic and anti-inammatory properties, 50,61 we wanted to determine whether FCD-3 or just 4 alone in solution was responsible for the cell killing effects observed. To this end, HeLa cells were incubated with FCD-3, 4 or core FCD-5 at two different concentrations (50 and 100 mg mL À1 with regard to the concentration of 4) for 2 h at 37 C. 62 Following the same process as before, the cells were then subjected to LED irradiation or kept in the dark for 60 min aer which cell viability and metabolism were measured aer 24 and 72 h and compared to controls (Fig. 5, S56 and S57 †). Generally, no signicant toxicity was observed in cells incubated with either 4 or FCD-5 either in the dark or under LED illumination, and only cells incubated with FCD-3 and subjected to LED-illumination showed an 80-90% reduction in cell viability as before. 63 To better understand the mechanism of toxicity, dihydroethidium (DHE) was used next to assess the production of cytosolic reactive oxygen species (ROS) in FCD-3-treated cells 1 h post illumination. In both HeLa and HDF, ROS production levels were similar to those of dark controls, except with exposure to FCD-3 of 500 mg mL À1 and 90 min illumination which diminished DHE cytoplasm uorescence by 50% in HeLa cells (Fig. 6). These results suggest that differences between HDF and HeLa responses to FCD-3 exposure/blue-LED illumination are not likely ROS associated.
Anaerobic glycolysis which produces adequate adenosine 5 0triphosphate (ATP) and biological components is the hall-mark of cancers to meet the rapid metabolic requirements. 64 To investigate whether ATP levels had been affected upon treatment with FCD-3 and LED irradiation, intracellular ATP levels were determined in HeLa cells using a luciferin-luciferase luminescence ATP detection assay. HeLa cells were exposed to FCD-3 at 1, 10, 50, 100 and 500 mg mL À1 and incubated for 2 h before 90 min blue-LED irradiation. ATP determination was conducted at 1 h, 1 day and 3 days post-illumination. Controls kept in the dark did not experience a signicant reduction in ATP levels aer exposure to FCD-3. However, in cultures irradiated with blue-LED, ATP levels were reduced in a concentration-dependent manner with 1 day and 3 day levels being diminished by 65% and 75%, respectively, at 1 mg mL À1 FCD-3, relative to dark controls (Fig. 7).
To explore if metabolic products derived from FCD-3 exposure to cells could induce toxicity, naïve HeLa and HDF cells were incubated with the supernatant and lysates of FCD-3 exposed cultures at different concentrations (1-500 mg mL À1 ) for 2 h before 90 min blue-LED illumination. FCD-3 suspended in the cell culture medium in the absence of cells was irradiated as the control and RM and live cell numbers were quantied aer 1 day of exposure. Cell lysates caused no toxic effects in either HeLa or HDF cells. Supernatants of irradiated HeLa cells were toxic only at the highest concentration of FCD-3 (500 mg mL À1 ), as determined by measuring cell viability and reductive metabolism ( Fig. S58 and S59 †), while no signicant effects were measured in HDF cells at all concentrations tested ( Fig. S60 and S61 †). Moreover, NMR spectroscopy analysis of pre-and post-90 min illumination of FCD-3 in water revealed no visible changes in the structure of the FCDs (Fig. S62 †). These results suggest that the cytotoxicity observed is not likely due to the generation of new metabolic or photochemical FCD by-products.

Conclusions
In conclusion, we have developed a facile, three minute largescale synthesis of a novel green-emissive carbon-based nanomaterial FCD-3 that can be used to target HeLa cancer cells preferentially over healthy epithelial HDF cells under mild activation conditions. The carbon-based probe contains a stable amorphous core and is decorated with 2,5-deoxyfructosazine 4 as the major component, which together give rise to the nanomaterial's unique properties. We demonstrate that FCD-3 is internalized into both HeLa and HDF cells partly through passive mechanisms and trafficked into the nucleus by interactions with DNA, albeit at signicantly distinct rates depending on the cell line. This differential uptake can be exploited in the LED-activated selective killing of HeLa over HDF cells aer 60 min of LED-illumination using as little as 50 mg mL À1 FCD-3. The killing is believed to be caused by an increase of intracellular localized temperature followed by ATP depletion. Interestingly, neither the FCD core 5 or surface component 4 are individually able to induce cancer cell death as observed for FCD-3. Furthermore, we show that metabolic products from FCD-3 treated cells do not elicit toxic effects on naïve cells. These novel metal-free bifunctional nanoprobes have the potential, with further development, to be effective theranostic agents in cancer therapy.

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