Poly(N,N-dimethylacrylamide)-coated upconverting NaYF4:Yb,Er@NaYF4:Nd core–shell nanoparticles for fluorescent labeling of carcinoma cells

Upconverting luminescent lanthanide-doped nanoparticles (UCNP) belong to promising new materials that absorb infrared light able to penetrate in the deep tissue level, while emitting photons in the visible or ultraviolet region, which makes them favorable for bioimaging and cell labeling. Here, we have prepared upconverting NaYF4:Yb,Er@NaYF4:Nd core–shell nanoparticles, which were coated with copolymers of N,N-dimethylacrylamide (DMA) and 2-(acryloylamino)-2-methylpropane-1-sulfonic acid (AMPS) or tert-butyl [2-(acryloylamino)ethyl]carbamate (AEC-Boc) with negative or positive charges, respectively. The copolymers were synthesized by a reversible addition-fragmentation chain transfer (RAFT) polymerization, reaching Mn ~ 11 kDa and containing ~ 5 mol% of reactive groups. All copolymers contained bisphosphonate end-groups to be firmly anchored on the surface of NaYF4:Yb,Er@NaYF4:Nd core–shell nanoparticles. To compare properties of polymer coatings, poly(ethylene glycol)-coated and neat UCNP were used as a control. UCNP with various charges were then studied as labels of carcinoma cells, including human hepatocellular carcinoma HepG2, human cervical cancer HeLa, and rat insulinoma INS-1E cells. All the particles proved to be biocompatible (nontoxic); depending on their ξ-potential, the ability to penetrate the cells differed. This ability together with the upconversion luminescence are basic prerequisites for application of particles in photodynamic therapy (PDT) of various tumors, where emission of nanoparticles in visible light range at ~ 650 nm excites photosensitizer.

www.nature.com/scientificreports/ a good efficiency. UCNP are typically synthesized by the precipitation of lanthanide salts, which results in polydisperse products with cubic α-phase structure, or thermal decomposition 15 and hydro(solvo)thermal methods allowing to produce uniform nanoparticles with hexagonal β-phase 16 . Both crystalline phases provide different upconversion efficiency, with the cubic phase yielding usually the lower performance. Therefore, the research is mostly focused on tuning the physicochemical properties of UCNP and solving problems associated with quenching that lowers the upconversion efficiency 17 . UCNP used for cell tracking and labeling compete with conventional water-soluble fluorescent markers developed for imaging of various cell organelles, specific proteins, or cell pathways [18][19][20] . Nevertheless, the advantage of UCNP consists in that they can be excited with near-infrared light (NIR) penetrating at 808 nm up to 5 cm deep into tissues, where it is upconverted to visible light 21 . This can be exploited for example in photodynamic therapy of tumors, the capillaries of which are typically fenestrated or immature enabling higher retention of photosenzitizers in the extracellular matrix. Another benefit of UCNP is the absence of photobleaching, spectrally distinct and narrow emissions, non-blinking, weak autofluorescence, high signal-to-noise ratio, and unique photostability. The weak point of nanoparticle-based labeling consists in the fact that the particles are often too large to readily diffuse across the negatively charged plasma cell membrane, which limits ability to interact with cognate intracellular targets. The mechanism of engulfment by cells is thus related to the size of nanoparticles and their aggregates. Involved can be pinocytosis (cell drinking), which is nonspecific, phagocytosis (cell eating) ingesting larger particles, or receptor-mediated endocytosis, when caveoline or clathrin proteins are engaged 22 . After the nanoparticle uptake, they may stay encapsulated by cell membranes in endosomes and lately in lysosomes, which protect them from reaching cytosol compartments. The nanoparticles are thus usually able to reach only the lysosomes, where they are trapped, without any chance to enter the inner cell environment. Moreover, majority of nanoparticles has only limited colloidal stability in the culture media due to electrostatic stabilization. In this regard, the steric stabilization by adsorbing or covalent binding hydrophilic polymers is preferred. Consequently, the nanoparticles are often decorated with charged molecules, such as saccharides, folic acid, various synthetic polymers, proteins, peptides 23,24 , or phospholipids 25 , to increase attractivity for cells. It is supposed that the larger and negatively charged particles are engulfed rather by phagocytic than nonphagocytic cells, which rather prefer smaller and positively charged nanoparticles 26,27 ; however, also opposite results were published, e.g., for A549 cell line 28 . Bioapplications of positively charged particles (coated with polyethyleneimine, chitosan, polylysine, polyarginine, etc.) are also accompanied with higher toxicity due to disruption of plasma membrane integrity 26 , while negatively charged particles can induce intracellular damage 29 . Negatively charged polymers supporting cellular uptake of the nanoparticles are exemplified by polystyrene sulfonic acid 30 and poly(γ-glutamic acid) 31 , while electroneutral polymers typically contain hydrophobic motives, e.g., esters of myristic acid 32 , or are based on silica. Polymer coatings for UCNP should also contain carboxyl, phosphate, (bis)phosphonate, or sulfate anchoring groups to facilitate attachment to the particle surface.
In this report, we investigated di-end-functionalized poly(N,N-dimethylacrylamide) copolymers containing both bisphosphonate anchoring groups binding to the surface of UCNP and amino or sulfonate groups supporting engulfment of the particles in human hepatocellular carcinoma HepG2, human cervical cancer HeLa, and rat insulinoma INS-1E cells. The ultimate goal was to design and develop new surface-engineered in vivo cell trackable UCNP excitable at 808 nm wavelength that excellently penetrate the tissues and are suitable for fluorescent labeling of carcinoma cells.

Preparation of fluorescently labeled P(DMA-AEC-Boc) and P(DMA-AMPS). Methanolic solu-
tions (2 ml) of P(DMA-AEC-Boc) or P(DMA-AMPS) (0.3 g) and sodium borohydride (20 mg) were stirred at room temperature (RT) for 2 h under an argon atmosphere to remove RAFT leaving group. Resulting intermediates were purified on a Sephadex LH-20 chromatographic column with methanol as an eluent under argon purging; the solvent was removed using a vacuum rotary evaporator. Afterwards, methanolic solutions of copolymers (150 mg; 33 mg/ml) and DY-615-maleimide (0.1 mg; 0.1 mg/ml) were added and the mixture was stirred at RT for 16 h. The fluorescently labeled P(DMA-AEC-Boc)-DY-615 or P(DMA-AMPS)-DY-615 polymers were purified by gel filtration on a Sephadex LH-20 column with methanol as an eluent; the solvent was then vacuumevaporated at RT. Amount of labeled DY-615 was determined by UV-Vis spectrophotometry at 621 nm (molar absorption coefficient ε = 200,000 l/mol·cm). Synthesis of NaYF 4 :Yb,Er core nanoparticles (C-UCNP). C-UCNP were synthesized according to previously published procedures 37,38 . Briefly, yttrium(III), ytterbium(III), and erbium(III) chlorides (1 mmol; 0.78/0.2/0.02 mol/mol/mol, respectively) and oleic acid (6 ml) were dissolved in octadec-1-ene (15 ml) at 160 °C for 30 min under an argon atmosphere. The mixture was cooled down to RT to allow addition of methanolic NaOH solution (2.5 mmol) and NH 4 F (4 mmol). The temperature was then increased to 70 °C to evaporate methanol and subsequently to 300 °C for 1.5 h to produce C-UCNP. They were separated by centrifugation (3,460 rcf) for 30 min, washed in hexane/ethanol mixture (1:1 v/v) twice (14 ml each), and dispersed in hexane.

Synthesis of CS-UCNP@Ner-PEG.
Surface of the CS-UCNP was modified by Ner-PEG according to an earlier published report 39 . Ner-PEG (3.5 mg) was added to an aqueous dispersion of CS-UCNP (6 ml; 1.7 mg/ ml) and the mixture was stirred at RT for 12 h. Resulting CS-UCNP@Ner-PEG were dialyzed against water using a cellulose membrane (MWCO 14 kDa) to remove excessive PEG-Ner. Characterization of nanoparticles. The morphology of nanoparticles was analyzed using a Tecnai Spirit G2 transmission electron microscope (TEM; FEI; Brno, Czech Republic) 38 . The particle size and distribution were determined by measuring at least 300 nanoparticles from four random TEM micrographs using ImageJ software. The average diameter of ellipsoidal nanoparticles was determined as follows: long axis (morphological descriptor MaxFeret) and short axis (morphology descriptor MinFeret) were measured and the average diameter was approximated as D = 1/2*(MaxFeret + MinFeret). Number-(D n ), weight-average diameter (D w ), and the uniformity (dispersity Ð) were calculated as follows:

Modification of CS-UCNP with
where N i and D i are number and diameter of the nanoparticle, respectively. The X-ray powder diffraction (XRD) measurements were performed using an Explorer powder diffractometer (GNR Agrate Conturbia, Italy) in the region 13-80 degree 2Θ.
The hydrodynamic nanoparticle diameter (D h ), size distribution (polydispersity PD), and ζ-potential were determined by dynamic light scattering (DLS) on a Zetasizer Ultra Instrument (Malvern Instruments; Malvern, UK) at 25 °C; D h and PD were calculated from the intensity-weighted distribution function obtained by CONTIN analysis of the correlation function embedded in Malvern software. 1 H and 31 P NMR spectra were recorded using a Bruker Avance III 600 spectrometer (Bruker; Billerica, MA, USA) equipped with a 5 mm diffusion probe-head. 1 H NMR conditions were as follows: 90° pulse width 10 μs, acquisition time 4.54 s, spectral width 7,212 Hz, relaxation delay 10 s, and 32 scans. 31 P NMR spectra were recorded in D 2 O at 22 °C with 90° pulse, width 18 μs, relaxation delay 15 s, spectral width 36,232 Hz, and acquisition time 0.9 s. The resulting spectra were processed in Topspin 4.1.0 software, where the integrated intensities were determined with an accuracy of ± 1%. During the measurements, temperature was maintained within ± 0.2 K using a BVT 3000 temperature unit.
Weight-(M w ), number-average molar mass (M n ), and M w /M n of the polymers were determined by the size exclusion chromatography (SEC) on a Shimadzu HPLC system (Tokyo, Japan) equipped with a UV-Vis diode array and OptilabrEX refractive index and DAWN EOS multiangle light scattering detectors (Wyatt; Santa Barbara, CA, USA). A TSK SuperAW3000 column was used with methanol/sodium acetate buffer (80/20 v/v) as a mobile phase (pH 6.5) at flow rate of 0.6 ml/min. FTIR spectra were recorded on a 100 T FTIR spectrometer (Perkin-Elmer; Waltham, MA, USA) using a Specac MKII Golden Gate single attenuated total reflection (ATR). The content of DY-615 in methanolic solution of polymers was determined using a Specord Plus UV-Vis spectrometer (Analytik Jena, Germany) at 621 nm using the molar absorption coefficient for DY-615 at 621 nm (ε = 200,000 l/mol cm). The elemental composition of particles was obtained from energy-dispersive X-ray (EDX) analysis (EDAX detector; Mahwah, NJ, USA).
The upconversion luminescence spectra of C-UCNP and CS-UCNP and their PDMA-or PEG-coated analogues (1 mg/ml) were measured in a Hellma 114F-QS cuvette (10 × 4 mm path length; Sigma-Aldrich) at RT using a FS5 spectrofluorometer (Edinburgh Instruments; Livingston, UK) equipped with continuous xenon lamps (150 W) and CW 808 and 980 nm infrared diode lasers as an excitation source with nominal laser power of 2 W (MDL-III-808 and MDL-III-980; beam size of 5 × 8 mm 2 ).
Cytotoxicity of nanoparticles. The cytotoxicity of particles was measured using a trypan blue exclusion test (Thermo Fisher Scientific). Briefly, HeLa, HepG2, and INS-1E cells were cultured in a cell medium at 37 °C for 48 h under 5% CO 2 humidified atmosphere and incubated with the particles (0.01, 0.02, 0.05, 0.1, and 0

Results and discussion
Reversible addition-fragmentation chain-transfer (RAFT) polymerization of DMA and its copolymerization with AEC-Boc and AMPS. As the starting UCNP are generally hydrophobic due to stabilization by OA, their surface hydrophilization is required. Here, poly(N,N-dimethylacrylamide) (PDMA) was selected as a basic coating polymer of the particles due to its excellent solubility in water, as well as in organic solvents, biocompatibility, and last but not least good reactivity 40 . To obtain PDMA with a controlled and narrow distribution of molar mass (M w /M n < 1.2), which is important for design of nanocarriers possessing uniform physicochemical properties and reproducible biological experiments, RAFT polymerization was used. The technique enables easy removal of thiocarbonyl end-group and subsequent conjugation of a therapeutic agent 41,42 . Therefore, DMA was copolymerized by RAFT polymerization with two reactive comonomers, AEC-Boc or AMPS (95/5 mol/mol) carrying amino and sulfo groups with positive and negative charges, respectively (Fig. 1). The charge is one of the important parameters that affect cellular internalization of the particles.
Polymerizations were terminated at 85-86% conversions (according to 1 H NMR), yielding M n ~ 11 kDa and a narrow molar mass distribution for both P(DMA-AEC-Boc) and P(DMA-AMPS). These values agreed with calculated M n,th (Table 1) and were sufficiently high to ensure a good steric stabilization of the particles in aqueous media.
Let us note that poor colloidal stability and aggregation of PDMA-coated particles was observed for M w < 8 kDa, whereas higher molar mass provided effective stabilization 36 . Moreover, the obtained molar masses of both PDMA-based polymers were lower than the renal excretion limit that is generally considered    . S1 a,b). Content of AMPS in the copolymer was calculated from signal 'd' (Fig. S2 a,b). Amount of reactive monomers in P(DMA-AEC-Boc) and P(DMA-AMPS) was 5.3 and 5.0 mol.%, respectively, which was in agreement with the monomer feed ratio (95/5 mol/mol). In the FTIR spectrum of P(DMA-AEC-Boc), peaks observed at ~ 1710 and 1220 cm −1 were assigned to ν(C=O) and ν(C-O) stretching vibrations of Boc group, respectively (Fig. 2a). The corresponding peaks of AMPS unit in P(DMA-AMPS) at 1205, 1180, and 1036 cm −1 were attributed to ν(S=O) stretching vibrations. Both 1 H NMR and FTIR spectra thus confirmed successful preparation of P(DMA-AEC-Boc) and P(DMA-AMPS) copolymers.  44 . In the next step, the PDMA (co)polymer was functionalized with Ale to ensure steric stabilization of the NaYF 4 :Yb,Er@NaYF 4 :Nd nanoparticles in biological media (Fig. 3a). It is an advantage that Ale contains bisphosphonate moieties with a strong binding affinity to a number of metal ions, such as alkaline earth 45,46 and transition metals 47 , as well as lanthanides 48 . The presence of phosphonate groups in the Ale-P(DMA-AEC-Boc)-DY-615 and Ale-P(DMA-AMPS)-DY-615 was confirmed by 31 P NMR spectroscopy (Fig. S3). In order to obtain positively charged particles, Boc-protected amino groups of Ale-P(DMA-AEC-Boc)-DY-615 polymer were removed (Fig. 3b) as confirmed by 1 H NMR (Fig. S1c) and FTIR spectroscopy due to disappearance of peaks at 1710 and 1220 cm −1 assigned to Boc groups (Fig. 2b).

Ale-P(DMA-AEC)-DY-615-, Ale-P(DMA-AMPS)-DY-615-, Ale-PDMA-, and PEG-Ner-modified CS-UCNP.
The uniformly-sized C-UCNP were prepared by a high-temperature (300 °C) coprecipitation of lanthanide chlorides in octadec-1-ene solvent in the presence of oleic acid as a stabilizer. According to TEM, the particles were spherical in shape with D n = 29 nm and a narrow size distribution (Ð = 1.01; Fig. 4a). Such a narrow distribution is important in terms of the same physicochemical and biological properties and reproducibility of the results. Further, C-UCNP were covered with NaYF 4 :Nd shell containing an additional sensitizer (Nd 3+ ) to enable excitation at 808 nm within the transparent NIR optical window of biological tissues and provide a bright NIR emission. The OA-stabilized CS-UCNP were ellipsoidal (Fig. 4b), in agreement with earlier described results 39 . Both C-UCNP and CS-UCNP were characterized also by XRD (Fig. S4). Despite the fact that the intensities of CS-UCNP diffractograms were higher than those of C-UCNP, they were similar, corresponding to the standard β-NaYF 4 known from literature (JCPDS card. No 28-1192). The full width at half maximum (FWHM) of several peaks for both samples was roughly the same and the crystal size from the first peak was estimated to 22 nm using the Scherrer formula: www.nature.com/scientificreports/ Here, K is the shape factor ~ 0.92 rad, Θ is the diffraction angle, and λ is the X-ray wavelength (0.154 nm). The only significant difference was the peak at 31.74 deg and three possible negligible peaks. The upconversion luminescence of C-UCNP and CS-UCNP was determined by emission at 980 and 808 nm (Figs. 5 and S5). The emission of both core and core-shell nanoparticles exhibited the characteristic emission at 409 ( 2 H 9/2 → 4 I 15/2 ), 525 ( 2 H 11/2 → 2 I 15/2 ), 542 ( 4 S 3/2 → 2 I 15/2 ) and 656 nm ( 4 F 9/2 → 2 I 15/2 ) typical for the transitions of Er 3+ ions in upconverting nanomaterials ( Fig. 5a and S5). While the C-UCNP did not show upconversion emission under the 808 nm excitation, the incorporation of Nd 3+ into the shell provided luminescence and excitation deep in the tissue.
As expected, compared to the C-UCNP, introduction of the NaYF 4 :Nd shell fourteen times increased emission intensity at 980 nm excitation with low power density (1 W/cm 2 ; Fig. S5a). This demonstrated that the shell protected dopants in the core from quenching. The transfer of nanoparticles from hexane into water slightly decreased the emission intensity of CS-UCNP (Figs. 5a and S5b). The TEM/EDX spectrum of C-UCNP exhibited main peaks of Na, Y, and F elements and weaker Yb peak and C and Cu peaks from the standard supporting TEM grid (Fig. 4c). The spectrum of CS-UCNP differed by the appearance of small peaks at 5.3 and 5.8 keV, which proved the presence of neodymium in the shell layer 38 (Fig. 4d). The next modification of CS-UCNP consisted of two steps: (i) carful removal of residual organic compounds (OA and octadec-1-ene) from particles by their washing with hexane, ethanol, and water and (ii) coating of particles with PDMA-or PEG-based (co)polymers. Coating of similar lanthanide-based nanoparticles by poly(N,N-dimethylacrylamide) copolymer was confirmed by FTIR analysis in our previous paper 49 . The resulting surface-modified particles varied in ζ-potential (Table 2), as the P(DMA-AMPS) and P(DMA-AEC) polymers contained sulfo and amino groups, respectively, rendering negative (− 9 mV) or positive ζ-potential (24 mV). In contrast, Ner-PEG and Ale-PDMA-DY-615 provided moderately positive surface charge (15 and 12 mV, respectively) to the particles. D h of neat CS-UCNP (230 nm) reflected rather the size of aggregates than that of individual nanoparticles.
The cell viability did not change after exposure to particles even at the concentration of 0.2 mg/ml that was higher than that used in other biological experiments. The emission spectra of Ale-P(DMA-AEC)-DY-615-, Ale-P(DMA-AMPS)-DY-615-, and PDMA-Ale-coated C-UCNP exhibited typical upconversion peaks at 530 and 650 nm (Fig. S7). After the engulfment of particles (0.15 mg/ml) by the cells, their components were also fluorescent. Nevertheless, due to the difference between the fluorescence of nanoparticles and cells, the particles were clearly detectable. It was obvious that the particles even in the cell milieu did not aggregate. Processing of selected segments from confocal micrographs was exemplified on the distribution of CS-UCNP@Ale-P(DMA-AEC)-DY-615 in the CellMask™-stained HepG2 cells (Fig. S8). When overviewing the segments of confocal micrographs of all particle-engulfed cells (Fig. 7), it was found that the core-shell nanoparticles with positive ζ-potential easily penetrated negatively charged cell membranes. The particles coated with PEG-Ner, Ale-P(DMA-AEC)-DY-615, and Ale-PDMA-DY-615 with the ζ-potential ranging 6-30 mV exhibited cellular uptake (Fig. 7a-l). In contrast, particles coated by Ale-P(DMA-AMPS)-DY-615 with negative surface charge (− 9 mV) seemed to be less prone to cell labeling (Fig. 7m-o).
During the experiments, the nanoparticles gradually penetrated the cells; after the first 2 h, they were mostly localized around the cells, not inside, but after 4 and 24 h, they adhered to the cell membranes entering then most    www.nature.com/scientificreports/ cells, respectively. As an example, the penetration of CS-UCNP@Ale-P(DMA-AEC)-DY-615 into HepG2 cells was shown on the confocal micrographs (Fig. 8), confirming that the particles were noncytotoxic and biocompatible.
In the cell cultures containing Ale-P(DMA-AMPS)-DY-615-, Ale-P(DMA-AEC)-DY-615-, and Ale-PDMA-DY-615-modified CS-UCNP in the absence of CellMask™ green, the particles and the polymer were green and red, respectively, proving that the coating remained firmly attached to the particle surface (Fig. 9). Some symmetrical shifts in the images can be ascribed to errors induced by switching between two different lasers.
To compare spreading of nanoparticles inside the cells and monitor the number of particle-internalized cells, the percentage of cell area occupied by nanoparticles and the percentage of cells containing the particles was determined (Fig. 10).
Considering both these parameters, CS-UCNP@Ale-P(DMA-AEC)-DY-615 were most spread inside the cells compared to other particle types and at the same time they penetrated all cells. In contrast, CS-UCNP@ Ale-P(DMA-AMPS)-DY-615 with negative ξ-potential internalized only relatively small number of cells.

Conclusions
Differently charged polymer-coated CS-UCNP, uniform in size with diameter of 29 nm, were synthesized by a high-temperature coprecipitation of lanthanide chlorides in a high-boiling organic solvent. This was followed by the modification of particles with four polymers: negatively charged sulfo group-containing P(DMA-AMPS), positively charged P(DMA-AEC), and two electroneutral PDMA and PEG as a control. Thanks to the PDMAbased coatings, the colloidal stability of particles in the cell culture medium was ensured. Optionally, the polymers were labeled with DY-615 and used as a coating of CS-UCNP to make fluorescent imaging of carcinoma cells possible, allowing at the same time to control the stability of both nanoparticles and coatings in the cell medium. All the particles, up to 0.2 mg/ml concentration, were very well tolerated by all three examined types of carcinoma cells, i.e., HeLa, HepG2, and INS-1E, without any sign of toxicity. The highest particle uptake in carcinoma cells was observed with CS-UCNP@Ale-P(DMA-AEC)-DY-615, followed by CS-UCNP@Ale-PDMA-DY-615, CS-UCNP@Ner-PEG, and neat CS-UCNP having the positive ζ-potential (12-30 mV). The CS-UCNP@ Ale-P(DMA-AMPS) were not significantly internalized by the carcinoma cells due to negatively charged cell membranes that prevented the mutual contacts with particles. It can be thus concluded that the Ale-P(DMA-AEC)-DY-615-coated CS-UCNP showed a favorable cellular uptake that makes them a suitable candidate for cell labeling and prospectively for PDT of various tumors.