Influence of the chirality of carbon nanodots on their interaction with proteins and cells

Characterization and concentration determination of the CNDs

Basic optical characterization of the CNDs

Different carbon nanodots (CNDs) were synthesized according to previously published protocols, i.e. N–CNDs³⁴, and R-CNDs and S-CNDs³⁵. Firstly, the optical properties of the CNDs were characterized. Briefly, the obtained lyophilized solids of CNDs were first weighted and then fully dissolved in filtered Milli-Q water to form stock solutions with a concentration of C_CNDs = 10 mg mL^–1. Then, the stock solutions were further diluted to C_CNDs = 100 µg mL^–1 with filtered sterilized water for absorbance and fluorescence spectra measurements (Supplementary Fig. 4) using an UV-Vis absorption spectrophotometer (Agilent 8453, Agilent technologies, Australia) and a fluorescence spectrometer (Fluorolog-3, Horiba Jobin Yvon, USA). In the absorption spectra the R-CNDs and S-CNDs displayed two absorption peaks, while the N–CNDs only had one peak, at the same position as the R-CNDs and S-CNDs at 280 nm. Concerning the fluorescence properties, the three CNDs demonstrated a slight excitation-dependent emission shift upon excitation at different wavelength λ_ex ranging from 330 to 440 nm.

Concentration-dependent absorption/fluorescence intensity measurements

In order to account for different absorption and emission properties of the different CNDs, the concentration-dependent adsorption and fluorescence spectra were measured for the different CND samples. For this objective, the CND solutions with concentrations ranging from C_CNDs = 0.78 µg mL^–1 to 400 µg mL^–1 were prepared by diluting the CND stock solutions with filtered Milli-Q water. Then, UV-Vis absorption spectra A(λ) and fluorescence spectra I(λ) (λ_ex = 405 nm) were recorded (Supplementary Fig. 26). The three different types of CNDs all possessed a dose-dependent absorbance and fluorescence behavior as expected, i.e. the adsorption and fluorescence linearly decreased with more diluted CNDs solutions. The absorption values A₂₈₀ = A(λ = 280 nm) plotted against the CND concentration are shown in Fig. 1 in the main manuscript. In order to probe the influence of the wavelength at which the absorption is measured for the error analysis, the same evaluation as performed for Fig. 1b in the main paper was also performed for the absorption A₄₀₅ at 405 nm of the CNDs (Supplementary Fig. 6). As A₄₀₅ ≪ A₂₈₀ all further evaluation was performed with A₂₈₀.

Concentration adjustment and fluorescence intensity correction

As the CNDs at different mass concentration C_CNDs (as determined by weighting) showed different absorption (Supplementary Fig. 6), the concentrations were adjusted to lead to the same absorption at 280 nm. This was done to allow further concentration determination via absorption measurements. For this, the adsorption values of the CND solutions were determined at 280 nm and the CND solutions with higher adsorption values (which was always the R-CNDs) were diluted with filtered Milli-Q water until achieving all samples had the same absorption A₂₈₀. The R-CND solution remained undiluted at concentration C_CNDs. In the following the adjusted concentrations of the S-CND and N–CND solutions were assumed to have the same concentration as the R-CND solution. Here the “same concentration” refers to equal absorption at 280 nm. These absorption–based concentrations are referred to as adjusted concentrations C′_CNDs. C′_CNDs(S-CNDs) = C′_CNDs(N–CNDs) = C′_CNDs(R-CND) = C_CNDs(R-CND). An example is shown in Supplementary Fig. 7. In order to determine the error in concentration determination absorption A(λ) and emission spectra I(λ) were recorded in dependence of the adjusted concentrations C_CNDs. The spectra are displayed for the five different batches of CNDs used in this study in Supplementary Figs. 18 and 19. The concentration-dependence of the absorption at 280 nm and of the fluorescence emission as derived from the spectra is plotted for all five batches in Supplementary Fig. 27. Based on Supplementary Fig. 23 error analysis was performed as described in Fig. 1 of the main manuscript, and the results are displayed in Supplementary Table 3. As can be seen in Supplementary Figs. 7 and 27 at the same adjusted concentration C′_CND the R-CND, S-CND, and N–CND samples have different fluorescence emission intensities. This needs to be considered when quantifying the uptake of CNDs by their fluorescence with flow cytometry and confocal microscopy. For flow cytometry was collected with a 450 nm/50 nm bandpass filter, and for confocal microscopy with an LP 420 nm long pass filter (Supplementary Fig. 7). In this way correction factors X taking into account the different fluorescence intensities at the same adjusted concentrations were defined. X_S/R = 〈I(S-CNDs)〉 〈I(R-CNDs)〉^–1 is the ratio of the integrated S-CND fluorescence and the integrated R-CND fluorescence, at the same adjusted concentration of S-CNDs and R-CNDs. The integration range was used emulating the flow cytometry and confocal microscopy filters. X_S/N = 〈I(S-CNDs)〉 〈I(N–CNDs)〉^–1 is the ratio of the integrated S-CND fluorescence and the integrated N–CND fluorescence, at the same adjusted concentration of S-CNDs and N–CNDs. The resulting values are enlisted for all five used batches in Supplementary Table 4.

Concentration determination by CND counting with atomic force microscopy (AFM)

Diluted CND solutions of concentration C_CNDs of R-CNDs and S-CNDs were drop-casted on mica substrates for AFM analysis. Images were acquired by tapping mode AFM (Nanoscope IIIa, VEECO Instruments) on a surface area A_scan of 25 μm² (Supplementary Figs. 28 and 29). The number N_CNDs of CNDs as identified in the image (by looking at the height profiles) was counted for samples prepared by two concentrations per typology of CNDs (R- and S-) and is displayed in Supplementary Table 1. The respective mean values 〈N_CNDs〉 and standard deviations ΔN_CNDs were then calculated. Taking together all determined four values for ΔN_CNDs〈N_CNDs〉^–1 leads to a mean value of ΔN_CNDs〈N_CNDs〉^–1 ≈ 0.21. The normalized error in NP counting corresponds to the normalized error in concentration determination Δc_CND c_CND^–1. Note that the real error even might be higher. This point is better rationalized if we consider the mass per CND to compare the theoretical and calculated number of CNDs per substrate area. Considering the similar average size of CNDs we calculated the average NP mass of CNDs considering their sphericity and the density of amorphous carbon (ρ_C = 3.50 g cm^–3)³⁵. By calculating the expected and measured number of particles on substrate, 〈n_CNDs〉 = 〈N_CNDs〉A^–1, is thus possible to make a direct comparison of the two quantities. As reported in Supplementary Table 1, the calculated and expected 〈n_CNDs〉 values are different by several order of magnitude and this result may be influenced by the drop casting deposition process⁵⁹ that could be cause of: (i) inhomogeneous distribution of NPs on the surface, (ii) formation of aggregates during the solvent evaporation. The following paragraph, treating the CNDs quantification trough transmission electron microscopy (TEM), further remarks the difficulty on calculating these NPs trough microscopy techniques using an analogous concept.

Concentration determination by CND counting with transmission electron microscopy (TEM)

Transmission electron microscopy (TEM) measurements to count CNDs were performed using a Jeol JEM-1011 instrument operating at 100 kV. 2 µL of the according CND solution (C_CNDs(R-CNDs) = 2.0 mg mL^–1; C_CNDs(S-CNDs) = 2.8 mg mL^–1) were drop-casted onto a copper grid (400 mesh, diameter 3.05 mm) coated with amorphous carbon. As can be observed from the TEM-micrographs shown in Supplementary Figs. 30 and 31, homogeneous coating was not achieved. For the R-CNDs different aggregates were observed, whereas for the S-CNDs some areas with dispersed CNDs were also found. It is not known, how much of the aggregates form during drying on the TEM grid. On both samples it was possible to differentiate single CNDs; however, due to the limited contrast, an exact determination of CND size was not possible. Based on micrograph analysis with ImageJ we obtained d_TEM ≈ 1.5 ± 0.4 nm for R-CNDs (N = 216 CNDs investigated) and d_TEM ≈ 2.4 ± 0.9 nm for S-CNDs (N = 255 CNDs investigated). These values are compatible with those obtained by AFM (d_AFM) and discussed in the main text. We emphasize, however, that due to the limited number of observed CNDs and limited contrast, the diameters as determined with TEM should be considered a rough estimate. We used the determined CND diameters to calculate theoretical concentrations, assuming sphericity. With the density of amorphous carbon ρ_C = 3.50 g cm^–3 (diamond has a similar density of ρ_C = 3.51 g cm⁻³) and the weight concentrations (C_CNDs(R-CNDs) = 2.0 mg mL^–1; C_CNDs(S-CNDs) = 2.8 mg mL^–1) we obtain a molar concentration of c_CNDs ≈ 540 µM for R-CNDs and c_CNDs ≈ 180 µM for S-CNDs. Because of the limited accuracy of CND diameter determination, also these concentrations must be considered as a rough estimate. In the TEM micrographs of S-CNDs we find 〈n_CNDs〉 = 〈N_CNDs〉A_scan^–1 ≈ 3014 CNDs µm^–2 on average (Supplementary Table 2). For R-CNDs we find 〈n_CNDs〉 ≈ 1920 CNDs µm^–2 on average on the micrographs. The accuracy of the numbers is limited by the contrast, depending on the micrograph. The standard deviations Δn_CNDs for both average numbers are very high (Supplementary Table 2), with the mean value of both standard deviation being 0.68. This mean value for Δn_CNDs〈n_CNDs〉^–1 would correspond to the uncertainty in concentration determination ΔC_CND C_CND^–1 = 0.68, underlining that the concentration determination with TEM is not feasible. Assuming a homogeneous coating of the whole TEM grid (area = 7.3 × 10⁶ µm²) with these densities, one can estimate 2.4 × 10¹⁰–2.8 × 10¹⁰ CNDs in the dried 2 µL that were drop-casted onto the grids. This would correspond to c_CNDs ≈ 10–20 nM solutions. As expected, this value is several orders of magnitude off the theoretical value, underlining that it is not feasible to obtain a meaningful CND concentration based on TEM analysis.

Concentration determination by using the nanoparticle tracking analysis (NTA)

Nanoparticle tracking analysis (NTA) was performed with a NanoSight LM10 (Malvern Panalytical) operated with a 405 nm laser. R-CND solution were diluted to c_CNDs = 18 µM (see the respective section in “Methods”) and S-CNDs to c_CNDs = 5.5 µM. As can be observed in Supplementary Fig. 5, only large aggregates were tracked by the system for both samples. The main population of CNDs is too small and scatters too weakly to be discernible with this technique. The CND concentrations (which are in fact aggregate concentrations) determined with NTA were n_CNDs ≈ 3.4 × 10⁷ CNDs mL^–1 for R-CNDs and n_CNDs ≈ 1.0 × 10⁷ CNDs mL^–1 for S-CNDs. This corresponds to concentrations in the femtomolar range, underlining that CNDs cannot be measured with NTA but also that the number of aggregates in the CND solutions seems negligible. Note that presence of agglomerates to a large extent can be ruled out by the FCS measurements shown in the respective section in “Methods”.

Other physicochemical characterization data

Additional standard characterization of the CNDs is provided in the form of Fourier–transform Infrared (FT-IR) spectra (KBr), shown in Supplementary Fig. 2, electronic circular dichroism (ECD) spectra shown in Supplementary Fig. 1 and X-ray photoemission spectroscopy (XPS) shown in Supplementary Fig. 3. FT-IR spectra were recorded on a Perkin Elmer 2000 spectrometer. ECD spectra were measured on a Jasco J-815. XPS spectra were measured on a SPECS Sage HR 100 spectrometer.

Influence of cell culture medium on the properties of the CNDs

While the characterization in the respective section about CND properties in “Methods” was carried out in water, uptake experiments of the CNDs took place in cell culture medium. Thus, it needed to be tested how the presence of cell culture medium affects the properties of the CNDs. For this, 400 µL of CND solutions (C′_CNDs = 200 µg mL^–1) were mixed with the same volume of either (i) Milli-Q water (as the negative control), (ii) RPMI (Roswell Park Memorial Institute) 1640 medium without phenol red (Thermofisher, USA) supplemented with 10% fetal bovine serum (FBS, Biochrom, UK), and (iii) RPMI 1640 medium without phenol red without serum. Thus, the final CND concentration was C′_CNDs = 100 µg mL^–1. After different incubation times of t = 0, 24, and 48 h, the CND solutions were characterized in a UV-Kuevette, ZH 8.5 mm Deckel (Sarstedt, Germany) with UV-Vis absorption spectroscopy and with fluorescence spectroscopy (λ_ex = 405 nm). In addition, the hydrodynamic diameters d_h of the CNDs in the different media were measured by dynamic light scattering (DLS, Malvern NANO ZS, England)⁶⁰. In Supplementary Figs. 8, 9 and 32 the absorption and fluorescence spectra of the three different types of CNDs are shown. In all three cases there is a slight fluorescence increase of the different types of CNDs after incubation in particular in serum containing RPMI 1640 medium. At this increase is similar, fluorescence intensities of the three different types of CNDs (R-CNDs, S-CNDs, and N–CNDs) can be also directly compared when the CNDs are exposed to cell culture medium.

The hydrodynamic diameters d_h of the CNDs after incubation with the different medium were measured after different time points by DLS. Data are shown in Supplementary Fig. 33. Due to the very small size of the CNDs and in the case of RPMI 1640 medium supplemented with 10% FBS due to the presence of proteins of similar size as the CNDs, the DLS values are unreliable and are not further interpreted in this study. In fact, the hydrodynamic diameter as detected in the plain media without proteins (yellow bars; water and RPMI 1640 medium without serum) most likely correspond to dust. The increase hydrodynamic diameters as detected in the serum-supplemented medium (yellow bars, RPMI 1640 medium with FBS) originate from serum proteins. Presence or absence of the CNDs (red and blue versus yellow bars) does not change the results, which demonstrates that it is not the CNDs which are detected here with DLS, which is due to their tiny size. Hydrodynamic radii r_h = d_h 2^–1 of the CNDs were instead measured with fluorescence correlation spectroscopy (FCS, see the respective section in “Methods”), where only the CNDs and CND−protein complexes, but not the free proteins provide signal.

Protein adsorption on CNDs

To explore whether the chiral surface of CNDs has an impact on protein adsorption, the interaction of the CNDs with different proteins, human serum albumin (HSA, CAS No. 70024-90-7, Sigma Aldrich, Germany), transferrin human (Tf, CAS No. 11096-37-0, Sigma-Aldrich, Germany), and alpha-2-macroglobulin (α2M, SRP6314, Sigma-Aldrich, Germany) was investigated with fluorescence correlation spectroscopy (FCS)^{39,40,41,61,62,63,64,65}. Measurements were carried out in a Confocal Light Scanning Microscope (CLSM) (LSM 880, Zeiss, Germany) with a Zeiss PlaN-Apochromat ×40/1.0 Water DIC (WD: 2.5 mm) objective with integrated FCS set-up (Zeiss). FCS studies were conducted with two solvents either in filtered Milli-Q water or in phosphate-buffered saline (PBS, Gibco, Invitrogen, Belgium). For measurements, proteins at different concentration were mixed with CNDs in either PBS or water, leading to a final variable protein concentration c_P (P = HSA, Tf, α2M) and a fixed CND concentration C_CND = 10 µg mL^–1 for batch #1 and 50 µg mL^–1 for batch #2 and batch #3. Before measurements all samples were incubated for 15 min and were then loaded to 35 mm petri dishes with glass bottom (Cat.No: 81218-200, ibidi GmbH, Germany) and were immediately covered by a cover glass (Product Code.10474379, Carl Zeiss™, Germany) with a thickness of 0.17 mm ± 0.005 mm. Subsequently, the lid of the glass bottle dish was assembled before FCS measurement. It is worth to mention, that the glass petri dish and the cover slide were continuously used through all measurements to exclude any possible deviation from their thickness, which could have probably resulted in an experimental error (the parameters ω₀ and S of the excitation volume as described below might vary). To make sure that the glass petri dish and cover slide were sufficiently clean and dried before carrying out the next measurement, they were washed by ethanol and Milli-Q water successively, gently wiped with soft tissue paper and dried thoroughly under room temperature (RT) for 5 min. The FCS set-up had to be calibrated. Before carrying out the measurements, the focal volume was calibrated at 488 nm laser excitation with the laser power of 0.5 (at the Zeiss LSM set-up) using a dye with a known diffusion coefficient D_Rho = 414 ± 1 µm² s^–1 (Rhodamine 6G)⁶⁶. Experimentally FCS determines diffusion times τ_D from an autocorrelation function G(τ) based on the fluorescence fluctuation of dyes diffusion in and out into the focus of the excitation. Here fluorescence fluctuations were recorded with 100 repetitions of each 10 s. An example of the autocorrelation function as obtained with Rhodamine 6G dissolved in water (c_Rho = 10 nM) is shown in Supplementary Fig. 34. This autocorrelation function was fitted with the following equation, using the FCS module implemented in the Zeiss ZEN software:

N is the average number of fluorophores within the effective detection volume, i.e. the volume of the excitation focus. M is the number of different fluorescent components in solution (e.g. if a mix of different fluorophores would be analyzed). In the present case M = 1, as there is either only Rhodamine 6G, or afterwards just CNDs in solution. f_i determines the contribution of the different fluorescent components to the autocorrelation function. As here there is only one component f₁ = 1. T is the fraction of the fluorescence decay from the triplet state of the fluorescent compound, and τ_T is the lifetime of the triplet state. For the Rhodamine 6G data shown in Supplementary Fig. 34 the mean of the fit parameters from three measurements were N = 0.15 ± 0.01, S = 4 ± 1, and τ_D = τ_Rho = 22.5 ± 0.5. For the CNDs, the contribution of fluorescence from the triplet state was neglected, i.e. T = 0 and τ_T = ∞. The cross–section of the excitation volume is considered as ellipsoid and S is the ratio of the axis of this ellipsoid (for a sphere it would be S = 1). Thus, the effectively applied fit-function reduced to:

By assuming the excitation volume as ellipsoid (Gaussian ellipsoid approximation), the diffusion time τ_D related to the corresponding diffusion coefficient D via the width of the excitation volume ω₀ by:

Using the experimentally determined value of the diffusion time τ_Rho of Rhodamine 6G and the literature value of its diffusion constant D_Rho, the width of the excitation volume was calculated to be:

By knowing ω₀ as determined for the applied experimental conditions, measured diffusion times τ_D could be related to diffusion constants D, also for the CNDs. In the following these measurements were applied to the different CND samples (different CNDs with different protein concentrations) and for each data point three independent measurements were carried out. In Supplementary Fig. 35 the data for S-CNDs (batch #1) as exposed to different concentrations of HSA in PBS are presented. The corresponding diffusion times τ_D and diffusion coefficients D as determined from the fit as provided in Supplementary Table 11. From the diffusion coefficients D the corresponding hydrodynamic radii were calculated according to the Stokes−Einstein equation:

k_B = 1.38 × 10⁻²³ J K^–1 is the Boltzmann constant, T = 298.15 K room temperature, and η is the solution viscosity. The solution viscosity was assumed to depend linearly on the protein concentration according to:

The protein mass concentration C_P relates to the molar protein concentration c_P by the molar mass of the protein MW(P). η₀ = 0.89 mPa × s is the viscosity of PBS, which is assumed to be the viscosity of water (at room temperature). η_i is the intrinsic viscosity of the proteins. Hereby the following values were used: HSA: M_W(HSA) = 66.5 kDa, η_i = 4.2 cm³ g^–1; Tf: M_W(Tf) = 80 kDa, η_i = 4.4 cm³ g^–1; α2M: M_W(α2M) = 725 kDa⁴⁰. In Supplementary Table 11 the conversion of diffusion coefficients D into hydrodynamic radii r_h is demonstrated. The resulting hydrodynamic radii r_h versus protein concentrations c_P are listed of the different types and batches of CNDs as recorded in water in Supplementary Fig. 36 and as recorded in PBS in Supplementary Fig. 10. In the following, all further discussion will be based on the results obtained in PBS. The data recorded in PBS with HSA show a saturation of the hydrodynamic radius, e.g. at high protein concentration the CND surface is completely saturated with proteins and thus the hydrodynamic radius does not increase further with raising concentrations. The behavior can be fitted with the Hill model³⁹. For this, the r_h(c_HSA) curves shown in Supplementary Fig. 10 were fitted with the following equation:

V_HSA is the volume of one HSA molecule. Assuming a triangular prism shape with 8.4 nm side length and 3.2 nm height we used V_HSA = 96 nm^3 39. V_CND is the volume of one CND, considering the CNDs as spheres, and the radius r_h(0) obtained from the FCS of the control sample, no protein:

r_h(0) is an experimentally determined value and is enlisted in Supplementary Table 5. The fit function had the following free fit parameters: r_h,0, N_max, K_D, and n. r_h,0 is the value from the fit for the hydrodynamic radius of the CNDs with no adsorbed proteins, i.e. r_h(c_HSA ≪ K_D). r_h,0 is a fit parameter, r_h(0) is an experimentally determined value. N_max is the number of HSA molecules bound per CND in saturation (i.e. c_HSA ≫ K_D), K_D is the dissociation coefficient, and n is the Hill coefficient³⁹. The resulting fit values from the curves shown in Supplementary Fig. 10 are presented in Supplementary Table 5.

Cell culture techniques

Two cell lines were used in this study: THP-1 monocytes and HeLa cells. The human monocytic leukemia cell line THP-1 (ATCC® TIB-202™) was obtained from American Type Culture Collection (ATCC, Manassas, VA, USA). THP-1 cells were cultured in suspension in RPMI 1640 medium (Sigma-Aldrich, #61870010) containing 10% heat inactivated fetal bovine serum (FBS, Biochrom, UK), 1 mM sodium pyruvate (Sigma-Aldrich, #S8636), 0.05 mM β-mercaptoethanol (Sigma-Aldrich, #M3148), 100 U mL^–1 penicillin and 100 μg mL^–1 streptomycin (P/S, Sigma-Aldrich, Germany) in a humidified incubator at 37 °C and 5% CO₂. For experimental usage, Phorbol 12-myristate 13-acetate (PMA, Sigma-Aldrich, #P1585) was applied to the THP-1 monocyte (cell passage less than 30) with a dosage of 150 nM¹⁴ for 3 days, inducing the differentiation from THP-1 monocytes to THP-1 macrophages. After stimulation, the THP-1 macrophages were exposed to the CNDs for in vitro uptake, toxicity, and colocalization studies (Supplementary Fig. 37). The human cervix cell line HeLa cells were purchased from ATCC and maintained in Dulbecco’s modified Eagle’s medium (DMEM, Thermofisher, USA) supplemented with 10% fetal bovine serum (FBS, Biochrom, Germany) and 100 U mL^–1 penicillin/streptomycin (P/S, Fisher Scientific, Germany) at 37 °C and 5% CO₂, until desired confluence was reached, before adding the CNDs.

Cell viability assays

The cell viability of THP-1 derived macrophages and HeLa cells after exposure to CNDs was evaluated by the resazurin assay^50,67. In viable cells there is a metabolic reduction of the non–fluorescent resazurin to the highly fluorescence of resorufin, and thus this fluorescence is assumed to be proportional to the number of living cells. In case of the addition of toxic materials the number of living cells will be decreased. As cell viability V the percentage of living cells in reference to a sample with untreated control cells is defined⁵⁰. Before evaluating the biocompatibility of the CNDs, several tests were performed to exclude the possibility that the fluorescent CNDs would interfere with the resazurin assay. In a first step, we investigated possible interference effects of CNDs with resazurin (i.e. without exposing the CNDs to cells), analyzing the concern that the CNDs alone could trigger the conversion of resazurin to resorufin. Briefly, several solutions were prepared as follows, including H₂O, RPMI 1640 medium supplemented with 10% FBS, 0.025 mg mL^–1 resazurin solution in RPMI 1640 medium with 10% FBS, C′_CND = 200 μg mL^–1 of R-CNDs or S-CNDs in H₂O, C′_CND = 200 μg mL^–1 of R-CNDs or S-CNDs in 10% FBS supplemented RPMI 1640 medium containing 0.025 mg mL^–1 resazurin (Sigma Aldrich, USA). Then 100 μL of each solution was loaded to a 96-well plate (Sarstedt, Germany) with 0.34 cm² growth area per well and the fluorescence spectra of each well was collected from 570 to 620 nm by a fluorimeter (Fluorolog-3, Horiba Jobin Yvon, USA) with excitation at 560 nm (Supplementary Fig. 38). Data show that the addition of CNDs to resazurin did not trigger fluorescence (i.e. conversion of resazurin to resorufin). In a second step, we investigated the interference effect of the fluorescence of the CNDs, which might interfere with the fluorescence of resorufin. THP-1 monocytes were seeded at a density of 34,000 cells/well with the medium volume V_medium = 0.136 mL per well in a 96-well plate with 0.34 cm² growth area per well and were differentiated to macrophages within in 3 days (Supplementary Fig. 37). Afterwards, the supernatant was removed and 100 μL of R-, S- or N–CNDs (C′_CND = 200 μg mL^–1) diluted in RPMI 1640 medium containing 10% FBS were added to the THP-1 derived macrophages. As control cells were exposed to medium without added CNDs. Cells were further incubated for 24 h in a humidified incubator at 37 °C, 5% CO₂. The next day, resazurin salt solution (Sigma Aldrich, USA) at a concentration of 0.25 mg mL^–1 was mixed with 10% FBS supplemented RPMI 1640 medium at a volume ratio of 1:10 as resazurin working solution. The control cells which had not been exposed to CNDs, the cells in each well were washed with 100 μL of phosphate-buffered saline (PBS, Gibco, Invitrogen, Belgium) and 100 μL of resazurin working solution was added. The CND-treated cells remained unchanged. Cells were further incubated at 37 °C and 5% CO₂ for 4 h. After this the fluorescence spectra of the different wells were collected from 570 to 620 nm at 560 nm excitation by a fluorimeter (Fluorolog-3, Horiba Jobin Yvon, USA; Supplementary Fig. 38b). As shown in Supplementary Fig. 38b, the significant fluorescence of resorufin from cells which have been treated with resazurin. The fluorescence from cells which had been treated with only CNDs was negligible in comparison to the resorufin fluorescence. This rules out interference of the viability test with the intrinsic CND fluorescence. For the viability tests slightly different protocol were used for the THP-1 monocytes and the HeLa cells. The THP-1 monocytes were seeded at a density of 34,000 cells/well with the medium volume V_medium = 0.136 mL per well in a 96-well plate with 0.34 cm² growth area and were differentiated to macrophages. On the fourth day, the supernatant in each well was removed and then CND solution (i.e. CNDs dispersed in medium with or without FBS supplement; V_medium = 0.136 mL) with a series of different concentrations was added for 24 or 48 h. In contrast, HeLa cells were seeded in 96-well plates with 0.34 cm² growth area at a density of 7500 cells/well in 0.1 mL DMEM medium supplemented with 10% FBS. On the following day, the old medium was removed then CND solution (V_medium = 0.1 mL; with or without FBS supplement) with a series of different concentrations was added for 24 or 48 h. For both cell lines, after the incubation time, phosphate-buffered saline (PBS, Gibco, Invitrogen, Belgium) (V_PBS = 0.1 mL) was used to wash the cells once, then 100 μL of resazurin working solution was added and further incubated for 4 h at 37 °C. The resazurin working solution was prepared by diluting the resazurin salt solution (Sigma Aldrich, USA) at a concentration of 0.25 mg mL^–1 ten times with 10% FBS supplemented medium (RPMI 1640 in case of THP-1 derived macrophages and DMEM in the case of HeLa cells). Afterwards, the fluorescence spectra of each well were collected from 570 to 620 nm with an excitation of 560 nm as described above. Subsequently, Matlab software was used for data analysis based on the fluorescence intensity at 590 nm, which was considered proportional to the number of living cells. The viability V represent the fluorescence intensity of cells treated with CNDs normalized to the fluorescence intensity of untreated control cells. All experimental conditions were recorded from triplicate independent experiments. As shown in Supplementary Figs. 39 and 40, both cell lines maintained high viability at different exposure doses of CNDs ranging from C′_CND 0.488 µg mL^–1 to 1000 µg mL^–1 for the different incubation times (24 and 48 h) in cell culture medium supplemented with or without 10% FBS. These results are consistent with the other biocompatibility tests regarding CNDs⁶⁸.

Time- and dose-dependent uptake studies based on flow cytometry

Time-dependent uptake of CNDs by HeLa cells

Uptake of the different CNDs by HeLa cells was investigated. Firstly, HeLa cells were seeded at a density of 40,000 cells well^–1 with 10% FBS contained DMEM medium of volume V_medium = 1 mL per well in 24-well plates (Sartstedt, Germany) with 1.9 cm² seeding area per well. On the next day, the medium in each well was removed and then CNDs diluted in DMEM medium supplemented with 10% or 0% FBS (V_medium = 0.5 mL) were added to the HeLa cells for specific time points (1, 3, 6, 24 or 48 h) at a concentration of C′_CNDs = 400 µg mL^–1. After the exposure time, cells were washed three times with 0.5 mL cold PBS, detached by addition of 0.05% trypsin–EDTA, isolated by centrifugation at 300 × g for 5 min, and finally re-suspended in 0.3 mL cold PBS for flow cytometer analysis (BD LSRFortessa™, BD Biosciences, US). The CND fluorescence signal I within each cell was collected with the flow cytometer with a 450/50 nm bandpass filter upon 405 nm excitation. 10,000 gated cells were counted and analyzed for each sample. Then, the Flowjo software was used to analyze the flow cytometry data. The recorded mean CND fluorescence per cell I was then background-corrected by subtracting the fluorescence of control cells which had not been exposed to CNDs, leading to the background-corrected mean CND fluorescence per cell:

Finally, as the different CNDs at the same concentration C′_CND have a different fluorescence the correction factor X as determined in Supplementary Table 4 was taken into account:

These effective fluorescence intensities per cell I′ were the values which were further compared. In Supplementary Fig. 23 the raw data of the flow cytometry measurements are shown. In Supplementary Fig. 22 the mean fluorescence intensity per cell data as extracted from those raw data are plotted. As additional parameter also the percentage of HeLa cells which had endocytosed CNDs P_cell was calculated from the flow cytometer data (Supplementary Fig. 23) by the Flowjo software. Here a fluorescence threshold was set to distinguish cells with fluorescence from internalized CNDs from the autofluorescence of cells. The gating strategy is shown in Supplementary Fig. 41. The results from n = 3 experiments are displayed in Supplementary Fig. 42.

Time- and dose-dependent uptake of CNDs by THP-1 derived macrophages

Apart from HeLa cells we selected THP-1 derived macrophages as cell model. Macrophages are important cells of the immune system, capable of distinguishing pathogens such as bacteria, cellular debris and foreign entities^69,70. Thus, they might be in particular sensitive concerning the surface properties of CNDs (such as different chirality of the CNDs) in regard of CND endocytosis. Briefly, THP-1 monocytes were seeded at a density of 100,000 cells/well with 10% FBS containing RPMI 1640 medium with volume V_medium = 0.4 mL per well in 48-well plates (Sartstedt, Germany) with 1 cm² growth area per well and they were differentiated into THP-1 derived macrophages within 3 days (see the respective section in “Methods”). Then, the supernatant was removed, and CND solution (V_medium = 0.4 mL) was added to the THP-1 derived macrophages at final concentrations ranging from C′_CNDs = 50 µg mL^–1 to 400 µg mL^–1 for different time points t, including 1, 3, 6, 24 and 48 h in RPMI 1640 medium supplemented with 10% or 0% FBS. After the different incubation time points, cells were washed three times with 0.5 mL cold PBS, detached by 0.05% trypsin–EDTA solution (Thermofisher, USA), isolated by centrifugation at 300 × g for 5 min, and were then re-suspended in 220 μL cold PBS. Subsequently, the re-suspended cells were analyzed by flow cytometry in the same way as described in the respective section in “Methods” for the HeLa cells. In the same way background subtraction and adjustments for the different fluorescence intensities of the CNDs was performed. The results for different incubation concentrations C′_CNDs and incubation times are provided in Supplementary Figs. 25 and 43.

Time-dependent uptake studies based on confocal microscopy

Uptake of CNDs by THP-1 derived macrophages was also quantified by confocal laser scanning microscopy (CLSM; LSM 510, Zeiss, Germany) with a Plan–Apochromat ×63/1.40 Oil DIC M27 objective. THP-1 monocytes were seeded at a density of 75,000 cells per well in complete RPMI 1640 medium with the volume V_medium = 0.3 mL per well in µ-Slide 8 Wells (ibidi GmbH, Germany) with 1 cm² growth area per well, and were differentiated into THP-1 derived macrophages within 3 days (see the respective section in “Methods”). On the fourth day, the medium was removed, and CND solution (V_medium = 0.3 mL) was added to cells at a concentration of C′_CNDs = 400 μg mL^–1 in RPMI 1640 medium supplemented with 10% or 0% FBS subsequently. At specific time points t similar to those chosen for flow cytometry experiments (see the respective section in “Methods”), cells were imaged by CLSM using a 405 nm laser as the excitation source and a LP 420 nm long pass filter for recording the fluorescence emission. Representative images are shown in Supplementary Figs. 52 and 53. After obtaining the CLSM images at each time point, quantitative analysis of cellular CND uptake was performed by utilizing a combination of free open–source software. In a first step, the images as obtained from the Zeiss microscope software were converted to TIFF format utilizing Matlab software. In a second step, Adobe photoshop CS6 was used to manually denote the perimeter of the cells. In a third step, the sum of the fluorescence intensities of all pixels belonging to a cell was calculated by the image analysis software Cellprofiler v2.2.0, and converted to the mean fluorescence per cell by dividing the summed up pixel intensities by the number of fluorescent cells^54,71. The intensity values were then corrected by the fluorescence difference between the different CND sample according to Supplementary Table 4:

More than 200 cells in at least 20 images from three independent experiments were analyzed for each time point (n = 3). No background correction was performed, as the background fluorescence was low. The results are shown in Supplementary Fig. 12 and Supplementary Table 8. In addition, from the microscopy data exemplary shown in Supplementary Figs. 52 and 53 the percentage P_cell of cells which had internalized CNDs was determined. Data are shown in Supplementary Fig. 54. This counting was performed manually, which was possible due to the low background. To achieve a low background first a control cell which had not been exposed to CNDs was imaged and the parameters of the confocal microscope (e.g. laser power and pinhole) were adjusted in a way that no fluorescence could be observed by the naked eye in the fluorescence images. With these settings then the cells with internalized CNDs were recorded.

Colocalization of CNDs with intracellular organelles

Colocalization studies of mitochondria or lysosome and CNDs

Colocalization studies of internalized CNDs with cell organelles, i.e. mitochondria and lysosomes, were carried out for THP-1 derived macrophages and for HeLa cells using Confocal Laser Scanning Microscopy (CLSM) (LSM 510, Zeiss, Germany) with a Plan–Apochromat ×63/1.40 Oil DIC M27 objective. Firstly, THP-1 monocytes were seeded at a density of 75,000 cells per well with complete RPMI 1640 medium volume V_medium = 0.3 mL supplemented with PMA at a concentration of 150 nM in µ-Slide 8 Wells (ibidi GmbH, Germany) with 1 cm² growth area per well. After 72 h incubation time in a cell culture incubator, the THP-1 monocytes had been differentiated into THP-1 derived macrophages. In the case of HeLa cells, 12,000 cells were seeded per µ-Slide 8 Well with the complete DMEM medium of volume V_medium = 0.3 mL per well and were cultured in a cell culture incubator at 37 °C in 5% CO₂ overnight. After this, for both cells type the previous medium was replaced with the V_medium = 0.3 mL CNDs dispersed in RPMI 1640 and DMEM medium supplemented with 10% or 0% FBS at a concentration of C′_CNDs = 400 μg mL^–1 for THP-1 derived macrophages and HeLa cells, respectively. After 24 or 48 h incubation time, mitochondria and lysosome were labeled with corresponding staining reagents as described in the following.

Immunostaining procedures

For mitochondrial staining, MitoTracker^R Deep Red^FM (Catalog No.: M22426, ThermoFisher Scientific)⁷² was used to specifically label the mitochondria. Briefly, cells were washed three times with 200 μL PBS and then 200 μL of pre-warmed (37 °C) MitoTracker^R Deep Red^FM in complete RPMI 1640 medium at a concentration of 400 nM was added and cells were further incubated in the incubator for 30 min at 37 °C. Afterwards, the staining solution was replaced with fresh pre-warmed RPMI 1640 medium without phenol red (Catalog No.: 11835030, ThermoFisher Scientific) and cells were observed using CLSM. A laser diode emitting at 405 nm and a bandpass emission filter BP 420–480 nm were used to visualize the CNDs. A helium–neon laser of 633 nm and long pass filter LP 650 nm were used for recording the fluorescence of MitoTracker^R Deep Red^FM. For lysosome staining, LysoTracker™ Green DND-26 (Catalog No: L7526, ThermoFisher Scientific) was selected as the lysosomal marker^73,74,75. Firstly, cells were washed three times with 200 μL PBS. Subsequently, 200 μL of pre-warmed (37 °C) LysoTracker™ Green DND-26 at a concentration of 1 μM in complete RPMI 1640 medium was added and cells were further incubated at 37 °C in 5% CO₂ for 30 min prior to imaging with CLSM. The excitation laser and emission collection setups were the same for the CNDs as that used in the colocalization studies of mitochondria and CNDs. An argon laser of 488 nm together with the BP 505–530 nm bandpass filter were used for observing the fluorescence of LysoTracker™ Green DND-26.

Calculation of Manders’ coefficients from the colocalization data

Based on the CLSM images, colocalization was quantified by quantitatively calculating Manders’ coefficients m₁ and m₂, which are indicators of the overlap degree between pixels from two different fluorescence channels ranging from 0 to 1 ^73,76,77. To achieve this purpose, Matlab and Cellprofiler v2.2.0 were used to calculate Manders’ coefficients^54,77,78. Briefly, the Matlab software was firstly used to subtract the background from the 8-bit grayscale TIFF images. Secondly, Cellprofiler v2.2.0 was used to identify pixels belonging to cells. Then the colocalization of CNDs and mitochondria or lysosomes for all pixels corresponding to cells was calculated. Thirdly, the below given equations were used to calculate Manders’ coefficients using Matlab. Hereby m₁ is the percentage of blue fluorescent pixels (i.e. CNDs) that overlapped with red or green fluorescent pixels (i.e. mitochondria or lysosomes). m₂ is the percentage of red or green fluorescent pixels (i.e. mitochondria or lysosomes) which overlap with blue fluorescent pixels (i.e. CNDs).

Hereby i = 1… N denotes all N pixels which belong to cells. I(B)_i, I(R)_i, and I(G)_i are the fluorescence intensities I at pixel i as obtained from the blue, red, and green channels of the fluorescence images. I(B)_i,coloc, I(R)_i,coloc, and I(G)_i,coloc are the fluorescence intensities I from the blue, red, and green channel only for pixels i where there is also green/red, blue, and blue fluorescence. m₁ or m₂ = 0 represents no overlap, while 1 denotes complete overlap of the fluorescence channels. As expected, the CNDs were largely localized inside lysosomes, and did not co-locate to a large amount with mitochondria. Selected fluorescence images and the resulting Manders’ coefficient calculations are shown in Supplementary Figs. 15 and 56–66.

Studies about the uptake pathway

Distinguishing CNDs adherent to the cell membrane from endocytosed CNDs

Flow cytometry based on standard fluorophores cannot distinguish trivially between CNDs adherent only on the outer cell membrane from actually endocytosed CNDs⁷⁹. Both scenarios however can be distinguished using confocal microscopy. To confirm that the majority of fluorescence signal from CNDs associated with cells originates from internalized CNDs z-stacks were recorded⁸⁰ with CLSM (LSM 510, Zeiss, Germany) with a PlaN-Apochromat ×63/1.40 Oil DIC M27 objective. Herein, Calcein–AM (Molecular Probes) was applied to label living cells (i.e. cells with esterase activity)⁵⁰, through enzymatic transformation from the non–fluorescent calcein AM to the highly fluorescent calcein⁸¹. Esterase activity only happens inside cells, and thus is a good label for the volume of the cell. In case the blue fluorescence CNDs overlapped with the green calcein AM fluorescence they can be considered internalized. The experiment was performed in the following way. On the first day, THP-1 monocytes were seeded at a density of 75,000 cells per well with complete RPMI 1640 medium (V_medium = 0.3 mL) in µ-Slide 8 Wells with 1 cm² growth area per well and were differentiated into THP-1 derived macrophages for 3 days. Then, the supernatant was removed and afterwards the CND (V_medium = 0.3 mL) dispersed in RPMI 1640 medium with or without serum were added to cells at a concentration of C′_CNDs = 400 μg mL^–1 for 24 h. Afterwards, cells were washed three times with PBS (200 μL each) followed by adding 200 μL pre-warmed (37 °C) Calcein AM (one component of the LIVE/DEAD™ Viability/Cytotoxicity Kit, Catalog No: L3224, Thermofisher Scientific) diluted in complete RPMI 1640 medium at a concentration of 1.3 μM. The cells were further incubated with the calcein staining solution at 37 °C for 30 min. Afterwards, the cells were washed with 200 μL PBS once and 300 μL of fresh pre-warmed RPMI medium without phenol red was added prior to imaging by confocal microscopy using z-stack analysis (Supplementary Figs. 67 and 68)⁸⁰. A laser diode emitting at 405 nm and a bandpass emission filter BP 420–480 nm were used for the excitation and emission collection of CNDs. An argon laser of 488 nm together with a bandpass emission filter BP 505–550 nm were used to visualize the fluorescence of calcein. In Supplementary Figs. 67 and 68, cross–sections through cells as indicated by the red and green solid lines are shown. These data clearly show that the blue fluorescence emitted from the CNDs originates from inside the cells, and thus corresponds to endocytosed CNDs.

Uptake of CNDs by THP-1 derived macrophages under the presence of inhibitors

The endocytosis pathway of CNDs can be investigated by evaluating the inhibition of certain internalization pathways by pharmacological/chemical inhibitors associated with⁸². In our case, we investigated the cellular uptake pathway of CNDs using the following chemical inhibitors: (i) Nocodazole⁸³ (an inhibitor of endocytosis of lager nanoparticles^82,84,85, CAS No.31430-18-9, Sigma-Aldrich, Germany), (ii) Bafilomycin A1⁸⁴ (an inhibitor of phagocytosis^86,87, CAS No. 88899-55-2, InvivoGen, France), (iii) Amiloride⁸⁸ (an inhibitor of micropinocytosis^89,90, CAS No. 17440-83-4, Sigma-Aldrich, Germany), and (iv) Chlorpromazine⁸⁸ (an inhibitor of clathrin–associated endocytosis^89,91, CAS No.69-09-0, Sigma-Aldrich, Germany). While the inhibitors were applied, still cytotoxicity experiments for the above inhibitors were carried out to ensure that at the used concentrations the inhibitors are not toxic. For experiments on the first day, THP-1 monocytes were seeded at a density of 34,000 cells per well with complete RPMI 1640 medium (V_medium = 0.136 mL) in 96-well plates with 0.34 cm² growth area per well and were differentiated into THP-1 derived macrophages. On the fourth day, the supernatant was removed and then each inhibitor diluted in RPMI 1640 medium supplemented with 10% or 0% FBS at several concentrations was added to the cells for 7 or 25 h. The exposure concentrations of each inhibitor are summarized in Supplementary Table 16. After the incubation time of 7 h or 25 h, the cell viability test was performed following the protocols described in the respective section in “Methods”. As shown in Supplementary Fig. 69, the four chemical inhibitors were non–toxic to THP-1 derived macrophages at the used concentration ranges. To ensure that there is also not reduction of cell viability of the THP-1-derived macrophages after exposing to CNDs pre-incubated with the cellular uptake inhibitors, also for this scenario a viability assay was carried out. Briefly, THP-1 monocytes were differentiated into THP-1 derived macrophages with an original seeding density of 34,000 cells per well with complete RPMI 1640 medium volume (V_medium = 0.136 mL) in 96-well plates. After 3 days, the previous medium was substituted with 0.136 mL of fresh RPMI 1640 medium containing 10% or 0% FBS containing optionally Nocodazole, Bafilomycin A1, Amiloride, or Chlorpromazine. The concentrations used for each inhibitor are described in Supplementary Table 17. Afterwards, cells treated with the inhibitors were incubated at 37 °C. Cells cultured at 37 °C without exposure to inhibitors were used as positive controls. Cells cultured at 4 °C (i.e. conditions where there is reduced endocytosis) without exposure to inhibitors were used as negative controls. After 1 h incubation time, the three types of CNDs were directly added at a final concentration of C′_CNDs = 400 μg mL^–1 and the 96-well plates were further incubated at the original conditions for another 6 h before the viability measurements. As shown in Supplementary Figs. 70 and 71, the exposure of the above inhibitors together with the CNDs did not reduce viability of the THP-1 derived macrophages. Thereafter, the cellular uptake pathway of CNDs in THP-1 derived macrophages was investigated. The experiment was conducted as follows. On the first day, THP-1 monocytes were seeded at a density of 100,000 cells per well with 10% FBS containing RPMI 1640 medium (V_medium = 0.4 mL) in 48-well plates (Sartstedt, Germany) with 1 cm² growth area per well. After 3 days, the cells were differentiated into THP-1 derived macrophages. Afterwards, the previous cell culture medium was replaced with fresh RPMI 1640 medium containing 10% or 0% FBS, supplemented optionally with Nocodazole, Bafilomycin A1, Amiloride, or Chlorpromazine at the concentrations described in Supplementary Table 17. The cells treated with the inhibitors were cultured at 37 °C for 1 h. Cells without exposure to inhibitors were cultured at 37 °C as positive controls. Cells without exposure to inhibitors were cultured at 4 °C (i.e. there is reduced endocytosis) as negative control. After the incubation time, the three types of CNDs were directly added at a final concentration of 400 μg mL^–1 and the plates were further incubated at the original conditions for another 6 h before flow cytometry analysis. The sample collection procedure, flow cytometry setups, and data analysis methods were the same as described in the respective section in “Methods”. As shown in Supplementary Figs. 72 and 73, the cellular uptake of the CNDs was blocked at 4 °C, suggesting that the internalization of the CNDs is an energy-dependent process⁵⁵. Presence of Bafilomycin A1 drastically reduced the uptake of CNDs. Thus, phagocytosis will be a major route of uptake of the CNDs^86,87. In case of the N–CNDs also presence of Chlorpromazine reduced uptake of the CNDs. Thus here also clathrin–associated endocytosis will be a relevant uptake pathway for N–CNDs^89,91. We point out again that the N–CNDs were partially agglomerated, which might be the reason for this additional pathway, which is most likely not related to the chirality of the CNDs. Addition of nocodazole did not reduce CND uptake, which is understandable as this has been reported to block in particular uptake of larger particles^84,85. Amiloride caused autofluorescence in the cells, which is why under presence of this inhibitor there is elevated fluorescence also in the presence of CNDs, which however is autofluorescence of the inhibitor.

Uptake of CNDs by Hela cells under the presence of inhibitors

The same experiments as done with THP-1 derived macrophages were also carried out for Hela cells. First, the cytotoxicity of CNDs pre-incubated with cellular uptake inhibitors was conducted. For that, 7500 HeLa cells in 0.1 mL complete DMEM medium were seeded in 96-well plates with a growth area of 0.34 cm² per well at 37 °C overnight. Afterwards, the old medium was removed and HeLa cells were incubated with fresh DMEM medium containing 10% or 0% FBS without or with inhibitors at the concentrations enlisted in Supplementary Table 17. Then, the inhibitor treated cells were incubated at 37 °C. As positive control cells without presence of inhibitors were used. As negative control cells without exposure to inhibitor were cultured at 4 °C instead at 37 °C. After 1 h incubation, the three types of CNDs were added to the cells at a final concentration of C′_CNDs = 400 μg mL^–1 (i.e. the inhibitors were not removed) and the plates were furtherly cultured at the original conditions (37 °C or 4 °C for the negative control) for another 6 h. Then the resazurin assay was applied. The results shown in Supplementary Figs. 74 and 75 demonstrated that the CNDs and chemical cellular uptake inhibitors did nor reduce the viability of the HeLa cells. In order to analyze the effect of the different inhibitors on the uptake of CNDs by Hela cells the same strategy as for the THP-1 derived macrophages was used. Briefly, 40,000 HeLa cells in 1.0 mL 10% or 0% FBS containing DMEM medium were seeded into 24-well plates with 1.9 cm² growth area per well at 37 °C overnight. On the following day, the medium was substituted with fresh DMEM medium supplemented with 10% or 0% serum without or with cellular uptake inhibitors at desired concentrations (Supplementary Table 17). The cells treated with inhibitors (or as positive control without inhibitors) were incubated at 37 °C for 1 h. As negative control cell which have not been treated with inhibitor were cultured at 4 °C for 1 h. After this incubation interval, the three types of CNDs were added to the cells (without removing the medium with the inhibitors) at a final concentration of C′_CNDs = 400 μg mL^–1. Then the plates were returned back with the original incubation conditions (37 °C or 4 °C for the negative control) for another 6 h before sample collection for flow cytometry analysis. The sample collection, flow cytometry setups and data analysis approach were used as described in the respective section in “Methods”. Note that here no correction for the different fluorescence emissions of the different types of CNDs was carried out, i.e. I and not I′ is plotted. As can be seen in Supplementary Figs. 76 and 77 incubation at 4 °C blocked CND uptake by cells, indicating that endocytosis of the CNDs is an energy consumed process. Besides, the three types of CNDs were taken up by the HeLa cells largely via a phagocytosis pathway, while N–CNDs could also be endocytosed by the HeLa cells via a clathrin–associated endocytosis pathway, similar to the results obtained for the THP-1 derived macrophages.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.