Hydrodynamic chromatography coupled to single-particle ICP-MS for the simultaneous characterization of AgNPs and determination of dissolved Ag in plasma and blood of burn patients

Abstract

Silver nanoparticles (AgNPs) are increasingly used in medical devices as innovative antibacterial agents, but no data are currently available on their chemical transformations and fate in vivo in the human body, particularly on their potential to reach the circulatory system. To study the processes involving AgNPs in human plasma and blood, we developed an analytical method based on hydrodynamic chromatography (HDC) coupled to inductively coupled plasma mass spectrometry (ICP-MS) in single-particle detection mode. An innovative algorithm was implemented to deconvolute the signals of dissolved Ag and AgNPs and to extrapolate a multiparametric characterization of the particles in the same chromatogram. From a single injection, the method provides the concentration of dissolved Ag and the distribution of AgNPs in terms of hydrodynamic diameter, mass-derived diameter, number and mass concentration. This analytical approach is robust and suitable to study quantitatively the dynamics and kinetics of AgNPs in complex biological fluids, including processes such as agglomeration, dissolution and formation of protein coronas. The method was applied to study the transformations of AgNP standards and an AgNP-coated dressing in human plasma, supported by micro X-ray fluorescence (μXRF) and micro X-ray absorption near-edge spectroscopy (μXANES) speciation analysis and imaging, and to investigate, for the first time, the possible presence of AgNPs in the blood of three burn patients treated with the same dressing. Together with our previous studies, the results strongly support the hypothesis that the systemic mobilization of the metal after topical administration of AgNPs is driven by their dissolution in situ.

Appendix

Characterization of the commercial AgNP standard dispersions

The morphological characterization of the mother AgNP standard dispersions was carried out by TEM using a Tecnai 12 G² instrument (FEI, USA). For the analysis, a 3 μL drop of each dispersion was deposited on a Formvar/carbon support on 200-mesh thick grid, let dry at room temperature and directly analyzed. The images were acquired at 120 kV high voltage and using a tungsten filament, twin optics and an Olympus side-mounted camera. The ImageJ software (National Institutes of Health, USA) was used for particle counting and shape characterization. Representative TEM images of the NP standards and a summary of their size/shape parameters are shown in Fig. 8 and reported in Table 3, respectively.

Fig. 8

TEM images of the AgNP standards (mother suspensions) used throughout the study. The nominal sizes are 10 nm (a), 20 nm (b), 40 nm (c), 60 nm (d) and 100 nm (e)

Table 3 Characterization of the AgNP standards (mother suspensions) used throughout the study

The z-potential was measured using a Zetasizer Nano (Malvern, UK) at 24 °C in DTS1070 cells pre-washed with a 60 μg mL⁻¹ citrate-water solution. For the analysis, each standard suspension was sonicated for 5 min and equilibrated for 120 s in the cell, and five replicate measurements of 10 to 100 readings were acquired. The z-potential of the NP standards is also reported in Table 3.

Total mass concentration of Ag in the standards was measured by ICP-MS previa dissolution in HNO₃ 5 % v/v and subsequent dilution in NH₄OH 2.8 % w/w. The analysis was carried out in full-quant mode by external calibration with Rh as internal standard.

Methods for μXRF and μXANES analyses

Standards and samples

Solid-state reference compounds for μXANES included the following: Ag⁰ foil, AgCl, Ag₂SO₄, AgNO₃, Ag₂O, Ag sulfadiazine (AgSD) and a fragment of the Acticoat Flex3™ intact dressing. Reference standards of 10 and 100 nm Ag⁰NPs were prepared from mother water suspensions (citrate stabilized, 20 μg mL⁻¹ as Ag) by deposition of a 20 μL drop between Ultralene® windows and microscopy slides, followed by rapid freezing and freeze-drying for 24 h. A standard of Ag bonded to GSH was prepared by incubating ionic Ag (from AgNO₃, 10 μg mL⁻¹ as Ag) in a water solution of GSH ~0.5 mg mL⁻¹, at 37 °C under gentle shaking for 2 h and in dark conditions, followed by freeze-drying as reported above. Standards of 10 nm Ag⁰NPs, ionic Ag and the Acticoat Flex3™ intact dressing (2.45-mg fragment) were also incubated in a water solution containing HSA (~0.5 mg mL⁻¹) and in the whole human plasma, freeze-dried as reported above and analyzed as unknown samples.

Instrumental parameters and data elaboration

The μXRF and Ag L_III-edge μXANES measurements were performed using the scanning X-ray microscope of beamline ID21 at the European Synchrotron Radiation Facility (ESRF, Grenoble, France), working at room temperature conditions. Detectors included a Si₃N₇ diode for I ₀ and an 80-mm active area silicon drift detector (Bruker) for the emitted fluorescence. Focusing was achieved using fixed curvature Kirkpatrick-Baez mirror optics. The photon flux was 3.6 × 10⁹ ph s⁻¹ at 3.42 keV with a beam size of 1.0 × 1.2 μm².

μXRF maps of signal intensities for individual elements (Ag, S and Cl) were collected for preliminary analysis to select optimal regions for the subsequent μXANES analysis. The μXRF maps were acquired with variable lateral resolution (0.5 to 2 μm) and integration time (100 ms). The raw data (counts) were elaborated using the PyMca software as follows: (i) correction for the settling time and conversion to counts per second (cps), (ii) deconvolution (batch fitting of the μXRF spectra) and (iii) normalization for the incident beam flux.

Batch Ag L_III-edge μXANES spectra of 30 s were collected and averaged for each spot of interest from 3.32 to 3.42 keV energy range with 0.5 eV steps. The beam position was slightly moved from one spectrum to another to avoid radiation damage. At least 10 spectra were averaged for each region of interest. After background removal and normalization, the spectra were calibrated by taking the first inflection point of at 3.3545 keV and then smoothed by interpolation with five iteractions. The μXANES spectra of Acticoat Flex3™ and HSA/plasma-incubated standards were treated by LCF using the Athena software with the following set of reference spectra as independent variables: Ag⁰ foil, 10 nm Ag⁰NPs, AgCl, Ag₂SO₄, AgNO₂, Ag₂O, AgSD and AgGSH. A linear term was allowed to compensate for small differences in data normalization, no energy shift was allowed, and weights and their sum were forced to sum to 1. The energy range used for fitting was 3.3345 to 3.4145 keV (e0 − 0.02 to e0 + 0.06). The quality of fitting was quantified by the normalized sum of squared residual NSS = (μ _experimental − μ _fit)² / Σ(μ _experimental)² × 100, where μ is the normalized absorbance. Linear combinations of one, two and three reference standards were examined. The best fit with n + 1 components was retained if NSS was decreased by more than 15 % as compared to the best fit using n components. Based on the results, four main Ag species were revealed in the samples: Ag⁰ foil, 10 nm Ag⁰NPs, AgCl and AgGSH; the other minority species were pooled as other. When two or more fits of equivalent quality (relative difference of NSS <10 %) were obtained with different combinations of such a minority species, proportions and NSS were expressed as mean percentage with standard deviation (SD) between parentheses, calculated for the equivalent fits.

The μXANES spectra were acquired in fluorescence mapping mode by scanning the beam with a 2 × 2 μm² step size and a 50 ms dwell time per pixel with 3 eV energy steps in the region from 3.320 to 3.341 KeV, 0.5 eV from 3.341 to 3.381 KeV and 1 eV from 3.382 to 3.42 KeV. This resulted in a total of 126 images recorded using a region of interest selective for Ag L3M4 and L3M5 emission lines, corrected for detector dead time (always kept below 20 %) and normalized by the incident beam flux. The stack of images was converted to an hdf5 file containing intensities and the energy values for each map to be processed using PyMca for extraction of μXANES spectra. The map was treated by moving merge of the spectra on 2 × 2 pixels areas and with 1 pixel step, in order to reduce the noise and improve the statistical representativity. After background removal and normalization, the spectra were calibrated and individually processed by LCF as above, but using only the reference spectra of 10 nm Ag⁰NPs, AgCl and AgGSH as independent variables. Based on visual inspection of the fits, an arbitrary threshold of NSS <0.1 was adopted to remove the pixels from the map with insufficient quality of the fit. A number of pixels in the upper right side of the map (see Fig. 4b) were discarded based on this criterion. These pixels were affected by the sharp change in intensity at the border of the analyzed particle coupled to beam drift caused by scanning the energy with the double-crystal monochromator. The retained pixels were re-processed by LCF testing all combinations in which one of the three reference standards was removed. Each standard was considered significant if its introduction decreased the NSS by more than 3 %. This threshold was calibrated a posteriori to guarantee that all pixels in the map had at least one significant component, and a coefficient equal to zero was assigned to the non-significant components. For each pixel, the coefficients were finally multiplied for the corresponding signal intensity of total Ag (from the μXRF map), also treated by the moving merge procedure, to obtain the absolute contribution of each species expressed in cps.