Methods of Granulocyte Isolation from Human Blood and Labeling with Multimodal Superparamagnetic Iron Oxide Nanoparticles

This in vitro study aimed to find the best method of granulocyte isolation for subsequent labeling with multimodal nanoparticles (magnetic and fluorescent properties) to enable detection by optical and magnetic resonance imaging (MRI) techniques. The granulocytes were obtained from venous blood samples from 12 healthy volunteers. To achieve high purity and yield, four different methods of granulocyte isolation were evaluated. The isolated granulocytes were labeled with multimodal superparamagnetic iron oxide nanoparticles (M-SPIONs) coated with dextran, and the iron load was evaluated qualitatively and quantitatively by MRI, near-infrared fluorescence (NIRF) and inductively coupled plasma mass spectrometry (ICP-MS). The best method of granulocyte isolation was Percoll with Ficoll, which showed 95.92% purity and 94% viability. After labeling with M-SPIONs, the granulocytes showed 98.0% purity with a yield of 3.5 × 106 cells/mL and more than 98.6% viability. The iron-loading value in the labeled granulocytes, as obtained by MRI, was 6.40 ± 0.18 pg/cell. Similar values were found with the ICP-MS and NIRF imaging techniques. Therefore, our study shows that it is possible to isolate granulocytes with high purity and yield and labeling with M-SPIONs provides a high internalized iron load and low toxicity to cells. Therefore, these M-SPION-labeled granulocytes could be a promising candidate for future use in inflammation/infection detection by optical and MRI techniques.


Introduction
Inflammation is the immune system s response to pathological agents (infection), damaged cells and toxins, and it aims to maintain integrity, defense against microorganisms and cellular and tissue repair [1][2][3]. Acute inflammation of the lower respiratory tract is a leading cause of mortality and morbidity worldwide, with the greatest prevalence in children younger than five years [4][5][6]. Sepsis is another important public health problem; it occurs in 30% to 50% of hospitalizations that culminate in death [7,8] due to exacerbated and inappropriate inflammation, and it can cause septic shock and organ failure [9]. In addition, the chronic inflammation process increases therapy and care costs, as well as mortality [10].
Thus, this in vitro study aimed to find the best (highest purity and yield) granulocyte isolation technique, with the aim of labeling the granulocytes with multimodal nanoparticles detectable by MRI-NIRF imaging, for qualitative and quantitative evaluations of their internalized M-SPION loads, a promising candidate for future research use in detecting inflammation/infection processes by optical and MRI techniques.

Evaluation of Granulocyte Isolation by Different Methods
Granulocytes were isolated by four different methods: method I-Percoll gradients, method II-Percoll with phosphate buffer solution (PBS), method III-Percoll with Hank s balanced salt solution (HBSS) and method IV-Percoll and Ficoll. Then, we assessed the purity of the granulocytes isolated by each method ( Table 1). The highest percentage mean purity was found in method IV, at 95.92% ( Figure 1A-iv), in which the granulocytes were deposited in the pellet, followed by method III, at 82.75% ( Figure 1A-iii), in which the granulocytes were concentrated in the white layer between the Percoll densities. Method I had a purity of 57.89% ( Figure 1A-i), and the granulocytes were concentrated in the white layer above the sediment of the red blood. Method II had a purity of 47.86% ( Figure 1A-ii), and all cells sedimented without visible separation. These results are depicted in the boxplot ( Figure 1B). The nonparametric ANOVA test (Kruskal-Walls test) showed a significant difference between methods (p < 0.001). The post hoc test showed a significant difference in all comparisons (p < 0.008; Table 2).  In the same samples used for the granulocyte purity analysis, a contaminant analysis was performed for each method ( Table 3). The lowest percentage mean contamination was found in  Figure 1A-ii). No granulocyte isolation was observed in method II, due to the unique pellet formation. These results are depicted in the boxplot ( Figure 1C). The nonparametric ANOVA test (Kruskal-Walls test) showed a significant difference between methods (p < 0.001). The post hoc test showed a significant difference in all comparisons (p < 0.003; Table 4).
Method III: Percoll with HBSS 10.950 6.212 12 Method IV: Percoll and Ficoll 0.887 0.615 8 SD: standard deviation.      Figure 2 compares some properties of the granulocytes isolated by methods III and IV. Panels A1 and B1 present forward-scatter versus side-scatter plots of the granulocyte isolation rates of method III (30.6%) and method IV (58.1%), respectively. Panels A2 and B2 show the respective proportions of isolated granulocytes that were double-positive for CD15/CD13. Panels A3 and B3 show the percentages of leukocytes that were positive for CD45 (98.5% and 99.5%, respectively). Panels A4 and B4 show the percentages of granulocytes that were double-negative for CD19/CD14 (94.3% and 97.6%, respectively). Panels A5 and B5 show the percentages of granulocytes that were double-negative for CD3/CD14 (86.5% and 98%, respectively).
In the fast panoptic staining, we detected the percentages of granulocytes in both methods (yellow arrows in Figure 2C3 and 2D3). The isolated granulocytes made up a mean of 6.3% of cells (eosinophils, basophils and immature neutrophils) in method III (blue arrows in Figure 2C3) and 3.2% in method IV (blue arrows in Figure 2D2). They made up 20.68% of lymphocytes and contaminants in method III (red arrows in Figure 2C3) and only 0.61% of lymphocytes and contaminants in method IV.
The granulocyte viability was analyzed after 4 h of isolation through membrane integrity by flow cytometry, which showed 98.3% of granulocytes isolated by method III were viable ( Figure 2E) and 94.0% of those isolated by method IV ( Figure 2F). Both were double-negative for Annexin-FITC and propidium iodine-phycoerythrin (PI-PE).
The purity analysis of isolated granulocytes and their respective contaminants in methods III and IV showed a proportional inverse relationship, in which method IV had the highest and the lowest contaminant rate, as depicted in the boxplot of Figure 2 (G and H). The purity (Table 5) and contaminant percentage (Table 6) were significantly different (p < 0.001) between methods III and IV. In relation to the isolated granulocyte yield, means of (4.99 ± 1.24) × 10 6 per mL and (3.51 ± 1.40) × 10 6 per mL were found by methods III and IV, respectively, without a significant difference between methods, as shown by a Student s t-test (p = 0.996). The isolated lymphocyte yield between methods III and IV were significantly different by the Mann-Whitney test (p = 0.025), with respective means of (0.33 ± 0.07) × 10 6 per mL and (0.09 ± 0.06) × 10 6 per mL.    The M-SPION optical characterization depicted in Figure 3 shows the excitation/emission spectrum of M-SPIONs. The spectrum shows the visible fluorescent spectrum peak (ex/em: 559.4/580.7 nm) and NIRF absorption/emission (ex/em: 757.0/777.4 nm).

Visualization of M-SPION Internalization into Granulocytes and Cellular Viability
After labeling the isolated granulocytes with M-SPIONs, internalization was verified through fluorescence ( Figure 4A,C,E-G) and brightfield microscopy ( Figure 4B,D). Figure 4A shows the labeling of granulocytes with M-SPIONs by fluorescence microscopy, in which the fluorescent rhodamine-coupled M-SPIONs are highlighted in red, and the labeling of granulocyte nuclei by 4′,6diamidine-2′-phenylindole dihydrochloride (DAPI) is shown in blue. An amplified view (inset of Figure 4A) allowed us to better visualize the presence of M-SPIONs internalized into granulocytes. In brightfield microscopy analysis, it was possible to visualize the granulocytes labeled with M-SPIONs while highlighting the iron of the intracellular region of granulocytes by Prussian blue staining, which better visualized the labeling of granulocytes in the inset of Figure 4B. Figures 4C and  4D show images corresponding to the same field of view of the granulocytes labeled with M-SPIONs by fluorescence and brightfield microscopy. Therefore, the magnetic and fluorescent properties of these nanoparticles allowed detection by both techniques. Figure 4E-G show the M-SPION internalization in each part. First, rhodamine was imaged by fluorescence microscopy ( Figure 4E), followed by the image corresponding to granulocyte nuclear staining by DAPI ( Figure 4F) and overlapping images ( Figure 4G), where M-SPIONs were observed in the granulocyte cytoplasm.
After 4 h of labeling, granulocyte viability was evaluated with M-SPION concentrations of 10, 30 and 50 µ g Fe/mL, which showed viability (double-negativity for Annexin-FITC and PI-PE) values  Figure 3B shows the polydispersion curve adjusted to a log-normal distribution with an average hydrodynamic diameter of 35.7 ± 0.3 nm, compatible with the manufacturer's specifications. For M-SPION stability evaluation, the polydispersion curves were acquired temporally until reaching 20 h. In this period, the M-SPION hydrodynamic diameter did not change, and there was no evidence of M-SPION agglomeration due to the interaction of M-SPIONs with the elements that contained culture medium.
To analyze the electrostatic interaction of granulocytes with M-SPIONs, the surface charge was measured using the same pH of 7.4 found in the cell labeling, and a zeta potential of approximately 32 mV was verified.

Visualization of M-SPION Internalization into Granulocytes and Cellular Viability
After labeling the isolated granulocytes with M-SPIONs, internalization was verified through fluorescence ( Figure 4A,C,E-G) and brightfield microscopy ( Figure 4B,D). Figure 4A shows the labeling of granulocytes with M-SPIONs by fluorescence microscopy, in which the fluorescent rhodamine-coupled M-SPIONs are highlighted in red, and the labeling of granulocyte nuclei by 4 ,6-diamidine-2 -phenylindole dihydrochloride (DAPI) is shown in blue. An amplified view (inset of Figure 4A) allowed us to better visualize the presence of M-SPIONs internalized into granulocytes. In brightfield microscopy analysis, it was possible to visualize the granulocytes labeled with M-SPIONs while highlighting the iron of the intracellular region of granulocytes by Prussian blue staining, which better visualized the labeling of granulocytes in the inset of Figure 4B. Figure 4C,D show images corresponding to the same field of view of the granulocytes labeled with M-SPIONs by fluorescence and brightfield microscopy. Therefore, the magnetic and fluorescent properties of these nanoparticles allowed detection by both techniques.

Quantification of M-SPIONs Internalized into Granulocytes by MRI, ICP-MS and NIRF
MRI, ICP-MS, and NIRF were used to quantify the iron mass internalized into granulocytes after M-SPION incubation at concentrations of 10, 30 and 50 µ g Fe/mL. The quantification results of the iron mass per cell and the number of nanoparticles per granulocyte are shown in Table 7 and Figure  5.   Table 7 and Figure 5.

MRI Quantification
The MRI quantification was begun by calculating the r2 value; for this purpose, the T2-weighted image of the phantom was obtained using different echo times (TE) and containing the following concentrations of M-SPIONs dispersed in 1% agarose at: 5, 10, 15, 20, 25, 30, 35, 40 and 50 µg Fe/mL ( Figure 5A). The contrast decay of well images was evidenced by the M-SPION concentration rise for a set time of TE. From the signal intensity of the MRI (in arbitrary units (AU)), the transverse relaxation curves were obtained as a function of TE (ms), as depicted in Figure 5A. The curves showed exponential decay behavior that varied with the M-SPION concentration, and the corresponding T2 parameters were obtained from the exponential adjustment. To obtain the r2 value, a curve of the inverse transverse relaxation time (1/T2) was constructed as a function of iron concentrations, and then a linear adjustment was obtained for the relaxivity value of r2 = (19.9 ± 0.9) × 10 −4 ms −1 µg Fe −1 mL for M-SPIONs, as depicted in the inset of Figure 5A.
For the iron load calculation of Equation (1) Table 7). Figure 5C shows the contrast of the MR images produced by M-SPIONs internalized into granulocytes.   Table 7).

ICP-MS Quantification
The calibration curve for the ICP-MS quantification is shown in Figure 5F Table 7).
The differences between the iron load values found in the MRI, NIRF and ICP-MS quantifications of the M-SPIONs internalized into granulocytes are shown in the radar graph of Figure 5H.

Discussion
Granulocytes are known for their high potential and efficiency in detecting and eradicating microbial infections through an immune response against invading pathogens [66]. Aiming at granulocytelabeling with nanoparticles for later use in inflammation/infection process detection, different methods of granulocyte isolation were tested, and their labeling with multimodal nanoparticles allowed their internalization and their qualitative and quantitative evaluations.
Among the methods of granulocyte isolation evaluated, the method that achieved the highest degree of purity was the use of Percoll gradients, followed by the Ficoll method (method IV). However, until now, no method using the combination of Percoll and Ficoll has been reported, only each technique separately. In addition, there is no consensus on the best technique for isolating granulocytes to achieve high purity. The methods applied for granulocyte isolation are diverse [29][30][31][33][34][35]53,[67][68][69][70], and their efficiency has been reported in different ways, such as cellular viability or cell isolation yield, cell functionality, number of target cells and contaminants after isolation, which makes it difficult to evaluate the methods for clinical application.
Method IV, which resulted in the highest purity of granulocyte isolation, was performed in two stages. The first stage consisted of the use of Percoll gradients with HBSS (method III), and the second stage used Ficoll on the granulocytes isolated by method III. This first stage of granulocyte isolation was used in some studies as a unique stage of isolation [29,31,32], and the purity analysis, granulocyte yield and cellular viability were reported as 81% purity in 4 × 10 6 cells/mL [31], 95% purity in 2.2 × 10 6 cells/mL [29] and 95% purity for the yield range of 1 to 2 × 10 6 cells/mL, as well as 95% cellular viability [32]. These results corroborate our findings of 86.5% purity in 5.0 × 10 6 cells/mL and 98.3% cellular viability. After the second stage of isolation with Ficoll (method IV), which yielded 98.0% purity, a 3.5 × 10 6 cells/mL isolation yield and 94.0% cellular viability were found. The studies that used only Ficoll for granulocyte isolation reported purity greater than 90% but with a low isolation yield, ranging from 0.6 to 1.0 × 10 6 cells/mL [51,53,68,71,72].
Considering these results, after the first stage of Percoll (method III), we were able to isolate more granulocytes than described in the literature. In addition to superior performance in terms of high viability, we found similar purity to those described in other studies, which used the same method of isolation with Percoll gradients with HBSS [29,31,32]. After the use of Ficoll (method IV), the granulocyte yield decreased by 30% but without a significant difference compared to method III, maintaining 94% cellular viability and 98% purity in the granulocyte isolation. The reduction in the number of contaminants was significant, especially regarding the number of lymphocytes present, which was reduced by almost 73% after method IV. The goal of this study was to isolate specific cells without contaminants, especially for clinical applications, since the cell viability remained adequate for application.
The influences of Percoll and Ficoll on the functional integrity of granulocytes after isolation have been evaluated by different studies [72][73][74], which have demonstrated unaltered phagocytic features through the analysis of the percentage and number of particles phagocytized.
For granulocyte labeling, nanoparticles must have appropriate physicochemical properties in relation to size, coating, zeta potential, concentration and stability; these proprieties contribute to efficient internalization, maintaining the cell viability of the granulocytes [75][76][77]. Efficient labeling with the contrast agent is important in granulocyte detection by different imaging techniques supporting various clinical applications, such as immune response, diagnosis and therapies [51,[78][79][80][81].
Regarding the nanoparticle property analysis during granulocyte labeling, the use of M-SPIONs with a 35 nm diameter yielded a zeta potential of + 32 mV [76], with adequate temporal stability of the M-SPIONs, and coating with dextran allowed adequate internalization (~85%) with high cellular viability (>98.6%). In addition, a high-intensity signal was detected by MRI, NIRF and brightfield/fluorescence microscopy at the nanoparticle concentrations used, and the labeling time of 4 h has been considered adequate for clinical applications [61].
For internalization, it is best for the nanoparticles to have a positive surface charge (zeta potential), given the negative surface charge of granulocytes [75,82] and the electrostatic process involved [77,83,84]. Therefore, granulocyte and leukocyte-labeling studies that have used SPIONs with a negative zeta potential (0 to −53 mV) [51,54,57,58,85,86] have reported the necessity of using chemical or physical transfection agents or other strategies that would contribute to SPION internalization into cells and boost the internalization yield [51,54,57,86].
Other relevant aspects of labeling are the incubation time of cells with SPIONs, the SPION concentration used and its corresponding viability. In our study, we found satisfactory viability values using a 4-h labeling time (99.3%, 99.6% and 98.6% for the respective M-SPION concentrations of 10, 30 and 50 µg Fe/mL). In studies on granulocyte and leukocyte-labeling with SPIONs, the incubation time was reported to be between 1 and 4 h for labeling [51,54,57,59,61,62,85,86]. For example, in the study of Tang et al. [61], an incubation time of 3 h was best for the clinical application of these cells.
Regarding the stability of nanoparticles in the labeling process, previous studies by our group [90][91][92] evaluated SPION stability before labeling in different media and demonstrated that supplementation with 10% fetal bovine serum (FBS) contributed stabilized SPIONs in the culture medium during the labeling process [90,91,93]. This type of strategy was also reported in other studies [54,58,59,85,87], with FBS concentrations ranging from 5% to 20% during the labeling process of granulocytes or leucocytes with SPIONs. The use of FBS helps avoid the nanoparticle agglomerations that can occur due to the interaction of SPIONs with a culture medium that is supplemented with proteins and electrolytes [91,93], as well as helping maintain the equilibrium of existing forces involved in the interaction between the nanoparticles, such as electrostatic forces, van der Waals forces, steric forces and magnetic forces modulated by the Brownian motion associated with nanoparticles [93].
In most studies, SPION internalization into granulocytes or leukocytes is confirmed using Prussian blue staining and MRI [51,54,57,61], because the composition of nanoparticles is based on iron oxide. However, as the nanoparticles used in our study have multimodal features (magnetic/dual fluorescence), it was possible to verify internalization using Prussian blue staining and MRI, as well as using fluorescence images in the visible spectrum (rhodamine) and infrared spectrum (IR750). The evaluation by brightfield and fluorescence microscopy showed that most granulocytes were labeled with M-SPIONs, corroborating other studies that reported similar results by brightfield microscopy evaluation [51,54], as the nanoparticles used in these studies only had iron oxide nuclei. The unlabeled granulocytes may be related to their maturation, because, in earlier stages of maturation, there are descriptions of the incomplete functionality of granulocytes for the phagocytosis process [94]. The granulocyte stages of maturation were not evaluated in our study, which can be considered one of the study limitations and an explanation of the unlabeled granulocytes. The isolation of granulocyte populations from each stage of granulocyte maturation present in venous blood and their marking could help to better understand this result.
This study lays the foundation for future applications of granulocyte labeling with nanoparticles for inflammation/infection process detection by in vivo imaging techniques. The nanoparticles used (M-SPIONs) can be detected by MRI and NIRF techniques. The MRI qualitative analysis showed that the hypointense signal of the granulocytes labeled at M-SPION concentrations of 10, 30 and 50 µg Fe/Ml, when compared with the control image (unlabeled granulocytes), had a signal reduction of 13%, 37% and 74%, respectively, compared to a TE of 48.6 ms and a repetition time (TR) of 3000 ms. The corresponding T2 values were 56.6 ms for the control, 47.5 ms (labeling with 10 µg Fe/mL), 36.8 ms (labeling with 30 µg Fe/mL) and 23.2 ms (labeling with 50 µg Fe/mL). These results show adequate internalization and, due to the M-SPIONs, a high r2/r1 ratio of 165.8 (r2 = (19.9 ± 0.9) × 10 −4 ms −1 µg Fe −1 mL and r1 = (1.2 ± 0.5) × 10 −5 ms −1 µg Fe −1 mL), indicating a strong contrast in T2-weighted images [95][96][97][98][99]. In relation to the NIRF qualitative analysis, the NIRF intensity signal was possible due to the M-SPIONs used for granulocyte-labeling having a fluorophore coupled to the infrared wavelength. The NIRF intensity signal of the lowest M-SPION concentration was on the order of 10 8 photons/s (labeling with 10 µg Fe/mL), and, at 30 and 50 µg Fe/mL, the SPION concentrations were six and eight times higher than the lowest concentration. Thus, it was possible to obtain MRI and NIRF signals in in vivo studies.
The quantitative analysis of the iron load internalized into granulocytes using the MRI, ICP-MS and NIRF techniques shows values ( Table 7) that are in accordance with the signal detection limitations as a function of sensitivity, spatial resolution, temporal resolution and the concentration of contrast agent [38,39]. In studies by Shanhua et al. [54,86], quantification analysis of the iron load of the SPIONs internalized into granulocytes was performed using the ICP technique, which showed an iron load of 0.1265 pg/cell (0.1265 ppm/106 cell) in the best strategy of internalization using LPS and 5 mg/mL SPIONs. Comparing these results with our quantitative analysis by the same technique revealed a much higher iron charge using a lower M-SPION concentration. With M-SPION concentrations of 10, 30 and 50 µg Fe/mL, the iron loads were 1.08, 2.56 and 6.83 pg/cell, respectively. Thus, the iron load internalized into cells in our study using the lowest M-SPION concentration was 10 times higher than the best condition of Shanhua et al. [54,86], and the highest M-SPION concentration yielded a result 54 times higher. The quantifications by MRI and NIRF techniques showed similar results, as depicted in Table 7. When the granulocyte viability after labeling was calculated, the highest M-SPION concentration used in our study showed 98% viability, whereas the studies that used 100 times more SPIONs reported 70% cellular viability after labeling [54,86].
One of the limitations of SPION exogenous-labeling of immune cells is the low sensitivity of MRI detection [50] compared to endogenous-labeling (MPIO) techniques, where MRI detection sensitivity is higher and provides high contrast [56,100] but has the disadvantage of false positive results in cases where hemorrhages and blood oxygen level-dependent (BOLD) effects induce signal voids in T2*-weighted images, as well as regarding toxicity, lack of biodegradability and accumulation in the reticuloendothelial system, which limit clinical use [50]. To improve the low sensitivity of exogenous-labeling used in our study, (i) the efficiency of isolation purification was purified, which is related to the increased yield of isolated cells, as well as their purity and (ii) the use of immunological cell-labeling strategies to increase internalized SPION load without affecting cell viability, which resulted in higher purity and with higher cell isolation yield than those obtained in the literature [29,31,32,51,53,68,71,72], as well as greater SPION load internalization compared to the literature [54,86,101], thus, increasing the SPION detection sensitivity by MRI techniques and their viability in future clinical applications. SPION detection by MRI using exogenous-labeling already reported in in vivo studies [101][102][103][104] and Krieg's study [101] showed that the quantification of the internalized SPION load on immune cells in the in vitro experiment was lower than that found in our study, using a 500-fold lower concentration in the labeling process, although even with a lower internalized SPION load, it was possible to detect the SPION signal (hypo-intensity image) by MRI in the in vivo study. Therefore, the M-SPIONs used point to a promising candidate for in vivo studies.
Therefore, our study showed that it is possible to isolate granulocytes with high purity and yield, and the labeling with M-SPIONs provided a high internalized iron load and low toxicity to cells. These multimodal nanoparticles with magnetic and fluorescent features are detectable in both the visible spectrum of fluorescence and the infrared spectrum and can be used in in vitro, ex vivo and in vivo studies for inflammation/infection process detection using various imaging techniques, such as MRI and NIRF. For these images, the acquisition of a single probe (M-SPION) is necessary, providing complementary information to the pathology diagnosis, considering the limitations of each technique. In addition, this study highlights a strong future for the in vivo investigation of inflammation/infection using multimodal nanoparticle-labeled granulocytes.

Granulocyte Isolation from Human Peripheral Blood
Granulocytes were obtained from healthy volunteers blood samples, and all subjects gave their informed consent for inclusion before participation in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of the Albert Einstein Hospital (CAAE-66749517.0.0000.0071).
From each volunteer (n = 12), 20 mL of peripheral venous blood was collected using a 20-G needle to avoid lysing cells, followed by the addition of 50 UI heparin sodium. In granulocyte isolation, the experimental design compared 4 different methods: Percoll density gradients, Percoll density gradients with PBS, Percoll density gradients with Hanks solution and Percoll gradients and Ficoll.
Each method is described in detail below, and Figure 6 represents the schematic differences. The granulocyte isolation methods were compared to find the one with the highest purity.

Isolation Method with PERCOLL Gradients and Ficoll
The fourth method consisted of improved granulocyte purification, which combined the result of the granulocyte isolation method (III) using Percoll density gradients with Hanks′ solution and another centrifugation with Ficoll Premium (17544602GE Healthcare, Brazil). Continuing the process described above, 4 mL of granulocytes isolated and suspended in DMEM were added carefully above 4 mL of Ficoll Premium and centrifuged at 400× g for 30 min at 22 °C without centrifugation braking. The pellet sediment was resuspended in 4 mL of DMEM supplemented with 10% FBS for granulocyte purity analysis, as depicted in Figure 6, method IV.

Visualization of Granulocyte Isolation Methods
For visual analyses, 50 µ L of isolated granulocytes was added to a cytofunnel with a cytocard and a microscope slide and centrifuged using a cytocentrifuge (Cientec, CT2000, Brazil) at 250× g for 3 min. After centrifugation, the granulocytes appeared as agglomerated cells in the middle of the microscope slide. The cells were dried at room temperature to perform fast hematology staining (fast panoptic, Interlab, Brazil) and visualized by convectional optical microscopy (Nikon, TiE).

Evaluation of Isolated Granulocyte Purity
First, granulocyte purity after each method was assessed through cell counting using an automated cell counter (Sysmex, Brasil), and the cells from the highest purity methods were characterized by flow cytometry (LSRFortessa, Bdsciencis). Isolated granulocytes were phenotyped by negative markers (CD14, CD19 and CD3) and positive markers (CD45, CD15 and CD13).

Isolation method with Percoll Density Gradient
First, the total leukocytes were separated from the venous blood, and 5 mL of voluven (Fresenius Kabi, Brazil) was added to the blood sample to increase the hemosedimentation speed and allowed to stand for 45 min. After this time, the leukocytes were now visible as a white layer between the red cells and plasma (buffy coat). The leukocytes were collected to begin the process of granulocyte isolation with Percoll density gradients, as described by de Vries et al. [67] and Roca et al. [69].
Two different densities of Percoll gradients (P4937, Sigma Aldrich, São Paulo, Brazil) were used for granulocyte isolation. Gradient solution 1 (G1, Figure 6) with 56% Percoll consisted of 1.0 mL of 9% NaCl, 5.5 mL of 100% Percoll and 3.4 mL of water for injection. Gradient solution 2 (G2, Figure 6) with 71% Percoll consisted of 1.0 mL of 9% NaCl, 9.0 mL of 100% Percoll and 1.9 mL of water for injection. The Percoll gradient was performed by slowly adding the following sequence: 3 mL of gradient solution 2 (G2), 3 mL of gradient solution 1 (G1) and 3 mL of the leukocyte mixture gently deposited on top, taking care to avoid mixing. The Percoll gradients were centrifuged at 300× g for 30 min at 22 • C without a centrifugation brake. After centrifugation, the supernatant was removed, and the white layer was carefully collected with granulocytes between the gradients and resuspended in 4 mL of DMEM (Sigma Aldrich, São Paulo, Brazil) supplemented with 10% FBS for granulocyte purity analysis, as depicted in Figure 6, method I.

Isolation method with Percoll Density Gradients and PBS
The second granulocyte isolation method followed the same total leukocyte separation process described above (Percoll density gradients). In this method, the Percoll gradients were constructed using different densities by diluting Percoll with PBS (Gibco ® , Carlsbad, CA, USA). Gradient solution 3 (G3, Figure 6) and gradient solution 4 (G4, Figure 6) contained 60% and 70% Percoll, respectively, as described in previous studies [35,70,105]. The Percoll solutions with PBS were obtained by slowly adding the following sequence: 3 mL of gradient 4 (G4), 3 mL of gradient 3 (G3) and 3 mL of the leukocyte mixture gently deposited on top, taking care to avoid mixing. These gradients were centrifuged at 100× g for 20 min at 22 • C without centrifugation braking. After centrifugation, the granulocyte-containing white layer between the gradient solutions was collected carefully and resuspended in 4 mL of DMEM supplemented with 10% FBS for granulocyte purity analysis, as depicted in Figure 6, method II.

Isolation Method with Percoll Density Gradients and Hanks Solution
The third method did not require previous leukocyte isolation. The gradients were constructed with Hanks balanced salt solution without calcium and magnesium 10× (HBSS, Sigma Aldrich, Brazil) and Percoll, as used in previous studies [32,33,106,107]. Gradient solution 5 (G5, Figure 6) with 60% Percoll consisted of 1.8 mL of Percoll 100%, 300 µL of HBSS and 900 µL of Milli-Q water, and gradient solution 6 (G6, Figure 6) with 70% Percoll consisted of 2.1 mL of Percoll 100%, 300 µL of HBSS and 600 µL of Milli-Q water.
The Percoll gradient with Hanks solution was obtained by slowly adding the following sequence: 3 mL of gradient 6 (G6), 3 mL of gradient 5 (G5) and 3 mL of venous blood, gently deposited on top, taking care to avoid mixing. These solutions were centrifuged at 400× g for 30 min at 22 • C without centrifugation braking. After centrifugation, the granulocytes were visible as a white layer between gradient 6 and gradient 5. The supernatant was removed, and the granulocyte layer collected was added to 4 mL of 1 × HBSS to recover the cellular osmolarity. Then, another centrifugation was performed at 150× g for 5 min at 22 • C without centrifugation braking. The pellet sediment was resuspended in 4 mL of DMEM supplemented with 10% FBS for granulocyte purity analysis, as depicted in Figure 6, method III.

Isolation Method with PERCOLL Gradients and Ficoll
The fourth method consisted of improved granulocyte purification, which combined the result of the granulocyte isolation method (III) using Percoll density gradients with Hanks solution and another centrifugation with Ficoll Premium (17544602GE Healthcare, Brazil). Continuing the process described above, 4 mL of granulocytes isolated and suspended in DMEM were added carefully above 4 mL of Ficoll Premium and centrifuged at 400× g for 30 min at 22 • C without centrifugation braking. The pellet sediment was resuspended in 4 mL of DMEM supplemented with 10% FBS for granulocyte purity analysis, as depicted in Figure 6, method IV.

Visualization of Granulocyte Isolation Methods
For visual analyses, 50 µL of isolated granulocytes was added to a cytofunnel with a cytocard and a microscope slide and centrifuged using a cytocentrifuge (Cientec, CT2000, Brazil) at 250× g for 3 min. After centrifugation, the granulocytes appeared as agglomerated cells in the middle of the microscope slide. The cells were dried at room temperature to perform fast hematology staining (fast panoptic, Interlab, Brazil) and visualized by convectional optical microscopy (Nikon, TiE).

Evaluation of Isolated Granulocyte Purity
First, granulocyte purity after each method was assessed through cell counting using an automated cell counter (Sysmex, Brasil), and the cells from the highest purity methods were characterized by flow cytometry (LSRFortessa, Bdsciencis). Isolated granulocytes were phenotyped by negative markers (CD14, CD19 and CD3) and positive markers (CD45, CD15 and CD13).

Granulocyte Viability after Isolation Methods
Granulocyte viability was analyzed 4 h after granulocyte isolation by the highest purity methods using flow cytometry (LSRFortessa, Bdsciencis). This time was considered plausible for granulocyte isolation, labeling and administration in future studies to detect inflammation and infection processes. The flow cytometry evaluation was performed with 10 6 isolated cells in 100 µL, 5 µL of annexin V-FITC, 5 µL of propidium iodide-phycoerythrin (PI-PE) (Sigma Aldrich, EUA) and 300 µL of lysis buffer (Excellyse Live, Czech Republic). The red blood cells were lysed for 20 min without light contact. These tubes were centrifuged at 500× g for 5 min, the supernatant was removed, and the resuspended sediment was analyzed by flow cytometry. Viable granulocytes were defined as annexin V negative and PI-PE negative.

Multimodal Superparamagnetic Iron Oxide Nanoparticles (M-SPIONs)
For granulocyte labeling, we used multimodal superparamagnetic iron oxide nanoparticles (M-SPIONs), with a crystalline-phase magnetic core identified as magnetite, a hydrodynamic size of 35 nm, an 8 nm iron oxide nucleus coated with dextran, a zeta potential of~+31 mV and a density of 1.25 g/cm 3 (Biopal, Molday ION™), half of which were conjugated with a fluorophore that emitted NIRF absorption/emission wavelengths in the 750/777 nm range and another half with a fluorophore that emitted rhodamine-B absorption/emission wavelengths in the 558/580 nm range.

M-SPION Optical Characterization and Analysis of Size Polydispersion, Stability, and Zeta Potential
The optical characterization of M-SPIONs dispersed in aqueous medium was performed at 50 µg Fe/mL using an RF-6000 spectrofluorophotometer (Shimadzu, Kyoto, Japan) to acquire the spectrum with excitation in the wavelength range from 520 to 800 nm and emission from 550 to 780 nm.
The polydispersion of hydrodynamic diameter (HD), temporal stability of HD and zeta potential of M-SPIONs (50 µg Fe/mL) were measured using dynamic light scattering (DLS) with Zetasizer Nano S equipment (Malvern, Reino Unido). The HD distribution curve was obtained using an angle of 173 degrees with 15 measurements in 5 s, maintaining a constant temperature at 37 • C. M-SPION HD stability analysis was performed in cell culture medium (DMEM +10% FBS) for 20 h. The zeta potential measurements (surface charge) were performed in the pH range from 7 to 9.

Labeling of Granulocytes with M-SPIONs
Isolated granulocytes were labeled with M-SPIONs using 10 6 cells per well (48-well plate) resuspended in 4 mL of DMEM (Sigma Aldrich, Brazil) with 10% FBS (Sigma Aldrich, Brazil). M-SPIONs were used at concentrations of 10, 30 and 50 µg Fe/mL and incubated for 4 h at 37 • C. After this period, the cells were washed three times with 0.9% saline solution and centrifuged at 300× g for 5 min at 22 • C to remove the M-SPIONs that were not internalized by the granulocytes.

Visualization of M-SPIONs Internalized in Granulocytes
The M-SPION internalization was analyzed by brightfield and fluorescence microscopy images due to the magnetic and fluorescence features of nanoparticles. The granulocytes were deposited on a microscope slide using a cytocentrifuge, as described in Section 4.2. Prussian blue staining was performed with 500 µL of a solution containing 5% potassium ferrocyanide (Sigma Aldrich, St Louis, MO, USA) and 5% hydrochloric acid (Merck, Darmstadt, Germany) for 15 min and washed once with deionized water. Then, Nuclear Fast Red staining was performed with a 1% solution (0.02 g of Nuclear Fast Red in 2 mL of deionized water) for 10 min for nuclear counterstaining, and the nuclei were quickly washed once more and analyzed by optical microscopy. Subsequently, fluorescence analysis was performed using diamidino-2-phenylindole (DAPI, Sigma Aldrich) to label the cell nuclei for 10 min, followed by washing once to register the M-SPION fluorescence image using an excitation/emission filter of 530/550 nm. Both image analyses were performed using a Nikon TiE fluorescence microscope (Tokyo, Japan).

Granulocyte Viability Evaluation after M-SPION Labeling
The granulocyte viability was evaluated after 4 h of labeling with M-SPION concentrations of 10, 30 and 50 µg Fe/mL using the same procedures performed in the flow cytometry analysis with annexin V-FITC and propidium iodide described in Section 4.4.

Quantification of M-SPION Internalized in Granulocytes
After granulocyte labeling, the M-SPIONs internalized in the cells were quantified using MRI, inductively coupled plasma mass spectrometry (ICP-MS) and NIRF techniques. For quantification of M-SPIONs internalized into the cell, a phantom for each well containing 2 × 10 6 granulocytes was dispersed in 1 mL of 1% agarose under the following conditions: unlabeled granulocytes and granulocytes labeled with 10, 30 and 50 µg Fe/mL of M-SPIONs. The MRI data were obtained in a similar way to r2, providing the T2 values for the respective concentrations studied. Finally, the M-SPION captured by granulocytes was determined using Equation (1): where 1/T 2 is the transverse relaxation rate proportional to the intracellular iron concentration, [M-SPION] is the concentration of intracellular iron on agarose gel and r 2 is the transverse relaxivity of M-SPIONs.
The complement of the results was used to calculate the number of nanoparticles internalized into granulocytes using Equation (2): where iron load is the iron loaded into granulocytes by internalization (mass), at_m is the atomic mass, ρ M−SPION is the iron oxide density, M Fe is the molecular weight of iron and ∅ M−SPION is the diameter of M-SPIONs.

Quantification of Iron Load by ICP-MS
The quantification of the iron load of the granulocyte labeling with M-SPIONs by ICP-MS used 2 × 10 6 cells/mL dispersed in 1 mL of DMEM and 1 mL of nitric acid (37%) for digestion over 4 h at 70 • C. The digested samples were diluted 10 times with Milli-Q ® water (EMD Millipore Corporation, Bedford MA, USA) and were analyzed with ICP-MS equipment (Perkin Elmer Nexion 350×, PerkinElmer Corporation, USA) to determine the iron content of each sample. Sample measurements were performed in quintuplicate, and quantification was based on a calibration curve using certified standard iron (NexION #N8145054) at the following concentrations: 5, 10, 20, 30 and 40 ppb.

Quantification of Iron Load by NIRF
The quantification of iron load by NIRF was performed after granulocyte labeling with 2 × 10 6 cells dispersed in 1 mL of DMEM placed in an Eppendorf tube. The NIRF signal of the triplicate samples was acquired by applying an excitation of 750 nm, registered in a range of emission of 810-875 nm, using an exposure time of 2.5 ms, binning of 4 and f/stop of 2. The sample images were acquired using IVIS ® Lumina LT Series III equipment (Xenogen Corp, PerkinElmer. CA, USA), and the fluorescence intensity signals were analyzed in absolute radiation units (photons/s). For the quantification analysis, the calibration curve was constructed using iron concentrations of 0.5, 1, 2, 8, 10 and 15 µg Fe/mL, and the fluorescence intensity signal of the samples was compared to a calibration curve to quantify the iron mass internalized by isolated granulocytes.

Statistics Analyses
The experimental data are presented as the mean and standard deviation. For comparison of the multiple isolation methods, differences between the means of the groups were tested by a nonparametric test for unpaired samples (Kruskal-Wallis test), following the post hoc analysis corrected by Bonferroni. The difference between two independent sample means was tested by a nonparametric Mann-Whitney test. Significant results were considered a p-value of less than 0.05. Statistical tests were performed using JASP v. 0.

Conflicts of Interest:
The authors declare no conflicts of interest.