Iron (II) chloride tetrahydrate (p.a.) and iron (III) chloride hexahydrate (p.a.), carboxymethyl cellulose sodium salt (with a low viscosity), sodium hydroxide (p.a.), N-(3-dimethylaminopropyl)-N′ethylcarbodiimide hydrochloride (≥99%), silver nitrate (p.a.), 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide methiodide, D-(+)-maltose monohydrate (p.a.), ammonia (25% v/v, p.a.), prostate-specific antigen from human semen (purity ≥95%), and anti-PSA (monoclonal antibody produced in rabbit) were purchased from Sigma Aldrich (San Jose, CA, United States) and utilized without further purification. Free prostate-specific antigen (fPSA) was purchased from Cedarlane (Buenos Aires, Argentina) and used without further purification.

Magnetite nanoparticles (MNPs) functionalized by CMC (carboxy methyl cellulose) were prepared according to the following protocol (Marková et al., 2016): First, 1 g of CMC was dissolved in 30 ml of deionized H₂O; the solution was heated up to 303 K. Next, 720 mg of FeSO₄.7H₂O was dissolved in 10 ml of deionized H₂O, acidified by the addition of 100 µl of 37% HCl. Both solutions were mixed and stirred at 300 rpm. Next, 7 ml of aqueous solution of NH₄OH (28%) was added to the mixture. The temperature was increased to 323 K over a period of 15 min, followed by the addition of 5 ml of 28% NH₄OH. The synthesized magnetic nanomaterial was separated by centrifugation at 20,000 × g and 293 K for 1 h. Finally, the magnetic pellet was resuspended in 60 ml of deionized H₂O and immersed in ultrasound bath for 15 min. This step was repeated two times. The obtained magnetic nanomaterial was centrifuged at 1,000 × g and 298 K for 20 min. In the last step, the supernatant containing carboxylated magnetic nanoparticles (labeled as cMNPs) was stored, and its concentration was adjusted to 1 g of magnetite/L.

Silver nanoparticles (d = 28 nm, AgNPs) were prepared according to protocol (Panáček et al., 2006). First, an aqueous solution containing 1 mmol of AgNO₃, 10 mmol of maltose, and 5 mmol NH₄OH was prepared. Next, the pH of the reaction mixture was adjusted to 11.5 ± 0.1 by the addition of NaOH. To obtain the MNPs@Ag composite, 10 µl of Ag NP suspension was added to the MNP suspension. The loading of Ag NPs was performed in a continuous flow for 15 min at room temperature.

Finally, immobilization of anti-PSA on the surface of silver nanoparticles was carried out. First, 1 ml of MNPs@Ag was mixed with 100 µl of EDC (final concentration 20 mmol) and stirred for 30 min. The nanocomposite was separated using external magnetic force, and adequately washed by 10 mmol PBS buffer. Next, anti-PSA was added to achieve a concentration of 0.2 ng/L, and its anchoring on the surface of the nanocomposite was performed under stirring at 300 rpm for 1 h at room temperature.

PSA and free PSA were analyzed using the following protocol: 10 µl of MNPs@Ag@anti-PSA was mixed with 10 µl of a standard solution of PSA or free PSA (c = 10 ng/ml). The mixture was incubated for 30 min at 293 K.

To demonstrate the method applicability, two volunteers (male), both aged 62 years, were selected. Whole blood samples (2 ml) were collected in Eppendorf tubes with sodium citrate as an anticoagulant. Samples were stored at −80°C before analysis and analyzed within 24 h after unfreezing. Samples for measurements were prepared by a dilution of 200 µl of unfrozen samples with 800 µl of deionized water to reduce sample density/viscosity. No further pretreatment of the samples was carried out.

Obtained samples were isolated using external magnetic force, the obtained droplet was deposited on a microscopy glass (2 × 8 cm), and the analyzed samples were dried out and concentrated in a coffee ring structure caused by capillary flow. The samples were analyzed by means of Raman spectroscopy at three spots on the ring. Obtained spectra were averaged and employed in the consecutive steps. Samples of the whole human blood were analyzed in an analogy to a methodology previously described in the section on Preparation of samples of whole human blood.

The determination of the PSA and free PSA, respectively, consisted of two consecutive steps. Intensity of the spectral band at 1,335 cm⁻¹ was used as a quantitative marker in both. In the first step, estimated concentration of PSA and free PSA, respectively, in the samples was determined from an external calibration assembled using standard solutions in a concentration range of 2–500 ng/ml. The second step was as follows: More precise/accurate values were determined using a standard addition method based on the three repeated additions of the standard of PSA or free PSA, respectively. The concentrations of the standard additions were derived from an approximate concentration of PSA/free PSA determined in the first step. Concentrations of additions 1–3 were set to cause 30%–50% increments in the measured signal.

Microscopic images were obtained by transmission electron microscopy (TEM) JEOL JEM 2100 microscope operated at 200 kV. Energy-dispersive X-ray spectroscopy (EDS) was performed in scanning (STEM) mode of high-resolution transmission electron microscope (HRTEM) TITAN G2 operated at 80 kV to obtain elemental maps (using Super-X system with four silicon drift detectors Bruker). STEM images were taken with an HAADF detector 3,000 (Fischione).

Zeta potential values ( ζ p ) were determined by the dynamic light scattering (DLS) method using the instrument Malvern Zetasizer Nano (Malvern Instruments Ltd., Worcestershire, United Kingdom). Fourier transform infrared (FT–IR) absorption spectra were measured on an FT–IR spectrometer Nicolet iS5 (Thermo Electron, United States) employing attenuated total reflection (ATR) and a ZnSe crystal, using 64 scans with a resolution of 2 cm⁻¹. The IR absorption spectra were obtained using ATR. The linear function was subtracted from the spectral data in order to correct the baseline. XPS data were obtained using a PHI VersaProbe II (Physical Electronics) spectrometer, equipped with an Al Kα source. The XPS spectra were analyzed using the MultiPak (ULVAC–PHI, Inc.) software package. The carbon C 1-s peak (284.4 eV peak) was used as a reference for all binding energies.

Raman spectral data were obtained on a DXR Raman microscopy (Thermo Scientific, United States) with an excitation laser operating at 633 nm (red). Raman spectra were measured in the range from 400 to 3,500 cm⁻¹. Spectral resolution of the applied combination of laser and optics was 1.0 cm⁻¹. Experimental parameters were as follows: the exposure time: 16 s, laser power 4 mW (measured on the sample surface), one experimental point was obtained by the averaging of 512 microscans. Interfering spectral background, including fluorescence and sample heating, was corrected using a subtraction of polynomial fit, where n = 3. Quantification of concentrations of proteins, including PSA and fPSA, was performed using the TQ-Analyst software package (Omnic, version 8, Thermo Scientific, Waltham, United States).

The prepared nanosensor, based on carboxymethylcellulose-stabilized magnetite–silver nanoclusters, was characterized by Raman and infrared spectroscopy, transmission electron microscopy, ζ-potential measurements, EDS analysis, elemental mapping, and XPS analysis. EDS spectrum (Figure 1A) confirmed the presence of Ag and Fe in the nanocomposite. The arrangement of nanoparticles was identified by microscopic technique, where the results from the TEM (Figure 1B) and comparison of dark-field STEM (Figure 1C) and elemental mapping (Figures 1D–G) reveal that composite is formed by 5- to 20-nm magnetite nanoparticles and 20- to 40-nm silver nanoparticles, which are attached to CMC, which presence was confirmed by zeta-potential and FT-IR measurements. The presence of free carboxyl groups of CMC on the nanocomposite surface, which plays a critical role for anti-PSA loading and good stability of the system (Kirby and Hasselbrink, 2004), was confirmed by the determination of very low values of zeta potentials of MNPs, MNPs@Ag, and MNPs@Ag@antiPSA (−52,5, −45,3, and −45,2 mV, respectively).

The comparison of FT-IR spectra of free CMC, MNPs, MNPs@Ag, and MNPs@Ag@antiPSA shows an effective anchoring of CMC on the surface of the nanostructures and its presence in all composites (Figure 2A). Spectra were interpreted in agreement with methods previously specified by Barth (Barth, 2007; Xue Li et al., 2014). All spectra involve characteristic peaks of CMC, namely, 1,030 (C-O stretch), 1,330 (bridge–O–stretch), 1,420 (bridge-O-stretch), 1,590 (-COO⁻ asymmetric stretch), 2,900 (C-H stretch), and 3,400–3,200 (O–H stretch). Moreover, a nonsymetric character of –COO⁻ peak at 1,590 cm⁻¹ was observed for MNPs@Ag-anti-PSA (Figure 2B) due to the presence of amide I band (∼1,650 cm⁻¹), which is characteristic of the presence of proteins (Miller et al., 2013). In accordance, the comparison of XPS spectra of MNPs@Ag and MNPs@Ag@anti-PSA (Figure 2C) confirmed the EDS analysis, and moreover, they revealed four times higher content of nitrogen in the sample of MNPs@Ag@anti-PSA (Figure 2D) that can be attributed to the presence of amine and amide groups of anti-PSA on the nanocomposite surface.

The synthesized nanomaterial was subjected to the analysis of standard solutions containing PSA and free PSA (c = 10 ng/ml). The resulting data can be seen in Figure 3A. Raman spectra obtained from the analysis of the model sample (sample label: MNPs@Ag@anti-PSA@PSA, MNPs@Ag@anti-PSA@fPSA) include spectral bands typical for proteins. Considerable vibrations of proteins included abundant spectral bands present at 650, 845, 1,003, and 1,290 cm⁻¹ (C-H deformation vibrations of proteins), and 1,450 cm ⁻¹ and 1,670 cm⁻¹ (labeled as amide I). Moreover, raw nanomaterial and functionalized sensor MNPs@Ag, stabilized by CMC (label: MNPs@Ag), were measured as spectral references. Both spectra do not contain any significant Raman bands.

Next, a set of standard solutions containing PSA and fPSA, respectively, was measured using MNPs@Ag@anti-PSA nanocomposite in the range of concentration levels from 5 to 500 ng/ml Figures 3B,C, where six concentration levels with five replicates were measured and classified. The resulting spectra of all six standards are shown in Figure 3D (PSA) and Figure 3E (free PSA). These data suggest that the Raman intensity of the spectral bands that are characteristic of PSA and free PSA varies with concentration. As a marker of quantification, we determined the amide vibration of the amide located at 1,335 cm⁻¹, and the obtained intensity of this band was applied to construct a calibration curve. The LOD values were statistically estimated by using reliability of calibration curves for PSA and free PSA, from the three-time evaluated standard deviation of the slopes. The calculated coefficient of determination (R ²) is 0.98 for PSA and 0.952 for free PSA, respectively. The resulting lower limit of detection (LOD) is 0.62 ng/ml for PSA and 0.49 ng/ml for free PSA, respectively.

The repeatability of the analytical method was classified on a set of five independent measurements (Figures 4A,B). It is clear from the results that all obtained spectra are fully comparable, with the statistical difference aimed at the previously selected peak being less than 9%. Then the stability of the nanomaterial was evaluated by regular measurements for 3 months, the spectral data obtained 3 months after the synthesis of the nanomaterial show a strong analytical signal of PSA and freePSA, confirming that the synthesized nanomaterial can be applied for a long time without significant signal loss.

We further concluded that Raman spectroscopy can be used to correctly distinguish between PSA and free PSA using our developed high-precision analytical sensor. Investigative study and supervised classification, namely, partial least squares discriminant analysis (PLSA), confirmed the possibility of using Raman spectroscopy to distinguish between the studied PSA and free PSA. Obtained data are shown in Figures 5A,B.

Finally, we studied the use of MNPs@Ag@anti-PSA nanocomposite for the analysis of human whole blood samples containing different concentrations of PSA and freePSA. Figure 6 shows the Raman spectrum obtained by analyzing a real sample and a standard mixture containing PSA and free PSA (c = 10 ng/ml in total). Subsequent detailed spectral analysis revealed new bands located at 1,350 and 1,525 cm⁻¹, which are not present in the spectrum of the real samples, thus, outlining minor nonspecific interactions. The total PSA and free PSA concentrations in the samples were calculated using the standard addition method and the synthetic analytical sensor, respectively, in order to minimize the influence of nonspecific interactions between the matrix components. For the measurement of the real samples containing blood, it is necessary to focus on possible hemolysis caused by hemoglobin and intracellular components during measurement. Using spectroscopic methods, we can find an oxyhemoglobin peak between 531 and 543 cm⁻¹, and due to the relatively wide absorption peak, hemolysis can affect the quantification of analytes in this range (Howanitz et al., 2015). Medipally et al. (2020) studied the effect of hemolysis using infrared spectroscopy and Raman spectra of blood plasma. They measured that hemolysis could affect the infrared and Raman spectra of blood plasma, but a larger study is needed to confirm the effect of hemolysis both in vitro and in vivo.

A total of two samples (both with a diagnosis of BHP) were evaluated. The PSA or free PSA concentrations in these samples were: 2.91 ng ml⁻¹ (PSA), 2.34 ng ml⁻¹ (free PSA) per first real sample, and 3.29 ng ml⁻¹ (PSA) and 2.76 ng ml⁻¹ (freePSA) per second real sample. For patients who have borderline concentrations, the ratio of free to total PSA (free/total PSA ratio) is determined (Hoffman et al., 2000).