2.1. Preliminary Experiments
OPA reacts with primary aliphatic amines in the presence of nucleophilic compounds (sulfite, mercaptoethanol, N-acetylcysteine) through a non-specific mechanism to form highly fluorescent derivatives. The well-documented advantages of OPA, the commercial availability, and the reasonable cost have established the latter as one of the most widely used derivatizing reagents. A handful of compounds such as histidine, hydrazine, glutathione, and histamine react with OPA through a different, unique mechanism, in the absence of nucleophilic additives in acidic or basic medium to also form fluorescent derivatives. When developing OPA-based methods for these analytes, the advantage of selectivity versus amino acids and biogenic amines is of key importance.
A representative off-line FL spectra from the reaction between OPA and histidine is shown in
Figure S1 (
Supplementary Materials). As can be seen in the spectra, the optimum excitation wavelength is λ
ex = 360 nm, which is perfectly suitable with the wavelength of the illumination UV lamp.
Figure S2 depicts a preliminary image of the plate at 0, 50, and 100 μM histidine (5 mM OPA, pH = 9.0, reaction time 30 min). A graphical depiction of the configuration of the experimental setup (lamp, plate, and smartphone), including the derivatization reaction, is shown in
Figure 1.
2.2. Development of the Assay for the Determination of Histidine
Chemical variables that were investigated and optimized included the amount concentration of the derivatizing reagent (OPA) in the range of 2 to 20 mM, the effect of the reaction pH in the range of 8 to 10 (100 mM phosphate buffer), and the practical reaction time in the range of 5–30 min. Due to the versatility and the high-throughput potentials of the assay, all experiments were carried out at 5 levels of histidine, namely 10, 25, 50, 75 and 100 μM. In this way, it is possible to study the effect of the variables not only on the sensitivity of the assay but on the linearity as well. The slopes of the respective calibration curves and the linearity were therefore used as optimization criteria for each set of experiments. The illumination and image capture conditions were kept constant in all cases. The volumes of the reagents and samples were also kept fixed at 50/50/150 μL for OPA, buffer, and sample, respectively, corresponding to a total volume of 250 μL per well.
Figure 2 depicts the effect of the pH on the assay reaction (at 5 mmol L
−1 OPA and 15 min reaction time). The pH value of 8 resulted in a higher sensitivity in terms of slope, while good linearity was obtained at all pH values (r
2 > 0.998). The effect of the amount concentration of the derivatizing reagent was evaluated in the range of 2–20 mM OPA. The pH was set at 8 and the reaction time at 15 min. As can be seen in
Figure 3, higher amounts of OPA had a ca 15% negative effect (quenching) on the reaction. The amount concentration of 2 mM was therefore selected for subsequent studies.
Generally, OPA offers rapid reactions with primary amines that are typically completed within a few minutes. On the other hand, the majority of the OPA derivatives suffer from moderate stability. The reaction time was confirmed to have a negligible effect on the performance of the assay for practical values > 10 min, while no stability issues were recorded up to 60 min. All images were therefore captured between 10 and 20 min.
Table 2 includes a synopsis of the examined variables, the investigated range, and the finally selected values.
2.3. Validation of the Fluorimetric Assay
The developed assay was validated for linearity, limits of detection (LOD), and lower limit of quantification (LLOQ), within and between-days precision, selectivity, matrix effect, and accuracy.
Linearity was obeyed in the range of 10 (LLOQ)–100 μM L
−1 (
n = 6). The regression equation was
where FI is the fluorescence intensity as measured by the smartphone and [histidine] is the amount concentration of the analyte in μM. The regression coefficient was
r2 > 0.998 and the percent residuals ranged between −8.2 and +6.7%. The LLOQ was set as the lowest level of the calibration curve (10 μmol L
−1 histidine), while the LOD was estimated to be 3.4 μM, using the standard deviation of the intercept criterion [
33].
The within-day precision was validated by preparing aqueous calibration curves using both the same and different 96-well plates (
n = 8). As can be seen in
Table 3, the relative standard deviation of the slopes was <7%. The between-days precision was validated similarly, by preparing two aqueous calibration curves per day, for four non-consecutive working days. The experimental results are also shown in
Table 3, verifying an acceptable RSD value of the slopes of better than 9%.
Selectivity and matrix effect are—without doubt—the most critical parameters when validating a non-separation method intended to be applied to biological material. In the case of this assay, increased selectivity could be expected due to the unique mechanism of the reaction between histidine and
o-phthalaldehyde, in the absence of nucleophilic compounds. This mechanism excludes most amino acids, ammonium, and biogenic amines from the formation of fluorogenic derivatives with OPA [
18]. All examined potential interferences were examined at 100 μM level + 50 μM histidine and each experiment was performed in triplicate. The following conclusions can be derived from the graphical results shown in
Figure 4:
- (i)
Ammonium and glycine do not seem to have any effect; this could be more or less expected due to the absence of an additional nucleophilic reagent [
34].
- (ii)
The same was observed for the biogenic amine methylamine due to the same reason as mentioned above.
- (iii)
On the other hand, the biogenic amine histamine that is similarly structured to histidine has a moderate positive interference at 100 μM level. Histamine is also known to react with OPA through the same mechanism, but the derivative is unstable under alkaline conditions (post-reaction acidification is required for stabilization) [
32].
- (iv)
The same moderate effect was observed for the peptide glutathione as well; although glutathione reacts with OPA in the absence of nucleophilic compounds through both the amine and the thiolic group, the optimal pH is >10. On the other hand, no interference was observed from glutathione disulfide, which is known to react with OPA at highly alkaline medium (pH > 12) [
35].
- (v)
Considering the extremely lower levels of glutathione and histamine in urine compared to histidine [
36,
37], the proposed assay can be considered as selective for the specific application.
- (vi)
The selectivity results were further evaluated statistically using the t-test. As expected, only histamine and glutathione produced statistically different results (p < 0.001), while the p-values for the rest of the selected interferences ranged between 0.2071 and 0.5543.
The potential matrix effect was validated using both histidine-free artificial urine and pooled human urine. The former was prepared as described in the experimental section and was spiked with known amounts of the analyte after proper dilution as shown in
Table 4. Pooled human urine matrix was prepared by mixing equal volumes of six individual samples (
n = 6). The pooled matrix was diluted 1:25 (the same dilution is applied to the analysis of real samples) and was spiked with known amounts of histidine in the range of 10–50 μM (
n = 4). As can be seen from the experimental results of
Table 4, both artificial and human urine matrices contribute acceptable matrix effects of <10% at dilution factors >25-fold. The aqueous calibration curve was therefore utilized for quantitative analysis in all subsequent experiments.
The accuracy of the assay was validated using both pooled and individual human urine samples. All samples were diluted 25-fold (see experimental section) and spiked with histidine at the 25 and 50 μM levels. The results that are tabulated in
Table 5 confirmed the acceptable accuracy of this high-throughput assay, with the percent recovery being in the range of 78–109%.
Finally, the robustness of the proposed method was evaluated at 50 μM histidine by small deliberate variations of critical chemical and geometrical parameters. The experimental conditions and findings are summarized in
Table 6. The method proved to be adequately robust with the percent recoveries being in the range of ±12% in all cases (
n = 3).
2.4. Assay of Histidine in Real Urine Samples
The evaluation of the applicability of the proposed high-throughput assay in real-world analyses was based on the determination of histidine in six human urine samples. The collection, preservation, and processing of the samples is described in detail in the experimental section below. As can be seen in the results of
Table 7, the concentration of the analyte ranged between 280 and 1540 μM.
Table 6 also contains the results from the analysis of the same samples by a corroborative HPLC method. The graphical comparison presented in
Figure 5 indicates good correlation between the proposed and reference methods, while representative urine chromatograms from the HPLC method can be seen in
Figure S2 (
Supplementary Materials). Additionally, the levels of histidine in the urine samples are in good correlation to the reference values for adults (>18 years,
n = 800) that were reported in the literature [
38].
Another series of experiments was carried out to evaluate the stability of the human urine samples at 0, 48, and 72 h; three temperatures were examined, namely 25 °C (room temperature), 4 °C (refrigeration), and −20 °C (freezer). The experimental percent recoveries confirmed the stability of the samples under the selected conditions, ranging between 91 and 114%.