Effective and repeatable chromatographic separation of 5 nucleotides in infant formula milk powder by ion-pair high-performance liquid chromatography–ultraviolet

A robust method using HPLC-UV was developed to improve the accuracy and repeatability of a quantitative method to detect 5 nucleotides (cytidine-mono-phosphate, uridine monophosphate, adenosine mono-phosphate, guanine monophosphate, and inosine mono-phosphate) in infant formulas. The results showed that efficient separation could not be achieved without strict conditions. The proposed method displayed a strong linear relationship (R 2 > 0.9999) of single nucleotide in infant formula milk powder in the range of 10 to 1,000 mg/kg, a steady recovery (80.0% ~110.0%) with relative standard deviation from 0.5% to 3.5%, under strict conditions of hydrophilic C 18 column with di-isopropyl at 62.5 ± 2.5°C (± standard deviation), 0.65 ± 0.1 mg/ mL tetrabutylammonium bisulfate, and mobile phase of pH of 2.75 ± 0.02. By applying this method on a series of milk products in the Chinese market, we found a few of them exceeded the legal limits of nucleotides.

Many methods have been reported to detect the 5 nucleotides in milk powder so far.There have been attempts to use both the C 18 chromatographic columns (widely used to separate nonpolar compounds; Gill and Indyk, 2007;Chen et al., 2021) and the amide chromatographic columns (commonly used to separate sugars; Mateos-Vivas et al., 2016;Ma et al., 2019) with HPLC to determine nucleotides.However, the lactose, which is more than 50% of the milk powder matrix, often flows out together with 1 or more of the 5 nucleotides (Chen et al., 2020) due to the similar structures.Hence, complex pretreatments, such as solid phase extraction (Ma et al., 2019), precipitated impurities (Viñas et al., 2010), or isotope internal standard method (Inoue et al., 2010;Gill et al., 2013) have generally been required in these methods.
Effective and repeatable chromatographic separation of 5 nucleotides in infant formula milk powder by ion-pair high-performance liquid chromatography-ultraviolet Yuanjia Chen, 1 Lin Luo, 1 Peiyan Feng, 1 Baojun Xu, 2 * and Xiaoqun Wei 1 * To simplify the pretreatment and avoid expensive reagents, we focused on the ion-pair reagent method based on C 18 columns by HPLC-UV.The phosphoric acid in 5 nucleotides combined with the ion-pair reagent to extend their retentions in C 18 chromatographic columns and to differentiate from lactose (Yang et al., 2010;Mateos-Vivas et al., 2015, 2016).Moreover, the UV detection incurs a low cost and will not be polluted by the ion-pair reagent.However, it is still a challenge to detect the 5 nucleotides in milk powder by HPLC-UV with C 18 chromatographic columns simultaneously.For one challenge, the accuracy of the methods is insufficient.This is because the binding force between the 5 nucleotides and the ion-pair reagent belongs to a nonchemical bond, and the interaction between the nucleotides and the column is greatly affected by the analysis conditions.Slight changes in the column packing (Cohen et al., 2010;Zhu et al., 2018), the pH value of the mobile phase (Studzińska and Buszewski, 2013), the concentration of the ion-pair reagent (Gil et al., 2007), or the column temperature will affect the separation effect (Johnsen et al., 2011).Another challenge is that the repeatability of the methods will be compromised due to the rapid loss of the column efficiency in conventional C 18 chromatographic columns with ionpair reagents (Chen et al., 2021), which is an important but easy to ignore problem.Therefore, improving the repeatability and accuracy of the determination of 5 nucleotides in infant formula based on a C 18 column with ion-pair reagents by HPLC-UV remains an urgent problem.
Considering this context, to quantify nucleotides more effectively and to explain their mechanisms, we investigated the retention time of the 5 nucleotides under more precise pH values (± 0.01), a greater range of column temperatures, and a broader range of ion-pairing reagent concentrations.These findings are expected to have a wide range of applications that involve the separation, preparation, and quantification of nucleotides.

Chemicals and Reagents
We purchased the CMP, UMP, AMP, GMP, and IMP (standard, purity of 99% or more) from Sigma.Tetrabutylammonium bisulfate (THS; purity of 98%), a reagent used in the mobile phase, was purchased from Aladdin.Acetonitrile, methanol, and phosphoric acid, were of HPLC grade (Thermo Fisher Scientific).Purified water was obtained from a Milli-Q Advantage A10 water purification system (Merck).Commercial sam-ples of infant formula were collected from the domestic market, and the imported samples were collected from the Guangzhou Customs.

Preparation of Standards
Standard solution (1.0 mg/mL) of each nucleotide was prepared by dissolving 25 mg (accurate to 0.001 g) of standard in water and diluting it with water to 25 mL.Intermediate solution was used to prepare mixture of 1 mL of 1.0 mg/mL CMP standard solution, 1 mL of 1.0 mg/mL UMP standard solution, 1 mL of 1.0 mg/ mL AMP standard solution, 1 mL of 1.0 mg/mL GMP standard solution, 1 mL of 1.0 mg/mL IMP standard solution, and 15 mL of water at concentration 50 μg/ mL.The mixed standard solutions were diluted with water to the required concentrations (50 μg/mL, 25 μg/mL, 10 μg/mL, 5 μg/mL, 2 μg/mL, 1 μg/mL, and 0.5 μg/mL) as standard working solutions.These stock solutions (valid for 3 mo) were stored in brown glass bottles at 4°C.

Preparation of Samples
An aliquot of 0.5 g of sample was put into a 50-mL centrifuge tube and mixed with 10 mL of water.The glacial acetic acid adjusted pH to 4 to 5 to precipitate the protein.Then, the tube was vortexed for 5 min, followed by centrifugation at 10,397 × g for 10 min at 20°C.The supernatant was transferred into the specimen bottle and then diluted with water to 10 mL.The diluted supernatant (1 mL) was filtered through 0.25μm membrane filters and kept for 60 h to allow for HPLC analysis.

HPLC-UV Analysis
All separation was performed on a 30A HPLC system equipped with a photo-diode array.An Agilent ZOR-BAX SB-AQ column (3 μm, 4.6 mm × 150 mm) was used at 65°C.The injection volume was 10 μL.The detection wavelength was set at 254 nm.The mobile phase consisting of solvent A (0.6 mg/mL THS) solution (pH 2.75, adjusted by 2 mol/L disodium hydrogen phosphate solution or phosphoric acid) and solvent B (acetonitrile) was used to deliver compounds in samples at a flow rate of 1.0 mL/min.The stepwise elution was as follows: 0 to 12 min, 0% B; 12 to 15 min, 0 to 100% B; 15 to 20 min, 100% B; 20 to 25 min, 0% B. This method was optimized based on published national standards (Ministry of Health of the People's Republic of China, 2012).

Method Validation
To validate a sensitive and reliable HPLC method for the quantification of nucleotides in milk powder, the studies on linearity, sensitivity, specificity, and precision were carried out.The linearity of the calibration curves was made by plotting 7 different concentrations (0.5, 2.0, 5.0, 10.0, 20.0, 25.0, and 50.0 μg/mL) and their peak area over the defined range of calibration.The limit of detection (LOD) and the limit of quantification (LOQ) were detected when the signal-to-noise ratio was 3 and 10 times of the chromatographic response, respectively.The calculation equations were listed in the following Eq. 1 and 2: where SD (b) is the standard deviation of the intercept and a is the slope of the calibration curve.
To assess matrix effects, the specificity was investigated through the spike-and-recovery experiments.Preparation and detection of the spiked samples were in accordance with optimized pretreatment and chromatographic conditions.We analyzed the peak areas of the standards spiked in sample matrix and determined the standard solvent to obtain the relative recovery rates, which could be calculated according to Eq. 3.Each concentration level was determined 6 times.The equation is as follows: where C 1 is the concentration determined in spiked sample, C 2 is the concentration determined in unspiked sample, and C 3 is the concentration of spiked analyte.

Statistical Analysis
The recovery tests were repeated 6 times to get the mean and calculate their relative standard variance.Other determinations were carried out in triplicates, and the resulting data were expressed as mean ± standard deviation.The significant differences between the means of different samples were calculated by one-way ANOVA.Duncan test was used to determine whether there were significant differences (P < 0.05) among the values.All data were analyzed in SPSS software (version 22, IBM Co.).The linear curves were obtained by fitting 6 data points in Origin software (version 8.6, Origin Lab Co.).

Method Development
Effect of Column Temperature on Nucleotide Retention Time.It was found that the ratio of water phase required for an ideal separation effect was 100% during the preliminary experiment.The Agilent ZORBAX SB-AQ column, a kind of hydrophilic C 18 chromatographic column bounded with di-isopropyl, was chosen for experiments.This column performance was still good in the presence of the ion pair because di-isopropyl side chain groups form steric hindrance to prevent the water phase from coming in contact with the silica gel.Generally, the separation effect of the compound is better with lower column temperatures.However, we found that the low temperature could not separate the nucleotides well through the experiment.Therefore, the effect of temperature on retention time was investigated by varying the column temperature from 35 to 80°C (10 calibrations) when the concentration of ion-pair reagent was 10 mmol/L and the pH was 2.75.As shown in Figure 1a, the retention time generally decreased with higher temperature, except for CMP.In addition, there was no overlap among the retention time curves of CMP, GMP, IMP, and UMP.However, the AMP curve intersected with that of GMP and IMP curves at 35 to 55°C and 70 to 80°C, respectively.Therefore, it was evident that the optimal column temperature was between 60 and 65°C at the established ion-pair reagent concentration and pH.
Higher temperatures could reduce the polarity of mobile phase (Johnsen, et al., 2011) and shorten retention time of 5 nucleotides.It was required to find out whether the AMP had larger polarity changes than other nucleotides when heating in other nonpolar chromatographic columns.As shown at Figure 1b, the NH 2 linked to the plane-structure C (4) changed from the p-π conjugate structure to the noncoplanar structure and caused the AMP to become less polar and remain longer when heating.Although other nucleotides had O as the fourth carbon of their nitrogenous base, it can be inferred from the results that the structures were more thermally stable.
Effect of Mobile Phase PH on Nucleotide Retention Time.As for the pH value, a wide range (or accurate to one decimal point) was permitted in most reports (Gill et al., 2013), which was not strict enough.We took the recommended range into the experiment but found it did not always work; therefore, a more precise pH value needed to be determined by controlling other factors as invariant.When the concentration of ion-pair reagent was 10 mmol/L and the column temperature was 60°C, the recommended range of pH values from 2.65 to 3.10 (14 calibrations) was performed in a comparative study of analytic separation.
We observed that the pH value had little effect on retention time of CMP, UMP, IMP, and GMP, and their separation did not interfere with each other (Figure 2a).However, AMP was sensitive to the pH value, and its retention time curve intersected the GMP and IMP curves between pH 2.65 to 2.90.Based on this result, a good separation of investigated nucleotides can be obtained with the pH value at 2.75 ± 0.02 or between 2.90 and 3.10.In addition, based on IMP and GMP retention time curves, it could be inferred that these 2 may intersect when the pH value is greater than 3.10.Therefore, the optimized pH results range from 2.75 ± 0.02 and 2.90 to 3.10, where the first range requires more precise control to achieve a good separation result.
Inspectors have changed the pH of the mobile phase for optimizing the shape of chromatographic peaks and prolonging the retention time of target compounds (Buszewski and Noga, 2012).Inspectors would be interested to discover the correlation between hydrogen icon concentration of mobile phase and the peak sequence in C 18 chromatographic column.A tautomeric structure model of AMP effected by H + was thus proposed.As shown in Figure 2b, the atom N (5) in the nitrogenous base would have charges in the mobile phase with a higher concentration of H + , which destroyed the conjugated π bonds on 5-membered ring and enhanced the polarity of AMP.Hence, the retention time of AMP was largely shortened, whereas retention times of other nucleotides remained the same.The AMP could be separated from other nucleotides and impurities in C 18 columns by adjusting the pH to control the retention time.
Effect of Ion-Pair Reagent Concentration on Retention Time of Nucleotides.In preliminary experiments, we discovered that the retention of nucleotides in the chromatographic column was very weak without ion-pair reagents in the mobile phase, and all nucleotides flowed out within 2 min.Therefore, the ionpair reagent THS was added to increase the bond force (Seifar et al., 2009), and the effect of THS was investigated by varying the concentration between 0.25 and 1.00 mg/mL.Similar to the effect of the pH value, the AMP interfered with the retention of GMP and IMP when the ion-pair reagent concentration was between 0.25 and 1.00 mg/mL (5 calibrations) at 65°C column temperature and 2.75 pH (Figure 3a).Judging from their retention time curves, the optimum concentration of THS for separation was 0.55 to 0.75 mg/mL.
It has been reported that the retention times of compounds with different structures were affected differently by changing the concentration of the ion pair (Qiao et al., 2018).This study found 4 nucleotides, each with a hydroxyl group of the bicyclic conjugate structure, had shorter retention times by increasing the ion-pair reagent concentration in the mobile phase, whereas the time of AMP remained the same.It should note that the enol group could transfer into a ketone group in the keto-enol tautomerism (Sheina et al., 1985).For example, when the tetrabutylammonium salt concentration increased in IMP (Figure 3b), the H atom of the hydroxyl group bonded to its adjacent ni- trogen atoms (3), and the original stable structure with conjugated π bonds on 6-membered ring was destroyed.The polarity of IMP increased, and its retention time in C 18 chromatographic column shortened; as a result, the retention time curve of AMP had crossed with those of IMP and GMP.

Validation of Method.
Linearity Range and LOD Standard working solutions were prepared and diluted to 0.5, 2.0, 5.0, 10.0, 20.0, 25.0, and 50.0 μg/mL for calibration curves.Using the concentration and the corresponding peak area as abscissa and ordinate to conduct the linear regression calculation, the calibration equations and determination coefficients were obtained.Calibration curves possess excellent linearity (R 2 > 0.99996) over the established concentration points, demonstrating good correlations between the relative peak areas and the nucleotide content in the samples (Table 1).According to the mathematical Eq. 1 and 2 mentioned above, the instrumental detection limit and the quantification limit of each analyte were similar.To simplify the experiments and calculations, the maximum result among the analytes was selected as the instrumental detection limit and quantification limit of the method.The method detection limit and quantification limit were multiplied by the dilution ratio of the sample in the pretreatment, and the results were 3 mg/kg and 10 mg/kg, respectively.Meanwhile, the linear range of the method was 10 to 1,000 mg/kg.The detection was appropriate because the limitations set by different countries were within the range.

Specificity and Precision
In the spike-and-recovery experiments, samples were spiked with the analyte at 3 different concentrations of 10.0, 20.0, and 60.0 mg/kg.Six replicates of each concentration level were studied.The chromatographic peaks of the nucleotides were symmetrical with the trailing factors ranging from 0.96 to 1.02 (Figure 4).Compared with the unsatisfied separation of the CMP peak in the original method, the resolution in the optimized method was 0.98, which was a better separation effect.Through the HPLC-UV test, it was found that the nucleotide spike-and-recovery rate was in the range of 80.0 to 110.0%, whereas the relative standard deviation varied from 0.5 to 3.5% (n = 6; Table 2).Both of the values were within the acceptable limits, as suggested by the AOAC (Sullivan, 2012); therefore, it could meet the requirement of analysis.
Authentic Samples To access the applicability and robustness of the proposed method, 17 samples of infant formula from different countries were analyzed by this newly developed method.The results reported in Table 3 show that 3 of 17 milk powder samples were not qualified.The total contents of nucleotides in one sample exceeded the maximum limit set by Chinese

CONCLUSIONS
The study revealed that nucleotides are susceptible to column temperature, pH, and concentration of ion reagent THS in chromatographic separation.The optimal conditions were concluded.Their structures changed under different chromatographic conditions, resulting in the failure of separation of AMP and IMP or GMP.The interference of the separation was due to the different nitrogenous base composition between the AMP and the other 4 nucleotides.In this method, the linear range of single nucleotide could be detected from 10 to 1,000 mg/kg, covering the requirement of conventional nucleotide addition (national standard).The recovery of standard addition was 80.0 to 110.0%, and the relative standard deviation ranged from 0.5 to 3.5% (n = 6).This method was applied to analyze com-mercial infant formula samples and achieved satisfactory results.The variation of 5 nucleotides summarized in this study under different conditions will provide the direction and theoretical basis for optimizing the separation, preparation and detection of nucleotides.Exceed or not: the detection value exceeds the maximum limit or not.
3 Identity value: the content of added nucleotides identified on the product label. 4 False or true: when the detection value <70% of the identity value, the label information is false; when the detection value ≥ the identity value, the label information is true.

Figure 1 .
Figure 1.Effects of HPLC column temperature on (a) retention time of 5 nucleotides, and (b) the potential change of chemical structure of AMP.The range of retention time at different temperatures represent the width of the nucleotides' chromatographic peaks.Symbols are represented as follows: (■) cytidine 5′-monophosphate, (•) uridine monophosphate, (▲) guanine monophosphate, (▼) AMP, and (♦) inosine monophosphate.The numbers on the rings are arranged according to the International Union of Pure and Applied Chemistry (IUPAC) nomenclature.

Figure 2 .
Figure 2. Effects of the HPLC mobile phase pH on (a) retention time of 5 nucleotides, and (b) the potential change of chemical structure of AMP.The range of retention time at different temperatures represent the width of the nucleotides' chromatographic peaks.Symbols are represented as follows: (■) cytidine 5′-monophosphate, (•) uridine monophosphate, (▲) guanine monophosphate, (▼) AMP, and (♦) inosine monophosphate.The numbers on the rings are arranged according to the International Union of Pure and Applied Chemistry (IUPAC) nomenclature.

Figure 3 .
Figure 3. Effects of ion-pair reagent concentration on (a) retention time of 5 nucleotides, and (b) the potential change of chemical structure of inosine monophosphate (IMP).The range of retention time at different temperatures represent the width of the nucleotides' chromatographic peaks.Symbols are represented as follows: (■) cytidine 5′-monophosphate, (•) uridine monophosphate, (▲) guanine monophosphate, (▼) AMP, and (♦) IMP.The numbers on the rings are arranged according to the International Union of Pure and Applied Chemistry (IUPAC) nomenclature.

Table 1 .
Chen et al.: CHROMATOGRAPHIC SEPARATION OF NUCLEOTIDES IN INFANT FORMULA MILK POWDER Standard curve equation, coefficient of determination, and linear range of nucleotides

Table 2 .
Recovery ratMateos-Vivas et al., 2016)des added at different levels in milk powder Ministry of Health of the People's Republic ofChina, 2012;Mateos-Vivas et al., 2016)had an unstable performance after 20 samples in the pre-experiment.Hence, this method has a strong practicality.

Table 3 .
Results of detecting the total contents of 5 nucleotides in unqualified infant formula sold on the market Maximum limit: the total contents of nucleotides set by Chinese government.