UPLC-MS/MS-Based Target Screening of 90 Phosphodiesterase Type 5 Inhibitors in 5 Dietary Supplements

The aim of individuals consuming health supplements is to attain a robust state through nutritional regulation. However, some unscrupulous manufacturers, motivated by profit, fraudulently incorporate drugs or unauthorized components with therapeutic effects into the product for instant product performance enhancement. The long-term use of these products may inadvertently inflict harm on human health and fail to promote nutritive healthcare. The illegal inclusion of these substances is prevalent in kidney-tonifying and sexuality-enhancing products. Developing effective analytical methods to identify these products and screen for illegal added ingredients can effectively prevent such products from reaching and remaining on the market. A target screening method for the detection and quantification of 90 phosphodiesterase type 5 inhibitors (PDE-5is) in 5 kinds of health products was developed and validated. The type of dietary supplements varied from tablets, capsules, and protein powder to wine and beverages. Sample preparation was completed with a one-step liquid phase extraction. The screening process of 90 PDE-5is was done efficiently within 25 min by ultra-high performance liquid chromatography tandem mass spectrometry (UPLC-MS/MS) using the dynamic multiple reaction monitoring (dMRM) technique. The LODs of 90 PDE-5is were detected at levels ranging from 25 to 85 ng/g or ng/mL. This novel targeting methodology was effective and can be applied to routine market supervision. Among 286 batches of samples, 8 batches were found to be positive. Three kinds of PDE-5is were first detected in healthy products. The screening method demonstrated herein will be a promising and powerful tool for rapid screening of PDE-5is.


Introduction
The global trend toward dietary supplements to encourage healthier living has consistently intensified appreciably [1,2].Considering the burgeoning demand for dietary supplements, potential risks also escalate concurrently.Throughout the previous decade, instances of suspected illegal adulterants of pharmaceuticals, unapproved drugs, and regulated substances have amplified markedly, posing a significant global challenge to the safety surveillance of dietary supplements [3].Phosphodiesterase type 5 inhibitors (PDE-5is) can selectively inhibit the activity of the PDE-5 enzyme and increase the concentration of cyclic guanosine monophosphate (cGMP) in cells, resulting in smooth muscle relaxation [4].At the earliest stage, it was used as a clinical drug for cardiovascular diseases [5].Later, it was found that it has a therapeutic effect on patients with erectile dysfunction (ED), and nowadays, it has been used as the first-choice drug for the treatment of ED [6][7][8].In pursuit of improving sexual ability, PDE-5is are often unethically added to various health products and dietary supplements.Most of these additions are unmarked and illegal, and taking these products without the guidance of a doctor is a great threat to health [9,10].
Several methods of PDE-5is detection or screening have been developed to effectively oversee and combat this illegal addition [11][12][13][14][15][16].Most of these methods rely on liquid chromatography tandem mass spectrometry (LC-MS/MS) [17][18][19][20][21][22][23].According to the differences in instruments, these methods are bifurcated into two main classifications: non-target screening methods based on high-resolution mass spectrometry and target screening methods based on tandem quadrupole mass spectrometry [24].Non-target screening utilizes a database of precise reference material MS and MS/MS profiles acquired on high-resolution mass spectrometry [25].The number of compounds screened hinges upon the contents of the database.Data processing consumes 70% of the whole screening work.Compound parameters, such as retention time and MS/MS spectrum are primary matching elements.Filtered parameters, such as the threshold of matching, are key factors affecting the screening results [26][27][28].
In contrast to the non-target methodology, the target screening approach is guided by optimizing acquisition parameters.The retention time and multiple reaction monitoring (MRM) transition channels of each compound have been fine-tuned to obtain the highest screening efficiency.The retention time and multiple reaction monitoring (MRM) transition channels for each compound have been meticulously curated to attain the utmost screening efficacy.The precise calibrators crafted by the reference materials are employed to construct the standard curve, facilitating not merely target screening but also precise quantification of the screened compounds [29].Over half of the target screening endeavor entails sample preparation and data acquisition.The established screening methodology by reference material data allows for swift sample examination and compound quantification.False positive rates are significantly lower for target screening methods compared to non-target screening methods.For the scrutiny of health-threatening compounds, target screening methods with superior accuracy are typically employed.Lee et al. simultaneously identified 38 PDE-5is in illicit erectile dysfunction products [30].Philippe et al. developed a prompt target assessment methodology for 71 active and 11 naturally occurring erectile dysfunction ingredients found in potentially tainted or counterfeit goods.Furthermore, they incorporated a high-resolution full scan procedure into this method, enabling subsequent identification via an untargeted approach to minimize the potential for false-positive outcomes [31].With the notable advancement in chromatographic separation and mass spectrometry acquisition, over 150 compounds can be target screened in a single analysis [32].Qie crafted a swift method for the simultaneous quantification of 160 drugs in urine or blood [33].Employing the power of the scheduled MRM mode, Yin and colleagues presented an extensive multiresidue analytical methodology for 210 drugs in pork within a concise 20 min [34].These case studies vividly underscore the broad application of the targeted screening strategy in food regulation.
To scrutinize all PDE-5is discovered, this research refined a target screening methodology to assess 90 of them (outlined in Table S1) across five food substrates or wholesome products, encompassing tablets, capsules, protein powder, healthful wine, and functional beverages.There was a significant incidence of PDE-5is incorporation into these wholesome products; this research approach streamlined existing laboratory methodologies and notably enhanced the food surveillance capabilities in the fight against illicit additions.

Extraction Solvent Optimization
Most PDE5-is were aromaticity compounds with weak polarity and limited solubility in pure water; organic solvents such as methanol, ethanol, acetone, and acetonitrile were often used for sample pre-treatment [35,36].Within this study, methanol and acetonitrile were utilized for sample dilution or extraction; all PDE5-is were identifiable in these two solvents.However, in blank matrix-spiked samples, some compounds such as hydroxythiovardenafil exhibited splitting chromatographic peaks in acetonitrile, particularly evident in samples of wine and beverages; however, it was ameliorated when methanol was employed as solvent.The splitting chromatographic peak in acetonitrile is shown in the upper part of Figure S1, and the ameliorated peak in methanol is shown in the lower part of Figure S1.

LC-MS/MS Optimization
For liquid chromatography separation, methanol and 0.1% formic acid aqueous solution were selected as the mobile phase due to the enhanced solubility of PDE-5is in methanol.Initially, a 10% organic phase was maintained for 1 min to ensure robust retention of the sample on the chromatographic column; subsequently, the proportion of the organic phase was incrementally increased to facilitate sequential elution of the compound for analysis.The proportion of the organic phase was elevated to 100% until 19 min and sustained for 3 min to ensure complete elution of all substances in the sample without accumulation on the chromatographic column to prevent contamination.To optimally separate 90 PDE-5i, three distinct types of columns were employed.In comparison to HSS T3 (2.1 × 100 mm, 1.8 µm, Waters, Milford, MA, USA) and BEH C18 (2.1 × 100 mm, 1.7 µm, Waters, Milford, MA, USA), separation on Zorbax Eclipse Plus C18 (3.0 × 150 mm, 1.8 µm, Agilent, Santa Clara, CA, USA) was superior, as this column was the longest, exhibiting superior resolution for basic PDE-5is compounds.Figure S2 presents the outcomes of LC separation, employing three different chromatographic columns.
For MS detection, the unique standard solution (100 ng/mL) of each compound was administered into the mass spectrometer, and the fragmentation voltage was refined in full scan mode to facilitate the precursor ion to attain the utmost response.The polarities of the electrospray ionization (ESI) source were also optimized; all PDE-5is exhibited superior response in positive mode except depiperazinothio sildenafil, vardenafil dimer, and N-phenylpropylethyl tadalafil.These three PDE-5is manifested superior response in negative mode.Subsequently, the acquiring mode was shifted to the product ion mode to optimize the collision energy, and the two most abundant product ions were chosen for quantification and qualitative validation.Currently, PDE-5is primarily comprise two structural classifications, akin to the two popular sex-enhancing drugs presently available.The first comprises structural analogs of sildenafil, yielding fragment ions with a m/z of 311.1, 113.1, or 99.1; the second, structural prostaglandins of Tadalafil, predominantly produce fragment ions at a m/z of 135.The subtle differences in structure engender disparities in their collision energies, necessitating individual optimization.
When all 180 transitions were configured in a single MRM method, the dwell time of each transition could only be allotted to 2 milliseconds when the cycle time was set to 400 milliseconds, which significantly impacted the sensitivity of the methodology.To mitigate this limitation, the dynamic multiple reaction monitoring (dMRM) modality was employed for data acquisition, wherein specific ion transitions were apprehended exclusively within a narrow retention time interval, thus potentially reducing the number of concurrent ion transitions and significantly augmenting sensitivity [37][38][39][40][41].The extract ion chromatograms (EICs) of the transitions of all PDE-5is are shown in Figure S3.Among these 90 compounds, 2-hydroxypropylnortadalafil and the vardenafil dimer need special attention because they both have tautomerism; there were two peaks in the respective EICs, which are shown in Figure 1.The two peak areas need to be added together for quantification.

Selectivity, Linearity, Sensitivity, and Matrix Effects
The chromatograms of blank matrix and spiked matrix samples at a concentration of 100 ng/mL were juxtaposed to illustrate the selectivity of the devised methodology.The findings are illustrated in Figure S4.Peaks in the chromatograms of the blank matrix did not obstruct the analysis of PDE-5is in all five matrices, which validated the outstanding selectivity of the method.
The calibration curves of all 90 PDE-5is were in accordance with the quantitative analysis requirement, demonstrating excellent regression coefficient values exceeding 0.99.The matrix effects exhibited considerable fluctuations across the five diverse matrices.For tablets, healthful wine, and functional beverages, the matrix effects were within the range of 85-115%, indicating a negligible impact on PDE-5is quantification.For capsules and protein powder, the matrix effects were less than 80%, establishing matrixmatched calibration curves to mitigate the influence of the matrix.The regression coefficients of these two matrix-matched calibration curves were superior to 0.99, reflecting superb linearity.The method LODs and LOQs were evaluated utilizing the spiked-in sample in a blank matrix.The LOD and LOQ of each of the 90 PDE-5is are detailed in Table 1.The LODs of these 90 PDE-5is were less than 50 ng/g or 30 ng/mL.These thresholds were relatively low and suggest that the developed methodology is sufficiently sensitive to quantify the PDE-5is in various matrices.

Selectivity, Linearity, Sensitivity, and Matrix Effects
The chromatograms of blank matrix and spiked matrix samples at a concentration of 100 ng/mL were juxtaposed to illustrate the selectivity of the devised methodology.The findings are illustrated in Figure S4.Peaks in the chromatograms of the blank matrix did not obstruct the analysis of PDE-5is in all five matrices, which validated the outstanding selectivity of the method.
The calibration curves of all 90 PDE-5is were in accordance with the quantitative analysis requirement, demonstrating excellent regression coefficient values exceeding 0.99.The matrix effects exhibited considerable fluctuations across the five diverse matrices.For tablets, healthful wine, and functional beverages, the matrix effects were within the range of 85-115%, indicating a negligible impact on PDE-5is quantification.For capsules and protein powder, the matrix effects were less than 80%, establishing matrix-matched calibration curves to mitigate the influence of the matrix.The regression coefficients of these two matrix-matched calibration curves were superior to 0.99, reflecting superb linearity.The method LODs and LOQs were evaluated utilizing the spiked-in sample in a blank matrix.The LOD and LOQ of each of the 90 PDE-5is are detailed in Table 1.The LODs of these 90 PDE-5is were less than 50 ng/g or 30 ng/mL.These thresholds were relatively low and suggest that the developed methodology is sufficiently sensitive to quantify the PDE-5is in various matrices.

Precision and Recoveries
The degrees of precision were computed through the ascertainment of mixed standard solutions spiked in 5 distinct blank matrices at concentrations of 15 ng/mL, 80 ng/mL, and 150 ng/mL.Six replicates of these solutions were quantified on a single day to assess intraday precision, and the identical samples were assessed continually for three consecutive days to appraise the inter-day precision.All intra-day degrees of precision were below 10%, and all inter-day degrees of precision were below 15%.All recoveries of the solutions formulated at 3 concentrations in 5 matrices fell within the range of 80-120%.The mean precision and recoveries of the solutions in five matrices at divergent concentrations are illustrated in Table 2.
Table 2.The average recoveries of 90 PDE-5is obtained in five matrices (tablets, capsules, protein powder, healthful wine, and functional beverages) at 15 ng/mL, 80 ng/mL, and 150 ng/mL.The average inter-day precision was obtained using six replicates of the same solution as the recoveries mentioned above.The inter-day precision was obtained by continuous determination for three days of the same solution as the intra-day precision mentioned above.

Method Application
The established method has played an important role in health product risk monitoring.From the perspective of product efficacy claims, samples can be segmented into five categories: boosting immunity, alleviating fatigue, reducing blood pressure, and amplifying sexual performance, and others.Within 286 cohorts of samples, PDE-5is were identified in 8 cohorts.Sildenafil was detected in 6 cohorts of tablet samples.It was a traditional illicit additive with a positive detection rate of 2.1%.In addition to 4 instances identified in samples that enhanced sexual efficacy, 2 events were identified in samples that mitigated fatigue.This suggests that illegal additions are becoming progressively more concealed, escalating the complexity of risk surveillance.2-Hydroxypropyl nortadalafil was detected positively in one of the nutritionally beneficial wine samples.N-Ethyltadalafil was identified in one of the capsule samples.These two inhibitors have not been detected in our previous screening endeavors, signifying that the illegal incorporation of PDE-5is has become progressively covert.For all screening specimens, the overall positive detection rate was 2.8%; it is imperative to persistently expand the spectrum of detection compounds for food risk evaluation.

Reagents and Standards
The names, molecular formulas, and CAS numbers of the exceptional 90 PDE-5is standards are detailed in Table S1.Reference standards (purity ≥ 98%) for the 90 PDE-5is were obtained from Dr. Ehrenstorfer Laboratories (Augsburg, Germany).Formic acid (LC-MS grade) was purchased from Merck Co. (Darmstadt, Germany).Methanol (MeOH, LC-MS grade) was purchased from Honeywell Burdick & Jackson (Muskegon, MI, USA).Ultrapure water (18.2MΩ) was obtained from a Milli-Q Advantage ultrapure water purification system.Healthy products such as vitality tablets, epimedium capsules, ginseng wine, protein power, and functional energy beverages were obtained from the National Institutes for Food and Drug Control (Beijing, China).

Sample Preparation
Liquid substances, such as wine and beverages, were precisely determined after dilution to an exact volume.For solid substrates, the tablets were meticulously crushed into a uniform powder and blended thoroughly.The shells of all capsules were gently eliminated, and the powder within was delicately mixed.Typically, 1 g (±0.05 g) of solid samples (tablets, capsules, and powder) or 1 mL of liquid samples (wine and beverage) were precisely transferred into a 10 mL volumetric flask and dissolved in 7 mL methanol, followed by gentle sonication for 10 min.The volume of all samples was precisely fixed to 10 mL after the solid matter was completely dispersed or dissolved.Then, 5 mL methanol was transferred to a centrifuge tube and centrifuged at 14,000 rpm for 5 min sequentially; the supernatant was collected and filtered through a 0.22 µm filter (NormJect, Tuttlingen, Germany) prior to analysis.

Preparation of Standard Solutions and Calibration Curve
Stock dilutions of the individual standards were meticulously prepared in methanol (ranging from 300 to 1000 µg/mL) and preserved at −80 • C. The working standard combination solution of 90 PDE-5is was formulated monthly at a concentration of 3 µg/mL in methanol and carefully stored at −20 • C. Blank instances were diligently processed, as outlined in Section 3.2, to yield a blank matrix solution.Subsequently, the matrix-matched calibration solution with the highest concentration (200 ng/mL) was prepared by fortifying the working solution (3 µg/mL) into a blank matrix solution.Finally, the remaining five matrix-matched calibration solutions were prepared by progressive dilution with the highest matrix-matched calibration solution (200 ng/mL).Mixed standard solutions with concentrations of 5, 10, 20, 50, 100, and 200 ng/mL were employed for constructing matrix-matched calibration curves.All calibrators were meticulously prepared prior to use.External standard calibration was utilized for the quantitative analysis.
MS analysis was executed on an Agilent 6460 triple quadrupole (QQQ) mass spectrometer furnished with an AJS electrospray ionization source (ESI).The fine-tuned ESI conditions were: drying gas (N2) temperature at 200 • C, drying gas flow at 14 L/min, sheath gas (N2) temperature at 250 • C, sheath gas flow at 11 L/min, capillary at 3000 V, and nebulizer gas at 40 psi.The acquisition was conducted in dynamic multi-reaction mode (dMRM) with a cycle duration of 500 ms.Two specific transitions for each PDE-5i were monitored over an automatic set delta retention time.The detailed parameters of the established methodology are illustrated in Table S2.

Method Validation
The optimized sample screening methodology was meticulously validated in terms of selectivity, linearity, quantification, precision, recovery, and sensitivity in various matrices [30,35,36].Selectivity was evaluated by comparing the chromatograms of 80 ng/mL standard analyte-spiked solutions and blank matrix, confirming the full resolution of all MRM extracted ion chromatograms (EICs) devoid of matrix interference.The peak area was correlated with specific spiked analyte concentrations within a blank matrix to formulate calibration curves.Calibration linearity was assessed using the coefficient of determination (R 2 ).Accuracy and precision were confirmed through the analysis of spiked replicates at 3 concentration levels (15 ng/mL, 80 ng/mL, and 150 ng/mL), serving as quality control (QC) samples.Precision was expressed as a relative standard deviation (RSD, %) of the measured concentrations in spiked replicates.The recoveries of analytes across diverse matrices were also calculated at identical concentrations.The method's reliability was gauged by establishing the limits of detection (LODs) and quantification (LOQs) of the spiked samples.When the signal-to-noise ratio (S/N) exceeding 3 for an individual target analyte in sequential dilutions was classified as LOD, it reached 10 signified LOQ.Matrix interference was assessed by comparing the measured peak area of analytes from QC samples to the predicted value using standards prepared in the solvent.

Conclusions
A targeted detection methodology for the concurrent quantification of 90 PDE-5is in five types of wholesome foods was efficiently designed and validated.The sample extraction procedures were straightforward and dependable, eliminating the necessity for a complementary clean-up protocol.The detection and quantitative evaluation of the 90 PDE-5is could be accomplished utilizing UHPLC-MS/MS, with a gradient chromatographic runtime of merely 25 min, executed in dMRM mode.This methodology provides an adept approach to target detection of PDE-5is in wholesome food with superior selectivity, precision, and sensitivity.Further expansion of this methodology could be achieved by augmenting the newly identified PDE-5i to enhance the capability of illicit addition detection.

Table 1 .
The LODs and LOQs of 90 PDE-5is obtained in five different matrices.

Table 1 .
The LODs and LOQs of 90 PDE-5is obtained in five different matrices.