Development and validation of an LC-MS/MS method for detection and quantification of in vivo derived metabolites of [Pyr1]apelin-13 in humans

[Pyr1]apelin-13 is the predominant apelin peptide isoform in the human cardiovascular system and plasma. To date, few studies have investigated [Pyr1]apelin-13 metabolism in vivo in rats with no studies examining its stability in humans. We therefore aimed to develop an LC-MS/MS method for detection and quantification of intact [Pyr1]apelin-13 and have used this method to identify the metabolites generated in vivo in humans. [Pyr1]apelin-13 (135 nmol/min) was infused into six healthy human volunteers for 120 minutes and blood collected at time 0 and 120 minutes after infusion. Plasma was extracted in the presence of guanidine hydrochloride and analysed by LC-MS/MS. Here we report a highly sensitive, robust and reproducible method for quantification of intact [Pyr1]apelin-13 and its metabolites in human plasma. Using this method, we showed that the circulating concentration of intact peptide was 58.3 ± 10.5 ng/ml after 120 minutes infusion. We demonstrated for the first time that in humans, [Pyr1]apelin-13 was cleaved from both termini but the C-terminal was more susceptible to cleavage. Consequently, of the metabolites identified, [Pyr1]apelin-13(1–12), [Pyr1]apelin-13(1–10) and [Pyr1]apelin-13(1–6) were the most abundant. These data suggest that apelin peptides designed for use as cardiovascular therapeutics, should include modifications that minimise C-terminal cleavage.

reported in metabolic diseases where it decreased adiposity, serum insulin and increased insulin sensitivity 16,17 ; and in renal diseases where it decreased acute renal injury and fibrosis 18 . It has recently emerged that apelin has pro-tumorigenic effects in various cancer models possibly by promoting angiogenesis and that inhibition of the apelin pathway was protective against tumour growth 14,19 . However, these beneficial effects of apelin peptides are limited by the rapid in vivo metabolism.
Previous studies investigating the metabolism of apelin peptides were largely conducted in plasma in vitro or in rodent models neither of which may represent metabolism in humans. These studies demonstrated that apelin peptides are very labile in plasma with a half-life of less than 1-5 minutes in vitro [20][21][22][23][24] . This plasma instability has to date been attributed the enzymatic activity of neprilysin 25 and angiotensin converting enzyme II (ACE2) [22][23][24] , and more recently plasma kallikrein 26,27 . Similarly, another recent study reported more rapid degradation of [Pyr 1 ]apelin-13 in rat and mouse plasma when compared to dog, monkey and human plasma in vitro 28 . The authors also confirmed their findings in vivo in rat and mouse, and identified N-terminal metabolites of the peptide, particularly apelin-7 that was most abundant 28 . This study therefore highlighted species differences in the repertoire of proteases circulating and present in rodent and higher mammalian systems. However, to date no studies have investigated the metabolism of apelin peptides in vivo in humans. The aim of this study was to develop a highly sensitive mass spectrometry based method for detection and quantification of apelin peptides in plasma. We used this method to measure intact [Pyr 1 ]apelin-13 and its metabolites generated in humans, following a constant 120 minutes infusion of the peptide. We found that [Pyr 1 ]apelin-13 was cleaved into smaller fragments from both termini but that the C-terminal was more susceptible. We identified the biologically active C-terminal cleavage product, [Pyr 1 ]apelin-13 (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12) , as the most abundant, as well as identifying novel metabolites including [Pyr 1 ]apelin-13 (1)(2)(3)(4)(5)(6)(7)(8)(9)(10) and [Pyr 1 ]apelin-13 (1)(2)(3)(4)(5)(6) .

Results
Precision and accuracy of the extraction and quantification method. An 8-point calibration line was generated for [Pyr 1 ]apelin-13 in human plasma (r 2 = 0.99, data not shown), with a lower limit of quantification (LLOQ) of 1 ng/ml. The relative errors (% RE) for all calibration standards were less than 20% at the LLOQ and below 15% at other levels, conforming with typical bioanalytical method validation guidelines 29 . The precision and accuracy of the QC samples showed that the method was robust and accurate. The LLOQ samples returned a coefficient of variation (%CV) of 8.0 and %RE of 15.5, whilst the other QC levels had %CV's below 6.1 and %RE's below 8.4. Representative chromatograms obtained from calibration standards 1 and 8 are shown in Fig. 1. obtained before infusion of [Pyr 1 ]apelin-13, no chromatographic peak was observed for the peptide ( Fig. 2A,B). Samples obtained at the end of the infusion (t = 120 minutes) showed strong peaks at 3.68 minutes corresponding to [Pyr 1 ]apelin-13 (Fig. 2C,D). The measured concentration of [Pyr 1 ]apelin-13 in these samples after 120 minutes was 58.3 ± 10.5 ng/ml. Additionally, data from the six donor control samples that did not receive [Pyr 1 ] apelin infusion showed that the endogenous levels of [Pyr 1 ]apelin in these samples were below the LLOQ (see Supplementary Fig. 1). The peak height obtained from the chromatogram of these donor samples had a maximum height that was 19.8% of that seen in the LLOQ and so was considered as blank for quantitative purposes based on the FDA method validation guidelines for demonstrating selectivity of an LC-MS methodology 30 .
Oxidation of the methionine residue in [Pyr 1 ]apelin-13 was identified, however since this modification was also observed in the extracted standards, it could not be ascertained if they occurred in vivo or as an artefact of the extraction process.
The discovery that [Pyr 1 ]apelin-13 was cleaved from both ends was unexpected since to date only cleavage from the C-terminus has been described 23,24 . Our findings may therefore better explain the extremely unstable nature of apelin peptides in plasma 6,28,31 . It is worth noting that the C-terminus was more susceptible to proteolytic activity than the N-terminus, whose metabolites were present at approximately 20-fold lower levels. This may partly be explained by the pyroglutamylation of the N-terminus which may protect this region from enzymatic activity to some degree. The N-terminus of [Pyr 1 ]apelin-13 also contains the RPRL motif critical for binding to the apelin receptor 32 , hence any cleavage from this direction is likely to profoundly affect the affinity of these N-terminal fragments for the receptor. A previous study showed in vitro that neprilysin cleaves [Pyr 1 ] apelin-13 between Arg 4 and Leu 5 and between Leu 5 and Ser 6 amino acids 25 , thereby making neprilysin the first enzyme identified to date that completely inactivates the peptide. Importantly, we have now shown in this study the presence of one of these proposed neprilysin cleavage products, [Pyr1]apelin-13 (6)(7)(8)(9)(10)(11)(12)(13) , in humans in vivo   (1)(2)(3)(4)(5) with 5.88 minutes retention time. These data were acquired using Orbitrap Mass spectrometer used for metabolite identification. The mass accuracy of the experimentally acquired monoisotopic peak was calculated for each potential metabolite, and is included along with the 13 C isotopic cluster for each peptide with their corresponding chromatogram. www.nature.com/scientificreports www.nature.com/scientificreports/ with additional evidence for cleavage of the scissile bond between Leu 5 and Ser 6 given by the detection of the C-terminal fragment, [Pyr 1 ]apelin-13 (1)(2)(3)(4)(5) . To date very few studies have investigated the metabolism of peptides in vivo in humans. Interestingly, like [Pyr 1 ]apelin-13, arginine vasopressin was also proposed to be cleaved in vivo from both the C-and N-termini, with carboxypeptidase and post-proline enzymes cleaving the C-terminus of arginine vasopressin, while aminopeptidases cleaved the N-terminal region [33][34][35] . In contrast, other in vivo studies of this nature identified only a single terminus cleavage of Peptide YY 36 , growth hormone-releasing hormone 37 , liraglutide, a glucagon-like peptide-1 (GLP-1) analogue 38 and big endothelin-1 39 .
Inhibition of degradative enzymes is a well-established strategy to generate therapeutic agents. ACE2 is an important member of the renin-angiotensin system that converts angiotensin-II to angiotensin 1-7, with the latter mediating vasodilatation, anti-proliferation, anti-apoptosis and anti-fibrotic effects 41 . In addition, ACE2 has been implicated in heart failure 42,43 , diabetic nephropathy 44,45 , acute lung failure 46 , lung injury induced by the lethal avian influenza A H5N1 virus 47 , respiratory syncytial virus 48 and severe acute respiratory syndrome (SARS) 46 . Recently, GSK developed a recombinant human ACE2, GSK2586881 for treatment of acute respiratory distress syndrome (ARDS) and showed that this molecule was well-tolerated in clinical trials 49 . Corroborating on this, apelin signalling induces ACE2 expression in failing hearts 12 , and protects against lung injury in experimental models of acute respiratory distress syndrome 50 , possibly by inhibiting the NF-κB pathway and components of the inflammasome 51 . Furthermore, apelin counteracts the effects of angiotensin-II signalling, which is negatively regulated by ACE2, suggesting that targeting ACE2 and apelin could be a potentially novel therapeutic strategy for treatment of lung injury related pathologies and heart failure.
The beneficial effects of apelin in heart failure are well characterised. Plasma apelin levels have been suggested to increase in early stages 5 of heart failure but decrease in late stages of the disease [52][53][54] . In support of this, heart failure therapies such has cardiac resynchronisation therapy used to treat refractory chronic heart failure were shown to increase plasma apelin suggesting that increased apelin levels are associated with improved therapeutic benefit 54 . Apelin administration increased stroke volume and contractility in failing hearts 11 , thereby improving the performance of the failing heart. Similarly, neprilysin inhibitors have emerged as a pivotal therapeutic strategy for clinical management of heart failure due to the role of neprilysin in the degradation of vasoactive peptides including natriuretic peptides and bradykinin 55 . Indeed, neprilysin inhibitors such as sacubitril are used for clinical management of heart failure 56 . Our data may therefore suggest that an additional benefit of neprilysin inhibitors in heart failure is to reduce apelin inactivation resulting in beneficial vasodilation, increased contractility and cardiac output. Building on these findings, further studies could substitute the amino acids at the neprilysin cleavage sites in [Pyr 1 ]apelin-13 with unnatural amino acids to improve its resistance to degradation. Indeed, it was recently shown that infusion of neprilysin resistant apelin-17 in an established mice model of abdominal aortic aneurysm ameliorated the adverse aortic remodelling and aneurysm formation 27 . Such a strategy was also demonstrated to significantly increase the resistance of [Pyr 1 ]apelin-13 and apelin-17 to ACE2 activity 22,23 , suggesting that this could potentially be a mechanism to improve plasma stability of apelin-based therapeutics for clinical indications. We have recently published on an another approach to stabilise apelin peptides in human blood using albumin domain (AlbudAb) antibody conjugated to an apelin analogue, MM202 and showed that this peptide was resistant to degradation yet retained biological activity at the human apelin receptor in vitro and in vivo 9 . Therefore, these strategies could in the near future result in the development of the first apelin-based therapeutics for treatment of human diseases. www.nature.com/scientificreports www.nature.com/scientificreports/ In conclusion, apelin peptides have protective roles in cardiovascular diseases, however, any potential therapeutic use is impaired by the poor plasma stability of the peptide. In this study, we have developed a highly sensitive method for detection and quantification of [Pyr1]apelin-13 in human plasma. For the first time in humans in vivo we have identified as the most abundant metabolite of [Pyr 1 ]apelin-13, the ACE2 cleavage product, [Pyr 1 ] apelin-13 (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12) that we have previously demonstrated retains significant biological activity in addition to the putative neprilysin metabolites [Pyr 1 ]apelin-13 (4)(5)(6)(7)(8)(9)(10)(11)(12)(13) and [Pyr 1 ]apelin-13 (6)(7)(8)(9)(10)(11)(12)(13) . Combined inhibition of ACE2 and neprilysin may be a novel strategy to enhance endogenous apelin levels in conditions in which the peptide is downregulated. Additionally, these data will inform the design of more stable apelin peptides for therapeutic use.

Material and Method
Materials.   (7)(8)(9)(10)(11)(12)(13) with retention time of 8.08 minutes, (C) [Pyr 1 ]apelin-13 (8)(9)(10)(11)(12)(13) with retention time of 9.35 minutes, (D) [Pyr 1 ]apelin-13 (10)(11)(12)(13) with retention time of 10.45 minutes, (E) [Pyr 1 ]apelin-13 (11)(12)(13) with retention time of 9.32 minutes. The mass accuracy of the experimentally acquired monoisotopic peak was calculated for each potential metabolite, and is included along with the 13 C isotopic cluster for each peptide with their corresponding chromatogram. www.nature.com/scientificreports www.nature.com/scientificreports/ healthy volunteers (3 male and 3 female, mean age 43.8 ± 6.9, with body mass index within the normal range of 23.0 ± 1.0) were recruited for infusion. Volunteers were fasted and were lying supine with their heads supported in a quiet, temperature controlled (23-25 °C) room for the duration of the study. Following a period of acclimatisation, the first sample of venous blood was obtained from the arm contralateral to the arm used for infusion of apelin. Vials containing [Pyr 1 ]apelin-13 were allowed to warm to room temperature and diluted with physiological saline to produce stock solutions, that were then filtered using a 0.2 µm Portex flat filter (Portex, UK) before undergoing serial dilutions with 0.9% sterile saline. There was no loss of apelin following this filtration procedure. Volunteers were infused with a concentration of 135 nmol/min of [Pyr 1 ]apelin-13, at a rate of 1 ml/min for 120 minutes, using a syringe pump, equipped with a 50 ml syringe and 16 gauge catheter. The second venous sample was obtained immediately after the end of the infusion. Blood samples were collected into 2.6 ml EDTA tubes, immediately put on wet ice and centrifuged for 7 minutes at ~4 °C, 4000 rpm and stored a −70 °C, prior to analysis. A previous study had used a concentration of up to 100 nmol/min for systemic infusion, where they obtained a therapeutic response in patients with pulmonary arterial hypertension and the highest dose was well tolerated 15 . The dose chosen of 135 nmol/min of [Pyr 1 ]apelin-13, was slightly higher in order to identify possible metabolites. Additional control samples were obtained from 6 donors (3 male and 3 female) within a similar age group who did not receive the apelin infusion for comparison. A full scan analysis of the peptide showed that the [M + 4 H] 4+ charge state was the predominant ion in the spectrum as previously described by Mesmin et al. 31 in their LC-MS/MS analysis of [Pyr 1 ]apelin-13. Therefore this was selected for fragmentation. A product ion spectrum was collected over a range of 100 to 1600 m/z and two ions were selected for SRM optimisation (m/z 424.6 and 408.55). The 424.6 ion corresponded to the b11 fragment and the 408.55 ion was derived from the loss of a methyl-sulphide group from the methionine on the b11 ion, as previously described by Mesmin et al. 31 26. Peptide peak areas were integrated using the TargetLynx program associated with Masslynx V 4.2 (Waters), and peak area ratios were generated against the corresponding stable isotope-labelled internal standard peptide peak. extraction of [pyr 1 ]apelin-13 from human plasma. Plasma samples were thawed on ice and 50 µl transferred into protein LoBind Eppendorf tubes containing 25 µl GuHCl (6 M). A 300 µl aliquot of 80% ACN in water (containing 25 ng/ml internal standard) was added to all plasma samples and vortexed before centrifuging at 12000 x g for 5 minutes to precipitate plasma proteins. The supernatant was transferred to a 1 ml protein LoBind 96-well plate and evaporated. Samples were reconstituted in 500 μl 0.1% FA (v/v) and loaded onto an Oasis HLB Prime µ-elution 96-well plate (Waters, Wilmslow, UK) and slowly extracted on a positive pressure manifold (Waters). The columns were washed with 200 µl of 5% methanol in water with 1% acetic acid (v/v) and eluted from the cartridge using 2 × 50 µl of 60% methanol in water with 10% acetic acid (v/v). The eluate was evaporated to dryness and reconstituted in 150 µl 0.1% FA (v/v) in water and 10 µl was injected onto the LC-MS/MS system. precision and accuracy of the extraction method. Blank plasma was pre-incubated at 37 °C for at least 2 hours, to degrade any endogenous [Pyr 1 ]apelin-13 and used to generate an eight point calibration line of custom synthesised [Pyr 1 ]apelin-13 over a range of 1-1000 ng/ml. A 50 µl aliquot of each calibration standard (1, 2, 5, 10, 50, 100, 900, and 1000 ng/ml) was extracted using the SPE method described above. Four levels of QC were also generated (1, 3, 100 and 800 ng/ml) and extracted six times in order to assess the precision and accuracy of the method. Calibration line followed a linear fit, and 1/x 2 weighting was applied. Recovery of the [Pyr 1 ]apelin-13 from plasma was assessed by analysing spiked solution before and after extraction at a concentration of 100 ng/ml. Plasma samples from six individuals were also extracted to assess the selectivity of the LC-MS/MS method.
Peptide Identification using high-resolution mass spectrometry. Samples were reanalysed on a high resolution mass spectrometer to identify potential metabolites from the administered [Pyr 1 ]apelin-13 peptide. A full scan analysis was performed using a ThermoScientific Ultimate 3000 LC system connected to a ThermoScientific Orbitrap Q-Exactive Plus mass spectrometer. Solvents used for the separation were A: 0.1% FA in water (v/v) and B: 0.1% FA in ACN (v/v). A volume of 30 μl of extract was injected onto a HSS T3 UPLC ™ column (2.1 × 50 mm; Waters, Elstree, UK) held at 60 °C and with a flow rate of 300 µL/min. A starting condition of 1% B was used to capture the more hydrophilic peptide metabolites, and these were eluted using a linear gradient up to 30% B over 16 minutes. The column was washed for 2 minutes at 90% B and returned to starting conditions for 2 minutes, totalling a run time of 20 minutes. Mass spectrometry was performed using positive electrospray mode with a needle voltage of 3 kV, gas settings of 55 and 10 for sheath gas and aux gas flow rates. The temperature of the gas was set at 350 °C and the transfer capillary at 350 °C and a s-lens value of 70 V. Full scan data were acquired over an m/z range of 250-1000, using a resolution of 70,000 and a maximum fill time www.nature.com/scientificreports www.nature.com/scientificreports/ of 100 ms. Acquired LC-MS data were interrogated for potential [Pyr 1 ]apelin-13 metabolites by searching for all potential cleavage products from the parent peptide in the RAW data files using the Qualbrowser software package (Thermofisher). The m/z values for these peptides at multiple charge states are displayed in Supplementary  Table 1. The potential [Pyr 1 ]apelin-13 metabolites that were manually identified were subsequently characterised, where 30 μl of sample was reinjected using a targeted MS/MS analysis. The potential [Pyr 1 ]apelin-13 (1)(2)(3)(4)(5)(6) and [Pyr 1 ]apelin-13 (1)(2)(3)(4)(5)(6)(7)(8)(9)(10) peptides were targeted using precursor ion m/z values of 370.214 (collision energy of 30) and 290.417 (collision energy of 22) respectively. The MS/MS analysis involved the same LC separation, but MS/ MS data were acquired at 17,500 resolution with an AGC of 1e6 ions, lowest m/z value of 100 and a max fill time of 200 ms.

Data availability
All data generated or analysed during this study are included in this published article.