Introduction

Idiopathic pulmonary fibrosis (IPF) is a pulmonary disease which leads to scarring, or fibrosis, of the lungs (1). There is no known cure to the disease at present. Currently, two medications, including pirfenidone (Esbriet®) and nintedanib (Ofev®), have been approved by the FDA to help slow the progression of IPF (1, 2). Serum amyloid P component (SAP), an endogenous human pentraxin-2, is expressed in hepatocytes and circulates in the blood of healthy adults (3, 4). SAP plays an important role in innate immunity and anti-inflammation (3, 4). It has emerged as a first-in-class natural modulator of fibrotic pathology with significant potential to treat a wide variety of human fibrotic diseases (5,6,7,8). Zinpentraxin alfa is a recombinant human pentraxin-2 (rhPTX-2), previously known as PRM-151, which has been evaluated for treatment of patients with IPF in various clinical trials (9,10,11,12,13).

Zinpentraxin alfa shares the same primary amino acid sequence with SAP. The molecules differ, however, due to the structures of the glycan at Asn32. While both proteins are mixtures with asialo-, monosialo-, and bisialo-glycans, a key difference is the type of linkage of sialic acid to galactose at the terminal end of the glycan. The zinpentraxin alfa protein expressed from the Chinese hamster ovary (CHO) cells consists of monosialo- and bisialo-variants, containing exclusively α2,3-linked terminal sialic acid residues (14). In contrast, SAP isolated from human serum contains only α2,6-linked sialic acid residues (15, 16). Using in vitro cell-based bioassays, zinpentraxin alfa has shown consistently higher potency and bioactivity than serum-derived SAP (16). In various clinical trials, use of zinpentraxin alfa was associated with a reduction in fibrocytes in IPF patients when administered locally or systemically to supplement SAP (9,10,11,12).

An immunoaffinity liquid chromatography mass spectrometry (LC–MS) method and an enzyme-linked immunosorbent assay (ELISA) were reported for quantification of SAP and zinpentraxin alfa. Immunoaffinity LC–MS based on a selected zinpentraxin alfa signature peptide was reported to measure circulating zinpentraxin alfa in rat and monkey plasma samples, which did not contain endogenous human SAP (17). The surrogate peptide selected for quantification is shared by both zinpentraxin alfa and SAP (17). Therefore, this non-specific LC–MS approach is not applicable to measure zinpentraxin alfa in human samples. ELISA was also reported to quantify zinpentraxin alfa in nonclinical and clinical studies (1, 2, 9, 10, 18, 19). It is worth noting that the ELISA reported was incapable of distinguishing zinpentraxin alfa from SAP because neither the capture reagent nor the detection reagent was sufficiently selective. In these studies, the pre-dose baseline SAP concentration was subtracted from the corresponding total hPTX-2 concentration at each time point post-dose to calculate the zinpentraxin alfa concentration for each patient. This was done under the assumption that SAP levels remained unchanged throughout the treatment course (9, 10). In fact, it is not yet clear whether there is a modulation of SAP over time following the zinpentraxin alfa administration. Therefore, we attempted to develop a novel approach to selectively measure the concentrations of zinpentraxin alfa, without the interference from endogenous SAP, to better assess drug-specific exposure.

Chemical derivatization has been applied to differentiate and quantify α2,3- and α2,6-linked glycopeptides (20,21,22,23). Chemical derivatization with a subsequent detection by matrix-assisted laser desorption ionization-MS (MALDI-MS) was applied to differentiate the derivatized α2,3-sialylated glycopeptides from the α2,6-sialylated glycopeptides (20, 21). However, due to the lack of a pre-separation step before MS detection, the signal of the target glycopeptides could be suppressed and/or interfered with by other peptides resulting in low sensitivity (20, 21). Chemical derivatization with a subsequent detection by a microchip capillary electrophoresis MS (CE-MS) was developed to specifically quantify the glycopeptides with α2,3-sialylated N-glycan (22). This method selectively labeled the α2,3- and α2,6-linked sialic acid residues by different chemicals, and the corresponding derivatized glycopeptides were detected by microchip CE-MS. Chemical derivatization followed by LC–MS was attempted for relative quantification of ratios between sialylated N-glycan isomers with α2,3 or α2,6 linkage(s) (23). While achieving high specificity, these CE-MS and LC–MS methods can be either time-consuming or provide relative quantification only (22, 23).

Here, we report the development and validation of an assay, which is rapid, sensitive, and enables the specific quantification of zinpentraxin alfa in human plasma. Both therapeutic zinpentraxin alfa and endogenous SAP are first enriched using immunoaffinity capture, and the assay selectivity can be achieved by the specific α2,3-sialidase hydrolysis of zinpentraxin alfa only. An on-bead tryptic digestion is subsequently performed, followed by the detection of the unique surrogate glycopeptide derived from the therapeutic. To the best of our knowledge, this is the first assay reported to effectively differentiate zinpentraxin alfa from its endogenous isomeric SAP in human plasma. This method has been successfully implemented in clinical trials evaluating zinpentraxin alfa in IPF patients.

Methods

Reagents and Chemicals

Dynabeads M-280 streptavidin was ordered from Life Technologies (Carlsbad, CA). Both New Zealand white rabbit plasma (K2 EDTA pooled gender) and human plasma (K2 EDTA pooled gender) were purchased from BioIVT Inc. (Westbury, NY). The reference material, zinpentraxin alfa, was manufactured by Roche (Penzberg, Germany). Biotinylated rabbit anti-zinpentraxin alfa monoclonal antibody (biotin-mAb) and biotinylated rabbit anti-zinpentraxin alfa polyclonal antibody (biotin-pAb) were prepared internally at Genentech. The clones of anti-zinpentraxin alfa monoclonal antibodies were developed by a procedure described previously (24). Stable isotope labeled peptides, SIL ESVT and SIL IVLG, were ordered from Elim Biopharmaceuticals, Inc. (Hayward, CA). The amino acid sequence of SIL ESVT is ESVTDHVNL*(13C6,15N1)ITPL*(13C6,15N1)EKPLQNFTLCFR. The SIL IVLG sequence is IVL*(13C6,15N1)GQEQDSYGGK. Solutions of SIL ESVT at 600 ng/mL and SIL IVLG at 300 ng/mL, respectively, were prepared from the 1.5 mg/mL stocks using 20% acetonitrile to serve as internal standards (IS). The SIL peptides were stored at − 70°C or colder until use.

Preparation of Stock, Calibration, and Quality Control Solutions

The calibration standards for the zinpentraxin alfa assay were prepared in pooled human plasma by spiking zinpentraxin alfa at the nominal concentrations of 5.00, 10.0, 20.0, 50.0, 120, 300, 400, and 500 μg/mL using low-protein binding tubes (Eppendorf, Hamburg, Germany). Quality control samples were prepared in pooled human plasma at the nominal concentrations of 5.00, 15.0, 200, and 375 μg/mL. For the total hPTX-2 assay, calibration standards were prepared in pooled rabbit plasma by spiking zinpentraxin alfa at the nominal concentrations of 5.00, 10.0, 20.0, 50.0, 120, 300, 400, and 500 μg/mL. Quality control samples were prepared in pooled rabbit plasma at the nominal concentrations of 5.00, 15.0, 200, and 375 μg/mL.

Immunoaffinity Capture and Specific Desialylation

An LC–MS sample preparation includes capture reagent immobilization, incubation, sialidase hydrolysis, and trypsin digestion. Briefly, the streptavidin magnetic beads were incubated with the biotinylated anti-zinpentraxin alfa antibody for 30 min. Subsequently, the beads were transferred into the sample vials and incubated for 1.5 h. The captured zinpentraxin alfa protein was then subjected to α2,3-sialidase hydrolysis for 2 h at 37°C. Finally, the proteins were digested on beads by trypsin overnight in ammonium bicarbonate buffer (pH 7.8) before injecting onto LC–MS for analysis. The detailed procedure was described in the supplemental materials.

Quantitative LC–MS

A SCIEX Triple Quad 6500 mass spectrometer (AB Sciex, Concord, ON) interfaced with a Shimadzu (Tokyo, Japan) Nexera HPLC system was utilized for the quantitative LC–MS analysis. The analytical column was Kinetex® XB-C18, 100 Å, 2.6 µm, 2.1 mm × 50 mm (Phenomenex, Torrance, CA). The mobile phase A was 0.1% formic acid in water, and the mobile phase B was 0.1% formic acid in acetonitrile. A 10-min LC gradient was applied, starting with 5% mobile phase B at 0–1.0 min, ramping up to 32% B within 7 min, then increasing to 95% B within 0.01 min and remaining at 95% B for 1 min, and returning back to 5% B within 0.01 min and maintaining at 5% B for 1.0 min. The flow rate was at 0.300 mL/min. The injection volume was 10 µL, and the column temperature was set at 65 ˚C. The target peptides were detected by the MS instrument equipped with an electrospray source and operated in the positive multiple reaction monitoring (MRM) mode. As shown in Table S1, for the specific zinpentraxin alfa drug assay, an asialo-glycopeptide, ESVTDHVNLITPLEKPLQNFTLCFR (ESVT-0SA), was monitored as the surrogate for quantification. A stable isotope-labeled analog peptide, SIL ESVT, was employed as the internal standard due to the unavailability of the corresponding SIL asialo-glycopeptide. For the total hPTX-2 assay, a surrogate peptide, IVLGQEQDSYGGK (IVLG), shared by both zinpentraxin alfa and SAP, was quantified with the corresponding stable isotope-labeled internal standard, SIL IVLG. Calibration curves were fit using a 1/x2 weighted, linear regression algorithm.

Clinical Plasma Sample Analysis

Samples were kept at − 70°C or colder until analysis. The plasma samples were quantified for zinpentraxin alfa and total hPTX-2 concentrations using the two methods developed. The SAP levels in pre-dose samples were equivalent to the measured total hPTX-2 concentrations. The SAP levels in post-dose samples were calculated by subtracting the measured zinpentraxin alfa concentration from the matching measured total hPTX-2 concentration.

Results

General Strategy for Selective Quantification of Zinpentraxin Alfa and Total hPTX-2

Zinpentraxin alfa and SAP share the exact same primary amino acid sequence (Fig. 1a). To accurately measure zinpentraxin alfa in human plasma while excluding or minimizing interference from SAP, we targeted the N-glycan structures. By employing a specific α2,3-sialidase, the α2,3-linked terminal sialic acid residues on zinpentraxin alfa should be selectively hydrolyzed and removed (Fig. 1b) while the α2,6-linked terminal sialic acid residues of SAP would remain unchanged (25). Subsequent proteolytic digestion with trypsin will release a single unique surrogate asialo-glycopeptide from zinpentraxin alfa, which can be effectively distinguished from the corresponding sialylated glycopeptides derived from SAP by the mass of sialic acid residue, for selective quantification of zinpentraxin alfa. In addition, a non-glycosylated peptide, shared by both zinpentraxin alfa and SAP, was selected as the surrogate peptide for measuring the concentrations of total human pentraxin-2 (i.e., zinpentraxin alfa + SAP). By subtracting the zinpentraxin alfa concentration from the total hPTX-2 concentration, the level of preexisting SAP was calculated and tracked over the treatment course. To enhance the assay sensitivity, immunoaffinity capture of zinpentraxin alfa, together with SAP, in human plasma samples was achieved by using a biotinylated rabbit anti-zinpentraxin alfa monoclonal antibody immobilized onto streptavidin-coated magnetic beads (Fig. 1b).

Fig. 1
figure 1

Strategy for simultaneous quantification of zinpentraxin alfa and total hPTX-2 (i.e., zinpentraxin alfa + SAP) by immunoaffinity capture LC–MS/MS. a Schematic of zinpentraxin alfa and SAP containing one N-linked biantennary glycan with two α2,3- and α2,6-linked terminal sialic acid residues, respectively. b The workflow depicting multiple sample preparation steps and quantification by LC–MS/MS via selected glycopeptide (for zinpentraxin alfa) and peptide (for zinpentraxin alfa + SAP) as surrogates, respectively

Full size image

Method Development and Optimization

The specificity of α2,3-sialidase to hydrolyze α2,3-sialylated linkage was demonstrated by comparing intact proteins of zinpentraxin alfa and SAP before and after the enzyme treatment. Three main species were observed for zinpentraxin alfa protein, representing asialo-, monosialo-, and bisialo-proteoforms (Fig. 2a). After the α2,3-sialidase treatment, only one single species was observed, indicating that all α2,3-linked sialic acid residues of monosialo- and bisialo-species were selectively hydrolyzed and removed, and both sialylated components were transformed to asialo-zinpentraxin alfa (Fig. 2b). The analysis of intact SAP protein revealed a predominant bisialo-species with a smaller abundance of monosialo- and a residual amount of asialo-SAP (Fig. 2c). The overall intact SAP profile remained unchanged after treatment with α2,3-sialidase, showing no cleavages of terminal sialic acid residues (Fig. 2d).

Fig. 2
figure 2

α2,3-sialidase specifically cleaves α2,3-linked terminal sialic acid residues in zinpentraxin alfa, but not the α2,6-linked ones in SAP. Deconvoluted spectra of a zinpentraxin alfa, b zinpentraxin alfa after 30 min of α2,3-sialidase treatment, c endogenous SAP, and d SAP after 17 h of α2,3-sialidase treatment

Full size image

Three main glycopeptides generated from tryptic digestion were detectable by LC–MS. Conventional tryptic digestion of zinpentraxin alfa produced asialo-, monosialo-, and bisialo-glycopeptides labeled as ESVT-0SA (m/z 1149, 4 +), ESVT-1SA (m/z 1222, 4 +), and ESVT-2SA (m/z 1295, 4 +), respectively (Fig. 3a). Following the α2,3-sialidase treatment and subsequent tryptic digestion, all glycopeptides were converted to a single form of asialo-glycopeptide (ESVT-0SA, m/z 1149), indicating an efficient hydrolysis of sialylated zinpentraxin alfa as intended (Fig. 3b). The corresponding product ion scans of these three glycopeptides were performed by the collision-induced dissociation (Fig. 4). The precursor ions at m/z 1149, m/z 1222, and m/z 1295 were selected for fragmentation, and the most abundant daughter ions identified were m/z 1410 (3 +), m/z 1410 (3 +), and m/z 1507 (3 +), respectively (Fig. 4). We were able to monitor the ion transitions of 1149/1410, 1222/1410, and 1295/1507, for asialo-, monosialo-, and bisialo-glycopeptides, respectively.

Fig. 3
figure 3

Detection of glycopeptides derived from zinpentraxin alfa following the treatment of α2,3-sialidase and/or trypsin. Full survey scan spectra of a glycopeptides generated by trypsin digestion and b glycopeptides generated by α2,3-sialidase hydrolysis and then trypsin digestion

Full size image
Fig. 4
figure 4

Product ion scans of a asialo-glycopeptide ESVT-0SA at m/z 1149, b monosialo-glycopeptide ESVT-1SA at m/z 1222, and c bisialo-glycopeptide ESVT-2SA at m/z 1295

Full size image

A characteristic glycopeptide and a common tryptic peptide were selected as surrogates for quantification of zinpentraxin alfa and total hPTX-2, respectively. The signature asialo-glycopeptide, ESVT-0SA (m/z 1149), was selected as the surrogate peptide for quantification of zinpentraxin alfa in human plasma samples. Ideally, a corresponding stable isotope labeled glycopeptide, SIL ESVT-0SA, should be adopted as the internal standard for the pharmacokinetic (PK) assay. However, the synthesis of a custom-made glycopeptide labeled at Leu9 and Leu13 failed due to unacceptable levels of impurities. Instead, we settled with an analog SIL peptide internal standard, SIL ESVT, which does not contain the glycan group at N19 (Figure S1a). The ESVT-0SA was detected at MRM 1149/1410 with a retention time of 7.11 min (Figure S1b) while the SIL ESVT was detected at MRM 747/885 with a retention time of 7.53 min (Figure S1c) using a 10-min LC gradient. In addition, a common tryptic peptide, IVLG, shared between zinpentraxin alfa and SAP, was selected as the surrogate for quantifying total hPTX-2. The product ion scan of IVLG and its MRM (697/1181) chromatogram are shown in Figure S2.

Initially, a biotinylated rabbit anti-zinpentraxin alfa polyclonal antibody was utilized as the immunoaffinity capture reagent bound to streptavidin magnetic beads. Coupling 4.0 μg antibodies to 0.25 mg beads afforded a working capture capacity of up to 800 μg/mL zinpentraxin alfa (Figure S3). Two different clones of anti-zinpentraxin alfa monoclonal antibody were later developed and tested, with no significant differences observed among the three capture reagents (Figure S4). Monoclonal antibody, clone #326, was selected for all subsequent experiments because of its robust performance and long-term availability. In addition, to develop a method that accurately reports the total concentration of endogenous and recombinant hPTX-2, an unbiased capture of both zinpentraxin alfa and SAP needs to be established. Quality controls prepared with zinpentraxin alfa or SAP at 15.0, 200, and 375 μg/mL in rabbit plasma, which was used as a surrogate matrix to human plasma due to its lack of endogenous human SAP, were tested for the capture recovery, respectively (Figure S5a). The capture recovery of zinpentraxin alfa and SAP appeared to be similar (Figure S5b). In addition, similar capture efficiency was demonstrated among the asialo-, monosialo-, and bisialo-proteoforms for both zinpentraxin alfa and SAP (Figure S5c).

The incubation step of zinpentraxin alfa with α2,3-sialidase was optimized to ensure the efficiency of hydrolysis. Twelve units of α2,3-sialidase were incubated with the immunoaffinity enriched zinpentraxin alfa out of human plasma for up to 4 h. Until 1.5 h of incubation, an increase of ESVT-0SA was observed corresponding to the decreases in ESVT-1SA and ESVT-2SA, respectively (Fig. 5a). The conversion to ESVT-0SA reached a plateau at and beyond 1.5-h incubation (Fig. 5b). The amount of α2,3-sialidase was also optimized, which showed that 12 units of the enzyme were sufficient to achieve the hydrolysis of enriched zinpentraxin alfa at 500 μg/mL in human plasma (Fig. 5c, Figure S6). Therefore, 12 units of α2,3-sialidase and 1.5-h incubation time were adopted for subsequent experiments. A 12-min LC gradient was initially applied with a retention time of around 8.42 min for ESVT-0SA during method development (Fig. 5a, Figure S6a). The potential crosstalk among ESVT-2SA, ESVT-1SA, and ESVT-0SA was verified, in which ESVT-2A and ESVT-1SA, mainly presented in SAP, could produce signals detectable in the ESVT-0SA channel due to collision-induced dissociation (CID). However, they were baseline separated at different retention times, i.e., 9.20 min and 8.79 min, respectively. Therefore, no interference to the quantification of zinpentraxin alfa was expected (Figure S6).

Fig. 5
figure 5

Optimization of the incubation time and enzyme amount for conversion of sialylated zinpentraxin alfa variants to asialo-zinpentraxin alfa by α2,3-sialidase. a Detection of the asialo-glycopeptide ESVT-0SA, monosialo-glycopeptide ESVT-1SA, and bisialo-glycopeptide ESVT-2SA before and after incubation, b relative abundance of zinpentraxin alfa variants vs. incubation time after immunoaffinity capture, c relative abundance of zinpentraxin alfa variants vs. amount of α2,3-sialidase following immunoaffinity capture. The samples were 500 μg/mL zinpentraxin alfa spiked into pooled human plasma

Full size image

Assay Validation for Zinpentraxin Alfa

A pre-existing trace amount of endogenous asialo-SAP was detected in blank human plasma (Fig. 2c). Since both calibration standards and unknown samples were in human plasma, potential impact by the matrix background was minimized, and an LLOQ chromatogram for the surrogate peptide, ESVT-0SA, with a sufficient S/N ratio was demonstrated (Fig. 6a). An analog internal standard, SIL ESVT, had to be utilized for quantification because of the lack of corresponding SIL ESVT-0SA (Fig. 6b). A good linearity (r2 = 0.9938) was observed (Fig. 6c). It is worth noting that a shorter 10-min LC gradient was adopted in validation, resulting in a retention time of 7.11 min for the analyte, ESVT-0SA, different from what was observed during method development (i.e., 8.42 min). In addition, a different retention time of 7.53 min was reported for the analog internal standard, SIL ESVT, due to its stronger hydrophobicity than that of the analyte, ESVT-0SA.

Fig. 6
figure 6

Quantification of zinpentraxin alfa. a Chromatograms of the surrogate glycopeptide, ESVT-0SA, in blank human plasma and zipentraxin alfa spiked in human plasma at LLOQ; b chromatogram of the surrogate peptide, SIL ESVT, of an analog internal standard only; c standard calibration curve of zinpentraxin alfa in human plasma

Full size image

The zinpentraxin alfa assay was validated using a sample volume of 10.0 μL. The dynamic range was set at 5.00–500 μg/mL zinpentraxin alfa in human plasma. Performance of representative standards across three batch runs during validation is summarized in Table S2. Accuracy and precision data obtained from individual runs and overall interday run summary are presented (Table S3). Samples above the ULOQ can be diluted up to tenfold using blank human plasma. Other parameters such as selectivity, recovery, matrix interference, hemolysis effect, hyperlipidemic effect, and injection carryover were determined to be within the acceptance criteria (data not shown). Zinpentraxin alfa was found to be stable in human plasma for at least five freeze/thaw cycles at both − 20°C and − 70°C, respectively, and on the bench-top for at least 6 h at room temperature (data not shown). Long-term storage stability of zinpentraxin alfa in human plasma at − 70°C has been established for at least 6 months so far. The processed samples remained stable for at least 189 h at 4°C, and the reinjection reproducibility was established for up to 163 h at 4°C. Whole blood samples were found to be stable in ice-water bath and at room temperature for 1 h during sample collection. An ISR evaluation was performed using an appropriate number of incurred study samples. The reanalysis results demonstrated that 90.9% of incurred samples tested met the acceptance criteria, confirming the robustness of the zinpentraxin alfa method.

Assay Validation for Total hPTX-2

To accurately quantify the total hPTX-2 concentration, which includes both the drug and endogenous SAP in human plasma, calibration curves were prepared in a surrogate matrix, rabbit plasma. Overlapped representative chromatograms of an LLOQ and a blank rabbit plasma sample were shown with a sufficient S/N ratio (Fig. 7a). A significant signal was observed for the pre-existence of endogenous SAP in blank human plasma (Fig. 7b). The corresponding internal standard, SIL IVLG, was utilized for quantification (Fig. 7c). A linear calibration curve (r2 = 0.9974) was observed for total hPTX-2 with a dynamic range of 5.00–500 μg/mL (Fig. 7d).

Fig. 7
figure 7

Quantification of total hPTX-2. a Chromatograms of the surrogate peptide, IVLG, in blank rabbit plasma and zipentraxin alfa spiked in rabbit plasma at LLOQ. b Detection of the surrogate peptide, IVLG, in blank human plasma. c Chromatogram of the surrogate peptide, SIL IVLG, of an internal standard only. d Standard calibration curve of total hPTX-2 in rabbit plasma

Full size image

Assay validation was successfully performed using 10.0-μL rabbit plasma spiked with zinpentraxin alfa at various concentrations. Performance of representative standards across three batch runs during validation is summarized in Table S4. Accuracy and precision data for individual runs and interday run summary are presented in Table S5. Other parameters such as selectivity, recovery, matrix interference, hemolysis effect, hyperlipidemic effect, and injection carryover were determined to be within the acceptance criteria. Stability of zinpentraxin alfa was established in rabbit plasma for at least five freeze/thaw cycles at both − 20°C and − 70°C, respectively, and on the bench-top for at least 6 h at room temperature (data not shown). Long-term storage stability of zinpentraxin alfa in rabbit plasma at − 70°C has been established for at least 6 months by now. The processed samples were found to be stable for at least 177.5 h at 4°C, and the reinjection reproducibility was confirmed for 149.5 h at 4°C. In addition, parallelism experiment confirmed a good linearity for serial dilutions from high concentrations of zinpentraxin alfa using surrogate rabbit plasma matrix (r2 = 0.9998, Figure S7).

Bioanalysis of Clinical Samples

We analyzed the plasma samples collected from a cohort of 14 patients in an open label extension study that was designed to evaluate the long-term safety and efficacy of zinpentraxin alfa. The participants with idiopathic pulmonary fibrosis were dosed intravenously over 60 min with 10.0 mg/kg zinpentraxin alfa at days 1, 3, and 5; week 4; and then every 4 weeks up to 24 weeks (totally nine dose cycles). On day-1 or dose-1, blood collections were taken pre-dose and at 1, 2, 4, 8, 12, and 24 h post-dose. On day-3, 27, 83, and 167 (dose-2, 4, 6, and 9), blood collections were taken pre-dose and at 2 h post-dose. Pharmacokinetic profiles of total hPTX-2, zinpentraxin alfa, and endogenous SAP were analyzed. As expected, the levels of total hPTX-2 and zinpentraxin alfa increased immediately after dosing and then slowly decreased within 72 h, while the SAP levels remained relatively constant (Fig. 8a). In addition, the concentrations of zinpentraxin alfa were quantified in 2-h post-dose samples of dose-1, 2, 4, 6, and 9. Among these doses, the levels of zinpentraxin alfa were comparable (Fig. 8b). This was also the case for the total hPTX-2 level in these 2-h post-dose samples (Fig. 8c). By subtracting the zinpentraxin alfa from the total hPTX-2 concentrations, we calculated the endogenous SAP levels for the 2-h post-dose samples of dose-1, 2, 4, 6, and 9 (Fig. 8d, slash pattern bar). It is worth noting that predose samples were also collected prior to each dose given at the aforementioned time points. The total hPTX-2 values obtained from the predose samples prior to each dose should represent the initial concentrations of endogenous SAP in these patients. The levels of total hPTX-2 measured in the predose samples are shown (Fig. 8d, dark bar). As indicated, the SAP levels measured in the predose samples and the values calculated for the 2 h post-dose samples appeared to have no significant differences (Fig. 8d).

Fig. 8
figure 8

Quantification of zinpentraxin alfa and total hPTX-2 in representative human plasma samples from a clinical study following repeat intravenous (IV) administration of zinpentraxin alfa at 10.0 mg/kg. a Zinpentraxin alfa and total hPTX-2 levels measured at pre-dose and various time points post-dose, b Zinpentraxin alfa levels measured at 2-h post-dose within different dose cycles. c Total hPTX-2 levels measured at 2-h post-dose within different dose cycles. d Comparison of measured total hPTX-2 levels at pre-dose and calculated SAP levels at 2-h post-dose from the corresponding samples

Full size image

Discussion

Distinguishing zinpentraxin alfa from the preexisting SAP in circulation is key to accurate exposure assessment of the therapeutic protein during the IPF treatment in clinical studies. Previously reported ELISA and LC–MS methods were applied to quantification of zinpentraxin alfa in nonclinical studies where no interference from human endogenous SAP was expected (17, 18). However, neither assay was amenable to differentiate zinpentraxin alfa from SAP in clinical studies, i.e., the LC–MS was based on a common surrogate tryptic peptide shared by both isomers (17) and the ELISA utilized reagents lacking selective structural recognition capability (18), respectively. The method reported here selectively converted the α2,3-linked sialylated zinpentraxin alfa to a single form of asialo-zinpentraxin alfa using α2,3-sialidase, which allowed the accurate quantification of zinpentraxin alfa in the presence of endogenous SAP. It is worth noting that SAP is a mixture containing monosialo-SAP, disialo-SAP, as well as a residual amount of asialo-SAP (Fig. 2c). The monosialo- and disialo-SAP with original α2,6-linkage should not interfere with the detection of targeted asialo-zinpentraxin alfa because of different masses. However, the residual amount of asialo-SAP has the same molecular weight as that of asialo-zinpentraxin alfa. Therefore, the calibration curve for the zinpentraxin alfa assay needs to be prepared in human plasma so that any potential interference from endogenous asialo-SAP will be compensated for the study samples.

The 25-mer asialo-glycopeptide, ESVTDHVNLITPLEKPLQNFTLCFR (containing a carbohydrate moiety, GlcNAc4Man3Gal2, at N following Q), derived from zinpentraxin alfa was selected as the surrogate for quantification, affording the required sensitivity and assay robustness. Initially, we attempted to identify a shorter surrogate glycopeptide to achieve greater sensitivity. A combination of trypsin and Lys-C failed to generate a shorter glycopeptide, PLQNFTLCFR. Another option of using trypsin and Glu-C to obtain a smaller glycopeptide, KPLQNFTLCFR, was also unsuccessful. It was reported that the cleavage efficiency of Lys-C at K\P would diminish due to the immediately following proline (26). In addition, the proximal side chain modifications, e.g., N-linked glycan, might also impact the cleavage efficiency of Lys-C at K\P and Glu-C at E\K.

Trypsin digestion was verified to make sure no unintended disruptions occurred to the linkage with terminal sialic acid residues. Zinpentraxin alfa treated by trypsin alone did not result in any cleavages of sialic acid residues over different incubation periods, represented by Fig. 3a. Only when treated with both α2,3-sialidase and trypsin sequentially, zinpentraxin alfa showed the loss of sialic acid residues and subsequent detection of asialo-glycopeptide exclusively (Fig. 3b). These results indicate that the trypsin, an enzyme that specifically cleaves the peptide bonds at the C-terminal side of lysine and arginine residues, did not hydrolyze the glycosidic bond that links the terminal sialic acid residue in the glycan structure.

We utilized an indirect method to determine the SAP levels in clinical samples. It was not possible to directly measure the endogenous SAP levels in clinical plasma samples using its characteristic α2,6-sialylated bisialo-glycopeptide as the surrogate. The α2,6-linked sialic acid residues in sialylated SAP species would not be hydrolyzed by α2,3-sialidase, and subsequently, the α2,6-linked bisialo-glycopeptide can be released in tryptic digestion. However, zinpentraxin alfa in plasma samples would be in a very high level after dosing (approximately 300 μg/mL). The SAP level in general subjects prior to dosing was around 30–50 μg/mL (3, 4). A small portion of remaining zinpentraxin alfa unhydrolyzed by α2,3-sialidase could result in a significant amount of α2,3-linked bisialo-glycopeptide, which would interfere with the chromatographic peak of α2,6-linked bisialo-glycopeptide from SAP. It is challenging to differentiate α2,3-linked and α2,6-linked bisialo-glycopeptides by reversed-phase LC–MS.

In previous clinical studies, due to unavailability of specific quantitative assays of zinpentraxin alfa, the drug levels were obtained indirectly by deducting the total hPTX-2 concentration at predose, which was equivalent to the preexisting SAP concentration, from the total hPTX-2 concentration at each time point (9, 10). The assumption was that the endogenous SAP would remain unchanged quantitatively throughout the course of the treatment (9, 10). There were no actual data available to confirm whether factors such as the dose level, dosing frequency, or length of the study might lead to any significant modulation of SAP over time. It was of interest to monitor the concentrations of SAP in addition to zinpentraxin alfa. In our current studies, SAP was determined indirectly by subtracting the measured zinpentraxin alfa concentration from the measured total hPTX-2 concentration. Our preliminary data suggest that endogenous SAP indeed remained largely unchanged in IPF patients throughout the treatment by zinpentraxin alfa (Fig. 8).

Our bioanalytical strategy for zinpentraxin alfa should have a broad application on quantification of general α2,3-sialylated glycoproteins while avoiding interference from the coexisting α2,6-linkage isomers. The sialylation linkage of glycoprotein isomers can play different cellular and physiological roles. For example, α2,3-sialylated zinpentraxin alfa was shown to have a higher potency and bioactivity than its α2,6-sialylated isomer, SAP (16). The avian influenza viruses mostly cause infection by binding α2,3-linked sialylation in the avian respiratory track, while the human tract is mostly enriched with α2,6-linked sialylation (27, 28). The α2,6-sialylated immunoglobulin G was reported to improve the anti-inflammatory effect over α2,3-sialylated isomer in treating a mouse arthritis model (29). With the increasing discovery of distinctive functions of the sialylated linkage isomers of glycoproteins, it becomes more important than ever to have specific quantification approaches. Multiple mass spectrometric methods have been developed to tackle this challenge (20,21,22,23). Our method provides an alternative to specifically determine the level of α2,3-sialylated glycoproteins without the interference from their sialylated linkage isomers.

Conclusion

The immunoaffinity LC–MS method reported here is believed to be the first pharmacokinetic assay to allow specific quantification of zinpentraxin alfa, a therapeutic recombinant hPTX-2 protein, in human plasma while avoiding potential interference from the native form of hPTX-2, i.e., SAP, which shares an identical amino acid sequence with the drug. This was enabled by incorporating an α2,3-sialidase treatment step to selectively hydrolyze the drug’s α2,3-linked terminal sialic acid residues without disrupting the α2,6-linked glycan structure unique to endogenous SAP. A signature asialo-glycopeptide derived only from zinpentraxin alfa facilitated its specific quantification in human plasma. Meanwhile, another common tryptic peptide shared by both zinpentraxin alfa and endogenous SAP served as a surrogate to quantify total hPTX-2. Both assays were successfully validated and implemented in the pharmacokinetic analysis of plasma samples collected from patients undergoing treatment of idiopathic pulmonary fibrosis. Preliminary data of total hPTX-2 and zinpentraxin alfa appeared to track closely with each other in parallel, indicating no significant changes to the endogenous SAP level during treatment. This bioanalytical approach should be applicable to the quantification of any glycoproteins with α2,3-linked sialic acid residues while avoiding potential interference from α2,6-linkage isomers.