Comparison of Analytical Methods for AntibodyâDrug

Article Cite This: Anal. Chem. XXXX, XXX, XXX−XXX

pubs.acs.org/ac

Comparison of Analytical Methods for Antibody−Drug Conjugates Produced by Chemical Site-Specific Conjugation: First-Generation AJICAP Yutaka Matsuda,† Veronica Robles,† Maria-Christina Malinao,‡ James Song,‡ and Brian A. Mendelsohn*,† †

Ajinomoto Bio-Pharma Services, 11040 Roselle Street, San Diego, California 92121, United States Phenomenex, Inc., 411 Madrid Avenue, Torrance, California 90501, United States

Downloaded via UNIV OF GLASGOW on September 3, 2019 at 19:32:06 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

‡

S Supporting Information *

ABSTRACT: Antibody−drug conjugates (ADCs) have become a major class of oncology biopharmaceuticals. Traditional ADCs have a stochastic distribution of cytotoxic drugs attached at several different sites on the antibody. The heterogeneous nature of stochastic ADCs results in a complex compositional analysis. To improve on traditional ADC technology, we have developed a chemical conjugation platform termed “AJICAP” for the site-specific modification of native antibodies using a class of IgG Fc affinity reagents. Here we report further investigation focusing on several analyses of a first-generation AJICAP-ADC (Angew. Chem., Int. Ed. 2019, 58, 5592−5597). For drug−antibody ratio (DAR) determination, we examined and compared six different analytical methods. To the best of our knowledge, this is the first report of a comparison of analytical techniques to measure the DAR for ADCs produced by a site-specific technology such as AJICAP. Furthermore, a rapid analytical process for confirmation of the site selectivity of AJICAP conjugation was established by SEC−Q-TOF-MS. The analytical strategy reported here can be applied to the DAR determination of site-specific ADCs.

A

This broad distribution of ADC compounds provides a challenge for analytical chemists to develop improved strategies14 for application to ADC analysis, including drug− antibody ratio (DAR) characterization and ADC stability. For DAR characterization, HPLC separation is commonly used because of the hydrophobicity of the payloads.15−19 Hydrophobic interaction chromatography (HIC) is the standard method for DAR determination because it can be performed on intact ADCs under nondenaturing conditions. However, HIC separation typically requires high initial salt concentrations and the low volatility of the salts in the mobile phases is therefore incompatible with MS analysis to identify different DAR species.19

ntibody−drug conjugates (ADCs) have become a major class of cancer biopharmaceuticals and remain a focus of interest for our research group.1,2 Four ADCs are currently on the market: Adcetris (brentuximab vedotin) from Seattle Genetics, Kadcyla (trastuzumab emtansine) from Genentech,3−5 and Mylotarg (gemtuzumab ozogamicin) and Besponsa (inotuzumab ozogamicin) from Pfizer. Presently, more than 60 clinical trials of ADCs are in progress.6,7 An ADC consists of a recombinant monoclonal antibody, which provides target specificity, covalently linked to cytotoxic payloads through chemical linkers. Various methods have been developed to construct these covalent bonds. The traditional synthetic approach is based on nonspecific drug conjugation using naturally present amino acid residues, such as lysines8 and reduced interchain cysteines.9,10 These ADCs have a heterogeneous distribution of cytotoxic drug molecules over many different sites on the antibodies,11,12 resulting in diminished efficacy and a lower therapeutic index.13 © XXXX American Chemical Society

Received: May 9, 2019 Accepted: August 15, 2019

A

DOI: 10.1021/acs.analchem.9b02192 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry Scheme 1. AJICAP Technology Overview

Scheme 2. Synthesis of Trastuzumab−AJICAP−MMAE (4)

ADCs still present a considerable challenge because of the complex structures and heterogeneity of ADCs. For example, the Källsten and Bergquist group have compared the performance of multiple DAR determination techniques, including MALDI-TOF-MS,30 and found that the heterogeneity of the reduced interchain ADCs created a challenge for DAR analysis. To overcome the issues present with traditional stochastic ADCs, several site-specific conjugation methods have been developed.31 We have developed a method that makes use of an Fc affinity peptide to produce Fc-selective conjugated ADCs.1 This platform, termed “AJICAP”, has been previously reported to enable a reliable and robust synthesis to afford sitespecific ADCs (Scheme 1); however, the DAR determination was performed by only Q-TOF-MS analysis. Here we report further investigation focusing on several analyses of an AJICAP-ADC, using easily created analytical methods. First, a comparison of several analytical methods for DAR analysis (HIC, RP-HPLC, RPLC−Q-TOF-MS, denaturing SEC−QTOF-MS, native SEC−Q-TOF-MS, and Ellman’s assay) was conducted. To the best of our knowledge, this is the first report to show an extensive comparison of DAR determination using a single batch of a site-specific ADC. In addition to characterization of the DAR, a rapid SEC−QTOF-MS analysis of the AJICAP-ADC was also performed,

Reversed-phase liquid chromatography (RPLC) is also a method commonly used for separation for DAR analysis, especially for cysteine-type ADCs.20,21 RPLC coupled to a quadrupole time-of-flight mass spectrometry (Q-TOF-MS) system is a very well known methodology for peak identification in the analysis of ADCs. RPLC−MS can theoretically be used to analyze all the DAR species of an ADC.21 Complete separation of different DAR species is not necessary for Q-TOF-MS analysis, so this method can be easily optimized. However, RPLC can be problematic because of the use of potential denaturing conditions, that is, high column temperatures and organic solvents in the mobile phases. Thus, this analytical method is not acceptable for several sensitive proteins. From a “non-denaturing” perspective, size-exclusion chromatography (SEC) can be considered an ideal approach for DAR analysis. SEC analysis is typically conducted with a neutral pH buffer at room temperature. Furthermore, SEC is compatible with MS analysis using a buffer solution with highly volatile salts. In 2012, the Valliere-Douglass group presented the first report of SEC−MS analysis for DAR characterization.22 Since this report, SEC−MS analysis has become popular for the rapid analysis of DAR distributions under mild conditions.22−29 Even though many analytical methods have been investigated, the biophysical characterization and quantitation of B


Article

Analytical Chemistry

at 25 °C using 1.1 M (NH4)2SO4, 25 mM NaH2PO4 (pH 6.0) (MPA), and 25 mM NaH2PO4 (pH 6.0) with 25% IPA (MPB), and the absorbance was monitored at 280 nm (reference wavelength at 450 nm). Each sample (1 mg/mL, 40 μL) was injected into the system and eluted over an 18 min run consisting of a 2 min isocratic hold at 0% MPB, a 6 min linear gradient from 0% to 50% MPB, a 3 min wash using 50% MPB, and a 7 min reequilibration at 0% MPB. HIC−HPLC Conditions B. HIC−HPLC analysis was performed on a MabPac HIC Butyl, 5 μm, 4.6 × 100 mm column (Thermo Scientific), connected to an Agilent 1260 HPLC system containing a binary gradient pump, temperature-controlled column compartment, autosampler, and a diode array detector. The system ran at 0.8 mL/min at 25 °C using 1.5 M (NH4)2SO4, 100 mM NaH2PO4 (pH 7.0) (MPA), and 100 mM NaH2PO4 (pH 7.0) (MPB), and the absorbance was monitored at 280 nm (reference wavelength at 450 nm). Each sample (1 mg/mL, 40 μL) was injected into the system and eluted over a 30 min run consisting of a 4 min isocratic hold at 16% MPB, a 16 min linear gradient from 16% to 100% MPB, a 4 min wash using 100% MPB, and a 11 min reequilibration at 16% MPB. RPLC−Q-TOF-MS. RPLC−Q-TOF-MS analysis was performed on a bioZen-Glycan, 2.1 × 100 mm, 2.6 μm column (Phenomenex), connected to an Agilent 1260 HPLC system with an Agilent 6550 Q-TOF system containing a binary gradient pump, temperature-controlled column compartment, autosampler, and a diode array detector. The system ran at 0.2 mL/min at 50 °C using 0.1% formic acid (FA), 0.01% TFA, 2% ACN, 98% water (MPA) and 0.1% formic acid, 0.01% TFA, 99.9% acetonitrile (MPB), and the absorbance was monitored at 280 nm (reference wavelength at 450 nm). Each sample (1 mg/mL, 5 μL) was injected into the system and eluted over a 15 min run consisting of a 5 min isocratic hold at 5% MPB, a 5 min linear gradient from 5% to 95% MPB, a 2 min wash using 95% MPB, and a 3 min re-equilibration at 5% MPB. Data evaluation was performed as previously reported.1 Denaturing SEC−Q-TOF-MS. Denaturing SEC−Q-TOFMS analysis was performed on a Biozen SEC-3, 2.1 × 50 mm, 1.8 μm column (Phenomenex), connected to an Agilent 1260 HPLC system with an Agilent 6550 Q-TOF system containing a binary gradient pump, temperature-controlled column compartment, autosampler, and a diode array detector. The system ran at 0.02 mL/min at 25 °C using 1% formic acid, 0.01% TFA, 50% acetonitrile, and 49% water, and the absorbance was monitored at 280 nm (reference wavelength at 450 nm). Each sample (1 mg/mL, 20 μL) was injected into the system and eluted over a 15 min run. Data evaluation was performed as previously reported.1 Native SEC−Q-TOF-MS. Native SEC−Q-TOF-MS analysis was performed on a Biozen SEC-3, 2.1 × 50 mm, 1.8 μm column (Phenomenex), connected to an Agilent 1260 HPLC system with an Agilent 6550 Q-TOF system containing a binary gradient pump, temperature-controlled column compartment, autosampler, and a diode array detector. The system ran at 0.02 mL/min at 25 °C using 150 mM NH4OAc (pH 6.0), and the absorbance was monitored at 280 nm (reference wavelength at 450 nm). Each sample (1 mg/mL, 20 μL) was injected into the system and eluted over a 15 min run. Data evaluation was performed as previously reported.1 Ellman’s Assay. Sample preparation and data evaluation were performed as previously reported.1

revealing that the drug linker was highly selectively linked to the antibody heavy chain. We believe that these studies not only demonstrate the reliability of the AJICAP technology to provide nextgeneration ADCs but also contribute to establishing a robust analytical methodology for site-specific ADCs in general.

■

MATERIALS AND METHODS Materials. Human IgG1 antibody trastuzumab (Herceptin) was purchased from Roche Pharmaceutical Co. Maleimide-C6valine-citrulline-monomethyl auristatin E (MC-VC-MMAE) was purchased from Abzena. The peptide reagent 1 was provided by Ajinomoto Co., Inc. All other chemicals were purchased from Sigma-Aldrich. Synthetic Procedure for Unpurified Trastuzumab− AJICAP−MMAE (4). Site-selective thiol group installation to trastuzumab proceeded to afford thiol-modified trastuzumab (3) as described in our previous paper (Scheme 2).1,2 After measuring the amount of thiol groups per antibody of compound 3 by Ellman’s assay [Table S1, Supporting Information (SI)], the thiol-modified trastuzumab 3 was conjugated with MC-VC-MMAE via the following method. To a buffered solution of thiol-modified trastuzumab 3 was added an MC-VC-MMAE solution [10 equiv in dimethylacetamide (DMA)] at 20 °C. After 2 h, N-acetylcysteine was added to the reaction mixture to quench excess drug linker. After 15 min, gel filtration was conducted to remove excess drug linker to obtain unpurified trastuzumab−AJICAP− MMAE (4). Preparative Hydrophobic Interaction Chromatography. Unpurified trastuzumab−AJICAP−MMAE (4) was purified by HIC using ToyoPearl Phenyl-650S columns attached to an Ä KTA Pure system.32 The trastuzumab− AJICAP−MMAE (4) containing two payloads was eluted using a linear gradient from 100% buffer A (50 mM sodium phosphate pH 7.0, 2 M NaCl) to 100% buffer B [50 mM sodium phosphate pH 7.0, 20% 2-propanol (IPA) v/v]. RP-HPLC of Reduced ADCs. Reduced samples were prepared as follows: 1.0 mg/mL of ADCs in 50 mM PBS and 10 mM ethylenediaminetetraacetic acid (EDTA), pH 7.4, was diluted to 0.6 mg/mL in 8 M guanidine HCl and reduced by addition of 1 M DL-dithiothreitol (DTT). The mixture was incubated at 80 °C for 5 min. RP-HPLC analysis was performed on an AdvanceBio RP-mAb Diphenyl, 2.1 × 100 mm, 3.5 μm column (Agilent), connected to an Agilent 1260 HPLC system containing a binary gradient pump, temperature-controlled column compartment, autosampler, and a diode array detector. The system ran at 0.4 mL/min at 70 °C using 0.1% trifluoroacetic acid (TFA) in water (mobile phase A, MPA) and 0.1% TFA in acetonitrile (ACN) (mobile phase B, MPB), and the absorbance was monitored at 280 nm (reference wavelength at 450 nm). Each sample (0.6 mg/mL, 20 μL) was injected into the system and eluted over a 35 min run consisting of a 2 min isocratic hold at 30% MPB, a 22 min linear gradient from 30% to 48% MPB, a 3 min wash using 95% MPB, and an 8 min reequilibration at 30% MPB. HIC−HPLC Conditions A. HIC−HPLC analysis was performed on a Tosoh Bio Butyl-NPR, 2.5 μm, 4.6 × 35 mm column (Tosoh Bioscientific), connected to an Agilent 1260 HPLC system containing a binary gradient pump, temperature-controlled column compartment, autosampler, and a diode array detector. The system ran at 0.8 mL/min C


Article

Analytical Chemistry Table 1. Drug−Antibody Ratios of Unpurified Compound 4 As Measured by Different Methods HIC−HPLC reduced RP-HPLC DAR = 0 DAR = 1 DAR = 2 DAR = 3 LC LC + MMAEa HC HC + 1 MMAEb HC + 2 MMAEc

100.0% 0.00% 29.0% 65.9% 5.15%

av DAR

1.52

SEC−Q-TOF-MS

conditions A

conditions B

RPLC−Q-TOF-MS

denaturing

native

6.15% 35.3% 51.0% 7.6%

5.67% 32.5% 59.1% 2.74%

8.07% 7.65% 84.2%

6.49% 8.82% 84.7%

12.1% 26.7% 51.9% 9.31%

1.59

1.59

1.76

1.78

1.58

a

LC + MMAE: light chain conjugated with one MMAE. bHC + 1 MMAE: heavy chain conjugated with one MMAE. cHC + 2 MMAE: heavy chain conjugated with two MMAEs

Table 2. Drug−Antibody Ratios of Purified Compound 4 As Measured by Different Methods HIC−HPLC reduced RP-HPLC DAR = 0 DAR = 1 DAR = 2 DAR = 3 LC LC + MMAEa HC HC + 1 MMAEb HC + 2 MMAEc

100.0% 0.00% 7.12% 91.3% 1.59%

av DAR

1.89

SEC−Q-TOF-MS

conditions A

conditions B

RPLC−Q-TOF-MS

denaturing

native

2.09% 97.0%

1.54% 98.5%

100.0%

100.0%

100.0%

1.96

1.99

2.00

2.00

2.00

a

LC + MMAE: light chain conjugated with one MMAE. bHC + 1 MMAE: heavy chain conjugated with one MMAE cHC + 2 MMAE: heavy chain conjugated with two MMAEs.

Figure 1. Analysis of trastuzumab−two peptides (2): (A) RPLC−Q-TOF-MS and (B) HIC−HPLC.

wavelength at 450 nm). Each sample (1 mg/mL, 20 μL) was injected into the system and eluted over a 15 min run. Data evaluation was performed as previously reported.1

SEC−Q-TOF-MS of Reduced ADCs. Reduced samples were prepared as follows: 1.0 mg/mL of ADC in 50 mM PBS and 10 mM EDTA, pH 7.4, was diluted to 0.6 mg/mL in 8 M guanidine HCl and reduced by addition of 0.5 M DTT and Nacetylcysteine. The mixture was incubated at 80 °C for 5 min. SEC−Q-TOF-MS analysis was performed on a Biozen SEC-3, 2.1 × 50 mm, 1.8 μm column (Phenomenex), connected to an Agilent 1260 HPLC system with an Agilent 6550 Q-TOF system containing a binary gradient pump, temperaturecontrolled column compartment, autosampler, and a diode array detector. The system ran at 0.02 mL/min at 25 °C using 1% formic acid, 0.01% TFA, 50% acetonitrile, and 49% water, and the absorbance was monitored at 280 nm (reference

■

RESULTS AND DISCUSSION

Ellman’s assay of thiol-modified trastuzumab (3) revealed the expected 1.93 average free-sulfhydryls per antibody, indicating that the conversion rate from trastuzumab to compound 3 was efficient (Table S1, SI). Analysis by RP-HPLC (Figure S5, SI) of 3 and our previous investigations1 strongly supported this high conversion rate. D


Article


Figure 2. RP−HPLC results for the AJICAP-conjugates: comparison of unpurified AJICAP-ADC (4) and purified AJICAP-ADC (4) (left) and comparison of trastuzumab (right, top column), thiol-modified trastuzumab 3 (right, middle column), and combined sample (right, bottom column).

Figure 3. HIC−HPLC results for trastuzumab AJICAP−MMAE (4): conditions A (left) and conditions B (right).

reliability of native SEC−Q-TOF-MS, while denaturing SEC− Q-TOF-MS provided a relatively inaccurate DAR distribution (this was observed even with a deglycosylated ADC sample derived from a more complex mixture of glycoforms). Other subtle differences in the DAR values shown in Table 1 can be attributed to impurities and the variability inherent in each method. Measured differences in DAR determination could possibly be due to method accuracy, sensitivity, or linearity. However, all the analyses of purified AJICAP-ADC gave almost the same DAR results (Table 2). RP-HPLC analysis is the most commonly reported analytical method for stochastic heterogeneous ADCs, but peak identification is sometimes challenging because of the wide drug distribution. However, site-specific technology, such as AJICAP, can simplify DAR determination analysis.35 The RPHPLC chromatograms of reduced trastuzumab−AJICAP− MMAE (4) showed two main peaks [Figure 2, left, and Figures S1 and S2 (SI)]. The highest peak (retention time = 13.9 min) corresponded to heavy chain conjugated with MMAE, and the second highest peak (retention time = 7.5 min) was assigned as unconjugated light chain. Additionally, three small peaks were observed in the RP-HPLC chromatogram. The smallest peak (retention time = 17.5 min) was identified as heavy chain conjugated with one extra MMAE, implying that an undesired side reaction with the heavy chain had occurred (Figure S3, SI).36 The two peaks observed between the unconjugated light chain and conjugated heavy chain eluted closely (retention times = 11.0 and 11.6 min). The peak with retention time = 11.0 min corresponded to unconjugated heavy chain, but the other peak (retention time = 11.6 min) did not match the retention time for stochastic ADC prepared using traditional cysteine-type conjugation (Figure S7, SI).37 To identify this minor peak from the AJICAP-ADC, further RP-HPLC analysis was conducted using unmodified trastuzumab and thiol-

The comparison of analytical methods for DAR characterization was conducted with both unpurified and purified trastuzumab−AJICAP−MMAE (4) (Tables 1 and 2). Slightly different results were obtained for the unpurified AJICAP-ADC using the five different analytical methods. The most significant difference among the results obtained from five different methods presented herein is the existence of a small amount of DAR = 3.0 compound. RP-HPLC, HIC, and native SEC−Q-TOF-MS indicated the presence of a minor quantity of DAR = 3.0 material. On the other hand, RPLC−QTOF-MS and denaturing SEC−Q-TOF-MS provided no corroborating evidence for antibody loaded with three drug molecules. To investigate this discrepancy in observations between analytical techniques, we also analyzed the intermediate trastuzumab−peptide conjugate (compound 2), which is the first intermediate compound in AJICAP conjugation (Figure 1). In addition to the major compound (2), smaller amounts of trastuzumab modified by a third equivalent of peptide reagent (1) were observed as a byproduct in both RPLC−Q-TOF (Figure 1A) and HIC (Figure 1B), a result similar to that in our previous study.1 The small amounts of trastuzumab antibody modified by three peptides converted to labeled antibody incorporating three available thiol moieties upon linker cleavage. These three free thiols were capable of reaction with MC-VC-MMAE and, indeed, did result in the generation of DAR 3 material during the payload conjugation step.33 These results implied that RPLC−Q-TOF-MS and denaturing SEC−Q-TOF-MS gave “semi-reliable” DAR results, as they did not indicate the presence of the minor population of DAR 3 species present in the unpurified conjugated ADC. In 2015, the Beck and Cianférani group reported data comparing denaturing SEC−Q-TOF-MS and native SEC−Q-TOF-MS using trastuzumab emtansine.34 Their results indicated the E


Article


Figure 4. Deconvoluted spectra of trastuzumab−AJICAP-ADCs: (A) RPLC−Q-TOF-MS of unpurified ADC 4, (B) denaturing SEC−Q-TOF-MS of unpurified ADC 4, (C) native SEC−Q-TOF-MS of unpurified ADC 4, (D) RPLC−Q-TOF-MS of purified ADC 4, (E) denaturing SEC−QTOF-MS of purified ADC 4, and (F) native SEC−Q-TOF-MS of purified ADC 4.

IPA-free HIC analysis (conditions B) to confirm the DAR distribution, showing identical results with condition A [Figure 3B and Figures S11 and S12 (SI)]. Four peaks were observed in both the chromatograms of the unpurified ADC, corresponding to unconjugated antibody (DAR = 0), antibody conjugated with MMAE (DAR = 1), antibody conjugated with two MMAE groups (DAR = 2), and antibody conjugated with three MMAE groups (DAR = 3); the presence of the latter peak indicated an undesired side reaction.36 Next, we examined several Q-TOF-MS methods based on different liquid chromatography (LC) methods. RPLC−QTOF-MS, the most common protein analytical method using LC−MS, gave only three deconvoluted peaks, corresponding to unconjugated antibody (DAR = 0), antibody conjugated with MMAE (DAR = 1), and antibody conjugated with two MMAE groups (DAR = 2) [Figure 4A and Figure S13 (SI)]. The mean DAR value derived from these peaks was 1.8, a little higher than the results obtained from HIC and RP-HPLC. Generally, RPLC−Q-TOF-MS systems require a desalting pretreatment for each sample because of the highly sensitive ion source, but the process for removal of salt from ADCs carries potential risks, including increasing the aggregation of the conjugates and/or loss of sample during the concentration process. Taking into consideration the disadvantages of RPLC−Q-TOF-MS, a SEC−Q-TOF-MS system is expected to be a powerful tool for user-friendly analysis. The SEC column can easily desalt samples without the need for any other pretreatments. We applied SEC−Q-TOF-MS analysis of

modified trastuzumab, which was identical to the starting material for payload conjugation [Figure 2, right, and Figures S4−S6 (SI)]. Analysis of the sample mixture of thiol-modified and unmodified trastuzumab showed an unconjugated heavy chain (HC) peak separate from the thiol-modified heavy chain peak, which was identical to the unassigned peak in the chromatogram of the AJICAP-ADC from the previous RPHPLC analysis. After purification, the unconjugated HC peaks were observed to shift slightly (Figure 1, left panel). This observation was rationalized by deviation from the lower limit of quantification (LLOQ) of this compound38 due to the relatively lower amount of these species post-purification. Since the relative quantities of unconjugated HC species were decreased by preparative HIC chromatography, they are more easily captured by column packing material than crude sample, resulting in a slight shift in the retention times. These RP-HPLC results supported the site-specificity of AJICAP conjugation, as no light chain conjugates were observed in the RP-HPLC chromatogram.36 The HIC analyses of trastuzumab−AJICAP−MMAE (4) using two different sets of conditions provided clear visual results because of their simple composition (Figure 3). First, IPA was used for mobile phase B component (conditions A) to provide good separation of DAR species [Figure 3A and Figures S8 and S9 (SI)], but the use of organic solvents such as IPA has potential risk to alter the diffusion of molecules, causing their retention time shift and poor peak resolution. From this “non-denaturing” perspective, we also attempted F


Article


Initial reaction condition screening studies can result in a larger sample set matrix generated from a design-of-experiments (DOE) optimization. From this perspective, RPLC−QTOF-MS has a greater advantage over both native and denaturing SEC−Q-TOF-MS methods due to the smaller sample required for analysis (typically ∼5 μg of ADC mixture). The comparison of Q-TOF techniques (Figure 4 and Table 3) indicates that each method can provide valuable analytical information when used appropriately. We recommend that all of these methods be used, but at different stages, to increase the efficiency given specific analytical objectives. In addition to these HPLC/LC−MS techniques, Ellman’s assay was conducted to confirm the remaining free thiol-toantibody ratio (Tables S1−S3, SI). Prior to the payload conjugation step, the free thiol-to-antibody ratio was determined to be 1.9 (compound 3), but the free-thiol value for the unpurified AJICAP-ADC sample post-payload conjugation was determined to be 0.33 (Table S2, SI). This result indicated that 1.6/1.9 of the free thiols per antibody reacted during the payload conjugation step. Finally, to check the selectivity of AJICAP conjugation, a rapid Q-TOF analysis was investigated. Samples were briefly pretreated and reduced by DTT (5 min exposure) and then injected into the denaturing SEC−Q-TOF-MS system without further manipulation. Unpurified trastuzumab−AJICAP-ADC was eluted at approximately 5 min [Figure 5 and Figure S20 (SI)]. Deconvolution spectra derived from these samples showed four peaks corresponding to unconjugated light chain, unconjugated heavy chain, heavy chain labeled with a free thiol group, and heavy chain conjugated with one MMAE payload. These results were consistent with the RP-HPLC results indicating the heavy chain selectivity of the AJICAP conjugation technology. In our recent previous study, a peptide mapping experiment of compound 3 (the precursor prior to conjugation with MC-VC-MMAE), indicated that AJICAP technology can install a thiol moiety in a site-specific manner in the Fc region of an antibody.1,39 These results support the conclusion that trastuzumab−AJICAP−MMAE (4) retains site-specific conjugation. Moreover, these analyses were produced in less than 10 min per sample, including 5 min pretreatment and 5 min for elution. This rapid analysis without purification is useful for the in situ monitoring of conjugation efficacy, especially for use in future manufacturing processes.

AJICAP-ADCs for in situ monitoring of the conjugation reaction. First, denaturing SEC−Q-TOF-MS using mobile phases containing TFA and FA was performed [Figure 4B and Figure S16 (SI)]. To avoid denaturing condition, ammonium acetate was selected as the mobile phase for SEC−Q-TOF-MS. Ammonium acetate solution is a well-known MS-compatible buffer, which has high volatility and a slightly acidic pH to stabilize the ADCs. Therefore, a SEC−Q-TOF-MS system using ammonium acetate termed “native Q-TOF-MS” was used. The DAR value produced by this native Q-TOF-MS method was 1.6, identical to the HIC results [Figure 4C and Figure S18 (SI)]. Regardless of the LC method, the DAR for the purified ADC was 2.0 [Figure 4D−F and Figures S14, S17, and S19 (SI)]. Next, the characteristics of these three Q-TOF-MS analysis methods were evaluated (Table 3). Table 3. Comparison of the Characteristics of Three Different Q-TOF-MS Analyses SEC−Q-TOF-MS RPLC− Q-TOF-MS required ADC amount (μg) accuracy pretreatment recommended purpose

denaturing

native

5

20

20

semireliable necessary reaction screening

semireliable not necessary reaction monitoring

reliable necessary accurate DAR determination

From the perspective of data accuracy, we found that native Q-TOF-MS is a more ideal analytical method for DAR determination due to the consistency of results compared to that of HIC analysis (which is a well-established approach to determine DAR). However, the ionization efficiency of ammonium cations is relatively weaker; therefore, this analysis requires a pretreatment sample preparation step such as gel filtration to remove interfering small molecules possessing stronger relative ionization efficiencies. On the other hand, denaturing SEC−Q-TOF-MS can separate ADC compounds and small molecules in an SEC column while the pretreatment sample preparation required for high-resolution native SEC−Q-TOF-MS is omitted. Furthermore, denaturing SEC−Q-TOF-MS is compatible with TFA/ FA mobile phases employed to enhance compound ionization. Considering the characteristics of denaturing SEC−Q-TOFMS, this technique has potential to find utility in conjugation reaction monitoring, mainly due to the simple sample preparation and shorter method time.

■

CONCLUSION Several analytical methods were employed for the DAR characterization at different points in the preparation of sitespecific ADC 4, which was produced utilizing AJICAP conjugation technology. Our results for the determined DAR

Figure 5. Rapid Q-TOF-MS analysis: deconvoluted mass spectra of reduced trastuzumab−AJICAP-ADC (4) (upper) and reduced trastuzumab (lower). G


Article


(6) Chudasama, V.; Maruani, A.; Caddick, S. Nat. Chem. 2016, 8, 114−119. (7) Chalouni, C.; Doll, S. J. Exp. Clin. Cancer Res. 2018, 37, 20. (8) LoRusso, P. M.; Weiss, D.; Guardino, E.; Girish, S.; Sliwkowski, M. X. Clin. Cancer Res. 2011, 17, 6437−6447. (9) Katz, J.; Janik, J. E.; Younes, A. Clin. Cancer Res. 2011, 17, 6428− 6436. (10) Senter, P. D.; Sievers, E. L. Nat. Biotechnol. 2012, 30, 631−637. (11) Behrens, C. R.; Liu, B. MAbs 2014, 6, 46−53. (12) Panowski, S.; Bhakta, S.; Raab, H.; Polakis, P.; Junutula, J. R. MAbs 2014, 6, 34−45. (13) Junutula, J. R.; Raab, H.; Clark, S.; Bhakta, S.; Leipold, D. D.; Weir, S.; Chen, Y.; Simpson, M.; Tsai, S. P.; Dennis, M. S.; Lu, Y.; Meng, Y. G.; Ng, C.; Yang, J.; Lee, C. C.; Duenas, E.; Gorrell, J.; Katta, V.; Kim, A.; McDorman, K.; Flagella, K.; Venook, R.; Ross, S.; Spencer, S. D.; Wong, W. L.; Lowman, H. B.; Vandlen, R.; Sliwkowski, M. X.; Scheller, R. H.; Polakis, P.; Mallet, M. Nat. Biotechnol. 2008, 26, 925−932. (14) Chen, T.; Chen, Y.; Stella, C.; Medley, C. D.; Gruenhagen, J. A.; Zhang, K. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2016, 1032, 39−50. (15) Cusumano, A.; Guillarme, D.; Beck, A.; Fekete, S. J. Pharm. Biomed. Anal. 2016, 121, 161−173. (16) Rodriguez-Aller, M.; Guillarme, D.; Beck, A.; Fekete, S. J. Pharm. Biomed. Anal. 2016, 118, 393−403. (17) Ouyang, J. Methods Mol. Biol. 2013, 1045, 275−283. (18) Haverick, M.; Mengisen, S.; Shameem, M.; Ambrogelly, A. MAbs 2014, 6, 852−858. (19) D’Atri, V.; Fekete, S.; Stoll, D.; Lauber, M.; Beck, A.; Guillarme, D. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2018, 1080, 37− 41. (20) Basa, L. Methods Mol. Biol. 2013, 1045, 285−293. (21) Fekete, S.; Molnar, I.; Guillarme, D. J. Pharm. Biomed. Anal. 2017, 137, 60−69. (22) Valliere-Douglass, J. F.; McFee, W. A.; Salas-Solano, O. Anal. Chem. 2012, 84, 2843−2849. (23) Chen, J.; Yin, S.; Wu, Y.; Ouyang, J. Anal. Chem. 2013, 85, 1699−1704. (24) Hengel, S. M.; Sanderson, R.; Valliere-Douglass, J.; Nicholas, N.; Leiske, C.; Alley, S. C. Anal. Chem. 2014, 86, 3420−3425. (25) Debaene, F.; Boeuf, A.; Wagner-Rousset, E.; Colas, O.; Ayoub, D.; Corvaïa, N.; Van Dorsselaer, A.; Beck, A.; Cianférani, S. Anal. Chem. 2014, 86, 10674−10683. (26) Valliere-Douglass, J. F.; Hengel, S. M.; Pan, L. Y. Mol. Pharmaceutics 2015, 12, 1774−1783. (27) Botzanowski, T.; Erb, S.; Hernandez-Alba, O.; Ehkirch, A.; Colas, O.; Wagner-Rousset, E.; Rabuka, D.; Beck, A.; Drake, P. M.; Cianférani, S. MAbs 2017, 9, 801−811. (28) Pacholarz, K. J.; Barran, P. E. EuPa Open Proteomics 2016, 11, 23−27. (29) Chen, T.-H; Yang, Y.; Zhang, Z.; Fu, C.; Zhang, Q.; Williams, J. D.; Wirth, M. J. Anal. Chem. 2019, 91, 2805−2812. (30) Källsten, M.; Hartmann, R.; Artemenko, K.; Lind, S. B.; Lehmann, F.; Bergquist, B. Analyst 2018, 143, 5487−5496. (31) Tsuchikama, K.; An, Z. Protein Cell 2018, 9, 33−46. (32) Bryant, P.; Pabst, M.; Badescu, G.; Bird, M.; McDowell, W.; Jamieson, E.; Swierkosz, J.; Jurlewicz, K.; Tommasi, R.; Henseleit, K.; Sheng, X.; Camper, N.; Manin, A.; Kozakowska, K.; Peciak, K.; Laurine, E.; Grygorash, R.; Kyle, A.; Morris, D.; Parekh, V.; Abhilash, A.; Choi, J.; Edwards, J.; Frigerio, M.; Baker, M. P.; Godwin, A. Mol. Pharmaceutics 2015, 12, 1872−1879. (33) Our research group has completed optimization studies to overcome this overreaction, and these advancements will be discussed in future literature. (34) Marcoux, J.; Champion, T.; Colas, O.; Wagner-Rousset, E.; Corvaïa, N.; Van Dorsselaer, A.; Beck, A.; Cianférani, S. Protein Sci. 2015, 24, 1210−1223.

values of the unpurified and chromatographically purified trastuzumab−AJICAP−MMAE (4) were 1.6 and 2.0, respectively. Multiple analytical methods using a single batch of a site-specific ADC allowed for their direct comparison and the evaluation of their use in DAR determination at different stages. Importantly, the measured DAR results varied by analytical method for the unpurified ADC 4 and intermediates during the synthesis of ADC 4, indicating that a combination of varying analytical methods can be used together to understand the average payload status for unpurified ADCs and intermediates.30,40 For the final ADC 4 sample which underwent purification, all utilized analytical techniques gave essentially the same results. We are confident that continued innovations and strategies in the field of protein analytical chemistry will aid in the further development of site-specific ADC technologies.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.9b02192. An experimental section and figures and tables, as described in the text (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Brian A. Mendelsohn: 0000-0002-6339-2377 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors wish to thank our colleagues from Ajinomoto BioPharma Services, Inc., and Ajinomoto Co., Inc., as follows: Dr. Samuel Janssen, Mr. Josh Toschi, and Mr. Keenan Moi for technical assistance with ADC analysis; Ms. Zhala Tawfiq for technical assistance with wild-type conjugation; Mr. Michiya Kanzaki and Dr. Tatsuya Okuzumi for many helpful comments and suggestions in this study; Mr. Kei Yamada for peptide reagent supply and many helpful initial studies on AJICAP technology; Dr. Kazutaka Shimbo and Ms. Natsuki Shikida for critical opinions on Q-TOF-MS analysis; Dr. Kazutoshi Takahashi for graphical assistance; and Mr. Brian Rivera from Phenomenex for manuscript draft review.

■

REFERENCES

(1) Yamada, K.; Shikida, N.; Shimbo, K.; Ito, Y.; Khedri, Z.; Matsuda, Y.; Mendelsohn, A. B. Angew. Chem., Int. Ed. 2019, 58, 5592−5597. (2) Our research group has completed optimization studies to streamline the conjugation reaction sequence for our site-specific approach, and these advancements will be discussed in future scientific literature. For this reason, we termed “first-generation AJICAP” the site-specific conjugation technology described in this paper. An initial study for first-generation AJICAP has recently been reported (see ref 1). (3) Thomas, A.; Teicher, B. A.; Hassan, R. Lancet Oncol. 2016, 17, e254−e262. (4) Beck, A.; Goetsch, L.; Dumontet, C.; Corvaia, N. Nat. Rev. Drug Discovery 2017, 16, 315−337. (5) Polakis, P. Pharmacol. Rev. 2016, 68, 3−19. H


Article

Analytical Chemistry (35) Xu, Y.; Jiang, G.; Tran, C.; Li, X.; Heibeck, T. H.; Masikat, M. R.; Cai, Q.; Steiner, A. R.; Sato, A. K.; Hallam, T. J.; Yin, G. Org. Process Res. Dev. 2016, 20, 1034−1043. (36) To clarify this observation, further analyses including peptide mapping are in progress. (37) Doronina, S. O.; Toki, B. E.; Torgov, M. Y.; Mendelsohn, B. A.; Cerveny, C. G.; Chace, D. F.; DeBlanc, R. L.; Gearing, R. P.; Bovee, T. D.; Siegall, C. B.; Francisco, J. A.; Wahl, A. F.; Meyer, D. L.; Senter, P. D. Nat. Biotechnol. 2003, 21, 778−784. (38) Xie, C.; Wang, Z. In Antibody−Drug Conjugates: The 21st Century Magic Bullets for Cancer; J. Wang, J., Shen, W.-C., Zaro, J. L., Eds.; AAPS Advances in the Pharmaceutical Sciences Series; Springer: Basel, Switzerland, 2015; Vol. 17, pp 97−115. (39) Hermanson, G. T. Bioconjugate Techniques, 3rd ed.; Elsevier: London, 2013; pp 240−246. It is commonly established that maleimide compounds such as MC-VC-MMAE react with high selectivity to thiol moieties. (40) The Källsten and Bergquist group also recommended at least two techniques to verify DAR values; see ref 30.

I


Comparison of Analytical Methods for AntibodyâDrug

Recommend Documents

Comparison of Analytical Methods for AntibodyâDrug