Native mass spectrometry and ion mobility characterization of trastuzumab emtansine, a lysine-linked antibody drug conjugate

Abstract

Antibody-drug conjugates (ADCs) are biochemotherapeutics consisting of a cytotoxic chemical drug linked covalently to a monoclonal antibody. Two main classes of ADCs, namely cysteine and lysine conjugates, are currently available on the market or involved in clinical trials. The complex structure and heterogeneity of ADCs makes their biophysical characterization challenging. For cysteine conjugates, hydrophobic interaction chromatography is the gold standard technique for studying drug distribution, the naked antibody content, and the average drug to antibody ratio (DAR). For lysine ADC conjugates on the other hand, which are not amenable to hydrophobic interaction chromatography because of their higher heterogeneity, denaturing mass spectrometry (MS) and UV/Vis spectroscopy are the most powerful approaches. We report here the use of native MS and ion mobility (IM-MS) for the characterization of trastuzumab emtansine (T-DM1, Kadcyla(®)). This lysine conjugate is currently being considered for the treatment of human epidermal growth factor receptor 2 (HER2)-positive breast cancer, and combines the anti-HER2 antibody trastuzumab (Herceptin(®)), with the cytotoxic microtubule-inhibiting maytansine derivative, DM1. We show that native MS combined with high-resolution measurements and/or charge reduction is beneficial in terms of the accurate values it provides of the average DAR and the drug load profiles. The use of spectral deconvolution is discussed in detail. We report furthermore the use of native IM-MS to directly determine DAR distribution profiles and average DAR values, as well as a molecular modeling investigation of positional isomers in T-DM1.

Keywords: Kadcyla®; T-DM1; antibody-drug conjugate (ADCs); drug to antibody ratio (DAR); ion mobility mass spectrometry; monoclonal antibody; native mass spectrometry.

Figure 1

Raw (left) and deconvoluted (right) electrospray ionization mass spectra obtained under denaturing conditions by direct infusion on a Q-TOF instrument of trastuzumab emtansine (A) before and (B) after deglycosylation. The insets show expanded views of the 45+ and 46+ charge states. The mass range considered for deconvolution is indicated by a double arrow. The asterisks indicate +220 Da linker adducts.

Figure 2

Raw (left) and deconvoluted (right) electrospray mass spectra obtained under native conditions by direct infusion on a Q-TOF instrument of trastuzumab emtansine (A) before and (B) after deglycosylation. The mass range considered for deconvolution is indicated by a double arrow. The asterisks indicate +220 Da linker adducts.

Figure 3

Raw (left) and deconvoluted (right) high-resolution mass spectra obtained under native conditions by direct infusion into an orbitrap instrument (nominal resolution: 17.5 k) of trastuzumab emtansine (A) before and (B) after deglycosylation. The corresponding signals obtained on the Q-TOF instrument are superimposed in (A) red and (B) pink for the 25+ charge state. The mass range considered for deconvolution is indicated by a double arrow. The asterisks indicate +220 Da linker adducts.

Figure 4

Raw (left) and deconvoluted (right) electrospray ionization mass spectra obtained under denaturing conditions by direct infusion on a Q-TOF instrument of trastuzumab emtansine (A) before and (B) after deglycosylation with 10 mM imidazole added for charge reduction. The mass range considered for deconvolution is indicated by a double arrow. The asterisks indicate +220 Da linker adducts.

Figure 5

Raw (left) and deconvoluted (right) high-resolution mass spectra obtained under native conditions by direct infusion into an orbitrap instrument (nominal resolution: 17.5 k) of trastuzumab emtansine (A–C) before and (D–F) after deglycosylation, in (A, D) the absence or (B–F) the presence of (B, E) 20 mM and (C, F) 40 mM imidazole. The mass range considered for deconvolution is indicated by a double arrow. The asterisks indicate +220 Da linker adducts.

Figure 6

Determining the drug-to-antibody ratio (DAR) of deglycosylated trastuzumab emtansine by ion mobility MS. (A) The arrival time distributions (ATDs) obtained for each Dn species of the 24+ charge state. (B) The drug load as measured from IM-MS. (C) Gaussian fits of the experimental drift traces of the 24+ charge state. The experimental resolution was 20, resulting in a full width of half maximum (FWHM) of 0.83 ms and a separation of the different DARs at 92.5% of the valley. (D) Theoretical Gaussian fits of Dn species generated with a FWHM of 0.23 ms corresponding to an IM resolution of 69 and resulting in separation at 50% of the valley. In order to obtain more symmetric ATDs, the number of drift time bins was increased virtually from 200/cycle to 400/cycle.

Figure 7

(A) Collision cross-section (CCS) as a function of drug-to-antibody ratio (DAR) measured for the different Dn species of glycosylated trastuzumab emtansine (T-DM1). (B) Molecular model of trastuzumab conjugated with four DM1 payloads (positional isomers Pi1 to Pi4). (C) Comparison of experimental and theoretical arrival time distributions (ATDs). The experimental ATDs for the D0 and D1 species of the 24+ charge state of glycosylated T-DM1 are represented in red and black lines, respectively. The theoretical values corresponding to the increment resulting from the conjugation of the four D1 positional isomers Pi1–Pi4 are represented respectively in green, purple, cyan, and orange. The increments were scaled down with the same scaling factor as measured between the experimental and theoretical CCS values for naked trastuzumab. In order to separate these isomers at 50% of the valley, a virtual resolution of ∼330 was used along with 400 bits/cycle.