Maytansinoid–BODIPY Conjugates - American Chemical Society

Mar 4, 2015 - ImmunoGen, Inc., 830 Winter Street, Waltham, Massachusetts 02451, United ... BODIPY as a conjugate with little through-space interaction...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/molecularpharmaceutics

Maytansinoid−BODIPY Conjugates: Application to Microscale Determination of Drug Extinction Coefficients and for Quantification of Maytansinoid Analytes Nathan Fishkin* ImmunoGen, Inc., 830 Winter Street, Waltham, Massachusetts 02451, United States S Supporting Information *

ABSTRACT: Determining drug to antibody ratios (DAR) for antibody−drug conjugates (ADCs) in early research and development can be hampered by difficulties in accurate weighing of the effector payload and subsequent determination of its extinction coefficient. Two maytansinoids, DM1 and DM4, potent antimitotic agents used in clinical ADCs, were derivatized with the compact fluorophore BODIPY FL using two different linker designs. We identified DM1−mal− BODIPY as a conjugate with little through-space interaction between the maytansinoid and BODIPY chromophores. The 1:1 stoichiometry between the maytansinoid and BODIPY makes the molar concentration of both components equal and the extinction coefficient of the maytansinoid in proportion with the known BODIPY chromophore according to Beer’s Law. By only derivatizing 50 μg of unpurified DM1 and analyzing about 25 μg of DM1−mal−BODIPY by UV−vis, we determined εDM1 252 nm and εDM1 280 nm as 26 355 ± 360 and 5230 ± 160 cm−1 M−1, respectively. These values are nearly identical to those accepted for DM1 based on weighing >100 mg of pure sample. Surprisingly, some of the maytansinoid−BODIPY conjugates that were synthesized were partially or completely fluorescence-quenched. The green fluorescence of quenched DM4−acetamide− BODIPY could be fully restored in the presence of an antibody designed to tightly bind maytansine. We exploited this observation to develop a simple “mix and read” fluorogenic immunoassay for detection of nanogram quantities of maytansinoids. KEYWORDS: BODIPY FL C1−IA, N-(4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-yl)methyl)iodoacetamide (BODIPY FL), N2′-deacetyl-N2′-(3-mercapto-1-oxopropyl)-maytansine, N2′-deacetyl-N2′-(4-mercapto-4-methyl-1-oxopentyl)-maytansine, drug quantification, homogeneous immunoassay, fluorogenic probe, extinction coefficient, antibody−drug conjugate (ADC), DM1, DM4



INTRODUCTION Antibody−drug conjugates (ADCs) are an emerging therapy platform that has been validated with regulatory approvals of Kadcyla (ado-trastuzumab emtansine)1 and Adcetris (brentuximab vedotin).2 Therefore, ADCs are the subject of continued development in targeted treatments for solid and liquid tumor indications. One of the key factors in the design of effective ADCs is the number of molecules of cytotoxic agent per antibody, typically referred to as the drug-to-antibody Ratio (DAR). DAR can markedly impact the potency of an ADC on the target cancer cell and its degree of off-target toxicity. The extent of drug loading can also have a significant effect on the biochemical characteristics of an ADC such as stability and protein aggregation, which can in turn affect in vivo pharmacokinetics and tumor accumulation properties. Therefore, it is essential to accurately determine and control DAR during the development of ADC therapeutics. Mass spectrometry (MALDI-TOF or ESI-MS) or UV spectroscopy is generally used to determine DAR of maytansinoid-containing ADCs. Calculations based on UV © XXXX American Chemical Society

absorption require the accurate determination of the molar extinction coefficient of the cell-killing agent. This typically requires a significant amount of pure reagent with little or no residual solvent impurities to ensure accurate sample weighing. While this is achievable in later-stage development and manufacturing, many ADC effector molecules in early stage lead screening and optimization are prepared in smaller quantities making accurate extinction coefficient determination, and therefore DAR determination, challenging for molecules with diverse chromophoric structures. Even within a chemical series, extinction coefficients can vary and introduce significant error into DAR calculations. In addition, effector candidates with limited solubility in fully aqueous buffer may have their absorbance properties change significantly with increasing Special Issue: Antibody-Drug Conjugates Received: December 17, 2014 Revised: February 23, 2015 Accepted: March 4, 2015

A

DOI: 10.1021/mp500843r Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 1. Structures of DM1, S-methyl DM1, and maytansinoid−BODIPY reagents used in this study.

methyl)iodoacetamide (D-6003) were purchased from Molecular Probes (Eugene, OR). Phosphate buffered saline (PBS) pH 7.4 was purchased from GIBCO. Nonbinding control chimeric antibody KTI and murine antimaytansine antibody (IgG1) were developed at ImmunoGen and purified by standard methods.5 DM1 and DM4 were prepared as previously described by Widdison, et al.6 Maytansinoid− BODIPY conjugates were analyzed on a Bruker ESQUIRE 3000 ion trap mass spectrometer used in line with an Alliance 2695 separation module outfitted with a 2996 PDA detector (Waters, Corp., Milford, MA). N,N-Dimethylacetamide (DMA) was purchased from Sigma and then further distilled. Compounds were purified using a Vydac C8, 250 × 4.6 mm, 5 μm column (Model #208TP104), 1.0 mL/min flow rate, absorbance detection at 252 nm, and fluorescence detection with 504 nm excitation, 520 nm emission. The mobile phase consisted of acetonitrile and water (with 0.025% acetic acid). A linear gradient of 15% to 100% acetonitrile over 25 min was used during the purification. The mass spectrometer used an electrospray ionization method with nebulizer gas flow of 25 psi, a drying temperature of 350 °C, and a heating gas flow of 8.0 L/min. Mass detection range was 50−2000 amu and was carried out in alternating positive and negative ion modes or in MS/MS mode. General Method for Preparing Maytansinoid−BODIPY Conjugates 1−4. A solution of BODIPY FL-maleimide or BODIPY FL-iodoacetamide in DMA (0.2 mL, 300 μM) was added to a solution of DM1 or DM4 (0.2 mL, 450 μM) in a

amounts of organic cosolvent needed to solubilize them, further confounding extinction coefficient determination. Therefore, we have developed a simple microscale method based on fluorophore derivatization for determining the extinction coefficients for ADC effector molecules without need for accurate sample weighing or high initial sample purity. This effector derivatization method is exemplified here with DM1 (Figure 1), the antimitotic maytansinoid employed in Kadcyla. The compact fluorophore BODIPY (derived from borondipyrromethene) is a laser dye with high quantum yield extensively used to track subcellular processes3 and as a sensitive indicator for pH, metal ions, redox potential, chemical reactions, and reactive nitrogen and oxygen species.4 It was chosen because it possesses a chromophore with red-shifted, nonoverlapping absorbance compared to DM1. Two commercially available 5,7dimethyl-substituted BODIPY analogues with different linkers and thiol reactive groups were reacted with two different maytansinoid thiols, and various levels of BODIPY fluorescence quenching was observed with the maytansinoid−BODIPY conjugates. The fully quenched conjugate DM4−acetamide− BODIPY was successfully employed in a “mix and read” 96-well plate immunoassay for sensitive quantitation of maytansinoid analyte.



EXPERIMENTAL SECTION Materials and Methods. BODIPY FL N-(2-Aminoethyl))maleimide (B-10250) and BODIPY FL C1−IA, N-(4,4difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-yl)B

DOI: 10.1021/mp500843r Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

methoxy group of the maytansinoid to the boron atom in BODIPY for the ten lowest energy conformers of 1 and 4 was calculated. These average distances were overlaid onto the stick model structures for the lowest energy conformer of 1 and 4 and shown in Figure 4B. Mix and Read Fluorogenic Assay to Quantify Maytansinoids in Aqueous Solution. Preparation of Antimaytansine−DM4−BODIPY Complex. A solution of antimaytansine antibody (50 μL, 7.8 mg/mL) was diluted in 1 mL of PBS, pH 7.4, and then mixed with DM4−acetamide− BODIPY 4 (10 μL of a 800 μM stock in 50:50 acetonitrile/ water) for a final concentration of 8 μM for 4 and 2 μM for antibody. The resulting complex was applied to a Sephadex G25 column (Illustra NAP10, GE Healthcare) equilibrated in 50 mM Tris·HCl, pH 7.0, containing 2 mM EDTA, with the purified antimaytansine/4 complex eluting in the first 1.5 mL volume and excess unbound 4 eluting in the second and third volumes. The purified complex was measured by UV−vis and found to have a protein concentration of 1.36 μM and 1.2 BODIPY linked per antibody, using an ε280 nm Ab = 225 000 cm−1 M−1 and ε504 nm BODIPY FL = 62 666 cm−1 M−1. S-Methyl DM1 Quantitation Curve Using Direct Competition Format. A stock solution of antimaytansine antibody (0.5 μM) was prepared in PBS, pH 7.4, containing 1% (w/v) bovine serum albumin (BSA), and 100 μL was added to each well in a Costar black-sided assay plate (Part #3603, Corning, NY). SMethyl DM1 analyte was serially diluted 2-fold in PBS, pH 7.4, in glass vials and then a fixed amount of 4 (250 μM in acetonitrile) was added to each S-methyl DM1 solution to give a final concentration of 0.51 μM of 4, and from 11 μM to 0.04 μM of S-methyl DM1 in each vial. Each of the probe/analyte mixtures (100 μL) was then added to a well on the assay plate containing the antimaytansine antibody. Final amounts of Smethyl DM1 ranged from 3.2 to 810 ng/well, and 4 was fixed at 55.6 ng/well and antibody at 7.5 μg/well. After incubation for 20 min at rt, BODIPY fluorescence was read on a SpectraMax 2 microplate reader (Molecular Devices, Sunnyvale, CA) with excitation at 485 nm and emission at 538 nm (530 nm cutoff). Fluorescence units were corrected by subtracting the residual fluorescence of control wells containing only 4. S-Methyl DM1 Quantitation Curve Using Competition with Purified Antimaytansine Antibody−4 Complex. Purified antimaytansine antibody/4 complex as described above was formulated at 0.42 μM in PBS, pH 7.4, and 1% (w/v) BSA, and 100 μL was added to each well of the assay plate. The S-methyl DM1 analyte was serially diluted 2-fold in PBS, pH 7.4, in glass vials to give a concentration range from 11 μM down to 0.04 μM. Each of the S-methyl DM1 dilutions (100 μL) were then added to the assay plate. The final amount of S-methyl DM1 analyte ranged from 3.2 to 810 ng/well, and the total amount of 4 complexed to antimaytansine antibody was fixed at 58 ng/ well. BODIPY fluorescence was measured as described above, at 30 min after sample mixing and then at 2 h intervals. Fluorescence signals remained constant after 36 h at rt, and the resulting measurements were used for the S-methyl DM1 quantification curve. Fluorescence units were also corrected by subtracting the residual fluorescence of control wells containing only uncomplexed 4.

mixture of HEPES buffer, pH 8.5, and DMA (1:1) in a glass vial under nitrogen at room temperature and allowed to shake gently for 2 h. The resulting orange solutions were then treated with 20 mM N-ethyl maleimide to quench unreacted maytansinoid thiol. Each reaction mixture was purified using a semipreparative C8 HPLC and the desired compound peaks were identified by their unique 252 nm/504 nm absorbance profile compared to the starting materials. The isolated compounds were characterized by MS, UV, and fluorescence analysis. Compound purity and MS data for 1−4 are shown in the Supporting Information. Calculation of Molar Extinction Coefficients of DM1 from the UV−Vis Spectrum of 1. Solutions of BODIPY FLmaleimide were prepared fresh in absolute ethanol with five independent weighings, from 7.8 to 17 mg, and the extinction coefficient of BODIPY FL at 504 nm was determined from these samples to be 83 340 cm−1 M−1, which is consistent with literature values.7 The A252 nm/A504 nm and A280 nm/A504 nm ratios for BODIPY FL-maleimide were determined to be 0.021 and 0.023, respectively, and these values were used as a corrective factor in eqs 1 and 3 to subtract BODIPY absorbance at 252 and 280 nm so as not to overestimate maytansinoid concentration. DM1−mal−BODIPY 1 has a 1:1 stoichiometry between maytansinoid chromophore and the BODIPY dye, thus CBODIPY = CDM1, and because C = A/ε, then ABODIPY 504 nm /83 340 cm−1 M−1 = [ADM1 252 nm − 0.021(ABODIPY 504 nm )]/εDM1 252 nm) (1)

Solving for εDM1 252 nm gives εDM1 252 nm = [83 340 cm−1(ADM1 252 nm ) − (83 340 cm−1) (0.021)(ABODIPY 504 nm )]/ABODIPY 504 nm

(2)

Then ABODIPY 504 nm /83 340 cm−1 = [ADM1 280 nm − 0.023(ABODIPY 504 nm )]/εDM1 280 nm) (3)

and solving for εDM1 280 nm gives εDM1 280 nm = [83 340 cm−1(ADM1 280 nm ) − (83 340 cm−1) (0.023)(ABODIPY 504 nm )]/ABODIPY 504 nm

(4)

Purified 1 in absolute ethanol was measured by UV−vis at four dilutions (0.4−2 OD at 504 nm), and the ADM1 252 nm, ADM1 280 nm, and ABODIPY 504 nm were input into eqs 2 and 4 to determine εDM1 252 nm and εDM1 280 nm, respectively. However, only absorbance values within the linear range of the spectrophotometer (0.08−1.5 OD) were used to calculate the final average εDM1 252 nm and εDM1 280 nm values reported in the results section. Molecular Dynamics Simulation of Compounds 1 and 4. Energy minimization of structures 1 and 4 using the MM2* force field was carried out in Chem3D Pro version 12.0 (CambridgeSoft), for an initial refinement of the 3D geometries. Then a molecular dynamics simulation was performed using a step interval of 2 fs, frame interval of 10 fs, cooling rate of 1 kcal/atom/ps, and target temperature of 300 K. For each compound, the average through-space distance from C20-



RESULTS Extinction Coefficient Determination of DM1 by Chromophore Derivatization and UV−Vis Analysis. We reasoned that if a maytansinoid could be derivatized in exactly

C

DOI: 10.1021/mp500843r Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics 1:1 stoichiometry with a noninteracting chromophore of known extinction coefficient then the concentration of both components would be equal and the extinction coefficient of the unknown component could be determined in proportion to the known chromophore according to Beer’s law. Four related maytansinoid−BODIPY conjugates were prepared (Figure 1) and DM1−mal−BODIPY 1 was found to be the best candidate for extinction coefficient determination of DM1 due to the lack of absorbance and fluorescence changes in BODIPY upon conjugation. This result suggested that the two chromophores do not interact (dipole−dipole coupling, etc.) when linked together and therefore do not significantly impact each other’s absorbance properties when proximal. The extinction coefficient for mal−BODIPY at 504 nm was independently determined by five separate weighings of dry powder and serial dilutions in ethanol (Figure 2) and the average A252 nm/

Figure 3. (A) UV−vis spectrum of compound 1 at 4 dilutions in ethanol; initial stock concentration = 1 mM. (B) Tabulated absorbance data from 1 at 252 and 280 nm for the maytansinoid chromophore and 504 nm for the BODIPY chromophore for all four dilutions. Calculated εDM1 252 nm and εDM1 280 nm are also tabulated for each dilution of 1. Although all ε values calculated vary by