Human IgG1 trastuzumab (Herceptin®) was purchased from Roche Pharmaceutical Company (Basel Switzerland). MCC-maytansinoid (catalog no: TCRS-1262) was purchased from Abzena (San Diego, CA, USA). PNGase F and GlycoBuffer 2 were purchased from New England Biolabs (Ipswich, Massachusetts, UK). All other chemical reagents were purchased from Sigma-Aldrich (St. Louis, MO, USA).

The SoloVPE spectroscopy system (C Technologies, Inc. Bridgewater, NJ, USA) provided ADC concentration.

Site-specific thiol intermediate was produced by a previously established procedure [13].

To a solution of site-specific thiol intermediate (6.8 mg/mL, 1.63 g) in a PBSE buffer (50 mM phosphate buffered saline (PBS), 10 mM ethylenediaminetetraacetic acid (EDTA), pH 7.4), dimethylformamide (DMF) (15 mL) and a 10-mM DMF solution of MCC-maytansinoid (10 eq., 11.7 mL) were added to form a mixture, which was then incubated at 20 °C. After 2 h, a small amount of the reaction mixture (0.5 mL) was sampled for the in-process control (IPC) analysis. Subsequently, the reaction mixture was quenched with an excess amount of a 50-mM aqueous solution of N-acetyl cysteine (NAC) and then incubated at 25 °C for 15 min. This reaction mixture was purified with a tangential flow filtration (TFF) system using a Sartocon Slice 200 Eco Hydrosart membrane (30 kDa; Sartorius) and diafiltration (DF) buffer as the conjugation buffer at an antibody concentration of 20 mg/mL. Next, buffer exchange of this solution was performed using a TFF system with a Sartocon Slice 200 Eco Hydrosart membrane (30 kDa; Sartorius) and DF buffer as the formulation buffer (20 mM histidine containing 5% trehalose, pH 5.2) at an antibody concentration of 5.7 mg/mL. Trastuzumab-AJICAP®-maytansinoid 4 (1.67 g, 98% yield) was then afforded in the formulation buffer.

Trastuzumab-AJICAP-maytansinoid (50 $\mu$L; 1 mg/mL in 50 mM Tris-HCl buffer, pH 7.6), PNGase F (2 $\mu$L; 500,000 units/mL), 10 $\mu$L of GlycoBuffer 2 (10$\times$), and 48 $\mu$L of water were incubated at 37 °C for 16 h. Then 5% v/v aqueous solution of trifluoroacetic acid as quencher was added to this reaction mixture and incubated at 37 °C for 30 min. The buffer exchange and sample concentration were carried out using 25 mM ammonium bicarbonate buffer with centrifugal concentrator (Vivaspin 500, 10,000 MWCO, Polyethersulfone (PES)).

Q-TOF MS analysis was performed using a PLRPS 1000 Å 8 $\mu$m column (Agilent) connected to an Agilent 1260 HPLC system, as well as with an Agilent 6550 Q-TOF system containing a binary gradient pump, temperature-controlled column compartment, autosampler, and diode array detector. The system ran at 0.3 mL/min at 80 °C using mobile phase A (MPA: 0.1% formic acid (FA), 0.01% TFA, 2% ACN, and 98% water) and mobile phase B (MPB: 0.1% FA, 0.01% TFA, and 99.9% acetonitrile). The absorbance was estimated at 280 nm using a reference wavelength of 450 nm. Intact or deglycosylated ADC (1 mg/mL, 5 $\mu$L) was injected into the system and eluted over a 15-min run consisting of a 2-min isocratic hold at 20% MPB, a 15-min linear gradient from 20% to 95% MPB, a 2-min wash using 95% MPB, and a 3-min re-equilibration at 20% MPB. Data acquisition was performed using MassHunter BioConfirm software (Agilent) to analyze the intact MS spectra for intact deconvolution, a mass range of 100,000–180,000 and a limited m/z range of 1000–5000 were set for standardization. The DAR was determined using DAR Calculator software (Agilent).

The EIC analysis was performed using Q-TOF MS system with m/z ratio of 1103.4 (M+).

The reductive pre-treatment was performed as previously reported [19].

RP-HPLC analysis was performed according to a previously reported study [19].

HIC HPLC analysis was performed according to a previously reported study [20].

Three sets of standard solutions of trastuzumab-AJICAP-maytansinoid (4) were prepared in a formulation buffer. All the samples were stored at various temperatures (–80, –20, 4, 25, and 37 °C) for 4 weeks. Aggregation was then analyzed using size-exclusion chromatography (SEC)-HPLC.

Aggregation analysis using SEC was performed as previously reported [20].

Using a well-established previously described approach [13], a site-specific thiol-modified mAb, the intermediate ADC precursor, was prepared from trastuzumab, which is the same antibody as that in T-DM1 upon completion of a gram-scale synthesis (Fig. 2). The final step in ADC preparation is the conjugation of this thiol intermediate with MCC-maytansinoid (Fig. 2A). This reaction completion was detected by IPC analysis using HIC HPLC, as per a previous good laboratory practice (GLP) material preparation [13]. Purification to remove residual drug linkers was performed in a straightforward manner using tangential flow filtration (TFF), which is a commonly used technique that can be applied to kilogram-scale ADC manufacturing [21, 22]. Compared to MC-VC-MMAE, which was previously used for preparing AJICAP-ADC, TFF could clear MCC-maytansinoid from the ADC composition. This can be attributed to the hydrophobicity difference of each drug linker. MC-VC-MMAE is relatively more hydrophobic than MCC-maytansinoid [23], thus, requiring careful monitoring of its clearance level. The removal rate of highly potent MCC-maytansinoid drug-linker was calculated by the extracted ion chromatogram (EIC) analysis, which showed that the rate was supposed to be near the detection limit in this analysis (data not shown).

Fig. 2.

Synthesis and analysis of trastuzumab-AJICAP-maytansinoid. (A) Synthetic scheme of the gram-scale ADC preparation. (B) Analytical summary of resulting ADC. (C) Aggregation comparison of T-DM1 (literature data) and trastuzumab-AJICAP-maytansinoid (tested data) at different temperatures after 4 weeks of storage. T-DMI, Trastuzumab-emtansine.

Three different analytical methods were used to determine the DAR of trastuzumab-AJICAP-maytansinoid (Fig. 2B). Q-TOF MS (Intact ADC: Supplementary Fig. 1, deglycosylated ADC: Supplementary Fig. 2), RP-HPLC (Supplementary Fig. 3) and HIC-HPLC (Supplementary Fig. 4) analyses revealed that this ADC has a high DAR value (1.8–1.9). The DAR results were slightly different among the three methods; however, this observation was also reported in several previous studies [24, 25, 26]. Aggregation percentage was within acceptance level (Supplementary Fig. 5).

Next, a thermal stability study of trastuzumab-AJICAP-maytansinoid was examined to estimate the suitable formulation and storage conditions towards forthcoming manufacturing processes (Fig. 2C). Several previous studies, including ours, had reported stability assessments results of T-DM1 in peer-reviewed literature; and we followed the same procedures for one of the processes applied to trastuzumab-AJICAP-maytansinoid [13, 16, 17, 18].

T-DM1 was found to be relatively stable at lower temperatures (4 °C) [11, 15]. At room temperature or higher, the aggregation percentage increased significantly; however, in the current study, trastuzumab-AJICAP-maytansinoid showed more resistance to high temperature than T-DM1. As described above, T-DM1 has a heterogeneous mixture of possibly more than 80 species [27], some of which may be unstable and prone to aggregation. In contrast, AJICAP-ADC is a low heterogeneous ADC; its conjugation position is known to be stable under thermal/blood circulation. This high stability of trastuzumab-AJICAP-maytansinoid reflects the stable linkage/conjugation site produced by the AJICAP technology.

A rat acute toxicity study was performed for the MTD determination of the ADCs (Fig. 3). Mortality or severe toxicity was observed with T-DM1 at a dose of 40 mg/kg; however, no toxicity was observed with AJICAP-ADC even at a dose of 120 mg/kg (Fig. 3A). Decreases in body weight or PLT levels and increased AST levels were not observed at doses of up to 120 mg/kg trastuzumab-AJICAP-maytansinoid and 20 mg/kg T-DM1. Additionally, Liver weight did not change significantly at a 120 mg/kg. Thus, the MTD of trastuzumab-AJICAP-maytansinoid and T-DM1 was estimated to be at least 120 and 20 mg/kg, respectively. Neutrophil, reticulocyte, lymphocyte and platelet count data supported these estimations, considering the mode of action of maytansinoids (Supplementary Fig. 6) [28, 29]. An exploratory rat safety toxicology study showed that the rats showed better tolerance to trastuzumab-AJICAP-maytansinoid than to T-DM1. Given the limited dosing in our study, the MTD of trastuzumab-AJICAP-maytansinoid was estimated to be at least 120 mg/kg, whereas the MTD of T-DM1 was estimated to be at least 20 mg/kg. This initial safety analysis indicated that the conjugation approach generated a stable ADC, although further biological/analytical studies are currently ongoing, such as PK. In previous report [15], a rat PK study showed that the trastuzumab-AJICAP-MMAE ADC is stable in blood circulation compared to native cysteine-based stochastic ADC. The measured levels of total antibody from trastuzumab-AJICAP-MMAE indicated a half-life similar to that expected for parent trastuzumab. Furthermore, in vitro binding assay using biolayer interferometry method concluded that trastuzumab-AJICAP-maytansinoid does not affect FcRn interactions [30]. These results suggest that site-specific conjugated ADCs at Lys248 result in a superior profile with regard to both safety and in vivo stability compared to the approved ADC, T-DM1.

Fig. 3.

Rat acute toxicity study to determine the MTD of ADCs. (A) Body weight change (N = 5 female SD rats). (B) Liver weight in rat serum at 2 days after administration (N = 5 female SD rats. (C) AST in rat serum at 2 d after administration (N = 5 female SD rats). (D) ALT levels in rat serum at 2 d after administration (N = 5 female SD rats). MTD, maximum tolerated dose; ADCs, antibody-drug conjugates; AST, aspartate aminotransferase; ALT, alanine aminotransferase. Significantly different from Histidine Buffer: * p 0.05, *** p 0.001. DT, Dunnett test (two-side); ST, Steel test (two-side).

Furthermore, in a previous study, we reported that the minimum effective dose (MED) of both ADCs was 5 mg/kg based on the NCI-N87 xenograft study. Thus, the TI of trastuzumab-AJICAP-maytansinoid was estimated to be qualitatively greater than that of T-DM1, considering their respective MTDs and MEDs (Table 1).