Impact of Maleimide Disubstitution on Chemical and Biological Characteristics of HER2 Antibody-Drug Conjugates.

Antibody-drug conjugates (ADCs) are the spearhead of targeted therapies. According to the technology used, the conjugation of a cytotoxic drug to an antibody can produce suboptimal heterogeneous species, impacting the overall efficacy. Herein, we describe the synthesis of HER2-targeting ADCs with three disulfide rebridging heads, allowing homogeneous and site-specific bioconjugation: dibromomaleimide (DBM), dithiomaleimide (DTM), and hybrid thio-bromomaleimide (TBM) chemical bricks to combine the properties of both previously used heads. The primary purpose of this study was to compare the reactivity of these three chemical bricks in the bioconjugation process. Then, the resulting ADCs were evaluated in terms of physicochemical stability, binding, and biological activity. We have demonstrated that the higher percentage of a drug-to-antibody ratio of 4 was obtained with TBM. Additionally, the reaction time was drastically reduced with TBM in comparison to DTM. The three ADCs showed good binding to HER2 and in vitro cytotoxicity, which validate the TBM structure as an attractive alternative scaffold for rebridging bioconjugation.

Introduction

Nowadays, there are more than 70 antibody–drug conjugates (ADCs)[1] in clinical development. Indeed, they are a promising class of medicine against cancer because they combine the specificity of a monoclonal antibody (mAb) with the cytotoxicity of a potent payload in such a way that this drug is delivered specifically to tumor cells overexpressing the targeted antigen. Despite significant efforts in development, only five ADCs are currently approved by the FDA. The first ADC to be approved in 2000 thanks to an accelerated approval was Mylotarg for acute myeloid leukemia, an anti-CD33 mAb attached through its lysines to a calicheamicin. In 2010, it was withdrawn from the market because of safety issues and rehabilitated thanks to a favorable risk–benefit balance in 2017. In 2011, Adcetris was approved by the FDA for Hodgkin lymphoma. It is obtained by the reduction of interchain disulfide bridges of an anti-CD30 mAb to fasten a maleimide linker to the free cysteines. In 2013, Kadcyla, in which an anti-HER2 mAb and DM1 are covalently linked via a lysine-based conjugation chemistry, reached the therapeutic market against breast cancer. Besponsa, which consists of a link between the lysines of an anti-CD22 mAb and a calicheamicin derivative, was approved in 2017 for acute lymphoblastic leukemia. More recently, in 2019, Polivy was approved with an accelerated procedure for large B-cell lymphoma. It combines an anti-CD79b mAb and vedotin. These five ADCs use nonspecific bioconjugation technologies, which are the most commonly used methods: stochastic lysine and cysteine modification. These methods present some drawbacks such as a limited control of the drug-to-antibody ratio (DAR) because of the random fixation on the amino acids (especially with the lysine residues, because of their greater abundance than reactive cysteine residues).[2] Thereby, the DAR distribution is heterogeneous, which is detrimental to the activity and pharmacokinetic/pharmacodynamics (PK/PD) profile of ADCs.[3] Indeed, there are unloaded species DAR 0 that compete with the loaded species because of their unchanged affinity for the target. Moreover, naked mAbs are less efficient than conjugated ones. Alongside DAR 0 species, the proportion of highly conjugated species is not negligible. These are more hydrophobic and at greater risk to be quickly eliminated,[4] decreasing the overall efficacy of the administrated dose.

To overcome stochastic bioconjugation drawbacks and control the number and position of the payload, the development of new site-specific technologies is required. Different bioconjugation techniques have been developed. For example, using molecular biology, antibody sequence engineering allows the introduction of orthogonal chemical handles, giving access to site-specific conjugation: reactive cysteine residues,[5] unnatural amino acids like p-acetylphenylalanine[6] or selenocysteine.[7] Another trend for site-specific bioconjugation is based on enzymatic ligation with microbial transglutaminase[8] for example. It is also possible to combine enzymatic ligation and the use of unnatural amino acids with an aldehyde tag strategy introducing a formylglycine residue that can be used for conjugation.[9] Although efficient, these strategies require mAb modification prior to bioconjugation which lead to increased cost and time of development. The potential number of relevant antibodies for ADC programs entails the need of technologies compatible with any off-the-shelf antibodies. One way to circumvent this hurdle is to use the rebridging of interchain disulfide bridges of the antibody with a bis-sulfone,[10] a pyridazinedione,[11] a divynilpyrimidine,[12] a 1,3-diacrylamide-1,3,5-triazinane[13] or a next-generation maleimide head of bioconjugation.[14−16] The latter will be described in this publication and permits a controlled DAR of 4 (for IgG1, possessing four disulfide bridges) either as full mAb (LHHL) or as half antibodies (LH) maintained together by weak bonds (Scheme ).

Scheme 1

Site-Specific Bioconjugation Using Rebridging Technology

The first step is the reduction of interchain disulfide bridges; then, the rebridging step is achieved by using a linker with a next-generation maleimide bioconjugation head.

The maleimide is often used to produce ADCs but it can present a loss of drug in the bloodstream through a retro-Michael reaction[17] as observed in Adcetris. In contrast, substituted maleimides such as dithiomaleimide (DTM) or dibromomaleimide (DBM) enable better stability.[16] Several substitution patterns were described in the literature[14,18] and we sought to explore the impact of this substitution in term of bioconjugation reaction and final ADC characteristics. In this work, we describe the synthesis of three linkers with different heads of bioconjugation: DBM, DTM, and mixed thio-bromomaleimide (TBM) heads. Corresponding ADCs were synthesized and evaluated both physicochemically and biologically.

Results and Discussion

Synthesis of Linkers

Linkers were designed to be sensitive to cathepsin B (Val-Cit-PABC trigger) and carry the monomethyl auristatin E (MMAE) cytotoxic payload, a representative combination of ADC linkers currently in clinical trials. In order to obtain the final linkers, bioconjugation heads were synthesized as caproic acid derivatives to allow them to be coupled to the commercial Val-Cit-PABC-MMAE motif.

The reaction between maleic anhydride and 6-aminocaproic acid in acetic acid to obtain 6-maleimidocaproic acid 1 proceeded nicely (Scheme ). Then, acid 1 was reacted, in a sealed tube, with bromine in acetic acid to obtain dibromomaleimido-aminocaproic acid 2a, which was used to synthesize the two other heads. According to the equivalents of thiophenol used, conversion of compound 2a, in basic conditions, led to both compounds 2b and 2c with excellent to poor yields because of purification difficulties. Finally, a peptide coupling was performed between compounds 2a–c and the commercially available trifluoroacetic acid (TFA) salt of H2N-Val-Cit-PABC-MMAE to afford, after purification by semi-preparative high-performance liquid chromatography (HPLC), the final linkers 3a–c in moderate to good yields.

Scheme 2

Synthesis of 3a–c Next-Generation Maleimide Linkers

Bioconjugation Reaction and Analysis

Trastuzumab (TTZ) was used as an mAb model for our study. It targets HER2 protein and is currently used in Kadcyla, one of the five approved ADCs on the market. It has four disulfide bridges, hence with the linkers 3a–c a DAR of 4 is expected, the optimal DAR value with the MMAE toxin.[10]

The bioconjugation of the final linkers 3a–c with trastuzumab was achieved in phosphate buffer saline (PBS) or borate buffer saline (BBS) after the reduction of its interchain disulfide bridges with TCEP (tris(2-carboxyethyl)phosphine)hydrochloride, a mild-reducing agent, to afford ADCs 4a–c.

The bioconjugation of ADCs 4a and 4b was performed as previously described[16,19] and gave satisfactory average DAR (Table ). Starting from the conditions used for the linker 3a, the bioconjugation of linker 3c allowed the obtention of the ADC 4c but with a low average DAR of 1.2 calculated by hydrophobic interaction chromatography (HIC). Hence, the bioconjugation of 3c was optimized thanks to a systematic screening of parameters in order to obtain an optimal ADC 4c (Scheme ). Primarily, different organic solvents, miscible with aqueous buffers, were tested: dimethyl sulfoxide (DMSO), N,N-dimethylformamide (DMF), and 1,4-dioxane. The use of DMF showed a slight increase of the average DAR with an average DAR of 1.7 in HIC compared to 1.2 with DMSO and 1.3 with 1,4-dioxane. DMF was chosen as the organic solvent for the future reactions. Then, the reaction time with the linker was increased from 1 to 5 h, but no improvement of the average DAR was observed. A reaction at a lower temperature, for example, 4 °C, was tested, leading to a decrease of the average DAR from 1.7 to 1.3. The reaction was not tested at 37 °C because of aggregation and deamidation risks at high temperatures.[20] The temperature was set to 20 °C and the linker equivalents were increased (from 8 to 15). It allowed an improvement of the DAR up to 2.7. Published conditions for bioconjugation of linker 3b consists in a double bioconjugation sequence: the protocol of reduction followed by bioconjugation was done twice.[19] Nevertheless, no positive impact on the average DAR was observed but a greater increase of linker equivalents (up to 30) led to an average DAR of 3.6. However, increasing the linker equivalents without changing its concentration induced a high percentage of organic solvent, which translated in the aggregation of the ADC. Hence, we decided to concentrate the solution of the linker from 1 to 2 mM in order to limit the aggregation, and this alternative permitted to reduce insoluble aggregates while improving the average DAR to 3.7. With these conditions in hand, a last step of optimization was conducted. The reaction time with the linker was increased (from 1 to 3 h) and the ADC 4c was obtained with an optimal average DAR of 4.0 without insoluble aggregates (Table ).

Table 1

Optimized Bioconjugation Conditions for Each Linker

ADC	linker	buffer, solvent	conditions of reduction	conditions of reaction with the linker	average DAR	yield (%)
4a	3a	PBS pH = 8.3, DMF	8 equiv TCEP (1 mM); 37 °C; 2 h	7.5 equiv linker (1 mM); 20 °C; 1 h; magnetic stirring	4.1	75
4b	3b	BBS pH = 8.0, DMSO	two reductions	two additions of linker	3.3	61
			1st: 6 equiv TCEP (1 mM); 37 °C; 1 h 15 min	1st: 6 equiv linker (1 mM); 4 °C; 16 h
			2nd: 3 equiv TCEP (1 mM); 4 °C; 1 h 15 min	2nd: 3 equiv linker (1 mM); 4 °C; 22 h
4c	3c	PBS pH = 8.3, DMF	10 equiv TCEP (1 mM); 37 °C; 1 h 45 min	30 equiv linker (2 mM); 20 °C; 3 h	4.0	72

Scheme 3

Average DAR Measured by HIC during the Bioconjugation Optimization of Linker 3c

To summarize, linker 3a required a shorter time of reaction (only 3 h) but needed magnetic stirring to get a satisfactory average DAR, whereas linker 3b requires almost 2 days of reaction to achieve a lower average DAR. It is noted that shorter times of reaction are correlated with higher temperatures. Double addition of TCEP and a linker was beneficial only for ADC 4b. Furthermore, the nature of the organic solvent was important. Indeed, the use of DMSO with bromide containing linker 3c was detrimental for the average DAR. ADC 4c was obtained with a short time of reaction but a higher quantity and an increased concentration of linker were needed to get the targeted average DAR of 4.0 (Table ).

The optimization of these bioconjugations has permitted to obtain the three ADCs 4a–c with a good percentage of DAR 4 species, a satisfactory average DAR, and good yields. We compared the DAR distribution of ADCs 4a–c using HIC analysis (Figure a). The profile of ADC 4b shows a higher heterogeneity of species than 4a and 4c. These latter ADCs (4a and 4c) are more homogeneous with a high proportion of DAR 4 (71 and 81% respectively) and with less than 5% of poorly loaded species (below DAR 3). However, ADC 4a has a higher percentage of DAR 5, which could be explained by a less-efficient rebridging or an additional noncovalent interaction between a linker and the covalent DAR 4 ADC.

Figure 1

Characteristics of each ADC. (a) DAR repartition of each ADC by HIC. Proportions are expressed as percentage of area under the curve (AUC). (b) Monomeric percentage of each ADC by SEC. (c) Repartition of species found by ESI–MS analysis. ND = not detected.

It was also noted from the electrospray ionization–mass spectrometry (ESI–MS) analysis (see Figure S8 in the Supporting Information) that despite the expected same resulting ADCs, the rebridging of ADC 4b is less effective (Figure c). Indeed, it was noticed that this ADC exhibited a lot of different fragments, after the denaturation induced by the ionization, with various loadings. Different fragments were observed, LHHL DAR 2-3-4, LHH DAR 0-1-2-3-4, and LH DAR 1-2, leading to lower average DAR per fragments. These results are consistent with the lower average DAR observed by HIC. ADCs 4a and 4c have similar ESI–MS profiles with less fragments found. It is noteworthy that a majority of LHHL DAR 4 and LH DAR 2 species (see Figures S7 and S9 in the Supporting Information) were obtained. The presence of LH DAR 2 could be explained by an intrachain instead of the interchain rebridging (Scheme ).[21]

Finally, we compared the proportion of monomer of each ADC with an size-exclusion chromatography (SEC) analysis (Figure b). ADCs 4b and 4c have a high percentage of monomer, which implies that the bioconjugation conditions used to produce them were mild and that they do not possess a strong propensity for aggregation. For ADC 4a, more high-molecular-weight species are observed (see Figure S2 in the Supporting Information) which can be due to the mechanical stirring during bioconjugation.

Stability of ADCs

Tests of stability, upon storage in PBS pH 7.4 without formulation, were performed for each ADC for 4 weeks at 4 °C, after purification by size exclusion using a PD-10 desalting column. Critical attributes evaluated were the average DAR, the percentage of DAR 4, and of monomeric species. First, the stability of the average DAR was evaluated by HIC (Figure a) and it was constant for all ADCs during these 4 weeks.

Figure 2

Stability of each ADC upon storage in PBS pH 7.4 without formulation for 4 weeks at 4 °C. (a) Stability by HIC in terms of average DAR for each ADC. (b) Stability by SEC in terms of percentage of monomer for each ADC. (c) Stability by HIC in terms of DAR species for 4a. (d) Stability by HIC in terms of DAR species for 4b. (e) Stability by HIC in terms of DAR species for 4c.

The proportion of each species was analyzed for the ADCs 4a–c (Figure c–e). It was notable that even if the average DAR is steady, the proportion of the different species varied over time. ADCs 4a and 4b showed less variations than ADC 4c. Indeed, ADC 4c presented a rapid decrease of its DAR 4 in favor of DAR 5. The nature of this DAR 5, observed with several different bioconjugation heads,[13,19,22] still needs a full understanding and characterization. The proportions of each DAR species appeared then to be the least steady over time for ADC 4c.

As shown previously, the percentage of monomer of the ADCs 4a–4c is high and no major modification of the aggregation was observed over 4 weeks (Figure b).

Proteins are highly complex molecules necessitating extensive formulation development, which is even more tricky when a hydrophobic payload is conjugated. The formulation is different for each commercial ADC; however, the trend is for slightly acidic buffer of low ionic strength with polysorbates.[23] Our stability profile in PBS 1× Gibco is satisfying considering the pH of 7.4, the ionic strength superior at 150 mM, and the lack of stabilizing polysorbates.

ELISA Assays

As expected,[14] enzyme-linked immunosorbent assay (ELISA) analysis (Figure c) demonstrated no difference in HER2-affinity for the three ADCs; similar Kd values (Figure d) were obtained in the same range as the native TTZ and the commercial T-DM1, used as controls.

Figure 3

In vitro assays for each ADC. Cytotoxicity assays on (a) MCF-7 (HER2 negative cell line) and (b) BT-474 (HER2 positive cell line). (c) HER2-affinity by ELISA analysis. (d) Values of Kd (nM) on HER2 antigen and IC50 on MCF-7 and BT-474 cell lines. NE = not evaluable.

Cytotoxicity Assays

The cytotoxicity of the ADCs was evaluated in vitro on MCF-7 (HER2 negative cell line from adenocarcinoma) and BT-474 (HER2 positive cell line from ductal carcinoma). The ADCs were compared to native TTZ, clinical treatment T-DM1, and MMAE (Figure a,b). The three ADCs demonstrated no cytotoxicity on MCF-7 cell lines, whereas the MMAE alone had a powerful cytotoxic effect on these cells (IC50 = 0.35 nM). ADCs 4a–c demonstrated excellent cytotoxicity on the BT-474 cell line, even better than the clinical treatment T-DM1 at lower concentrations. The cytotoxicity of ADCs 4a and 4c is equivalent to the control MMAE; the cytotoxicity of ADC 4b is slightly less potent, which could be imputed to its lower average DAR. However, IC50 values for ADCs 4a–c are all in a low nanomolar range (Figure d). These results demonstrate that our ADCs are specific to HER2-positive cells and at least as efficient in vitro as the clinical reference T-DM1.

Conclusions

The purpose of this work was to design a new rebridging maleimide head having a combination of advantages of DBM and DTM heads without their drawbacks. The DBM permitted a high percentage of the wanted DAR 4 with an ease of implementation of the bioconjugation but is also associated with a higher percentage of DAR 5. As for the DTM, it allowed a diminution of the DAR 5 species but the overall distribution is more heterogeneous. Its bioconjugation needed a longer time of reaction than DBM. The DTM brick led to a better percentage of monomer than DBM probably because of a milder stirring during the bioconjugation reaction. Our hypothesis was that the designed TBM brick would combine the properties and advantages of the DBM and DTM. Thanks to this design, a high percentage of DAR 4 (even greater than with DBM) was obtained in a short reaction time as for the DBM. No aggregation was observed with the TBM head either. This is in agreement with the hypothesis that the bromide is a better leaving group than the thiophenolate, probably because of its reduced steric hindrance. Even if, on small proteins, it is described that the kinetic is similar for both,[24] the more hindered antibody might react faster with a small leaving group. Hence, we think that the bromide reacts first and ensures good reactivity, which reduces considerably the reaction time of the bioconjugation process. In a second time, the slower reaction of the hindered thiophenol part permitted to temper the reactivity of the head in order to increase the proportion of DAR 4 and to keep a high percentage of monomer. All three ADCs conserved good HER2 binding and a high cytotoxic effect on HER2-positive cell lines only. Taken together, these results show that the TBM brick is an attractive alternative scaffold for rebridging bioconjugation.

Material and Methods

Starting materials for the linker synthesis were purchased from commercial suppliers in the highest purity grade available. TFA.H2N-Val-Cit-PABC-MMAE was obtained from Levena Biopharma. Acros Organics and Carlo Erba were respectively providers of dry and classical solvents. The Hospital Pharmacy of the Tours Teaching Hospital supplied Trastuzumab (Herceptin, Genentech) and T-DM1 (Kadcyla, Roche). The 1H (300 MHz) and 13C (75 MHz) NMR spectra were recorded with a Bruker UltraShield 300 spectrometer. Chemical shifts are reported in parts per million (ppm, δ), and are referenced to solvent signal. Coupling constants are reported in hertz (Hz). Semipreparative HPLC was carried out on a Gilson PLC 2050 system ARMEN V2 (pump) equipped with ECOM TOYDAD600 (UV) for UV detection at 254 nm at 25 °C; a Waters XBridge C-18; 5 μm (250 mm × 19.00 mm) column was used; compounds were eluted with 0.1% TFA in water (solvent A), and acetonitrile (solvent B); as a gradient from 20 to 100% B over 32 min then 100% B for 6 min at 17.1 mL/min. Flash purifications were carried out by chromatography on silica gel columns on an ISCO purification unit, Combi Flash RF 75 PSI, with Redisep flash silica gel columns (60 Å, 230–400 mesh, grade 9385).

Mass Analysis

High-resolution accurate mass (HRAM) measurements for small molecules were carried out in positive mode with an ESI source on a Q-TOF mass spectrometer (accuracy tolerance: 2 ppm) by the “Fédération de Recherche” ICOA/CBM (FR2708) platform.

Mass spectrometric analysis of ADCs was carried out by the “Fédération de Recherche” ICOA/CBM (FR2708) platform as previously described.[19]

HIC Analysis

Samples of ADCs were diluted to 1 mg/mL with PBS pH 7.4 and filtered on 0.22 μm. Loading of 50 μg of each ADC onto a MAbPac HIC-butyl, 5 μm, 4.6 × 100 mm from Thermo Scientific, connected to a Waters Alliance (e2695) equipped with a photodiode array detector (2998) set for detection at 280 nm. Samples were run with a linear gradient from 100% buffer A [1.5 M ammonium sulfate, 50 mM sodium phosphate, 5% isopropanol (v/v), pH 7.0] to 20% buffer B [50 mM sodium phosphate, 20% isopropanol (v/v), pH 7.0] for over 2 min then to 85% buffer B over 30 min and held for 1 min at a flow rate of 1 mL/min. The column temperature was maintained at 25 °C.

SEC Analysis

Samples of ADCs were diluted to 1 mg/mL with PBS pH 7.4 and filtered on 0.22 μm; 50 μg was injected onto an AdvanceBio SEC 2.7 μm, 7.8 × 300 mm from Agilent Technologies, connected to a Waters Alliance (e2695) equipped with a photodiode array detector (2998) set for detection at 280 nm. Samples were eluted with an isocratic gradient (1 mM potassium phosphate monobasic, 155 mM sodium chloride, 3 mM sodium phosphate dibasic, 3 mM sodium azide, pH 7.0) over 24 min at a flow rate of 1 mL/min. The column temperature was maintained at 25 °C.

HER2 binding of ADCs (from 0.01 to 100 μM) was performed as previously described.[25]

Cells were obtained from American Type Culture Collection. A frozen stock vial of cells was thawed quickly in a 37 °C water bath and cells were washed twice with complete growth medium (DMEM/F12 supplemented with 8% fetal bovine serum, 100 μg/mL l-glutamine, 100 μg/mL penicillin G sodium, 100 μg/mL streptomycin sulfate) and plated at a density of at least 10 000 cells/cm2 in 150 cm2 cell culture flasks. Cells were maintained at 37 °C in a humidified atmosphere with 5% CO2 for at least 1 week. Then, cells were harvested and seeded in 96-well plates at a density of 1.25 to 2.5 × 103 cells/well for cytotoxicity assays. Cells were incubated 48 h at 37 °C prior to addition of the test molecules and selected comparison compounds and vehicle (if DMSO, final concentration did not exceed 0.5%).

Test drugs were added at desired final concentrations and further incubated for 72 h (±2 h).

Final concentrations of ADCs, control groups, and MMAE in culture wells will be 225, 75, 25, 8.33, 2.78, 0.926, 0.309, 0.103, 0.034, 0.011 nM.

Cell viability was assessed at cells’ incubation (D-2), before drugs’ addition (D0) and 72 h (±2 h) after drugs’ addition by measuring ATP cell content using CellTiter-Glo Luminescent Cell Viability Assay (Promega) according to the manufacturer’s instructions. Luciferase activity was measured on a luminometer (PerkinElmer EnVisionTM).

Each concentration of compounds was tested in triplicate and two different experiments were performed.

Chemical Synthesis

Compound 1

Maleic anhydride (203 mg; 2.07 mmol; 1.0 equiv) and 6-aminocaproic acid (272 mg; 2.07 mmol; 1.0 equiv) were solubilized in acetic acid (2.0 mL) and stirred at reflux for 16 h. Acetic acid was co-evaporated with toluene under reduced pressure. The resulting solid was dissolved in dichloromethane, washed twice with brine, dried over magnesium sulfate, and concentrated under reduced pressure. Compound 1 (307 mg; 67%) was obtained as an orange solid.

1H NMR (300 MHz, CDCl3) spectrum was in accordance with previous literature.[15]

Compound 2a

Compound 1 (200 mg; 0.95 mmol; 1.0 equiv) and sodium acetate (171 mg; 2.08 mmol; 2.2 equiv) were solubilized in acetic acid (2.7 mL) in a tube at 0 °C. Bromine (107 μL; 2.08 mmol; 2.2 equiv) was added, the tube was sealed, and the mixture was stirred at 135 °C for 4 h. After cooling down to 0 °C, water (10 mL) was added and the solution was extracted with ethyl acetate (3 × 15 mL). Organic phases were combined, washed with sodium thiosulfate (2 × 15 mL), dried over magnesium sulfate, and concentrated under reduced pressure. Residual acetic acid was co-evaporated with toluene under reduced pressure. After purification by flash chromatography (SiO2, cyclohexane/ethyl acetate, 40:60) compound 2a (273 mg; 78%) was obtained as a yellow solid.

1H NMR (300 MHz, DMSO-d6) spectrum was in accordance with previous literature.[15]

Compound 2b

Under inert atmosphere, thiophenol (31 μL; 0.232 mmol; 2.2 equiv) was dissolved in dry dichloromethane (2 mL) with sodium acetate (26 mg; 0.242 mmol; 2.3 equiv). After cooling down to 0 °C, a solution of compound 2a (45 mg; 0.105 mmol; 1.0 equiv) in dry dichloromethane (0.4 mL) was added dropwise. The mixture was stirred for 30 min; then, the reaction was quenched with hydrochloric acid 1 M at 0 °C until pH 2 was reached. The solution was extracted with dichloromethane (3 × 10 mL). Organic phases were combined, dried over magnesium sulfate, and concentrated under reduced pressure. The resulting orange oil was solubilized in cyclohexane. After purification by flash chromatography (SiO2, dichloromethane/methanol, 94:6) compound 2b (49 mg; 94%) was obtained as a yellow solid.

1H NMR (300 MHz, CDCl3) spectrum was in accordance with previous literature.[14]

Compound 2c

Under inert atmosphere, compound 2a (50 mg; 0.14 mmol; 1.0 equiv) was dissolved in methanol (0.6 mL) and sodium acetate (12.1 mg; 0.15 mmol; 1.05 equiv) was added to the mixture. Then, a solution of thiophenol (15.9 μL; 0.16 mmol; 1.1 equiv) in methanol (0.2 mL) was added dropwise. The resulting mixture was stirred at room temperature for 3 h. After cooling down to 0 °C, the reaction was quenched with hydrochloric acid 1 M until pH 2 was reached. The solution was extracted with ethyl acetate (3 × 10 mL). Organic phases were combined, dried over magnesium sulfate, and concentrated under reduced pressure. A first purification by chromatography was realized (SiO2, dichloromethane/methanol, 95:5). The resulting residue was solubilized in DMF and purified by semipreparative HPLC (tR = 30.4 min) to obtain compound 2c (18.9 mg; 35%) as a yellow oil.

1H NMR (300 MHz, CDCl3): δ (ppm) 7.61–7.51 (m, 2H), 7.50–7.34 (m, 3H), 3.53 (t, J = 7.2 Hz, 2H), 2.34 (t, J = 7.4 Hz, 2H), 1.72–1.51 (m, 4H), 1.39–1.28 (m, 2H).

13C NMR (75 MHz, CDCl3): δ (ppm) 178.01, 167.00, 155.30, 135.04, 132.07 (2 C), 130.14, 129.42 (2 C), 129.13, 128.55, 39.62, 33.55, 28.24, 26.16, 24.17.

Compound 3a

Under inert atmosphere, in the dark, compound 2a (4.4 mg; 0.012 mmol; 1.5 equiv) was dissolved in dry acetonitrile (218 μL); then, N-ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline (EEDQ; 3.2 mg; 0.013 mmol; 1.6 equiv) was added. The mixture was stirred at room temperature for 30 min before the addition of TFA.Val-Cit-PABC-MMAE (10 mg; 0.008 mmol; 1.0 equiv) in solution in DMF (218 μL) followed by N,N-diisopropylethylamine (1,4 μL; 0.008 mmol; 1.0 equiv). The resulting solution was stirred at room temperature for 1 h before dilution by two with DMF and purification by semipreparative HPLC (tR = 25.3 min) to give compound 3a (6.0 mg; 50%) as a white solid.

1H NMR (300 MHz, DMSO-d6): δ (ppm) 10.00 (s, 1H), 8.10 (d, J = 7.4 Hz, 1H), 7.90 (d, J = 8.6 Hz, 1H), 7.82 (d, J = 8.6 Hz, 1H), 7.32–7.15 (m, 7H), 5.99–5.95 (m, 1H), 5.41 (s, 2H), 5.11–4.94 (m, 2H), 4.77–4.60 (m, 1H), 4.49–4.36 (m, 2H), 4.29–4.16 (m, 2H), 4.02–3.90 (m, 2H), 3.25–3.12 (m, 10H), 3.02–2.92 (m, 4H), 2.88–2.83 (4H), 2.78–2.71 (m, 2H), 1.52–1.45 (9H), 1.30–1.14 (m, 11H), 1.06–0.93 (m, 9H), 0.93–0.72 (m, 28H).

HRAM (ESI): calcd m/z for C68H104Br2N11O15 [M + H]+, 1472.6074; 1472.6064 observed.

Compound 3b

Under inert atmosphere, compound 2b (2.6 mg; 0.006 mmol; 1.5 equiv) was dissolved in dry (83 μL). Then, hexafluorophosphate azabenzotriazole tetramethyl uronium (HATU; 3.0 mg; 0.008 mmol; 2.0 equiv) and 2,6-lutidine (1.67 μL; 0.014 mmol; 3.5 equiv) were added and the mixture was stirred at room temperature for 10 min. TFA.Val-Cit-PABC-MMAE (5.0 mg; 0.004 mmol; 1.0 equiv) was added and the resulting solution was stirred at room temperature for 24 h before dilution by two with dimethylsulfoxide and purification by semipreparative HPLC (tR = 22.5 min) to give compound 3b (5.0 mg; 81%) as a yellow solid.

1H NMR (300 MHz, DMSO-d6) spectrum was in accordance with the previous report of the compound.[19]

HRAM (ESI): calcd m/z for C80H113N11O15S2 [M + H]+, 1532.7932; 1532.7920 observed.

Compound 3c

Under inert atmosphere, in the dark, compound 2c (2.6 mg; 0.007 mmol; 1.5 equiv) was dissolved in dry acetonitrile (120 μL); then, EEDQ (1.8 mg; 0.007 mmol; 1.6 equiv) was added. The mixture was stirred at room temperature for 30 min before the addition of TFA.Val-Cit-PABC-MMAE (5.5 mg; 0.004 mmol; 1.0 equiv) in solution in (120 μL) followed by N,N-diisopropylethylamine (0,8 μL; 0.004 mmol; 1.0 equiv). The resulting solution was stirred at room temperature for 1 h before dilution by two with and purification by semipreparative HPLC (tR = 21.1 min) to give compound 3c (3.8 mg; 57%) as a yellow oil.

1H NMR (300 MHz, DMSO-d6): δ (ppm) 10.00 (s, 1H), 8.16–8.04 (m, 2H), 7.94–7.89 (m, 1H), 7.82 (d, J = 9.1 Hz, 2H), 7.69–7.62 (m, 1H), 7.61–7.53 (m, 3H), 7.51–7.38 (m, 3H), 7.29–7.23 (m, 4H), 7.17–7.10 (m, 1H), 6.01–5.92 (m, 2H), 5.59–5.45 (m, 6H), 5.15–5.87 (m, 4H), 4.86–4.57 (m, 4H), 4.49–4.38 (m, 4H), 4.33–4.12 (m, 4H), 4.03–3.78 (m, 4H), 3.53–3.43 (m, 4H), 3.26–3.10 (m, 15H), 3.02–2.79 (m, 7H), 2.75–2.69 (m, 1H), 2.30–1.91 (m, 15H), 1.75–1.38 (m, 9H), 1.25–1.21 (m, 3H), 1.10–0.92 (m, 5H), 0.81 (m, 12H).

HRAM (ESI): calcd m/z for C74H108BrN11O15 [M + H]+, 1502.7003; 1502.6998 observed.

ADC Synthesis

PBS buffer preparation: 5 mL of 10× Gibco PBS was dissolved in deionized water (45 mL) in a volumetric flask. Sodium chloride (73.0 mg, 25 mM) and ethylenediaminetetraacetic acid (18.6 mg, 1 mM) were added to the solution. The pH was adjusted to 8.3 with sodium hydroxide 0.1 M.

BBS buffer preparation: Na2B4O7.10H2O (528.2 mg) was dissolved in deionized water (27.7 mL). Sodium chloride (62.9 mg, 25 mM) and ethylenediaminetetraacetic acid (16.0 mg, 1 mM) were solubilized in 22.3 mL of hydrochloric acid 0.1 M; then, the two solutions were mixed together. The pH was adjusted to 8.0 with hydrochloric acid 0.1 M.

Synthesis of ADC 4a

To trastuzumab (5.0 μg/μL, 250 μg, 50 μL) in PBS buffer pH 8.3 was added TCEP (1 mM, 13.5 μL, 8 equiv) and the reaction was incubated at 37 °C for 2 h. Linker 3a in DMF (1 mM, 12.7 μL, 7.5 equiv) was added under magnetic stirring and the resulting solution was mixed at 20 °C for 1 h (600 rpm). The crude ADC was purified by a PD-10 desalting column (Sephadex G-25 M, GE HealthCare) in PBS Gibco pH 7.4 to afford the desired ADC 4a (75%).

Synthesis of ADC 4b

To trastuzumab (5.0 μg/μL, 250 μg, 50 μL) in BBS buffer pH 8.0 was added TCEP (1 mM, 10.1 μL, 6 equiv) and the reaction was incubated at 37 °C for 1 h 15 min. Linker 3b in DMSO (1 mM, 10.1 μL, 6 equiv) was added and the resulting solution was mixed at 4 °C for 16 h (600 rpm). An iteration of incubation with TCEP (1 mM, 5.0 μL, 3 equiv) at 4 °C for 1 h 15 and mixing with linker 3b (1 mM, 5.0 μL, 3 equiv) in DMSO at 4 °C (600 rpm) for 22 h was done. The crude ADC was purified by a PD-10 desalting column (Sephadex G-25 M, GE Healthcare) in PBS Gibco pH 7.4 to afford the desired ADC 4b (61%).

Synthesis of ADC 4c

To trastuzumab (5.0 μg/μL, 250 μg, 50 μL) in PBS buffer pH 8.3 was added TCEP (1 mM, 16.9 μL, 10 equiv) and the reaction was incubated at 37 °C for 2 h. Linker 3c in DMF (2 mM, 25.3 μL, 30 equiv) was added and the resulting solution was mixed at 20 °C for 3 h (600 rpm). The crude ADC was purified by a PD-10 desalting column (Sephadex G-25 M, GE Healthcare) in PBS Gibco pH 7.4 to afford the desired ADC 4c (72%).