Stable and potent selenomab-drug conjugates

Summary

Selenomabs are engineered monoclonal antibodies with one or more translationally incorporated selenocysteine residues. The unique reactivity of the selenol group of selenocysteine permits site-specific conjugation of drugs. Compared to other natural and unnatural amino acid and carbohydrate residues that have been used for the generation of site-specific antibody-drug conjugates, selenocysteine is particularly reactive, permitting fast, single-step, and efficient reactions under near physiological conditions. Using a tailored conjugation chemistry, we generated highly stable selenomab-drug conjugates and demonstrated their potency and selectivity in vitro and in vivo. These site-specific antibody-drug conjugates built on a selenocysteine interface revealed broad therapeutic utility in liquid and solid malignancy models.

eTOC Blurb

Li et al. harness the high reactivity of the 21st natural amino acid selenocysteine to rapidly and efficiently assemble antibody-drug conjugates that reveal excellent stability, potency, and selectivity in diverse in vitro and in vivo models of human cancers.

Introduction

Antibody-drug conjugates (ADCs) are among the most promising next-generation antibody therapeutics for cancer therapy and are being pursued by an increasing number of biotech and pharma companies (Chari et al., 2014; de Goeij and Lambert, 2016; Drake and Rabuka, 2015; McCombs and Owen, 2015; Polakis, 2016). The concept of ADCs is the targeted delivery of a highly cytotoxic drug for selective (as opposed to systemic) chemotherapy, resulting in a higher therapeutic index (Adair et al., 2012; Sievers and Senter, 2013). The Food and Drug Administration (FDA) approvals of brentuximab vedotin (Adcetris®) for the therapy of Hodgkin lymphoma and anaplastic large cell lymphoma and ado-trastuzumab emtansine (Kadcyla®) for HER2+ breast cancer therapy were milestones that established the therapeutic utility of ADCs. Brentuximab vedotin is a chimeric mouse/human anti-human CD30 mAb in IgG1 format conjugated to monomethyl auristatin E (MMAE). MMAE is a synthetic analogue of the antimitotic pentapeptide dolastatin 10, isolated from the marine mollusk Dolabella auricularia, which inhibits cell division at subnanomolar concentrations by blocking tubulin polymerization. Each MMAE drug is linked to the antibody through an enzymatically cleavable linker (Doronina et al., 2003). The linker also contains a maleimide group that, affording a succinimide thioether, reacts with the thiol (SH) groups of up to eight reduced cysteines (Cys) that are normally engaged in four interchain disulfide bridges in the IgG1 hinge region. Random Cys conjugation results in an ADC mixture having a drug-to-antibody ratio (DAR) ranging from 0:1 to 8:1, with an average ratio of 4:1 (Senter, 2009). Ado-trastuzumab emtansine is based on the FDA-approved humanized anti-human HER2 mAb trastuzumab (Herceptin®) randomly conjugated to an average of 3.5 maytansine drugs through the ε-amino group of lysines (Lys) and a non-cleavable linker (Lewis Phillips et al., 2008; LoRusso et al., 2011). Like MMAE, maytansine blocks tubulin polymerization. In addition to these two FDA-approved ADCs, several dozen ADCs are currently in clinical trials and the majority are based on random Cys or Lys conjugation (Beck and Reichert, 2014; Chari et al., 2014; Mullard, 2013).

The random conjugation of the drug to Cys and Lys residues in the antibody results in a complex mixture of ADC species with different pharmacokinetic (PK) and pharmacodynamic (PD) properties (Boswell et al., 2011; Hamblett et al., 2004; Jackson et al., 2014). For example, ADC species with high DARs are less soluble and less stable than ADC species with low DARs but can potentially deliver more drugs. In addition, heterogeneous ADCs require complex mixture analyses to warrant batch-to-batch consistency. Homogeneous ADCs, by contrast, have a defined DAR with distinct pharmacological properties and, ideally, no batch-to-batch variability (Agarwal and Bertozzi, 2015; Hu et al., 2016; Panowski et al., 2014). Site-specific conjugation technologies have facilitated the manufacturing of homogenous ADCs. Side-by-side studies of conventional heterogeneous ADCs and next-generation homogeneous ADCs demonstrated that they are similar in potency but that site-specific conjugation can improve safety (Junutula et al., 2010; Junutula et al., 2008). In other words, the therapeutic index, which compares the concentration at which an ADC becomes toxic to the concentration at which it is effective, is often higher for homogeneous as compared to heterogeneous ADCs, permitting higher dosing.

Several site-specific conjugation technologies for the generation of ADCs have been developed in recent years (Behrens and Liu, 2014). These methods are based on modifying natural or engineered amino acid or carbohydrate residues in the antibody component by chemical or enzymatic reactions. Chemical site-specific conjugation of drug molecules requires the introduction of a unique chemical reactivity in the antibody, which is displayed by engineered natural, generally Cys (Casi et al., 2012; Dornan et al., 2009; Jeffrey et al., 2013; Junutula et al., 2010; Junutula et al., 2008), or engineered unnatural amino acids (Axup et al., 2012; Jackson et al., 2014; Tian et al., 2014; Zimmerman et al., 2014) or, alternatively, by engineered carbohydrates (Okeley et al., 2013; Zhou et al., 2014; Zhu et al., 2014). In certain cases, the unique chemical reactivity is generated enzymatically (Drake et al., 2014; Rabuka et al., 2012). By contrast, enzymatic site-specific conjugation of drugs, for example by transglutaminase (Jeger et al., 2010; Strop et al., 2013) or sortase (Beerli et al., 2015), does not necessitate the display of unique chemical reactivity, but simply the presence of engineered peptide sequences.

Antibodies with engineered reactive Cys residues are known as thiomabs and the corresponding ADCs as thiomab-drug conjugates (Panowski et al., 2014). Thiomab-drug conjugates established that homogeneous ADCs have a higher therapeutic index than heterogeneous ADCs (Junutula et al., 2010; Junutula et al., 2008). Subsequent studies revealed that the stability of thiomab-drug conjugates was inversely related to the solvent accessibility of the engineered Cys residues (Shen et al., 2012). Similar in concept to thiomabs, we have developed a site-specific conjugation technology based on antibodies with engineered selenocysteine (Sec) residues (Cui et al., 2012; Hofer et al., 2009; Hofer et al., 2008; Vire et al., 2014), which we named selenomabs.

We originally developed selenomabs for the generation of chemically programmed antibodies (Cui et al., 2012; Hofer et al., 2008), which are molecularly defined covalent compositions of a variable small molecule component that serves as a targeting moiety with an invariable antibody component that serves as a carrying moiety (Rader, 2014). Using drug surrogates, we subsequently provided evidence that selenomabs may be suitable for the generation of homogeneous ADCs (Hofer et al., 2009; Li et al., 2015; Li et al., 2014; Patterson et al., 2014; Thomas et al., 2012). In eukaryotes, Sec (also known as the 21st natural amino acid) (Hatfield and Gladyshev, 2002) is encoded by the stop codon UGA and its translational incorporation requires the presence of a Sec incorporation sequence (SECIS) in the 3′ untranslated region (UTR) of the mRNA. The Sec incorporation machinery plays an essential role in mammalian cells by governing the expression of ~25 natural selenoproteins (Labunskyy et al., 2014; Reeves and Hoffmann, 2009) and can be harnessed for the expression of recombinant selenoproteins. Due to the different atomic structure and composition of selenium (³⁴Se, 79 Da) compared to sulfur (¹⁶S, 32 Da), the selenol (SeH) group of Sec (pKa 5.2) is more nucleophilic than the thiol group of Cys (pKa 8.3). Taking advantage of their different chemical reactivities, we showed that under mildly acidic (pH 5.2) and reducing (0.1 mM dithiothreitol (DTT)) conditions small molecules having electrophilic functionalities can be selectively conjugated to engineered C-terminal Sec residues without modifying Cys residues or any other amino acids or carbohydrates in selenomabs (Hofer et al., 2009; Hofer et al., 2008). Indeed, corresponding thiomabs with engineered C-terminal Cys residues are not conjugated under these conditions (Hofer et al., 2008). Consequently, selenomab conjugations avoid the extra step of reformation of disulfide bridges in the antibody required for thiomab conjugation (Hofer et al., 2008; Li et al., 2014). It should be noted that due to the competition of Sec incorporation and termination at the UGA codon, our first-generation selenomabs in IgG or scFv-Fc format contain a heavy chain heterodimer with one rather than two C-terminal Sec residues (Hofer et al., 2009). Compared to other natural and unnatural amino acid and carbohydrate residues that have been used for the generation of site-specific ADCs, Sec is particularly reactive, permitting fast, single-step, and complete reactions under near physiological conditions in the absence of catalysts. In the current study we demonstrate that selenomabs can be utilized to generate selenomab-drug conjugates having excellent stability, potency, and selectivity in diverse in vitro and in vivo models of human cancers.

Results

Human plasma stability of selenomab-fluorescein conjugates linked through either maleimide or iodoacetamide

To find suitable linker functionalities for making selenomab-drug conjugates, we first compared the human plasma stability of selenomab-fluorescein conjugates based on either maleimide-selenol or iodoacetamide-selenol adducts. Maleimide and iodoacetamide derivatives of fluorescein were conjugated to humanized anti-human HER2 mAb trastuzumab in scFv-Fc format with a C-terminal Sec residue (Li et al., 2015) using optimized conjugation conditions reported previously (Li et al., 2014) and the resulting selenomab-fluorescein conjugates were incubated in human plasma at 37° C. The maleimide-selenol conjugate at this site was unstable in human plasma and revealed a rapid exchange reaction with human serum albumin (HSA) (Figure 1). This exchange reaction between a succinimide selenoether and a succinimide thioether likely proceeded through a retro-Michael/Michael reaction sequence (Alley et al., 2008; Shen et al., 2012). After 8 h, only traces of the scFv-Fc-Sec/maleimide-fluorescein conjugate were still detectable (Figure 1). In marked contrast, ~80% of the scFv-Fc-Sec/iodoacetamide-fluorescein conjugate remained intact after 10 days (Figure 1). Based on this finding, we chose iodoacetamide for drug attachment to selenomabs.

Figure 1. Human plasma stability of selenomab-fluorescein conjugates.

Anti-HER2 scFv-Fc-Sec/maleimide-fluorescein versus anti-HER2 scFv-Fc-Sec/iodoacetamide-fluorescein conjugate stability in human plasma at 37° C. (A) Fluorescent (top) and Coomassie stained (bottom) SDS-PAGE gels are shown. Antibody and HSA bands are indicated by arrows. Molecular weights from a pre-stained protein ladder are shown on the left. (B) Summary of selenomab-fluorescein conjugate stability from 3 independent experiments (mean ± SD) with band intensities quantified by NIH ImageJ software. Note that 100% remaining maleimide conjugate refers to the intensity of the band at time point “0” in (A), left and 100% remaining iodoacetamide conjugate refers to the intensity of the band at time point “0” in (A), right. (C) Scheme of the reaction between scFv-Fc-Sec and iodoacetamide-fluorescein (top) or maleimide-fluorescein (bottom). See also Figure S2.

Generation of selenomab-drug conjugates and evaluation of their in vitro activity and human plasma stability

We synthesized three monomethyl auristatin F (MMAF) derivative peptides (CN27, CN28, and CN29) with a non-cleavable iodoacetamido-caproyl linker at their N-terminus and variable C-termini (Figure 2A). MMAF differs from MMAE (the drug component of brentuximab vedotin) by having a C-terminal phenylalanine. Following Sec-selective conjugation of the three compounds to the anti-HER2 scFv-Fc-Sec, the in vitro cytotoxicities of the conjugates were measured following incubation with HER2-high breast cancer cell lines (Table S1) SK-BR-3 and KPL-4 in comparison to HER2-low breast cancer cell line (Table S1) MCF-7 for 72 h at 37° C. As shown in Figure 2B and Table S2, all three conjugates revealed potent and specific activity toward HER2-high breast cancer cell lines. The non-targeting control ADC anti-CD79B scFv-Fc-Sec/CN29, which revealed an IC₅₀ of 0.3 nM when tested against CD79B+ B-cell line Ramos (data not shown), was inactive when tested against the breast cancer cell lines (Figure 2B). We also conjugated CN27, CN28, and CN29 to mouse anti-human CD138 mAb B-B4 in scFv-Fc format with or without C-terminal Sec residue under Sec-selective conjugation conditions and tested the in vitro cytotoxicities of these conjugates on CD138 expressing multiple myeloma cell lines U266 and H929. All three anti-CD138 selenomab-drug conjugates showed potent activity with double digit picomolar IC₅₀ values, whereas the antibody without Sec was inactive, validating our Sec-selective conjugation conditions (Figure S1 and Table S3). When the three unconjugated compounds were tested against breast cancer and multiple myeloma cells, CN29 revealed the lowest potency with triple digit nanomolar IC₅₀ values (Table S2 and Table S3), conceivably reflecting a weakened ability to penetrate the plasma membrane as compared to CN27 and CN28. Owing to the approximate 1,500-fold lower activity of its unconjugated compared to its conjugated state, CN29 was chosen for all in vivo studies with the rationale to diminish any systemic activity from premature drug release. To evaluate its stability in human plasma, we compared the cytotoxicity of anti-HER2 scFv-Fc-Sec/CN29 toward HER2-high and HER2-low breast cancer cells before and after incubation in human plasma for 3 days at 37° C. The two samples revealed virtually identical activity (Figure S2), confirming the high stability of our selenomab-drug conjugates.

Figure 2. Drugs and in vitro activity of HER2-targeting selenomab-drug conjugates.

(A) Structural formulas of iodoacetamido-caproyl-MMAF derivatives CN27, CN28, and CN29. (B) Cytotoxicity of anti-HER2 scFv-Fc-Sec conjugated to CN27, CN28, and CN29 following incubation with HER2-high breast cancer cell lines SK-BR-3 and KPL-4, and HER2-low breast cancer cell line MCF-7 for 72 h at 37° C (mean ± SD of triplicates). Anti-CD79B scFv-Fc-Sec/CN29 served as non-targeting control ADC. See also Figure S1, Figure S2, Figure S4, Table S1, Table S2, and Table S4.

Evaluation of the HER2-targeting selenomab-drug conjugate in a human breast cancer xenograft mouse model

Breast cancer xenograft studies were conducted using KPL-4 cells with CD-1 nude mice. Mice bearing established tumors (~100 mm³) were treated every four days with an intravenous (i.v.) injection of 1 mg/kg and 3 mg/kg of anti-HER2 scFv-Fc-Sec/CN29, 3 mg/kg unconjugated anti-HER2 scFv-Fc-Sec, and the benchmark ADC ado-trastuzumab emtansine at 1 mg/kg for a total of four treatments. Significant tumor regression and growth inhibition was observed for anti-HER2 scFv-Fc-Sec/CN29 at both doses (Figure 3). Four of the five mice in the 3 mg/kg group were tumor free at the end of the experiment on day 54, six weeks after the last treatment.

Figure 3. In vivo activity of the HER2-targeting selenomab-drug conjugate.

Human breast cancer cell line KPL-4 was xenografted into the mammary fat pads of female CD-1 nude mice, grown to ~100 mm³, randomized into 5 groups comprising 5 mice each, and treated with i.v. (tail vein) injections of the indicated ADCs and controls four times every four days at the indicated doses. Mean ± SD values are plotted; the p values (t-test) compare the treatment groups to the vehicle alone group.

Evaluation of the CD138-targeting selenomab-drug conjugate in a human multiple myeloma metastasis mouse model

The in vivo efficacy of anti-CD138 scFv-Fc-Sec/CN29 was evaluated in NSG mice that had been injected i.v. with firefly luciferase (ffluc)-expressing human multiple myeloma cell line U266 (U266-ffluc), which secretes a human IgEλ mAb. Tumor bearing mice were treated by i.v. injection every four days for a total of four cycles of 3 mg/kg anti-CD138 scFv-Fc-Sec/CN29 and 3 mg/kg unconjugated anti-CD138 scFv-Fc-Sec. The selenomab-drug conjugate significantly inhibited tumor growth compared to the unconjugated selenomab and vehicle control as measured by human lambda light chain titers (Figure 4A) and in vivo bioluminescence imaging (Figure S3). Accordingly, anti-CD138 scFv-Fc-Sec/CN29 significantly extended the survival time of tumor bearing mice (Figure 4B).

Figure 4. In vivo activity of the CD138-targeting selenomab-drug conjugate.

Human multiple myeloma cell line U266-ffluc was i.v. (tail vein) injected into NSG mice. After 4 weeks, the mice were randomized into 3 groups comprising 8–9 mice each and treated with i.v. (tail vein) injections of 3 mg/kg of the ADC, the unconjugated antibody, and vehicle alone. (A) Human lambda light chain concentrations correlating with tumor burden were determined by ELISA at the indicated time points and plotted as mean ± SD values. The p value (t-test) compares the ADC group to the unconjugated antibody group. (B) Kaplan-Meier survival curve. The p value (log-rank test) compares the ADC group to the unconjugated antibody group. See also Figure S3 and Table S3.

Analytical characterization of selenomab-drug conjugates

The DAR of the ADCs with C-terminal Sec residue was determined by hydrophobic interaction chromatography (HIC)-HPLC. The aggregation levels in unconjugated antibodies and ADCs were determined by both HIC-HPLC and size exclusion chromatography (SEC)-FPLC. Less than 10% of aggregates were found in ADCs by SEC-FPLC (Figure S4 and Table S5). Approximately 60% of both anti-HER2 and anti-CD138 scFv-Fc-Sec/CN29 had the expected DAR=1 whereas 40% remained unconjugated (Figure S4 and Table S5). Presumably due to its low abundance, given the predominantly asymmetric dimerization of scFv-C_H2-C_H3-Sec polypeptides (~10%) with terminated scFv-C_H2-C_H3 polypeptides (~90%) (Hofer et al., 2009; Hofer et al., 2008), a species with a DAR=2 was not detected by HIC-HPLC. Electrospray ionization high resolution mass spectrometry (ESI-HRMS) confirmed that the species with a DAR=1 was the most abundant followed by the species with a DAR=0 for both anti-HER2 and anti-CD138 scFv-Fc-Sec/CN29 (Table S4).

Generation and analytical characterization of a second-generation selenomab-drug conjugate with higher drug-to-antibody ratio

Based on the analytical characterization of the first-generation of selenomab-drug conjugates with a DAR=0.6, we concluded that a second-generation with a DAR=2 would require moving the C-terminal Sec residue to an upstream position that provides higher reactivity and warrants exclusion of terminated scFv-Fc polypeptides from dimerization. We chose Serine 396 (Kabat numbering) for substitution with Sec (Ser396Sec) due to its accessibility in an exposed loop in the N-terminal region of C_H3. Termination at this position results in an scFv-C_H2-(Ser396stop) polypeptide that cannot dimerize with the scFv-C_H2-(Ser396Sec)C_H3 polypeptide. Thus, symmetric scFv-Fc(Ser396Sec) selenomabs purified by Protein G affinity chromatography contain two Sec residues (Figure S5A). Furthermore, previous studies with thiomabs based on Ser396Cys suggested the suitability of Ser396Sec for conjugation (Li et al., 2014; Patterson et al., 2014; Shen et al., 2012). Indeed, Sec-selective conjugation of CN29 to scFv-Fc(Ser396Sec) (Figure S5B) yielded >90% of selenomab-drug conjugate with a DAR=2 as determined by HIC-HPLC and no aggregates were detected by SEC-FPLC (Figure S6A and Table S5). ESI-HRMS confirmed that the species with a DAR=2 was by far the most abundant with no unconjugated selenomab detectable (Figure S6B). A minor peak with an observed mass of 98 Da above the DAR=2 species is likely a DAR=2 species with posttranslational modification(s) or incomplete enzymatic deglycosylation. Similar minor species have also been reported for thiomabs (Bhakta et al., 2013).

Evaluation of the in vitro cytotoxicity and human plasma stability of the second-generation selenomab-drug conjugate

We next compared the in vitro cytotoxicities of anti-HER2 scFv-Fc-Sec/CN29, anti-HER2 scFv-Fc(Ser396Sec)/CN29, anti-HER2 scFv-Fc(Ser396Sec)/MMAF, and ado-trastuzumab emtansine toward human breast cancer cell lines expressing different levels of HER2. As expected, the IC₅₀ value of the second-generation selenomab-drug conjugate with DAR=2 was found to be significantly lower (0.052 nM toward KPL-4 cells and 0.14 nM toward MDA-MB-361/DYT2 cells) than that of the first-generation selenomab-drug conjugate with DAR=0.6 (0.50 nM toward KPL-4 cells and 8.4 nM toward MDA-MB-361/DYT2 cells) and similar to the control conjugate using parental drug MMAF as payload (0.050 nM toward KPL-4 cells and 0.11 nM toward MDA-MB-361/DYT2 cells) as well as the benchmark ADC ado-trastuzumab emtansine (0.066 nM toward KPL-4 cells and 0.12 nM toward MDA-MB-361/DYT2 cells) (Figure 5). Anti-HER2 scFv-Fc(Ser396Sec)/CN29 revealed less off-target cytotoxicity than ado-trastuzumab emtansine toward HER2-low breast cancer cell line MCF-7 and HER2-negative breast cancer cell line MDA-MB-468 (Figure 5). Importantly, iodoacetamide-mediated conjugation of fluorescein (Figure S7A) and CN29 (Figure S7B) at the Ser396Sec sites was found to be as stable in human plasma as previously determined for the C-terminal Sec site.

Figure 5. In vitro activity of the second-generation HER2-targeting selenomab-drug conjugate.

Cytotoxicity of anti-HER2 scFv-Fc-Sec/CN29, anti-HER2 scFv-Fc(Ser396Sec)/CN29, anti-HER2 scFv-Fc(Ser396Sec)/MMAF and ado-trastuzumab emtansine following incubation with HER2-high breast cancer cell lines KPL-4 and MDA-MB-361/DYT2, HER2-low breast cancer cell line MCF-7, and HER2-negative breast cancer cell line MDA-MB-468 for 72 h at 37° C (mean ± SD of triplicates). Anti-CD79B scFv-Fc(Ser396Sec)/CN29 served as non-targeting control ADC. See also Figure S5, Figure S6, Figure S7, and Table S1.

Evaluation of the in vivo activity of the second-generation selenomab-drug conjugate in a human breast cancer xenograft mouse model

Breast cancer xenograft studies were conducted using KPL-4 cells with NOD/SCID/IL-2Rγ^null (NSG) mice. Mice bearing established tumors (~150 mm³) were treated every week with an intravenous (i.v.) injection of 5 mg/kg of anti-HER2 scFv-Fc-Sec/CN29, non-targeting anti-CD79B scFv-Fc-Sec/CN29, anti-HER2 scFv-Fc(Ser396Sec)/CN29, and non-targeting anti-CD79B scFv-Fc(Ser396Sec)/CN29 for a total of four treatments. The second-generation selenomab-drug conjugate showed superior efficacy against tumor growth compared to the first-generation selenomab-drug conjugate (Figure 6A). The non-targeting selenomab-drug conjugates of either generation were inactive (Figure 6A).

Figure 6. In vivo activity and pharmacokinetics of the second-generation HER2-targeting selenomab-drug conjugate.

(A) Human breast cancer cell line KPL-4 was xenografted into the mammary fat pads of female NSG mice, grown to ~150 mm³, randomized into 5 groups comprising 6 mice each, and treated with i.v. (tail vein) injections of the indicated ADCs and controls four times every week at 5 mg/kg. Mean ± SD values are plotted. (B) Four female CD-1 mice were injected i.v. with 6 mg/kg of anti-HER2 scFv-Fc(Ser396Sec)/CN29. The plasma concentrations of total antibody and intact ADC were quantified by ELISA at the indicated time points, using HER2 ECD for capturing and either anti-His mAb (total antibody) or anti-MMAF mAb (intact ADC) for detection. Shown are mean ± SD values for each time point. See also Figure S5, Figure S6, Figure S7, and Table S6.

Pharmacokinetics of second-generation selenomab-drug conjugate

We performed a pharmacokinetic (PK) study with anti-HER2 scFv-Fc(Ser396Sec)/CN29 in mice to examine its in vivo stability and distribution. Female CD-1 mice were injected i.v. with 6 mg/kg of the ADC. Blood samples were withdrawn at indicated time points (Figure 6B) over a period of two weeks and plasma was prepared. The ADC concentration in the plasma was measured with a sandwich ELISA using immobilized HER2 extracellular domain (ECD) for capture and then either anti-His tag mAb or anti-MMAF mAb for detection of total antibody and intact ADC, respectively. The ADC that was administered to the mice was used as standard. The plasma concentrations of the ADC were plotted as μg per mL to allow comparison between the levels of intact ADC and total antibody. The circulating concentrations of intact ADC and total antibody overlapped, revealing high stability of our selenomab-drug conjugates in vivo (Figure 6B). Both sets of data were analyzed for their PK parameters by two-compartment modeling. Table S6 shows that elimination half-life (t_1/2), area under the curve (AUC), clearance (CL), and steady state volume of distribution (V_ss) of intact ADC and total antibody were similar, once again confirming the fitness of iodoacetamide as linker functionality.

Discussion

We established a platform for generating site-specific ADCs with a Sec interface and refer to this novel class of ADCs as selenomab-drug conjugates. Iodoacetamide-based selenomab-drug conjugates were found to have excellent stability in human plasma in vitro and in circulation in mice in vivo. In addition, anti-HER2 and anti-CD138 selenomab-drug conjugates with MMAF derivatives as drug component revealed potent and specific in vitro and in vivo activity in models of human breast cancer and multiple myeloma, respectively. Finally, we showed that positioning Sec in a C_H3 loop of the antibody component affords selenomab-drug conjugates with DAR=2 and enhanced potency both in vitro and in vivo. Collectively, these stable and potent selenomab-drug conjugates have the potential to find wide therapeutic applicability with respect to cancer targets and indications.

The stability of the conjugate plays a major role in the PK and PD properties of ADCs and is influenced by linker composition, conjugation chemistry, and conjugation sites on the antibody component (Gordon et al., 2015; Ross and Wolfe, 2016). Because only a low percentage of an ADC finds its target in vivo, the stability of the linker that connects antibody and drug component is important for the safety of ADCs.

Our previous studies predominantly used maleimide derivatives for Michael addition with Sec under conditions that prevented concurrent conjugation to Cys (Cui et al., 2012; Hofer et al., 2009; Hofer et al., 2008; Thomas et al., 2012; Thomas et al., 2008; Vire et al., 2014). However, similar to maleimide-thiol adducts (succinimide thioethers) at solvent accessible Cys residues (such as Ser396Cys) in the antibody component (Alley et al., 2008; Shen et al., 2012), maleimide-selenol adducts (succinimide selenoethers) rapidly decay in human plasma by undergoing a retro-Michael/Michael reaction sequence that transfers the payload from the antibody to HSA, presumably to its free Cys residue at position 34 (Figure 1) (Patterson et al., 2014). This exchange reaction not only lowers the activity of the ADC due to drug loss but also increases systemic toxicity mediated by both free drug and HSA-conjugated drug. Thus, maleimide is not a suitable electrophilic functionality for conjugating drugs to selenomabs. We recently showed that replacing maleimide with methylsulfone phenyloxadiazole affords highly stable thiomab conjugates at position Ser396Cys (Patterson et al., 2014). In the same study, we also investigated selenomab conjugates and found that methylsulfone phenyloxadiazole linkers are significantly more stable than maleimide linkers. However, the comparison of thiomab (Ser396Cys) and selenomab (C-terminal Sec) conjugates with maleimide and methylsulfone phenyloxadiazole linkers revealed the anticipated lower stability of Sec compared to Cys adducts (Patterson et al., 2014). Therefore, we set out to investigate alternative conjugation chemistries that address the instability of selenomab conjugates. Here we show that selenomab conjugates with iodoacetamide linkers are exceptionally stable regardless of whether the Sec residue is at the C-terminus or at position Ser396Sec.

Iodoacetamide is used routinely as a Cys alkylation agent for peptide mapping (Lundell and Schreitmuller, 1999). However, cross-reactivity with methionine (Met), Lys, and histidine (His) (Mendoza and Vachet, 2009) diminishes the utility of iodoacetamide for selective Cys conjugation, such as for the generation of thiomab-drug conjugates. By contrast, the much higher reactivity of Sec under mildly acidic conditions allows selective Sec conjugation to iodoacetamide in the presence of Cys, Met, Lys, and His. Indeed, as we predicted in a previous study with iodoacetamide derivatives of drug surrogates (Li et al., 2014), we show here that iodoacetamide derivatives of MMAF can be selectively conjugated to selenomabs to provide highly potent selenomab-drug conjugates. Attempted conjugation to the same antibody without Sec residue did not yield active ADCs (Figure S1). Thus, combining the nucleophilic Sec with the electrophilic iodoacetamide functionality affords a uniquely tailored conjugation chemistry for selenomab-drug conjugates.

In addition to augmenting the stability of ADCs, lower toxicity can also be achieved by diminishing the cell permeability of systemically released drugs. Iodoacetamido-caproyl-MMAF derivative CN29, which contains a water soluble polyethylene glycol (PEG) group at the C-terminus of the pentapeptide, was prepared as its trifluoroacetic acid (TFA) salt to further increase its hydrophilicity. When compared side-by-side as free drug with the more hydrophobic iodoacetamido-caproyl-MMAF and its derivatives CN27 and CN28 (Figure 2A and Scheme S3), CN29 is substantially less toxic, yet displays similar potency when conjugated to the HER2- and CD138-targeting selenomabs (Table S2, Table S3, and Figure 5). Thus, ADCs with CN29 as drug component can be anticipated to have higher therapeutic indices than the corresponding ADCs with CN27 or CN28 and are preferred for in vivo studies.

Our previous studies exclusively used first-generation selenomabs with one (Cui et al., 2012; Hofer et al., 2009; Hofer et al., 2008; Li et al., 2015; Patterson et al., 2014; Thomas et al., 2012; Thomas et al., 2008; Vire et al., 2014) or two C-terminal Sec residues (Li et al., 2014). This position mirrors the C-terminal location of the Sec residue in the natural selenoprotein thioredoxin reductase 1 (TXNRD1). The 3′-untranslated region (UTR) of the cDNA of human TXNRD1, which harbors the SECIS element, is used in our selenomab expression cassettes. Thus, our first-generation selenomabs use a natural distance between UGA codon and SECIS element, which was reported to be critical for efficient Sec incorporation (Martin et al., 1996; Turanov et al., 2013; Wen et al., 1998). However, as noted for differently positioned engineered Cys residues in thiomabs (Patterson et al., 2014; Shen et al., 2012), the solvent accessibility of the Sec residue is likely to affect both conjugation efficacy and stability. Our finding that first-generation selenomab-drug conjugates only reach DAR=0.6 when using the iodoacetamido-caproyl-MMAF derivatives as drug components prompted us to move the Sec residue to an upstream position (Ser396Sec) with higher solvent accessibility as shown by correspondingly designed thiomabs (Ser396Cys) (Junutula et al., 2008; Patterson et al., 2014). These second-generation selenomabs had the added advantage of containing two rather than one Sec residue following Protein G affinity chromatography (Figure S5) and indeed revealed higher conjugation efficacy with DAR=2. Second-generation selenomabs also provide additional opportunities for engineering thio-selenomabs which combine engineered Sec with engineered Cys residues for bioorthogonal conjugation to two different drugs (Li et al., 2015). Combining the Ser396Sec mutation with an engineered Cys in the light or heavy chain of antibodies in IgG1 format would afford a dual drug loading platform for a combined DAR=4.

Despite its low DAR=0.6, HER2-targeting scFv-Fc-Sec/CN29 was at least as potent as the FDA-approved ADC ado-trastuzumab emtansine (DAR of 3.5) when tested at 1 mg/kg in vivo in a CD-1 nude mice xenograft model of breast cancer (Figure 3). Although the two ADCs have different molecular weights (~110 kDa vs. ~150 kDa) and carry different drugs (MMAF derivative vs. maytansine), they use the same antibody component in combination with a non-cleavable linker. However, the linker of ado-trastuzumab emtansine harbors a succinimide thioether (Lewis Phillips et al., 2008) which is potentially less stable than the iodoacetamide-selenol adduct. In fact, at high concentrations in vitro ado-trastuzumab emtansine also killed HER2-negative cells, which we did not observe with either HER2-targeting scFv-Fc-Sec/CN29 or HER2-targeting scFv-Fc(Ser396Sec)/CN29 (Figure 5). This off-target toxicity of ado-trastuzumab emtansine, presumably caused by drug release, has also been observed in other studies (Lewis Phillips et al., 2008). Thus, we attribute the high on-target activity and the low off-target toxicity of our selenomab-drug conjugates to both higher linker stability and lower cell permeability of the released drug.

By increasing DAR=0.6 to DAR=2, the second-generation selenomab-drug conjugate showed significantly improved efficacy both in vitro and in vivo compared to the first-generation selenomab-drug conjugate (Figure 5 and Figure 6A). It also showed excellent stability in vivo judged by the similar serum concentrations of ADC and total antibody in our PK studies, indicating negligible deconjugation of the payload (Figure 6B).

A remaining challenge of selenomab-drug conjugates is the currently inefficient Sec incorporation via UGA stop codon and SECIS element from TXNRD1. Although our yields from transiently transfected HEK293 cells (4 mg/L scFv-Fc-Sec and 2 mg/L scFv-Fc(Ser396Sec) provided sufficient material to conduct the extensive in vitro and in vivo studies described here, it is clear that these yields need to approach the five to ten times higher yields of the corresponding antibodies without Sec residues in order to facilitate clinical translation of selenomab-drug conjugates. This should be possible by optimizing the Sec incorporation machinery. For example, based on the observation that noncanonical SECIS elements (GGGA) of protozoan parasites and its AUGA mutants mediated highly efficient Sec insertion in mammalian cells (Novoselov et al., 2007) we replaced the TXNRD1 SECIS element used in first-generation selenomabs with the AUGA mutant of the GGGA-type Toxoplasma gondii Selenoprotein T 3′UTR for our second-generation selenomabs. Co-transfection with a mammalian expression vector encoding human SECIS binding protein 2 (SECISBP2) resulted in a 50% increase in the yield of scFv-Fc(Ser396Sec), demonstrating that further tuning of the endogenous Sec incorporation machinery in mammalian cells has the potential to overcome current manufacturing challenges.

STAR METHODS

KEY RESOURCES TABLE

The table highlights the genetically modified organisms and strains, cell lines, reagents, software, and source data essential to reproduce results presented in the manuscript. Depending on the nature of the study, this may include standard laboratory materials (i.e., food chow for metabolism studies), but the Table is not meant to be comprehensive list of all materials and resources used (e.g., essential chemicals such as SDS, sucrose, or standard culture media don’t need to be listed in the Table). Items in the Table must also be reported in the Method Details section within the context of their use. The number of primers and RNA sequences that may be listed in the Table is restricted to no more than ten each. If there are more than ten primers or RNA sequences to report, please provide this information as a supplementary document and reference this file (e.g., See Table S1 for XX) in the Key Resources Table.

Please note that ALL references cited in the Key Resources Table must be included in the References list. Please report the information as follows:

• REAGENT or RESOURCE: Provide full descriptive name of the item so that it can be identified and linked with its description in the manuscript (e.g., provide version number for software, host source for antibody, strain name). In the Experimental Models section, please include all models used in the paper and describe each line/strain as: model organism: name used for strain/line in paper: genotype. (i.e., Mouse: OXTR^fl/fl: B6.129(SJL)-Oxtr^tm1.1Wsy/J). In the Biological Samples section, please list all samples obtained from commercial sources or biological repositories. Please note that software mentioned in the Methods Details or Data and Software Availability section needs to be also included in the table. See the sample Table at the end of this document for examples of how to report reagents.

• SOURCE: Report the company, manufacturer, or individual that provided the item or where the item can obtained (e.g., stock center or repository). For materials distributed by Addgene, please cite the article describing the plasmid and include “Addgene” as part of the identifier. If an item is from another lab, please include the name of the principal investigator and a citation if it has been previously published. If the material is being reported for the first time in the current paper, please indicate as “this paper.” For software, please provide the company name if it is commercially available or cite the paper in which it has been initially described.

• IDENTIFIER: Include catalog numbers (entered in the column as “Cat#” followed by the number, e.g., Cat#3879S). Where available, please include unique entities such as RRIDs, Model Organism Database numbers, accession numbers, and PDB or CAS IDs. For antibodies, if applicable and available, please also include the lot number or clone identity. For software or data resources, please include the URL where the resource can be downloaded. Please ensure accuracy of the identifiers, as they are essential for generation of hyperlinks to external sources when available. Please see the Elsevier list of Data Repositories with automated bidirectional linking for details. When listing more than one identifier for the same item, use semicolons to separate them (e.g. Cat#3879S; RRID: AB_2255011). If an identifier is not available, please enter “N/A” in the column.

  • A NOTE ABOUT RRIDs: We highly recommend using RRIDs as the identifier (in particular for antibodies and organisms, but also for software tools and databases). For more details on how to obtain or generate an RRID for existing or newly generated resources, please visit the RII or search for RRIDs.

Please see the sample Table at the end of this document for examples of how reagents should be cited. To see how the typeset table will appear in the PDF and online, please refer to any of the research articles published in Cell in the August 25, 2016 issue and beyond.

Please use the empty table that follows to organize the information in the sections defined by the subheading, skipping sections not relevant to your study. Please do not add subheadings. To add a row, place the cursor at the end of the row above where you would like to add the row, just outside the right border of the table. Then press the ENTER key to add the row. You do not need to delete empty rows. Each entry must be on a separate row; do not list multiple items in a single table cell.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Goat anti-human HER2 polyclonal antibodies	R&D Systems	Cat#AF1129; RRID: AB_354617
Goat IgG	Jackson ImmunoResearch Laboratories	Cat#005-000-003; RRID: AB_2336985
Alexa Fluor 488-conjugated donkey F(ab′)2 anti-goat polyclonal antibodies	Jackson ImmunoResearch Laboratories	Cat#705-546-147; RRID: AB_2340430
Mouse anti-His tag monoclonal antibody	Qiagen	Cat#34660
HRP-conjugated goat anti-mouse polyclonal antibodies	Jackson ImmunoResearch Laboratories	Cat#115-035-146; RRID: AB_2307392
Mouse anti-MMAF monoclonal antibody	Epitope Diagnostics	Cat#MAB30700
Bacterial and Virus Strains
Biological Samples
Chemicals, Peptides, and Recombinant Proteins
5-Iodoacetamido-fluorescein	Marker Gene	Cat#M0638
Fluorescein-5-maleimide	Thermo Fisher Scientific	Cat#F150
CN27	This paper	N/A
CN28	This paper	N/A
CN29	This paper	N/A
Iodoacetamido-caproyl-MMAF	This paper	N/A
See chemistry procedures for synthesis of additional compounds	This paper	N/A
DMSO anhydrous	Thermo Fisher Scientific	Cat#D12345
DTT	Fisher Scientific	Cat#BP172-5
D-luciferin	Caliper Life Sciences	Cat#122796
Bovine serum albumin	Sigma-Aldrich	Cat#A7030-100G
PNGase F	New England Biolabs	Cat#P0705S
Human HER2 extracellular domain	Novus	Cat#NBP1-94702
Matrigel	BD Bioscience	Cat#356237
Anti-HER2 scFv-Fc-Sec	This paper	N/A
Anti-CD138 scFv-Fc-Sec	This paper	N/A
Anti-CD79B scFv-Fc-Sec	This paper	N/A
Anti-HER2 scFv-Fc(Ser396Sec)	This paper	N/A
Anti-CD79B scFv-Fc(Ser396Sec)	This paper	N/A
Anti-HER2 scFv-Fc-Sec/CN27	This paper	N/A
Anti-HER2 scFv-Fc-Sec/CN28	This paper	N/A
Anti-HER2 scFv-Fc-Sec/CN29	This paper	N/A
Anti-CD138 scFv-Fc-Sec/CN27	This paper	N/A
Anti-CD138 scFv-Fc-Sec/CN28	This paper	N/A
Anti-CD138 scFv-Fc-Sec/CN29	This paper	N/A
Anti-CD79B scFv-Fc-Sec/CN29	This paper	N/A
Anti-HER2 scFv-Fc(Ser396Sec)/CN29	This paper	N/A
Anti-HER2 scFv-Fc(Ser396Sec)/MMAF	This paper	N/A
Anti-CD79B scFv-Fc(Ser396Sec)/CN29	This paper	N/A
Ado-trastuzumab emtansine biosimilar	Levena Biopharma	Custom conjugation
Critical Commercial Assays
CellTiter 96 AQueous One Solution Cell Proliferation Assay	Promega	Cat#G3580
Human Lambda ELISA Quantitation Set	Bethyl Laboratories	Cat#E80-116
Deposited Data
Experimental Models: Cell Lines
Human: SK-BR-3	ATCC	Cat#HTB-30
Human: MCF-7	ATCC	Cat#HTB-22
Human: MDA-MB-468	ATCC	Cat#HTB-132
Human: U266	ATCC	Cat#TIB-196
Human: H929	ATCC	Cat#CRL-9068
Human: MDA-MB-361/DYT2	Gregory P. Adams, Fox Chase Cancer Center; referenced in this paper: Yang et al., 1998	N/A
Human: KPL-4	Naoto T. Ueno, University of Texas MD Anderson Cancer Center; referenced in this paper: Kurebayashi et al., 1999	N/A
Experimental Models: Organisms/Strains
Mouse: CD-1	Charles River Laboratories	Strain Code: 022
Mouse: CD-1 nude	Charles River Laboratories	Strain Code: 086
Mouse: NOD/SCID/IL-2Rγnull (NSG)	The Jackson Laboratory	Stock No: 005557
Oligonucleotides
Recombinant DNA
Software and Algorithms
FlowJo	Tree Star	http://docs.flowjo.com/vx/faq/general-faq/tree-star-flowjo/
GraphPad Prism	GraphPad Software Inc	http://www.graphpad.com/scientific-software/prism/
Phoenix WinNonlin	Pharsight	https://www.certara.com/software/pkpd-modeling-and-simulation/phoenix-winnonlin/
Other

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Christoph Rader (crader@scripps.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Cell lines

Human breast cancer cell lines SK-BR-3, MCF-7, and MDA-MB-468, and human multiple myeloma cell lines U266 and H929 were obtained from American Type Culture Collection (ATCC). Human breast cancer cell line MDA-MB-361/DYT2 (Yang et al., 1998) was kindly provided by Dr. Gregory P. Adams (Fox Chase Cancer Center; Philadelphia, PA) based on a Material Transfer Agreement (MTA) with Georgetown University (Washington, DC). Human breast cancer cell line KPL-4 (Kurebayashi et al., 1999) was kindly provided by Dr. Naoto T. Ueno based on an MTA with the University of Texas MD Anderson Cancer Center (Houston, TX) and with permission from Dr. Junichi Kurebayashi (Kawasaki Medical School; Kurashiki, Japan). Breast cancer cells were cultured in DMEM medium and multiple myeloma cells were cultured in RPMI 1640 medium, both supplemented with 10% fetal bovine serum (FBS), 100 μg/mL streptomycin, and 100 U/mL penicillin at 37° C in an atmosphere of 5% CO2 and 100% humidity.

Mice

All procedures were approved by the Institutional Animal Care and Use Committee of The Scripps Research Institute and were performed according to the NIH Guide for the Care and Use of Laboratory Animals. CD-1 and CD-1 nude mice were purchased from Charles River Laboratories. NOD/SCID/IL-2Rγ^null (NSG) mice were purchased from The Jackson Laboratory. All mice used in this study are female and 7 weeks old when the experiments started.

METHOD DETAILS

Antibody cloning, expression, and purification

Anti-HER2 mAb trastuzumab, anti-CD138 mAb clone B-B4 (Kraus et al,, 2015), and anti-CD79B mAb (Chen et al., 2012) in scFv-Fc format containing one C-terminal Sec residue was generated as described previously (Li et al., 2014). The Fc region of this construct contains the hinge-C_H2-C_H3 sequence of human IgG1, followed by a TGA codon, six His codons, a TAA codon, and a SECIS element from the 3′-UTR of the cDNA of human TXNRD1. Based on this construct, the anti-HER2 and anti-CD79B scFv-Fc containing two Sec residues was generated by moving the TGA codon from the C-terminus to position 396 in the C_H3 domain to replace the original serine residue. Antibody expression and purification followed the procedures previously described (Li et al., 2014). Yields for antibody containing one and two Sec residues reached 4 mg/L and 2 mg/L, respectively, compared to 20 mg/L for the corresponding antibody without Sec residues. Selenomabs in scFv-Fc(Ser396Sec) format were also generated by replacing the TXNRD1 SECIS element with the AUGA mutant of the GGGA-type Toxoplasma gondii Selenoprotein T 3′UTR. Co-transfection with SECIS binding protein 2 (SECISBP2) improved the yield to 3 mg/L (Li et al., 2017).

Flow cytometry assay

To test the HER2 density on the five breast cancer cell lines by flow cytometry, cells were harvested using TrypLE (Life Technologies), and 2 × 10⁵ cells were distributed in each well of a V-bottom 96-well plate (Corning). The cells were incubated with 2 μg/mL of polyclonal goat anti-human HER2 IgG (R&D Systems) or control polyclonal goat IgG (Jackson ImmunoResearch Laboratories) for 30 min on ice and washed three times with 200 μL flow cytometry buffer (PBS, 1% (w/v) BSA, 0.01% (w/v) NaN₃, pH 7.4), followed by incubation with AlexFluor® 488-conjugated polyclonal donkey anti-goat IgG (Jackson ImmunoResearch Laboratories) for 20 min on ice. After washing twice with ice-cold flow cytometry buffer, the cells were analyzed using an LSR II Flow Cytometer (Becton-Dickinson). Data were analyzed using FlowJo software (Tree Star).

Antibody conjugation

For selective conjugation at the Sec interface, each antibody was buffer-exchanged to 100 mM sodium acetate (pH 5.2) and concentrated to 4 μM (~0.5 mg/mL) using a 30-kDa cutoff centrifugal filter device. The proteins were reduced by incubation with 0.1 mM DTT for 20 min at room temperature, followed by incubation with a 10-fold molar excess of fluorescein derivatives or drugs in reaction buffer supplemented with 1–2% DMSO for 1 h at room temperature. DTT and unreacted compounds were removed by using a PD-10 desalting column (GE Healthcare). The conjugates in PBS were stored at 4° C for short term use and at −80° C in aliquots for long term use. Antibody concentrations were determined with the Bio-Rad Protein Assay, using a known concentration of an unconjugated antibody as standard.

In vitro cytotoxicity assay

Cells were plated in 96-well plates at 5 ×10³ cells per well for breast cancer cells and 2.5 ×10⁴ cells per well for multiple myeloma cells. Breast cancer cells were allowed to adhere overnight. Serial dilutions of ADCs and free drugs were added to the cells at concentrations ranging from 0 to 1 μM. After incubation for 72 h, the cell viability was measured using the CellTiter 96 AQueous One Solution Cell Proliferation Assay (Promega) following the manufacturer’s instructions. The cell viability was calculated as a percentage of untreated cells (≡ 100%). The IC₅₀ values were determined using logistic non-linear regression analysis with GraphPad Prism software.

Human plasma stability

Anti-HER2 scFv-Fc-Sec/fluorescein and anti-HER2 scFv-Fc(Ser396Sec)/fluorescein conjugates at a concentration of 1 mg/mL were diluted 1:1 (v/v) into human plasma and incubated at 37° C for the duration of the time course study. After 0, 8, 12, 24, 48, 72, 96, 144, and 240 h, 1-μL aliquots were removed and loaded on a NuPAGE Novex Bis-Tris 4–12% gradient gel (Life Technologies) in NuPAGE LDS Sample Buffer (Life Technologies). Following SDS-PAGE and prior to staining with SimplyBlue SafeStain (Life Technologies), a picture of the gel was taken under blue light illumination (Life Technologies) to record the fluorescence. Anti-HER2 scFv-Fc-Sec/CN29 and anti-HER2 scFv-Fc(Ser396Sec)/CN29 conjugates at a concentration of 1 mg/mL were diluted 1:1 (v/v) into human plasma and incubated at 37° C for 72 h before the cytotoxicity assay. Control samples (0 h) were prepared by diluting the ADCs in human plasma immediately before the cytotoxicity assay.

In vivo human breast cancer xenograft model

CD-1 nude mouse model

KPL-4 cells (5×10⁶ per mouse) in a 1:1 mixture of PBS and BD Matrigel (BD Bioscience) were inoculated subcutaneously into the mammary fat pad of 7-weeks old female CD-1 nude mice (Charles River Laboratories). When tumors reached ~100 mm³, the mice were randomly assigned to 5 groups of 5 mice each and treated with anti-HER2 scFv-Fc-Sec/CN29 at 1 mg/kg or 3 mg/kg, or with unconjugated anti-HER2 scFv-Fc-Sec at 3 mg/kg, or with vehicle (PBS) alone, or with ado-trastuzumab emtansine biosimilar (Levena Biopharma) at 1 mg/kg by i.v. (tail vein) injection every 4 days for a total of 4 cycles. The tumor size was monitored every other day using caliper measurement.

NSG mouse model

KPL-4 cells (5×10⁶ per mouse) in a 1:1 mixture of PBS and BD Matrigel (BD Bioscience) were inoculated subcutaneously into the mammary fat pad of 7-weeks old female NOD/SCID/IL-2Rγ^null (NSG) mice. When tumors reached ~150 mm³, the mice were randomly assigned to 5 groups of 6 mice each and treated with anti-HER2 scFv-Fc-Sec/CN29, anti-CD79B scFv-Fc-Sec/CN29, anti-HER2 scFv-Fc(Ser396Sec)/CN29, and anti-CD79B scFv-Fc(Ser396Sec)/CN29 at 5 mg/kg, or with vehicle (PBS) alone by i.v. (tail vein) injection every week for a total of 4 cycles. The tumor size was monitored twice weekly using caliper measurement.

In vivo human multiple myeloma metastasis mouse model

NOD/SCID/IL-2Rγ^null (NSG) mice were inoculated with luciferase expressing U266 cells (5×10⁶ per mouse) via i.v. (tail vein) injection. Four weeks after tumor inoculation, the mice were randomly assigned to 3 groups of 8–9 mice each and treated with anti-CD138 scFv-Fc-Sec/CN29 or unconjugated anti-CD138 scFv-Fc-Sec at 3 mg/kg or with vehicle (PBS) alone by i.v. (tail vain) injection every 4 days for a total of 4 cycles. Tumor burden was monitored (i) by weekly in vivo bioluminescence imaging using a Xenogen IVIS Imaging System (Caliper Life Sciences) following i.p. injection of 100 μL of 25 mg/mL D-luciferin (Caliper Life Sciences) and (ii) by weekly measurement of human lambda light chain concentrations using the Human Lambda ELISA Quantitation Set (Bethyl Laboratories). Mice were euthanatized at the first sign of hind leg paralysis.

Pharmacokinetic study

Female CD-1 mice (~25 g; Charles River Laboratories) were injected i.v. (tail vein) with anti-HER2 scFv-Fc(Ser396Sec)/CN29 at 6 mg/kg. Blood was collected at 5 min and 12, 24, 48, 72, 96, 144, 240, and 336 h post-injection with heparinized capillary tubes. Plasma was obtained by centrifuging the samples at 2,000 g for 2 min in a microcentrifuge. The samples were stored at −80° C until analysis. The concentrations of total antibody and intact ADC in the plasma samples were measured using ELISA. Each well of a 96-well Costar 3690 plate (Corning) was incubated with 100 ng recombinant human HER2 extracellular domain (Novus) in 25 μL carbonate/bicarbonate buffer (pH 9.6) at 37° C for 1 h. After blocking with 150 μL 3% (w/v) BSA/PBS for 1 h at 37° C, samples or standards (anti-HER2 scFv-Fc(Ser396Sec)/CN29) were added. Mouse anti-His tag mAb (Qiagen) and HRP-coupled goat anti-mouse IgG pAbs (Jackson ImmunoResearch Laboratories) were used for the measurement of total anti-HER2 scFv-Fc(Ser396Sec) concentration. Mouse anti-MMAF mAb (Epitope Diagnostics) and HRP-coupled goat anti-mouse IgG pAbs (Jackson ImmunoResearch Laboratories) were used for measurement of intact ADC concentrations. The concentration of antibody in the samples was extrapolated from a four-variable fit of the standard curve. Pharmacokinetic parameters were analyzed using Phoenix WinNonlin PK/PD Modeling and Analysis software (Pharsight).

Hydrophobic interaction chromatography

HIC-HPLC was carried out using a TSKgel Butyl-NPR column (4.6 mm ID × 3.5 cm, 2.5 μm, Tosoh Bioscience). Mobile phase A buffer was an aqueous solution of 1.5 M ammonium sulfate, 25 mM sodium phosphate (pH 7.0); mobile phase B buffer was 75% (v/v) aqueous solution of 25 mM sodium phosphate (pH 7.0) and 25% (v/v) isopropanol. Samples were diluted 1:1 in mobile phase A buffer and loaded on the column at a flow rate of 0.8 mL/min with a gradient of 0% B to 100% B over 12 min for anti-HER2 and anti-CD138 scFv-Fc-Sec/CN29 conjugates and over 30 min for the anti-HER2 scFv-Fc(Ser396Sec)/CN29 conjugate. Conjugation efficacy and DAR were determined by integration of the observed peaks at 280 nm.

Size exclusion chromatography

Analytical SEC measurements were performed on an ÄKTA FPLC system (GE Healthcare) equipped with a Superdex 200 10/300 GL column (GE Healthcare). In each run, 30 μg samples were analyzed on the column with a flow rate of 0.5 mL/min PBS. Aggregate levels were determined by integration of peak areas at 280 nm.

Mass spectrometry

All samples were deglycosylated with PNGase F (New England Biolabs) overnight at 37 °C prior to analysis by direct-infusion MS (DI-MS). The intact mass analysis was performed using a syringe pump system coupled to a Q Exactive Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo Scientific). Sample antibody cations (2 μM in 30% acetonitrile, 0.1% formic acid) were converted to gas-phase ions by electrospray ionization (ESI) at a flow rate of 3 μL/min. HRMS data were acquired using a continuous survey scan (1,600–3,000 Th) at a resolution of 17,500 or 35,000 (at 200 m/z) with a maximum injection time of 50–100 ms and a 1–5 × 10⁵ ion count target. An in-source collision-induced dissociation (CID) of 100.0 eV was applied to increase the desolvation of the proteins and hence the ionization efficiency. Deconvolution was manually performed by averaging the molecular masses obtained for all detectable peaks in the spectra.

Solid-phase syntheses of iodoacetamido-caproyl-MMAF and its derivatives CN27, CN28, and CN29

Chemistry materials

All reagents and resins were purchased from commercial sources unless otherwise referenced. Abbreviations: Dap = Dolaproline; Dil = Dolaisoleucine; TFA = trifluoroacetic acid; TIS = triisopropylsilane; DIEA = diisopropylethylamine; HATU = 1-[bis (dimethylamino) methylene]-1H-1, 2, 3-triazolo [4, 5-b]pyridinium 3-oxid hexafluoro-phosphate; HOBt = hydroxybenzotriazole.

CN27

Pre-swollen Sieber resin (50 μmol) was Fmoc-deprotected with 20% piperidine in DMF (2 mL × 20 min × 2 times) (Scheme S1). The resin was agitated with a freshly prepared solution of Fmoc-Phe-OH (5 equiv), HATU (5 equiv), HOBt (1 equiv) and DIEA (10 equiv) in DMF (1.5 mL) at room temperature (3 h), then drained and washed with DMF. Unfunctionalized resin was capped with a solution of acetic anhydride (10:10:80 v/v Ac₂O: DIEA: DMF) at room temperature (1 h), then washed thoroughly with DMF to yield Resin 1. This resin was Fmoc-deprotected with 20% piperidine in DMF (2 mL × 20 min × 2 times), then agitated with a freshly prepared solution of Fmoc-Dap-OH (Doronina et al., 2009; Nelson and Burke, 2012) (1.5 equiv), HATU (1.5 equiv), HOBt (1 equiv) and DIEA (4 equiv) in DMF (1.5 mL) at room temperature (overnight) to yield Resin 2. Resin 2 was Fmoc-deprotected with 20% piperidine in DMF (2 mL × 20 min × 2 times), then agitated with a freshly prepared solution of Fmoc-Dil-OH (Doronina et al., 2009; Nelson and Burke, 2012) (1.5 equiv), HATU (1.5 equiv), HOBt (1 equiv) and DIEA (4 equiv) in DMF (1.5 mL) at room temperature (overnight) to yield Resin 3. Resin 3 was Fmoc-deprotected with 20% piperidine in DMF (2 mL × 20 min × 2 times), then agitated with a freshly prepared solution of Fmoc-Val-OH (5 equiv), HATU (5 equiv), HOBt (1 equiv) and DIEA (10 equiv) in DMF (1.5 mL) at room temperature (3 h) to yield Resin 4. Resin 4 was Fmoc-deprotected with 20% piperidine in DMF (2 mL × 20 min × 2 times), then agitated with a freshly prepared solution of Fmoc-N-methyl-Val-OH (5 equiv), HATU (5 equiv), HOBt (1 equiv) and DIEA (10 equiv) in DMF (1.5 mL) at room temperature (3 h) to yield Resin 5. Resin 5 was Fmoc-deprotected with 20% piperidine in DMF (2 mL × 20 min × 2 times), then agitated with a freshly prepared solution of Fmoc-caproic acid (5 equiv), HATU (5 equiv), HOBt (1 equiv) and DIEA (10 equiv) in DMF (1.5 mL) at room temperature (3 h) to yield Resin 6. Finally, resin 6 was Fmoc-deprotected with 20% piperidine in DMF (2 mL × 20 min × 2 times), then agitated with a freshly prepared solution of iodoacetic anhydride (Radford et al., 2010) (0.5 M in DMF, 300 μL, 3 equiv) at room temperature (1 h) (Scheme S1). The resin was washed sequentially with DMF, CH₂Cl₂, and Et₂O and then dried in vacuo. The resin was treated with dilute TFA solution (2.5:2.5:95 v/v TFA: triisopropylsilane (TIS): CH₂Cl₂; 2 mL × 5 min × 3 times) to cleave peptide product and the resulting solution was concentrated under a positive stream of argon. The residue was precipitated from cold Et₂O and purified by HPLC to afford product CN27 (3 mg). MS m/z: 1010.5 (M-H), 1012.6 (M+H), 1034.6 (M+Na), 507.1 (MH₂²⁺).

CN28 and CN29

Pre-swollen 2-chlorotrityl chloride resin (0.35 mmol) was agitated with a solution of diamine [1, 4-diaminobenzene for 7a and 2, 2′-(ethane-1, 2-diylbis (oxy)) bis (ethan-1-amine) for 7b] (3.5 mmol, 10 equiv) and DIEA (3.5 mmol, 10 equiv) in DMF (4 mL) at room temperature (1 h), then drained and washed (Scheme S2). The diamine-functionalized resins 7a and 7b were dried in vacuo, stored cold and dispensed as necessary with an assumed loading matching that of the original unfunctionalized resin. A portion of diamine resin 7 (50 μmol) was treated according to the procedure outlined in Scheme S1 to provide the following products: From resin 7a (R = phenyl), CN28 (15.5 mg). MS m/z: 1103.6 (M+H), 1101.5 (M-H), 552.7 (MH₂²⁺). From resin 7b (R = PEG), CN29 (16.5 mg). MS m/z: 1143.7 (M+H), 1141.6 (M-H), 572.7 (MH₂²⁺).

Iodoacetamido-caproyl-MMAF

Pre-swollen 2-chlorotrityl chloride resin (0.035 mmol) was functionalized with Fmoc-Phe-OH (5 equiv) in CH₂Cl₂ with DIEA (10 equiv) overnight. The resin was then washed with MeOH, CH₂Cl₂, and DMF (3 × 5 mL each). The Phe-functionalized chlorotrityl resin (Scheme S3) was then treated according to the procedure outlined in Scheme S1. Following cleavage, the crude product was purified by preparative HPLC to afford iodoacetamido-caproyl-MMAF (16 mg, 45% overall yield). MS m/z: 1011.5 (M-H), 1014.3 (M+H), 507.4 (MH₂²⁺).

QUANTIFICATION AND STATISTICAL ANALYSIS

Cell viability data are presented as mean ± SD of triplicates. Breast cancer xenograft tumor growth curves are plotted as mean ± SD (n=5 or 6). Multiple myeloma xenograft tumor growth data are presented as mean ± SD (n=8 or 9). IC₅₀, survival data, and statistical analysis were performed using GraphPad Prism. Statistical values including the exact n and statistical significance are also reported in the Figure Legends. Statistical significance was defined as p < 0.05 and determined by t-test (two-tailed, unpaired, uneven variance) or log-rank test. ADC and total antibody concentration in mouse serum are plotted as mean ± SD (n=4) using a variable slope (four parameter) non-linear fit.

Supplementary Material

2

Scheme S1. Solid-phase synthesis of CN27.

Scheme S2. Solid-phase synthesis of CN28 and CN29.

Scheme S3. Solid-phase synthesis of iodoacetamido-caproyl-MMAF.

Significance.

Antibody-drug conjugates are an emerging class of highly potent and selective cancer therapeutics. Site-specific conjugation technologies have enabled the generation of homogenous antibody-drug conjugates that are superior to heterogeneous antibody-drug conjugates due to a more consistent manufacturability and a higher therapeutic index permitting higher dosing. Here we show that the high reactivity of the 21st natural amino acid selenocysteine can be harnessed to rapidly and efficiently assemble antibody-drug conjugates that reveal excellent stability, potency, and selectivity in diverse in vitro and in vivo models of human cancers. We anticipate broad therapeutic utility with respect to cancer targets and indications for homogeneous antibody-drug conjugates built on a selenocysteine interface.

Highlights.

• Selenomab-drug conjugates are antibody-drug conjugates based on selenocysteine

• Iodoacetamide-based selenomab-drug conjugates have excellent stability

• Selenomab-drug conjugates are potent and specific cancer therapeutics

• The location of selenocysteine in selenomab-drug conjugates impacts their potency