Incorporation of Hydrophilic Macrocycles Into Drug-Linker Reagents Produces Antibody-Drug Conjugates With Enhanced in vivo Performance.

Antibody-drug conjugates (ADCs) have begun to fulfil their promise as targeted cancer therapeutics with ten clinical approvals to date. As the field matures, much attention has focused upon the key factors required to produce safe and efficacious ADCs. Recently the role that linker-payload reagent design has on the properties of ADCs has been highlighted as an important consideration for developers. We have investigated the effect of incorporating hydrophilic macrocycles into reagent structures on the in vitro and in vivo behavior of ADCs. Bis-sulfone based disulfide rebridging reagents bearing Val-Cit-PABC-MMAE linker-payloads were synthesized with a panel of cyclodextrins and crown ethers integrated into their structures via a glutamic acid branching point. Brentuximab was selected as a model antibody and ten ADCs with a drug-to-antibody ratio (DAR) of 4 were prepared for biological evaluation. In vitro, the ADCs prepared showed broadly similar potency (range: 16-34 pM) and were comparable to Adcetris® (16 pM). In vivo, the cyclodextrin containing ADCs showed greater efficacy than Adcetris® and the most efficacious variant (incorporating a 3'-amino-α-cyclodextrin component) matched a 24-unit poly(ethylene glycol) (PEG) containing comparator. The ADCs bearing crown ethers also displayed enhanced in vivo efficacy compared to Adcetris®, the most active variant (containing a 1-aza-42-crown-14 macrocycle) was superior to an analogous ADC with a larger 24-unit PEG chain. In summary, we have demonstrated that hydrophilic macrocycles can be effectively incorporated into ADC reagent design and offer the potential for enhanced alternatives to established drug-linker architectures.

Modulation of drug-linker design to enhance in vivo potency of homogeneous antibody-drug conjugates.

Antibody-drug conjugates (ADCs) are a promising class of anticancer agents which have undergone substantial development over the past decade and are now achieving clinical success. The development of novel site-specific conjugation technologies enables the systematic study of architectural features within the antibody conjugated drug linker that may affect overall therapeutic indices. Here we describe the results of a systematic study investigating the impact of drug-linker design on the in vivo properties of a series of homogeneous ADCs with a conserved site of conjugation, a monodisperse drug loading, a lysosomal release functionality and monomethyl auristatin E as a cytotoxic payload. The ADCs, which differed only in the relative position of certain drug-linker elements within the reagent, were first evaluated in vitro using anti-proliferation assays and in vivo using mouse pharmacokinetics (PK). Regardless of the position of a discrete polymer unit, the ADCs showed comparable in vitro potencies, but the in vivo PK properties varied widely. The best performing drug-linker design was further used to prepare ADCs with different drug loadings of 4, 6 and 8 drugs per antibody and compared to Adcetris® in a Karpas-299 mouse xenograft model. The most efficacious ADC showed complete tumor regression and 10/10 tumor free survivors at a single 0.5mg/kg dose. This study revealed drug-linker design as a critical parameter in ADC development, with the potential to enhance ADC in vivo potency for producing more efficacious ADCs.

In Vitro and In Vivo Evaluation of Cysteine Rebridged Trastuzumab-MMAE Antibody Drug Conjugates with Defined Drug-to-Antibody Ratios.

The conjugation of monomethyl auristatin E (MMAE) to trastuzumab using a reduction bis-alkylation approach that is capable of rebridging reduced (native) antibody interchain disulfide bonds has been previously shown to produce a homogeneous and stable conjugate with a drug-to-antibody ratio (DAR) of 4 as the major product. Here, we further investigate the potency of the DAR 4 conjugates prepared by bis-alkylation by comparing to lower drug loaded variants to maleimide linker based conjugates possessing typical mixed DAR profiles. Serum stability, HER2 receptor binding, internalization, in vitro potency, and in vivo efficacy were all evaluated. Greater stability compared with maleimide conjugation was observed with no significant decrease in receptor/FcRn binding. A clear dose-response was obtained based on drug loading (DAR) with the DAR 4 conjugate showing the highest potency in vitro and a much higher efficacy in vivo compared with the lower DAR conjugates. Finally, the DAR 4 conjugate demonstrated superior efficacy compared to trastuzumab-DM1 (T-DM1, Kadcyla), as evaluated in a low HER2 expressing JIMT-1 xenograft model.

Bridging disulfides for stable and defined antibody drug conjugates.

To improve both the homogeneity and the stability of ADCs, we have developed site-specific drug-conjugating reagents that covalently rebridge reduced disulfide bonds. The new reagents comprise a drug, a linker, and a bis-reactive conjugating moiety that is capable of undergoing reaction with both sulfur atoms derived from a reduced disulfide bond in antibodies and antibody fragments. A disulfide rebridging reagent comprising monomethyl auristatin E (MMAE) was prepared and conjugated to trastuzumab (TRA). A 78% conversion of antibody to ADC with a drug to antibody ratio (DAR) of 4 was achieved with no unconjugated antibody remaining. The MMAE rebridging reagent was also conjugated to the interchain disulfide of a Fab derived from proteolytic digestion of TRA, to give a homogeneous single drug conjugated product. The resulting conjugates retained antigen-binding, were stable in serum, and demonstrated potent and antigen-selective cell killing in in vitro and in vivo cancer models. Disulfide rebridging conjugation is a general approach to prepare stable ADCs, which does not require the antibody to be recombinantly re-engineered for site-specific conjugation.

A new reagent for stable thiol-specific conjugation.

Many clinically used protein therapeutics are modified to increase their efficacy. Example modifications include the conjugation of cytotoxic drugs to monoclonal antibodies or poly(ethylene glycol) (PEG) to proteins and peptides. Monothiol-specific conjugation can be efficient and is often accomplished using maleimide-based reagents. However, maleimide derived conjugates are known to be susceptible to exchange reactions with endogenous proteins. To address this limitation in stability, we have developed PEG-mono-sulfone 3, which is a latently reactive, monothiol selective conjugation reagent. Comparative reactions with PEG-maleimide and other common thiol-selective PEGylation reagents including vinyl sulfone, acrylate, and halo-acetamides show that PEG-mono-sulfone 3 undergoes more efficient conjugation under mild reaction conditions. Due to the latent reactivity of PEG-mono-sulfone 3, its reactivity can be tailored and, once conjugated, the electron-withdrawing ketone is easily reduced under mild conditions to prevent undesirable deconjugation and exchange reactions from occurring. We describe a comparative stability study demonstrating a PEG-maleimide conjugate to be more labile to deconjugation than the corresponding conjugate obtained using PEG-mono-sulfone 3.

Fab-PEG-Fab as a potential antibody mimetic.

IgG antibodies have evolved to be flexible so that they can bind to epitopes located over a wide spatial range. The two Fabs in an IgG antibody are linked together as if each Fab is at the end of a linear, flexible molecule. PEG was used as a scaffold molecule to link two Fabs together to give Fab-PEG-Fab molecules, or FpFs. Preparation of FpFs was achieved with reagents that undergo site-specific conjugation at each PEG terminus by bis-alkylation with the two cysteine thiols from a disulfide bond. This allowed each Fab to be conjugated to the PEG scaffold in essentially the same region that each Fab is linked in an IgG. Fabs were sourced directly (e.g., ranibizumab) or monoclonal IgG antibodies were proteolytically digested to obtain the Fabs. This allowed the resulting FpFs to be directly compared to parent IgGs. PEG scaffolds of 6, 10, and 20 kDa were used to make the corresponding FpFs. Dynamic light scatting data suggested the resulting FpFs were similar in size to an IgG antibody and about half the size of a 20 kDa PEGylated-Fab. The solution size of PEG-conjugated proteins is known to be dominated by the extended solution structure of PEG, so it is thought that the smaller size of the FpFs is due to interactions between the two Fabs. Anti-VEGF and anti-Her2 FpFs were prepared and evaluated. The FpFs displayed similar apparent affinities to their parent IgGs. Slower dissociation rates were observed for the anti-VEGF FpFs compared to bevacizumab. The anti-VEGF FpFs also displayed in vitro anti-angiogenic properties comparable to or better than bevacizumab. These first studies indicate that FpFs warrant further examination in a therapeutic indication where the presence of the Fc may not be required.

Comparative binding of disulfide-bridged PEG-Fabs.

Protein PEGylation is the most clinically validated method to improve the efficacy of protein-based medicines. Antibody fragments such as Fabs display rapid clearance from blood circulation and therefore are good candidates for PEGylation. We have developed PEG-bis-sulfone reagents 1 that can selectively alkylate both sulfurs derived from a native disulfide. Using PEG-bis-sulfone reagents 1, conjugation of PEG specifically targets the disulfide distal to the binding region of the Fab (Scheme 2 ). PEG-bis-sulfone reagents 1 (10-40 kDa) were used to generate the corresponding PEG-mono-sulfones 2 that underwent essentially quantitative conjugation to give the PEG-Fab product 4. Four Fabs were PEGylated: Fab(beva), Fab'(beva), Fab(rani), and Fab(trast). Proteolytic digestion of bevacizumab with papain gave Fab(beva), while digestion of bevacizumab with IdeS gave F(ab')(2-beva), which after reaction with DTT and PEG-mono-sulfone 2 gave PEG(2)-Fab'(beva). Ranibizumab, which is a clinically used Fab, was directly PEGylated to give PEG-Fab(rani). Trastuzumab was proteolytically digested with papain, and its corresponding Fab was PEGylated to give PEG-Fab(trast). Purification of the PEGylated Fabs was accomplished by a single ion exchange chromatography step to give pure PEG-Fab products as determined by silver-stained SDS-PAGE. No loss of PEG was detected post conjugation. A comparative binding study by SPR using Biacore with low ligand immobilization density was conducted using (i) VEGF(165) for the bevacizumab and ranibizumab derived products or (ii) HER2 for the trastuzumab derived products. VEGF(165) is a dimeric ligand with two binding sites for bevacizumab. HER2 has one domain for the binding of trastuzumab. Binding studies with PEG-Fab(beva) indicated that the apparent affinity was 2-fold less compared to the unPEGylated Fab(beva). Binding properties of the PEG-Fab(beva) products appeared to be independent of PEG molecular weight. Site-specific conjugation of two PEG molecules gave PEG(2×20)-Fab'(beva), whose apparent binding affinity was similar to that observed for PEG-Fab(beva) derivatives. The k(d) values were similar to those of the unPEGylated Fab(beva); hence, once bound, PEG-Fab(beva) remained bound to the same degree as Fab(beva). Biacore analysis indicated that both Fab(rani) and PEG(20)-Fab(rani) did not dissociate from the immobilized VEGF at 25 °C, but ELISA using immobilized VEGF showed 2-fold less apparent binding affinity for PEG(20)-Fab(rani) compared to the unPEGylated Fab(rani). Additionally, the apparent binding affinities for trastuzumab and Fab(trast) were comparable by both Biacore and ELISA. Biacore results suggested that trastuzumab had a slower association rate compared to Fab(trast); however, both molecules displayed the same apparent binding affinity. This could have been due to enhanced rebinding effects of trastuzumab, as it is a bivalent molecule. Analogous to PEG-Fab(beva) products, PEG(20)-Fab(trast) displayed 2-fold lower binding compared to Fab(trast) when evaluated by ELISA. The variations in the apparent affinity for the PEGylated Fab variants were all related to the differences in the association rates (k(a)) rather than the dissociation rates (k(d)). We have shown that (i) Fabs are well-matched for site-specific PEGylation with our bis-alkylation PEG reagents, (ii) PEGylated Fabs display only a 2-fold reduction in apparent affinity without any change in the dissociation rate, and (iii) the apparent binding rates and affinities remain constant as the PEG molecular weight is varied.

Site-specific PEGylation at histidine tags.

The efficacy of protein-based medicines can be compromised by their rapid clearance from the blood circulatory system. Achieving optimal pharmacokinetics is a key requirement for the successful development of safe protein-based medicines. Protein PEGylation is a clinically proven strategy to increase the circulation half-life of protein-based medicines. One limitation of PEGylation is that there are few strategies that achieve site-specific conjugation of PEG to the protein. Here, we describe the covalent conjugation of PEG site-specifically to a polyhistidine tag (His-tag) on a protein. His-tag site-specific PEGylation was achieved with a domain antibody (dAb) that had a 6-histidine His-tag on the C-terminus (dAb-His(6)) and interferon α-2a (IFN) that had an 8-histidine His-tag on the N-terminus (His(8)-IFN). The site of PEGylation at the His-tag for both dAb-His(6)-PEG and PEG-His(8)-IFN was confirmed by digestion, chromatographic, and mass-spectral studies. A methionine was also inserted directly after the N-terminal His-tag in IFN to give His(8)Met-IFN. Cyanogen bromide digestion studies of PEG-His(8)Met-IFN were also consistent with PEGylation at the His-tag. By using increased stoichiometries of the PEGylation reagent, it was possible to conjugate two separate PEG molecules to the His-tag of both the dAb and IFN proteins. Stability studies followed by in vitro evaluation confirmed that these PEGylated proteins retained their biological activity. In vivo PK studies showed that all of the His-tag PEGylated samples displayed extended circulation half-lives. Together, our results indicate that site-specific, covalent PEG conjugation at a His-tag can be achieved and biological activity maintained with therapeutically relevant proteins.

Poly(2-methacryloyloxyethyl phosphorylcholine) for protein conjugation.

The water-soluble, biocompatible polymer poly(2-methacryloyloxyethyl phosphorylcholine) (PMPC) was evaluated for protein conjugation. PMPC is a zwitterionic polymer that is able to form a more compact conformation in aqueous solution than poly(ethylene glycol) (PEG). While a terminally functionalized N-hydroxysuccinimide derivative of PMPC was not efficient for conjugation to an amine moiety on interferon-alpha2a (IFN), we found that a bis-thiol specific derivative of PMPC could be conjugated after reduction of the disulfide bonds in IFN. Utilizing PMPC that displayed a similar hydrodynamic volume to 20 kDa PEG, we evaluated the in vitro antiviral and antiproliferative activity and pharmacokinetics of a PMPC-IFN conjugate. As a hygroscopic zwitterionic polymer, PMPC is able to form a compact conformation in aqueous solution, which was found to be more compact than PEG. This suggests that PMPC protein conjugates may display different plasma elimination characteristics than PEG protein conjugates. PMPC-IFN displayed marked resistance to antibody binding in Western blot analysis with a polyclonal anti-IFN antibody while displaying comparable in vitro antiviral and antiproliferative activity to PEG-IFN. During an in vivo pharmacokinetic study, the absorption t(1/2) for PMPC-IFN was considerably extended compared to the native IFN and 20 kDa PEG analogue. This is also consistent with the SDS-PAGE result where an apparent reduction in mobility through a hydrated medium was observed. The elimination t(1/2) was also vastly extended over the native IFN and twice the value of 20 kDa PEG-IFN. This suggests that tissue migration of PMPC-IFN occurs more slowly than the 20 kDa PEG-IFN despite their similarity in hydrodynamic volume, leading to an an improved depot effect, which could explain the longer elimination t(1/2). In this study, we demonstrate a potential use of PMPCylation as a novel tool for enhancing the pharmacokinetic profile of therapeutic proteins in ways that complement PEGylation.

Disulfide bridge based PEGylation of proteins.

PEGylation is a clinically proven strategy for increasing the therapeutic efficacy of protein-based medicines. Our approach to site-specific PEGylation exploits the thiol selective chemistry of the two cysteine sulfur atoms from an accessible disulfide. It involves two key steps: (1) disulfide reduction to release the two cystine thiols, and (2) bis-alkylation to give a three-carbon bridge to which PEG is covalently attached. During this process, irreversible denaturation of the protein does not occur. Mechanistically, the conjugation is conducted by a sequential, interactive bis-alkylation using alpha,beta-unsaturated-beta'-mono-sulfone functionalized PEG reagents. The combination of: - (a) maintaining the protein's tertiary structure after reduction of a disulfide, (b) bis-thiol selectivity of the PEG reagent, and (c) PEG associated steric shielding ensure that only one PEG molecule is conjugated at each disulfide. Our studies have shown that peptides, proteins, enzymes and antibody fragments can be site-specifically PEGylated using a native and accessible disulfide without destroying the molecules' tertiary structure or abolishing its biological activity. As the stoichiometric efficiency of our approach also enables recycling of any unreacted protein, it offers the potential to make PEGylated biopharmaceuticals as cost-effective medicines.

Site-specific PEGylation of protein disulfide bonds using a three-carbon bridge.

The covalent conjugation of a functionalized poly(ethylene glycol) (PEG) to multiple nucleophilic amine residues results in a heterogeneous mixture of PEG positional isomers. Their physicochemical, biological, and pharmaceutical properties vary with the site of conjugation of PEG. Yields are low because of inefficient conjugation chemistry and production costs high because of complex purification procedures. Our solution to these fundamental problems in PEGylating proteins has been to exploit the latent conjugation selectivity of the two sulfur atoms that are derived from the ubiquitous disulfide bonds of proteins. This approach to PEGylation involves two steps: (1) disulfide reduction to release the two cysteine thiols and (2) re-forming the disulfide by bis-alkylation via a three-carbon bridge to which PEG was covalently attached. During this process, irreversible denaturation of the protein did not occur. Mechanistically, the conjugation is conducted by a sequential, interactive bis-alkylation using alpha,beta-unsaturated beta'-monosulfone functionalized PEG reagents. The combination of (a) maintaining the protein's tertiary structure after disulfide reduction, (b) the mechanism for bis-thiol selectivity of the PEG reagent, and (c) the steric shielding of PEG ensure that only one PEG molecule is conjugated at each disulfide bond. PEG was site-specifically conjugated via a three-carbon bridge to 2 equiv of the tripeptide glutathione, the cyclic peptide hormone somatostatin, the tetrameric protein l-asparaginase, and to the disulfides in interferon alpha-2b (IFN). SDS-PAGE, mass spectral, and NMR analyses were used to confirm conjugation, thiol selectivity, and connectivity. The biological activity of the l-asparaginase did not change after the attachment of four PEG molecules. In the case of IFN, a small reduction in biological activity was seen with the single-bridged IFN (without PEG attached). A significantly larger reduction in biological activity was seen with the three-carbon disulfide single-bridged PEG-IFNs and with the double-bridged IFN (without PEG attached). The reduction of the PEG-IFN's in vitro biological activity was a consequence of the steric shielding caused by PEG, and it was comparable to that seen with all other forms of PEG-IFNs reported. However, when a three-carbon bridge was used to attach PEG, our PEG-IFN's biological activity was found to be independent of the length of the PEG. This property has not previously been described for PEG-IFNs. Our studies therefore suggest that peptides, proteins, enzymes, and antibody fragments can be site-specifically PEGylated across a native disulfide bond using three-carbon bridges without destroying their tertiary structure or abolishing their biological activity. The stoichiometric efficiency of this approach also enables recycling of any unreacted protein. It therefore offers the potential to make PEGylated biopharmaceuticals as cost-effective medicines for global use.

Site-specific PEGylation of native disulfide bonds in therapeutic proteins.

Native disulfide bonds in therapeutic proteins are crucial for tertiary structure and biological activity and are therefore considered unsuitable for chemical modification. We show that native disulfides in human interferon alpha-2b and in a fragment of an antibody to CD4(+) can be modified by site-specific bisalkylation of the two cysteine sulfur atoms to form a three-carbon PEGylated bridge. The yield of PEGylated protein is high, and tertiary structure and biological activity are retained.

PEGylation of native disulfide bonds in proteins.

PEGylation has turned proteins into important new biopharmaceuticals. The fundamental problems with the existing approaches to PEGylation are inefficient conjugation and the formation of heterogeneous mixtures. This is because poly(ethylene glycol) (PEG) is usually conjugated to nucleophilic amine residues. Our PEGylation protocol solves these problems by exploiting the chemical reactivity of both of the sulfur atoms in the disulfide bond of many biologically relevant proteins. An accessible disulfide bond is mildly reduced to liberate the two cysteine sulfur atoms without disturbing the protein's tertiary structure. Site-specific PEGylation is achieved with a bis-thiol alkylating PEG reagent that sequentially undergoes conjugation to form a three-carbon bridge. The two sulfur atoms are re-linked with PEG selectively conjugated to the bridge. PEGylation of a protein can be completed in 24 h and purification of the PEG-protein conjugate in another 3 h. We have successfully applied this approach to PEGylation of cytokines, enzymes, antibody fragments and peptides, without destroying their tertiary structure or abolishing their biological activity.

Polyvalent dendrimer glucosamine conjugates prevent scar tissue formation.

Dendrimers are hyperbranched macromolecules that can be chemically synthesized to have precise structural characteristics. We used anionic, polyamidoamine, generation 3.5 dendrimers to make novel water-soluble conjugates of D(+)-glucosamine and D(+)-glucosamine 6-sulfate with immuno-modulatory and antiangiogenic properties respectively. Dendrimer glucosamine inhibited Toll-like receptor 4-mediated lipopolysaccharide induced synthesis of pro-inflammatory chemokines (MIP-1 alpha, MIP-1 beta, IL-8) and cytokines (TNF-alpha, IL-1 beta, IL-6) from human dendritic cells and macrophages but allowed upregulation of the costimulatory molecules CD25, CD80, CD83 and CD86. Dendrimer glucosamine 6-sulfate blocked fibroblast growth factor-2 mediated endothelial cell proliferation and neoangiogenesis in human Matrigel and placental angiogenesis assays. When dendrimer glucosamine and dendrimer glucosamine 6-sulfate were used together in a validated and clinically relevant rabbit model of scar tissue formation after glaucoma filtration surgery, they increased the long-term success of the surgery from 30% to 80% (P = 0.029). We conclude that synthetically engineered macromolecules such as the dendrimers described here can be tailored to have defined immuno-modulatory and antiangiogenic properties, and they can be used synergistically to prevent scar tissue formation.

Narrow Molecular Weight Distribution Precursors for Polymer-Drug Conjugates.

Atom transfer polymerization has been used to prepare a narrow molecular weight distribution (MWD), active ester homopolymer that acted as a precursor to prepare families of narrow MWD polymer-drug conjugates during preclinical studies.