The concept of antibody-drug conjugates (ADCs) as targeted cancer therapeutics combines the unique targeting properties of a monoclonal antibody to deliver a high potency of a cytotoxin attached via an appropriate linker. As the number of ADCs entering the clinic and achieving regulatory approval continues to grow, there remains a concerted research effort to explore alternative drug-linker designs with the aim of enhancing ADC efficacy and safety. A wide array of structural options for the linker component connecting the antibody to the cytotoxic drug exist. Recent attention has focused on the use of site-specific conjugating linkers to improve ADC homogeneity and a variety of both cleavable and non-cleavable linkers have been employed to maximize toxin delivery to tumor cells (Frigerio and Kyle, 2017; Zhou, 2017; Dal Corso et al., 2019).

Recent research has focused on the inclusion of polymeric portions within ADC linker structure to overcome some of the inherent issues surrounding ADC design such as hydrophobicity, aggregation, instability, insufficient drug-loading and reduced circulatory half-life. Typically, the polymers have been amphiphilic in nature, being water-soluble, synthetic and substantially non-antigenic with particular focus on polyalkylene oxides such as poly(ethylene glycol) (PEG). Since the first PEGylated protein was approved in 1990, PEG has become one of the most widely used polymers in biopharmaceutical applications with numerous products entering clinical trials and receiving marketing approval (Veronese and Mero, 2008; Turecek et al., 2016). PEG features within two of the currently approved ADCs, sacituzumab govitecan (Trodelvy^®) and loncastuximab tesirine (Zynlonta^®), and its utility within ADC reagents has been widely investigated, mainly to improve reagent solubility and to reduce aggregation of the final ADC. The correlation between ADC hydrophobicity and accelerated plasma clearance, particularly for conjugates with high drug-loading, has highlighted the value of incorporating a hydrophilic polymer moiety to mitigate the hydrophobic effects of drug-linkers (Hamblett et al., 2004; Lyon et al., 2015). Recent studies have shown that reagent design, and particularly the positioning of the PEG within the ADC architecture, is an essential consideration when developing the optimal ADC. Some groups describe the use of linear PEG spacers which separates the drug moiety from the antibody (Tiberghien et al., 2016; Bryden et al., 2018; Yang et al., 2019). Others have highlighted an improved reagent format whereby the PEG exists as a pendant molecule which is attached to the reagent via a branching linker (Zhao et al., 2008; Lyon et al., 2015; Burke et al., 2017; Shao et al., 2018; Shao et al., 2020). Our approach has been to use glutamic acid as a branching point to suitably position a PEG chain and a cleavable linker-payload and these reagent structures have been shown to produce efficacious DAR 4, 6 and 8 ADCs (Pabst et al., 2017). ADCs with branched formats have shown improvements in pharmacokinetic (PK) behavior and enhanced in vivo efficacy when compared to ADCs where PEG exists as a spacer segment between the antibody and drug molecules (Lyon et al., 2015; Burke et al., 2017; Pabst et al., 2017). A common feature with these pendant structures is the polymer tethered to the linker has a single point of attachment leaving a free end group in solution. The unattached end of the polymer chain in PEG is typically a methoxy group. It has been suggested that when PEG is arranged in a pendant configuration, it allows for more effective masking of the hydrophobic drug-linker portion of the ADC (Lyon et al., 2015).

Alternative polymers to PEG have also been integrated into ADC reagent design. Submonomer solid phase synthesis has allowed the production of monodisperse polysarcosine (PSAR) oligomers containing 6–24 repeat units (Viricel et al., 2019). Trastuzumab-glucuronide-MMAE based ADCs containing PSAR in branched (n = 6, 12, 18 and 24 units) and spacer (n = 12 units) formats were compared in vivo alongside structurally analogous ADCs with either no polymer or PEG (n = 12 units) in the branched format. ADCs containing the branched reagent structure had improved PK and efficacy compared to the ADCs with no polymer and where the spacer configuration was employed. The branched PSAR(12u) ADC was also found to be more efficacious than a PEG(12u) containing variant. Polyacetal polymers, as part of the Fleximer^® technology platform, have also found utility in the development of ADCs with significantly higher drug-to-antibody ratios (DAR ≥20) compared to more common conjugation approaches, which typically produce DARs between 2 to 4 (Yurkovestskiy et al., 2015). Other polymeric reagent architectures have been reported (Zhang et al., 2017; Marcinkowska et al., 2019; Shi et al., 2020), however PEG, with its long-established history and application in the enhancement of biopharmaceuticals, remains at the forefront in the design of more efficacious ADCs. In contrast to the active research interest in the use of polymers in ADC reagents, there are no reports of ADC reagents incorporating pendant macrocyclic moieties where both ends of the polymer chain are tethered to the reagent linker and the pharmacological impact of introducing such a structural constraint.

Macrocyclic molecules, comprising twelve or more atoms in a ring structure, are long established in medicine as drug molecules and drug delivery vehicles. Cyclodextrins, macrocycles constructed of cyclic arrangements of glucopyranose units, have been extensively investigated in drug delivery applications (Challa et al., 2005; Loftsson et al., 2005; Jansook et al., 2018; Kim et al., 2020). Three naturally occurring cyclodextrins (α-, β- and γ-), containing 6, 7 or 8 glucopyranose repeats connected in cycles via α-1,4-glycosidic linkages respectively, are known. As drug carriers, cyclodextrins have been applied to a wide range of biologics including genes, peptides, proteins and oligonucleotides (Irie and Uekama, 1999; Redenti et al., 2001; Ortiz Mellet et al., 2011; Haley et al., 2020). Small molecules, including cytotoxic agents, however, have been the predominant research focus, with over 35 cyclodextrin containing pharmaceutical products gaining market approval (Loftsson and Brewster, 2010; Kurkov and Loftsson, 2013; Gidwani and Vyas, 2015). Crown ethers, cycles consisting of oxyethylene repeat units, are macrocyclic analogues of PEG which have been shown to have a wide variety of activities, ranging from ion complexation and micelle formation (Gokel et al., 1988) to antimicrobial activity (Febles et al., 2016). Substantial research demonstrating the drug delivery potential of crown ethers has also been reported (Jansen et al., 2002; Muzzalupo et al., 2007; Chehardoli and Bahmani, 2019).

With cyclodextrins and crown ethers established in (bio)pharmaceutical applications, the aim of this study was to determine whether these macrocycles may be used as hydrophilic substitutes for the polymeric element present within ADC reagent structures to generate an efficacious drug molecule. In the present study, we demonstrate the effect that integrating such hydrophilic macrocycles into drug-linker reagent structures has on the in vitro and in vivo properties of a series of homogeneous DAR 4 ADCs. The ADCs investigated incorporate the anti-CD30 antibody brentuximab, with the prodrug Val-Cit-PABC-MMAE and either cyclodextrin or crown ether macrocycles linked via a glutamic acid branching point. Conjugation to the antibody was achieved using an established bis-sulfone based disulfide rebridging linker, previously shown to produce stable and homogeneous DAR 4 ADCs (Badescu et al., 2014; Bryant et al., 2015). Analogous ADCs with linear PEGs were prepared as comparators and the biological performance of the conjugates produced were benchmarked against the clinically approved ADC, brentuximab vedotin (Adcetris^®).

Cluster of differentiation 30 (CD30) is a type I transmembrane protein of the tumor necrosis factor receptor (TNFR) superfamily expressed on the medullary of the thymus gland and on a subset of activated T and B cells (Romagnani et al., 1998; van der Weyden et al., 2017). CD30 expression is not normally detected in healthy tissue outside of the immune system, however high levels of CD30 expression are found in Hodgkin lymphoma (HL) and ALCL, making it an attractive target for antibody-based immunotherapy. Numerous naked and conjugated CD30-targeting mAbs have entered clinical development but despite promising results from animal models, early candidates were unsuccessful (Younes, 2011; Schirrmann et al., 2014). Clinical success was eventually achieved with brentuximab vedotin (SGN-35), an ADC comprising a chimeric anti-CD30 IgG1 antibody conjugated to the potent cytotoxin, MMAE, via an enzymatically cleavable MC-Val-Cit-PABC linker. Conjugation of the linker-payload to the antibody is achieved over two steps: 1) partial reduction of the interchain disulfide bonds of the antibody followed by 2) thiol-maleimide conjugation to the unpaired cysteines liberated during reduction. The nature of the conjugation process results in a heterogeneous mixture of differently loaded species ranging from DAR 0 to DAR 8, with an average drug loading of approximately 4. Phase II clinical trials involving brentuximab vedotin as a single agent in patients with relapsed or refractory HL and ALCL resulted in objective response rates of 75 and 86% respectively (Pro et al., 2012; Younes et al., 2012). Based on these impressive results the FDA granted accelerated approval of brentuximab vedotin in 2011 and the ADC has since been marketed as Adcetris®.

Ongoing efforts to improve ADC efficacy have included focussing upon the design of linker-payload reagent architectures. In our previous work we investigated the inclusion of PEG in ADC reagent design, where structures containing branching architectures were revealed to enhance ADC efficacy over structures with no PEG component and where the PEG is included as a spacer within the ADC backbone (Pabst et al., 2017). Other studies have corroborated these results including research from Lyon and coworkers (Lyon et al., 2015; Burke et al., 2017). In the present work, we designed and synthesized eight new reagents using hydrophilic macrocycles in place of the linear PEG component. Each reagent contained identical bis-sulfone disulfide rebridging conjugating groups, glutamic acid branching points connecting Val-Cit-PABC-MMAE as linker-payload. The reagents designed varied only in the nature of the hydrophilic side-chain moiety appended to the branching point. Six new reagents containing cyclodextrins, two reagents for each of α-, β- and γ-cyclodextrin variants with mono-amino substituents in the 3′ or 6′ positions (Figure 1, reagents 1a-f) were produced. Two further reagents bearing 1-aza-crown ethers were synthesized, the smaller macrocycle incorporated into the reagent template was a 24-crown-8 based ring and the larger macrocycle was a 42-crown-14 derivative (Figure 1, reagents 2a-b). Two reagents with 8- and 24-unit linear PEG chains (Figure 1, reagents 3a-b) were also synthesized so comparator ADCs could be prepared. Each of the ten disulfide rebridging reagents described were conjugated to brentuximab via reaction at pre-reduced interchain disulfide bonds to produce ten DAR 4 ADCs for biological evaluation. ADCs 4a-f were prepared using the cyclodextrin based reagents, ADCs 5a-b were synthesized using the crown ether reagents and the linear PEG reagents were used to prepare ADCs 6a-b.

Table 1; Supplementary Figure S1 shows in vitro cell-viability data for the conjugates produced against Karpas-299 cells. Within each figure both Adcetris^® and conjugate 6b were used as internal controls as comparators for the test ADCs. From Supplementary Figures S1A–S1C, it can be determined that the cell killing potency of each test ADC is higher than the free payload, MMAE, with each conjugate displaying similar IC50 values in the same range as conjugate 6b and Adcetris^®. These data suggest that although the in vitro assay is a reliable test for determining the overall activity of an ADC, as has been demonstrated in previous studies it may not be the best predictor of in vivo efficacy (Hamblett et al., 2004; Pabst et al., 2017).

Figure 2 shows that all of the cyclodextrin-containing ADCs display improved in vivo efficacy over the clinical comparator Adcetris^® within this Karpas-299 xenograft tumor model. These data were in contrast to the in vitro data obtained, which showed that all of the cyclodextrin conjugates were of similar or slightly poorer potency to Adcetris^®. Figure 2 also shows that 3′-amino-α-cyclodextrin containing conjugate 4d in particular had outstanding efficacy in this model, equivalent to that of the PEG containing conjugate 6b. These data are very encouraging and supportive of future development of cyclodextrin-containing ADCs.

From Supplementary Figure S1C, it can be seen from the in vitro data that all of the PEG-containing ADCs; 5a, 5b, 6a and 6b have comparable IC50s to that of Adcetris^®. Once again, this is not reflected within the in vivo efficacy results in Figure 3. It is notable that conjugates 5b and 5a are particularly efficacious in vivo, more so than Adcetris^® and even conjugate 6b, which is again in contrast to the in vitro IC50 data. Conjugate 5b appears to be the best of these ADCs, which as with the cyclodextrin-containing ADCs, is very encouraging for further development of crown-ether containing ADCs in the future.

In conclusion, we tested whether hydrophilic macrocycles (cyclodextrin or crown ethers) may be used as hydrophilic substitutes for the polymeric element present in known ADC structures. Within current ADC structures, it is hypothesized that the linear polymer structure, imparts greater efficacy to ADCs in vivo by forming a ‘shield’ around the hydrophobic linker-payload structure, whereby the hydrophobicity of the conjugate is ‘masked’, resulting in a slower clearance of the ADC from the subject (Lyon et al., 2015). It may be considered therefore that in restricting the size and mobility of the polymeric element of the ADC, one may expect decreased shielding of the linker-payload and therefore poorer retention within the subject, resulting in reduced efficacy. However, we have shown that this is not the case for the cyclodextrin- or crown ether- containing ADCs tested in this work.